New Frontiers in Comparative Sociology (International Studies in Sociology and Social Anthropology)

New Frontiers in Comparative Sociology

International Studies in Sociology and Social Anthropology Series Editor

David Sciulli, Texas A&M University Editorial Board

Vincenzo Cicchelli, Cerlis, Paris Descartes-CNRS Benjamin Gregg, University of Texas at Austin Carsten Q. Schneider, Central European University Budapest Helmut Staubmann, University of Innsbruck

VOLUME 109

New Frontiers in Comparative Sociology Edited by

Masamichi Sasaki

LEIDEN • BOSTON 2009

The contents of this volume has previously been published in Volumes 1–6.3 of Brill’s journal Comparative Sociology. This book is printed on acid-free paper. Library of Congress Cataloging-in-Publication Data New frontiers in comparative sociology / edited by Masamichi Sasaki. p. cm. — (International studies in sociology and social anthropology ; 109) Includes bibliographical references and index. ISBN 978-90-04-17034-6 (hardback : alk. paper) 1. Sociology. I. Sasaki, Masamichi S. II. Title. III. Series. HM585.N459 2008 301—dc22

2008034492

ISSN 0074-8684 ISBN 978 90 04 17034 6 Copyright 2009 by Koninklijke Brill NV, Leiden, The Netherlands. Koninklijke Brill NV incorporates the imprints Brill, Hotei Publishing, IDC Publishers, Martinus Nijhoff Publishers and VSP. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA. Fees are subject to change. printed in the netherlands

CONTENTS Introduction ................................................................................ Masamichi Sasaki

1

PART ONE

METHODS IN COMPARATIVE SOCIOLOGY Strategies in Comparative Sociology ......................................... Mattei Dogan Methods for Assessing and Calibrating Response Scales across Countries and Languages ............................................ Tom W. Smith, Peter Ph. Mohler, Janet Harkness, and Noriko Onodera

13

45

PART TWO

RADICAL SOCIAL CHANGE The Transition to Capitalism in China and Russia .................. Erich Weede Social Structure and Personality during the Process of Radical Social Change: A Study of Ukraine in Transition .... Melvin L. Kohn, Valeriy Khmelko, Vladimir I. Paniotto, and Ho-fung Hung

97

119

PART THREE

VALUES, CULTURE AND DEMOCRACY A Theory of Cultural Value Orientations: Explication and Applications ............................................................................ Shalom H. Schwartz

173

vi

contents

Islamic Culture and Democracy: Testing the ‘Clash of Civilizations’ Thesis ................................................................ Pippa Norris and Ronald Inglehart

221

The Cultural-Economic Syndrome: Impediments to Democracy in the Middle East .............................................. Brigitte Weiffen

251

PART FOUR

INSTITUTIONS IN COMPARATIVE PERSPECTIVE Running Uphill: Political Opportunity in Non-Democracies ... Maryjane Osa and Cristina Corduneanu-Huci Does a Strong Institution of Religion Require a Strong Family Institution? .................................................................. Kristen R. Heimdal and Sharon K. Houseknecht

277

313

PART FIVE

SOCIAL PROCESSES Globalization and Income Inequality in the Developing World ................................................................... Margit Bussmann, Indra de Soysa, and John R. Oneal

353

English as an International Language in Non-Native Settings in an Era of Globalization ....................................... Masamichi Sasaki, Tatsuzo Suzuki and Masato Yoneda

379

A New Test of Convergence Theory ......................................... Robert M. Marsh

405

Notes on Contributors ................................................................ Index ...........................................................................................

449 457

INTRODUCTION Masamichi Sasaki New Frontiers in Comparative Sociology is a collection of notable papers from the journal Comparative Sociology, gathered from its first six volumes spanning the period 2002–2007. Choosing from among all the journal’s outstanding papers published during that period was, of course, not an easy task. Several members of the journal’s editorial board were polled for their suggestions. This work is the result of that sometimes agonizingly difficult selection process. Equally difficult was selecting a title for this work. Among numerous potential titles, the idea of new frontiers stood out, as it suggested leading-edge work in the burgeoning science of comparative sociology. Indeed, the discipline is literally bursting with astute analyses of a globalizing world in transition. Given that not all scholars and interested laypersons are well acquainted with the topic of comparative sociology, this book was viewed as an opportunity to enlighten otherwise unfamiliar readers about the cogency of comparative sociology to the new world order. New Frontiers in Comparative Sociology has been organized into five parts: Methods in Comparative Sociology; Radical Social Change; Values, Culture and Democracy; Institutions in Comparative Perspective; and Social Processes. Taken together, all the articles in this book serve to highlight one or more aspects of comparative sociology—some theoretical, some methodological, and some substantive. Some compare social entities in subjective, case-study fashion, while others report on rigorous social research and analyses. Thus, all contribute in one form or another to describing the many and varied facets of the exciting “new” science of comparative sociology. Methods in Comparative Sociology There were several reasons for choosing Mattei Dogan’s “Strategies in Comparative Sociology” as the lead chapter of this work. Principal among these is Dogan’s broad stroke portrait of comparative sociology. Dogan details 15 strategies (methods, methodologies, techniques)

2

masamichi sasaki

for conducting comparative sociology, but first he gets at the root of the concept of comparison and what it means in various scientific contexts. In the social sciences, he points out, “There is not a single sociological theory that has not been invalidated in some cases: in the social sciences there are very few paradigms.” This, then, begs varied and abundant methodologies, both in terms of actual scientific research and statistical methods, or techniques, and also in terms of general strategies. To put this in context, he describes some excellent analogies from comparative architecture. Dogan then proceeds to delve into brief but concise descriptions of his 15 strategies. They are: (1) “comparing by replication of single case studies”; (2) “comparison by ideal types and by empirical typologies”; (3) “binary comparison”; (4) “comparing similar countries”; (5) “comparing contrasting countries by functional equivalence”; (6) “conceptual homogenization of a heterogeneous domain”; (7) “worldwide statistical comparisons”; (8) “cross-national comparisons of intra-national diversities”; (9) “longitudinal, diachronic and asynchronic comparisons”; (10) “comparison of causal relationships staggered over time”; (11) “comparison by composite indices”; (12) “comparison by scoring and scaling as a substitute for formal statistics”; (13) “comparing ecological environments”; (14) “comparing mini-states and mega-cities”; and (15) “anomaly, deviance, exceptionalism and uniqueness in comparative perspective.” Dogan concludes by pointing out that one must carefully select the appropriate and relevant strategies for a given comparative endeavor. In “Methods for Assessing and Calibrating Response Scales across Countries and Languages,” Tom Smith, Peter Mohler, Janet Harkness and Noriko Onodera address the very real challenges inherent in comparative attitude surveys. How does one design an attitudinal questionnaire that is useful in multiple languages, countries and cultures? How does one achieve the levels of comparability required to justify drawing comparative conclusions across nations, cultures, and peoples? This chapter focuses in particular on the construction and use of response categories for questions. The authors go to great lengths to explore the uses and linguistic nuances of unipolar and bipolar response categories such as “agree/disagree,” “important/not important,” and “in favor of/against” (along with appropriate positive and negative modifiers for increasing or decreasing the intensity of the response).

introduction

3

The authors explore several issues, including “how response categories influence the reported distributions of results,” measuring response category intensities, and using alternative response scales. These inquiries are carried out using American and German pilot studies and a Japanese replication. Extensive results are presented with a view toward seeking optimal techniques for constructing attitudinal survey questionnaires and their respective response categories. Finally, the authors propose numerous areas for further investigation with the aim of “achieving equivalence in cross-cultural, multiple-language surveys.” Radical Social Change from Socialism to Capitalism In “The Transition to Capitalism in China and Russia,” Erich Weede takes us on a comparative socio-economic tour of post-communist Russia and China. Relying heavily on pre-communist and post-communist economic statistics, Weede explores the relative failures of Russia vis-àvis the two countries’ socio-economic well-being. Indeed, in nearly every instance, Russia lags behind China—on per capita GDP, on numerous growth indices, on privatization (of agriculture in particular), on governance, on foreign investment, on capital flight, and so on. Along the way, Weede continually asks why. Inequalities are worse in Russia. Russia is far more ethnically heterogeneous, whereas China is much more homogeneous. The author notes, for instance, that “Only China has overcome collective agriculture, not yet Russia.” China has managed to establish market-preserving federalism, whereas Russia’s federalism is still “market-hampering.” China has proactively promoted foreign trade for many years, whereas Russia rests on the laurels of its depleting natural resource exports. All in all, Weede makes convincing arguments for the superior socio-economic performance of China over Russia in their post-communist years. At the same time, the author gives us a solid example of a two-country comparative case study. In “Social Structure and Personality during the Process of Radical Social Change: A Study of Ukraine in Transition,” Melvin Kohn, Valeriy Khmelko, Vladimir Paniotto, and Ho-Fung Hung report on a subset of a massive cross-national comparative effort. This chapter focuses on analyses of surveys conducted in the Ukraine just shortly after the dissolution of the Soviet Union and then again three to threeand-a-half years later—much of which is then comparatively juxtaposed

4

masamichi sasaki

with analyses of similar investigations in Poland, the United States and, to a lesser extent, Japan. The surveys analyzed were conducted face-to-face among Ukrainian men and women in 1992–1993 and again in 1996. Kohn’s group reports a number of very surprising findings. (It must always be kept in mind that these surveys were conducted during a period of radical social change.) For example, the investigators found that “the overtime correlations—the stabilities—of two underlying dimensions of personality—self-directedness of orientation and a sense of well-being or distress—were startlingly low” (emphasis added). Indeed, these and many other similar findings “flew in the face of ” those of myriad previous studies. For persons not to experience changes in their levels of well-being or distress when their social positions changed strikingly was at first so implausible that the researchers expended a great deal of effort to turn up any methodological flaws, none of any consequence of which were found. Among the many other interesting findings was that “change in none of the component dimensions of social stratification . . . is significantly related to change in either self-directedness of orientation or intellectual flexibility. . . .” Indeed, the list of such “startling” findings could go on and on. The substantive complexity of one’s work emerges as one of a number of interesting variables. Nonetheless, it is difficult to summarize this exceptional comparative study in just a few short paragraphs. The implications for further cross-national case studies and comparisons are enormous and the authors urge such pursuits. Values, Culture, and Democracy In “A Theory of Cultural Value Orientations: Explication and Applications,” Shalom Schwartz presents a theory of “seven cultural value orientations that form three cultural value dimensions.” Much of the work compares the theory to those of Hofstede and Inglehart. Schwartz draws on data from 73 countries to validate his seven orientations. Ultimately Schwartz’s theory “yields three broad dimensions”: autonomy versus embeddedness (where autonomy subdivides into intellectual and affective autonomy), egalitarianism versus hierarchy, and mastery versus harmony. All three of these cultural dimensions “contribute uniquely to the explanation of important social phenomena.” Schwartz describes numerous analyses of the data, which reveal a circular pattern to the

introduction

5

seven orientations, which in turn explicates their common and opposing characteristics (interdependent as opposed to orthogonal) and ultimately boils down to the three dimensions. Throughout the work, Schwartz compares his findings to those of Hofstede and Inglehart. Though simpler, their two approaches mesh and agree with Schwartz’s exceptionally well. For instance, they all yield similar world cultural regions: African, Confucian, East-Central European, English-Speaking, Latin American, South Asian, and West European. Considering the differences in the approaches, the fact that these same regions emerge in nearly all instances is quite remarkable. All orientations were examined closely for their associations with socio-economic development, demographics, and attitudes and behavior (moral, political, etc.). Also notable is confirmation of findings regarding “countries as cultural units,” thus supporting “the idea of national cultures.” The positioning of ethnicity in this complex equation is also described in some detail. Opportunities for future research abound, and Schwartz’s work suggests numerous avenues for such pursuits. Pippa Norris and Ronald Inglehart, in “Islamic Culture and Democracy: Testing the ‘Clash of Civilizations’ Thesis,” explore Huntington’s hypothesis about the clash of civilizations from political and social perspectives. They begin by detailing Huntington’s thesis, which proposes that, while culture does matter, it is attitudes toward and expressions of democracy that lie at the root of events such as 9/11. Norris and Inglehart examine evidence from the World and European Values Surveys for the period 1995 to 2001, focusing on attitudes toward four key political values (democratic performance, democratic ideals, religious leaders, and strong leaders) and four key social values (gender equality, homosexuality, abortion, and divorce). Cultural regions of the world are extensively described so that a comparison can be made between regions characterized by Islamic culture and other regions. The authors’ findings do not support the “core components” of Huntington’s “clash of civilizations” thesis, i.e., “societal values in contemporary societies are rooted in religious cultures; the most important cultural division between the Western and Islamic world relates to differences over democratic values; and, in the post-Cold War era, this ‘culture clash’ is at the source of much international and domestic ethnic conflict.” To the contrary, Norris and Inglehart found that there is hardly any difference at all between the Islamic world and the West in terms of political attitudes toward democracy in practice, attitudes

6

masamichi sasaki

toward democratic ideals, and overall disapproval of strong leaders. Indeed, they found that the so-called democratic “clash” was in fact between the post-Communist states and most of the rest of the world (including “both Western and Islamic nations”). They next point out that while support for religious political leaders was stronger in the Islamic societies, this was also the case in a number of other, non- or minimally Islamic, cultural regions (such as sub-Saharan Africa and Latin America). Finally, in studying the results of the four key social attitudinal indicators, Norris and Inglehart found that “there is a substantial cultural cleavage, although one underestimated by Huntington, in social beliefs about gender equality and sexual liberalization.” Here the West is far more liberal with respect to gender equality and sexual liberalization than all other cultural groups, and especially the Islamic nations. At first glance, Brigitte Weiffen’s “The Cultural-Economic Syndrome: Impediments to Democracy in the Middle East,” would seem to be a study in direct contradiction to the preceding chapter by Norris and Inglehart. This is not the case, however, as Weiffen is studying the democratically political realities of Muslim and non-Muslim states, whereas Norris and Inglehart are addressing the political attitudes and aspirations of survey respondents to the ideas and ideals of democracy. Weiffen proposes that there is a “cultural-economic syndrome” which afflicts Muslim states, this based on a relatively overt “resistance to democratization” in the Middle East. Culturally, this resistance is attributed to the role of Islam, and economically it is attributed to oil wealth. Weiffen sets out to show that these two factors “mutually reinforce each other” to create the syndrome which hinders or retards the emergence of democratic regimes in the Middle East. Weiffen analyzes data from diverse sources based upon a matrix of religious orientation and democratic aspirations. These in turn are analyzed within the context of oil-wealthy states and non-oil-wealthy states. She concludes that “in countries where oil wealth and Islamic cultural tradition are at work, religious doctrine, political authoritarianism and [oil] wealth . . . mutually reinforce each other in blocking the democratic option.” In those Muslim states without oil wealth, democratization is a much more likely phenomenon. Finally, Weiffen spends quite a bit of time speculating upon what will happen to Middle Eastern states as their oil wealth runs dry. In so doing, she makes clear that “Islam does not inherently make democracy impossible. It hinders democracy mainly as long as

introduction

7

Islamic doctrine is interpreted by autocratically-minded leaders or would-be autocrats.” Institutions in Comparative Perspective In “Running Uphill: Political Opportunity in Non-Democracies,” Maryjane Osa and Cristina Corduneanu-Huci describe a fascinating and expansive analysis of factors affecting social mobilization in non-democratic autocracies. They studied 24 cases in 15 “stable non-democratic regimes” “to determine conditions of political opportunity in high-risk authoritarian contexts” using Ragin’s Boolean method of Qualitative Comparative Analysis (QCA). This methodology greatly extends the power of otherwise relatively anecdotal comparative case studies. Osa and Corduneanu-Huci began their analysis by selecting four previously identified factors for mobilization: state repression, elite divisions, influential allies, and media access. They later found that adding social networks increased the model’s robustness considerably. The analysis produced four “prime implicants, or specific combinations of political opportunity variables resulting in mobilization, and two prime implicants associated with its non-occurrence.” Among their specific results, they found that increasing repression contributes to mobilization when it is in conjunction with increased media access and a unified elite but no influential allies. Decreasing repression in conjunction with media access, a divided elite, and social networks also contributes to mobilization. Among the six cases where mobilization did not occur, two prime implicants were identified: (a) decreasing repression, unified elites, lack of media access and social networks, but with an influential ally present; and (b) unified elite, increasing repression, lack of media access, no allies, no social networks. Finally, in a reduction of the conditions to their most minimal level, media access and social networks emerged. What is interesting is that there was no one single necessary condition for political opportunity. The prime implicants identified differing combinations of the five independent variables studied. The authors present a number of specific policy implications as a consequence of their findings, such as supporting “cultural and educational exchanges,” and subsidizing “uncensored broadcasts such as Radio Free Europe.” External allies alone, however, are not sufficient to serve as a condition to create political opportunities for social mobilization.

8

masamichi sasaki

Kristen Heimdal and Sharon Houseknecht, in “Does a Strong Institution of Religion Require a Strong Family Institution?” explore the relationships between the institutions of family and religion, principally by using World Values Survey data from 1990. Their main focus is whether one of these institutions has primacy over the other. They emphasize that this cross-sectional, cross-national (i.e., non-longitudinal) work is exploratory in nature; that is, “the goal here is not to see whether strength in family is a predictor of strength in religion, but rather to see whether strong family seems to be a requisite [i.e., ‘necessary but not sufficient’] for strong religion.” Forty-one countries were included in the analyses. One of the key features of the analyses is that they look at both attitudinal and behavioral measures of family and religion. Among the contextual variables included in the analyses are Catholic predominance, communist/communist transition status, level of democracy, level of development, degree of urbanization, per capita GDP, and educational level. The authors found that their original hypotheses were “compellingly” supported. Family strength, at least at the time of this study (circa 1990), was not seen to have declined as much as some other investigators had suggested, nor was religious decline seen as inevitable. Finally, the authors caution that the available data have certain limitations, which suggest one of many solid opportunities for further research. For example, longitudinal studies, they note, would be of immense value. Social Processes In “Globalization and Income Inequality in the Developing World,” Margit Bussmann, Indra de Soysa and John Oneal look at income equality in 72 countries as juxtaposed with these countries’ levels of foreign direct investment (a proven measure of globalization) over the period 1970 to 1990. Other political, economic, and social factors are brought into the study as well. For many years it was thought that globalization negatively impacted income distributions in developing countries, toward increases in income inequalities. However, some studies have found this not to be the case. Indeed, there is a substantial amount of controversy when studying results related to income inequality and its commonly associated social indicators. Drawing on data from UNCTAD and the World Bank, as well as income inequality data from Deininger and Squire, the authors studied

introduction

9

“the influence of foreign investment to gauge the consequences of globalization.” They looked not only at the Gini index but also at the income of the poorest 20% of each country. They found “no evidence that globalization has adversely affected national income inequality.” With one minor exception, this was an across-the-board conclusion. Contrary to earlier thinkers who claimed globalization was harming the developing world, the authors’ findings reinforce recent studies that suggest globalization “has reduced global income inequality.” In conclusion, though, the authors point out that “our analyses make clear the limits of our understanding of the determinants of income inequality: We know more about what does not affect the distribution of incomes than what does.” In “English as an International Language in Non-Native Settings in an Era of Globalization,” Masamichi Sasaki, Tatsuzo Suzuki, and Masato Yoneda underscore that language is an “unquestionable prerequisite for human communication,” and is thus intrinsic to sociology. The apparent dominance of English is addressed in some detail, and attitudes toward that dominance are examined through data from Tokyo’s National Language Research Institute’s 1996–1998 cross-national surveys in 25 nations where English is not the native language. The surveys asked about one’s preference (and tendency) for one’s mother tongue in talking with foreigners inside one’s own country. The surveys also asked about what language or languages would be essential for international communication in the future, as well as for communication within one’s own country. The surveys also asked about what languages one would like one’s children to learn. Respondents were asked to agree or disagree about English as the world’s dominant or most influential language. Finally, respondents were asked what they thought about English’s dominance. Using cross tabulations and correspondence analyses, the authors identified clusters of nations: (A) speakers tend to use their native language, do not think English dominance is good and advocate greater use of other languages; (B) speakers tend to use their native language, do not think English dominance is good but see no alternative; and (C) use English when talking with foreigners and who think English’s dominance is good. In conclusion, the authors validated that English is or is becoming the dominant international language and that there are no immediate competing languages. This does not mean, however, that non-native

10

masamichi sasaki

speakers are happy about using English. As the data clearly show, there is a great deal of ambivalence and discontent about using English. Ultimately, the results suggest a plethora of opportunities for further study. In “A New Test of Convergence Theory,” Robert Marsh sets out to test Marion Levy’s proposition of convergence theory as related to modernization theory. That is, Marsh takes data from 148 non-modernized societies and 52 modernized societies and sets out to determine degrees of convergence or divergence on a number of relevant parameters including economic development, capitalist market economies, demographics, technology, political democracy, cognitive modernization, health, income equality, gender particularism-universalism, and information and communications. The principal proposal is that already modernized societies will show less variation on the variables used in the analysis than will non-modernized societies; i.e., that modernized societies will show increasing “structural uniformity” as opposed to greater structural variation in non-modernized societies. In the two-part analysis, Marsh first compares already modernized and non-modernized societies on 51 variables, and then he compares, within the modernized societies only, shifts in structural uniformity across time (generally between 10 and 20 years, depending upon data available for the specific variables). In the first part of the analysis, he found that the modernized societies showed more convergence on 49 of 51 variables. In the second part, he found that, on 32 of the 45 variables, modernized societies became more convergent over time. Marsh describes how the results support Levy’s general modernization theory across a broad range of social, economic, demographic and political parameters. In conclusion, Marsh roughly fits his results into four domains and shows their relative congruence with Alex Inkeles’ institution domains. Marsh emphasizes that these findings are important for current theories of globalization. Convergence, he states, “is not some uniform process operating throughout the world. Rather, the fact of great variability among the less developed societies means that while some are converging toward the patterns of the developed world, others are diverging.”

PART ONE

METHODS IN COMPARATIVE SOCIOLOGY

STRATEGIES IN COMPARATIVE SOCIOLOGY Mattei Dogan Introduction There is no such thing as comparative chemistry or contextual physics. In the natural sciences, the chain of causality is everywhere identical. In experimental physics or chemistry discoveries have an universal validity. On the contrary, the social sciences, because of the diversity and idiosyncrasy of human societies, are contextual and relativistic. This is true for all living species, as has been demonstrated by the great comparativists Lamarck and Darwin. “Truth on this side of the Pyrénées, error on the other side”, said Montaigne several centuries ago. The best way to comprehend such a biological and social diversity is the comparative method. But in the social sciences, there are very few theoretical explanations with universal validity or applicability. Indeed, John Stuart Mill had admitted that the methods of concomitant variations could not be applied to social realities. There is not a single sociological theory that has not been invalidated in some cases: in the social sciences there are very few paradigms (Dogan 2002). Such a diversity of causal relationships requires a variety of methods. The word “method” has two meanings. The first refers to the technicalities of analysis, such as sampling in survey research, classification of aggregate data, multiples regressions, ecological inference, and so on. The second designates the general approach, the stratagem of conducting the investigation. Here we are exclusively interested in the general strategies, leaving aside the research techniques. In comparative research, the various strategies are complementary. They can be combined and used successively at various stages of the analysis and synthesis. A concrete example could be useful, chosen deliberately from outside the social sciences. Let’s make an incursion in comparative architecture, focusing on religious monuments. We may start by comparing Gothic cathedrals in Europe. This would be a comparison between similar cases, the common trait being the Gothic style. But we will soon discover that in spite of the same basic structure, no two cathedrals are identical. The basic model is easily recognizable,

14

mattei dogan

but each cathedral has its own form, size, decoration and stained-glass windows. Here, the comparison underscores these detailed differences within the category of gothic cathedrals. Suppose now that we compare the Gothic style and the Roman style. We could this time adopt the strategy of comparison between contrasting cases, since the dome of the Saint Peter’s Basilica in Rome and the towers of Notre Dame in Paris are fundamentally different from each other, aesthetically speaking. We can go further and include in the comparison Protestant temples, patriarchical Orthodox churches, Buddhist temples, mosques, Greek temples, pyramids of Incas and pyramids of pharaohs. What is the common denominator of all these buildings? Clearly enough, the willingness to offer a symbolic “home” to deities and to honour the Gods. We arrive then at the concept of “sanctuary”. Around this concept we homogenize a heterogeneous domain: a common belief despite the architectural diversity. This is what can be called conceptual homogenization of an heterogeneous domain. This rapid incursion into comparative architecture enables us to distinguish three kinds of strategies in comparative research. There are many others. Before inventorying them, it is necessary to stress that an important decision has to be taken from the outset: the need to delineate the field of research. Overall comparisons in the classical tradition are becoming increasingly rare, because the social sciences are now more analytical and functional. Today, almost all comparative studies deal with segments, with parts of a society. Overall analyses in the tradition of Montesquieu, Spencer, or Weber are becoming increasingly rare, because the progress of knowledge leads sociologists increasingly to define and limit their field of investigation. A similar specification of research is to be found in physics. The intellectual profile of Nicolaus Copernicus, Leonardo da Vinci, or Isaac Newton no longer corresponds to that of the leading scientists of today. The discipline matures by dividing the social reality it studies. This is not to say that the holistic perspective has been abandoned. Some great comparativists have helped to keep it alive. But one has only to consult the bibliographies devoted to comparative analysis to note the overwhelming predominance of sectoral comparisons. Very few studies attempt to compare, in their entirety, vast political and social structures. The division of the system into segments is the normal course of the comparative approach. Confronted with the complexity

strategies in comparative sociology

15

of the political system, unless he opts for pure theory, the researcher is led to make a choice, to divide, to select the phenomenon on which to center the comparison. The distinction between segmentation and the global approach is a matter of degree. Between the restrictive sectoral study and the global approach that loses itself in abstract theory, there lies a progression from the particular to the general. In contrasting these two facets, it is the overall method that we wish to emphasize. Comparing always involves extracting a small or large sector from a society or political system. But there is a considerable distance between, for instance, the analysis of the political behaviour of workers in two countries and the study of the aggregative functions of parties in twenty countries. Once the comparativist has delimited the object to compare, he has to make a second important decision: the choice of countries to be included in the comparison. This choice depends on many parameters, that will appear progressively in the fifteen following strategies that we shall now review. 1. Comparing by Replication of Single Case Studies In bibliographies and citation indices concerning comparative research, the majority of studies mentioned deal with only one country (Sigelman and Gadbois 1983). The main reason is that even if they are not directly comparative they help comparative research in the sense that they contain a significant theoretical-conceptual component. Another reason is the confusion accepted by many scholars and institutions between truly comparative studies and foreign area studies. On this point, Giovanni Sartori is intransigent: “I must insist that as a ‘one case’ investigation the case study cannot be subsumed under the comparative method though it may have comparative merit” (Sartori 1994:23). Nonetheless, the case study is advocated by many comparativists. Harry Eckstein long ago defined its “merit”: “case studies are first and foremost, part and parcel of theory-building, not of theory controlling” (Eckstein 1975). In this sense, Sidney Verba is right to say that one can validly explain a particular case only on the basis of general hypotheses (Verba 1967:114). All the rest is less relevant, and so is of no use. Alfred Grosser puts it differently: “In a certain sense, no monograph is scientific. There is science only if the analysis of a specific subject is conceived straightaway as a case study: that is to say if one

16

mattei dogan

asks the subject questions deduced from a comparative, even though brief, view of similar subjects” (Grosser 1972:137). For Henry Teune “even single-country case studies, if theoretically framed, can be used to support generalizations. Such cases also can be important first steps to selecting other relevant cases to elaborate a theoretical problem” (Teune 1990:45). B. Guy Peters concurs: “scholars may utilise the case method as their fundamental basis for methodology but then themselves accumulate a number of cases that create a theoretical whole. These scholars had a common theoretical framework, which they then applied to a series of cases. This purposive selection would not meet the canons of experimental or statistical methodology, but it still permitted these scholars to make reasonable theoretical statements with a strong comparative basis” (Peters 1998:141). Many researchers have tested or developed general models within the framework of a single country. Here we have in mind the works of Rene Dumont on India, David Apter on Ghana, James Coleman on Nigeria, Fred Riggs on Thailand, Michael Hudson on Lebanon, and Lucian Pye on Burma. If we consider this last case, we see how, by pondering the problems of this new Asian country, Pye made certain theoretical observations that have since been widely recognized and discussed in studies on political development and communication. No modern society can take shape unless complex and efficient large organizations develop; but the case of Burma makes it clear that such organizations cannot be established in the absence of informal communication between citizens; that is, in the absence of an adequate social organization. These works exemplify how a case study may bring to light significant factors and variables neglected in more inclusive comparisons. Limiting the analysis to a single country has the advantage of allowing the researcher to study the subject in depth. The case study becomes “heuristic”, as Harry Eckstein says, when it contributes to the refinement of a theory. To study Canada as a consociational democracy means distinguishing between explanatory elements that become integrated into the cumulative knowledge of this type of democracy. To note, for example, how little resistance the consociational model offers to excessive governmental responsibilities, as in Lebanon, leads to a better understanding of the rules of the game in such a system. These examples illustrate how the case study, far from passively depicting social features, contributes actively to their explanation. “When explanations are drawn from such single-country

strategies in comparative sociology

17

studies—explanations that have theoretical or potential applicability to other contexts—such studies clearly contribute to the goals of crossnational analysis” (Mayer 1983:175). “Indirectly, case studies can make an important contribution to the establishment of general propositions and thus to theory-building in political science” admits Arendt Lijphart (1971:691), who distinguishes between six types of case studies: nontheoretical, interpretative, hypothesis-generating, theory confirming, theory infirming, and deviant cases. A study covering a single country could become, retrospectively, truly comparative if it is replicated in one or several other countries, and if the replication is focusing on relationships between variables and is not limited to descriptive facts. Most comparativists make a clear distinction between descriptive facts and the replication of relationships (Nowak 1977:17). What does the word “replication” mean? In the hard sciences it means the exact repetition of another’s research design and experiments to assess if the same conclusions can be reached. In the social sciences, a case study is replicative if it is “consciously patterned after methods, hypotheses or measures that had previously been employed in another study” (Sigelman and Gadbois 1983:279). The nature of the problem is an important factor for deciding whether a case study will be of value for the comparativist. Studies focused on structural or systemic data have rather good prospects because the political system already provides a universal matrix; that is, it exhibits a generally relevant set of issues and allows a replication from one particular experience to another. Such replication may become more difficult when the field considered in the monograph implies the kind of intimate approach that only history can provide. Case studies dealing with segments of the political system, such as parties or parliaments, are generally more relevant to a comparative perspective than, say, analyses devoted to ideologies. 2. Comparison by Ideal Types and by Empirical Typologies There are two types of typologies: deductive and inductive. The deductive approach consists in building abstract types. Max Weber is the classic representative of this method. The art of constructing ideal types implies a profound knowledge of the reality and a great capacity to synthesize. Max Weber’s typology of three kinds of legitimacy—charismatic,

18

mattei dogan

rational and traditional—is one of the most cited typologies in sociology and political science. Today, it is nevertheless obsolete. In the Weberian typology of legitimacy, the concept of charismatic leadership plays a crucial role. But if one tries to apply the concept of charisma to contemporary leaders—without stretching it too far—one finds, through empirical research, only a handful of genuine cases during the last few decades, and even fewer cases of traditional hereditary legitimacy: only three or four, if ceremonial kings deprived of real political power are excluded. Consequently, two of the three “boxes” of the Weberian typology are almost empty for the contemporary world. The third “box”, legal-rational-bureaucratic rulership, is overfull with about 180 contemporary independent countries, and it is also diluted, since it amalgamates a large variety of regimes: Latin American bureaucratic authoritarianism, Scandinavian neocorporatist democracies, African tyrannical regimes. The theoretical discomfort becomes even more acute when the empiricist finds that most regimes included in this third “box” are not even legitimate. Thus the old Weberian typology does not include the majority of contemporary regimes. This classical typology must be updated by adding a fourth type, reserved to semi-legitimate regimes, and a fifth, for totally illegitimate rulerships. On closer inspection, the researcher finds that these two new types of authority are still too heterogeneous; the choice is then between multiplying the number of types and the distinction of several sub-types. Bearing in mind that, according to classical theory, legitimacy is not a paragraph in a constitution but a belief in the minds of the people, the researcher needs empirical data. He then discovers that the legitimacyillegitimacy dichotomy is too rigid and that another concept is needed, one that can be operationalized in empirical research by gradually measuring the notions of confidence and trust. Hence the right questions to ask are: how much confidence, by whom, in which domain? Many examples of deductive typologies could be given, starting with Aristotle. Most of them are still alive in the literature (Tonnies: community and society; Durkheim: organic and mechanical solidarity; Redfield: folk societies and urban societies; Parsons has suggested several seminal but very abstract typologies). But in recent decades the trend has been toward empirically grounded typologies. Classification is an old undertaking in all sciences. A typology is a multidimensional classification. The most simple typology results from

strategies in comparative sociology

19

the crossing of two dimensions. Typologies of social actors are often elaborated in a single context, whereas typologies concerning systems, regimes, and societies are conceived from the outset in an international framework. Someone who observes individuals or groups can compare them without looking beyond national frontiers. He can do a noninternational comparison, build a typology of leaders or voters, taking into consideration a single country, or city. This becomes difficult, if not impossible, when the analysis deals with groups, institutions or structures existing in limited numbers only. International comparisons become more valuable when the objects of analysis are classes or parties than when the study deals with families or individuals; they are more useful for understanding pressure groups and unions than for distinguishing between the leaders of these groups. Typologies require an extension of the field across national boundaries when the number of cases is insufficient: the typology of political systems falls naturally into the hands of the comparativist. The value of typologies of regimes, like typologies of social actors, depends on the amount of debate they generate. It is essential that a certain consensus evolves in order that the typology become a real instrument of comparisons. The way the reflection on authoritarian types of government gradually emerged from discussion and confrontations illustrates the point. The democracy-totalitarianism dichotomy lost most of its analytical interest with the increase in the number of “hybrid” countries. As many countries of the Third World became independent, comparativists studying these new countries rapidly found that the concept of totalitarianism was inadequate, if only because of the absence of a technical infrastructure permitting the control of individuals. There was little real analogy between tyrannical African or Asian countries and Stalinist or Nazi regimes. Leo Strauss has rightly defined totalitarianism by two elements. Contrary to the classical tyranny, he wrote, the totalitarian regime possesses technology and ideology. This means that the will to mobilize the population totally—the ideological factor—is not sufficient to transform the new state into a totalitarian state. For that, the development of the country must be at a level that enables it to penetrate deep into the society. It is necessary that the central government be in possession of the infrastructures, the means, and mechanisms, such as the media or educational system, for effectively controlling employment, incomes, travel, voluntary associations, and military or police forces.

20

mattei dogan

The great diversity of newly independent states engendered many typologies that often overlapped. What is remarkable is the consensus that has finally been reached among the greatest comparativists. The typology elaborated by Edward Shils marked a pioneering stage in this direction. Shils distinguished two intermediary types between the extreme poles of democratic and totalitarian regimes—the tutelary democracies characterized by the hypertrophy of the executive, and the modernizing oligarchies whose dominant trait is domination by military or bureaucratic groups, unconcerned with democratizing the country. To these four types, Shils added a fifth, which is rapidly disappearing: the traditional oligarchy. James S. Coleman distinguished between three types of developing countries: competitive, semi-competitive, and authoritarian; the orientation toward modernization introduced a second axis that permitted the elaboration of five types (Coleman 1960). More complex, more ambitious, and more abstract than typologies of actors, the global typologies have a crucial place in comparative research. From Aristotle to Max Weber, history has been marked by these constructions, the best of which were true tools in the progress of sociological knowledge. It is because the analyst tries to fill the voids left by conceptual framework that he is led to formulate new ones. There is no better generator of concepts than a good typology (Dogan and Pelassy 1990:178–9). 3. Binary Comparison There are two kinds of binary comparisons: explicit and implicit. An explicit binary comparison is a comparison between two countries chosen according to a clear hypothesis, crossing analogies against differences. A binary analysis may be aimed at covering two countries in their entirety, but such an attempt may result in a series of parallel studies that are not directly comparative, or in a series of analyses by sectors. The need to “segment” reality appears clearly in such attempts. A good example is the comparison of modernization in Japan and Turkey (Ward and Rustow 1964). The choice of the pair of countries is of crucial importance. Some pairs are interesting, and others meaningless. The pair Japan and the United States is an instructive example. Here is how Lipset justifies this binary comparison: Japan and the United States are two of the foremost examples of industrial success in the contemporary world, and they took very different

strategies in comparative sociology

21

paths to reach that position. Efforts to account for America’s past success have emphasized that it had fewer encrusted pre-industrial traditions to overcome, in particular, that it had never been a feudal or hierarchically state church dominated society. All of Europe and, of course Japan were once feudal, organized in terms of monarchy, aristocracy and fixed hierarchy, with a value system embedded in religious institutions. (Lipset 1994:153)

Similar justifications can be found for the pair junkers and samurai, chosen by Reinhard Bendix, and for the pair Japan and China by Marion J. Levy. Binary comparison permits a kind of detailed confrontation that is almost impossible when the analysis encompasses too many cases. Binary comparison sometimes seems the best way to undertake a study that leaves out neither the specific nor the general. Comparing two countries naturally enhances one’s interest in each one; in particular it stresses the main characteristics and the originality of each situation. But binary comparison can be used not only for increasing our knowledge of two different systems. It can also contribute to an understanding of more general phenomena. In the latter cases, the two countries considered are thought of as contrasted illustrations of a theoretical reflection. When the sociologist compares the British and French industrial revolutions, he proposes an analysis worthy of consideration not only for those interested in France and Great Britain but also for those who study the dynamic of industrialization. When a comparativist contrasts the political attitudes of the working class in Britain and France, he tries to identify variables that can explain the more or less developed propensity to political radicalism. In France, workers valued the political arena as the unique site where change is to be obtained; whereas, the “pragmatism” of the British working class gave union representatives the decisive role. Why did the samurai in Japan become agents of the central power and modernization, whereas in Germany the Junkers became a conservative force? In attempting to answer this question, Reinhard Bendix was able to bring to light some phenomena of general significance. It was partly because Japan had withdrawn into itself that its aristocracy, unlike that of Germany, did not feel threatened; it was partly because the samurai had been deprived of private lands that they so easily adapted to city and administrative life. Structural factors such as the openness of a country and the connections people have with the land can have important effects on the behaviour of members of a society. In such cases, binary comparison may provide general illustrations of

22

mattei dogan

the way in which development, modernization, or national integration come about. Binary comparison is often used for countries that show contextual similarities, even if the aim of the analysis is to bring out differences in one or more specific fields. To analyze comparatively the recruitment and tenure of Cabinet ministers in France and Britain, considered as opposing systems, might show some analogies between the two countries, for instance in the stability of a “governmental nucleus” (Dogan and Campbell 1957:313–45). Conversely, a study of political cleavages in France and Italy, in contexts considered similar, might demonstrate that various social strata do not distribute themselves similarly between political parties in the two countries. A pairing of France and Italy has a natural appeal because the two countries share many features. The France-Italy pairing has stimulated lots of comparisons between the two communist parties, which were the strongest in the Western world. Britain and the United States share other characteristics, which have encouraged many comparisons on policy-making processes, since Ostrogorski first contrasted them. It is more attractive to use pairs of countries like France and Italy, Morocco and Tunisia, Norway and Sweden, or Uruguay and Costa Rica than to compare Finland and Bolivia or Brazil and Pakistan. Some pairs will produce a great deal of interest, while others will give only meagre results. A comparison of England and Japan, as two insular nations or two maritime powers, might be very meaningful, but an attempt to compare Switzerland and Chad, as two countries having no direct access to the sea, would be of little interest. Of course, the comparativist has the liberty to establish original pairs based on his own conception of relevance. It would be relevant to compare India and China in the framework of a study of the choices available to overpopulated Asian countries as they try to solve problems connected with demography, underemployment, and famine. On the other hand, someone interested in power structures or mobilization of the masses in totalitarian regimes would no doubt find it more meaningful to compare Nazi Germany and the USSR. For those interested in European fascism, Germany could pair with Italy. With England, Germany forms a pair often used by those studying the industrial revolution. Germany and France can be studied together in the framework of an analysis of social stratification. These examples show the range of possibilities open to binary comparisons (Dogan and Pelassy 1990:126–8).

strategies in comparative sociology

23

A binary comparison may be implicit when a foreign country is seen in relation to the observer’s own country. By a kind of dialectical process, the view from afar strengthens the perception of our own society. One knows one’s own country better when one knows other countries too. Some characteristics of the French Society appeared more clearly to Tocqueville when he observed the American Society. Lucian Pye has perceived the “non-Western political process”, comparing it implicitly to his own Western culture. As Charles C. Ragin notes, “many area specialists are thoroughly comparative because they implicitly compare their chosen case to their own country or to an imaginary but theoretically decisive idealtypic case” (Ragin 1987:4). The implicit binary comparison is not always immune from ethnocentrism. 4. Comparing Similar Countries Geographical contiguity is evidently neither the only nor necessarily the best means to define a relatively homogeneous universe. Yet, geographical contiguity often implies certain shared cultural, economic, social, historical similarities. The regional approach presents important advantages. This strategy ensures in the most natural way a control over those variables that the observer would like to keep constant so as to better analyse others variables. The comparison between similar countries in many cases overlaps with the regional approach. Studies on Western Europe, Latin America, the Middle-East and Tropical Africa have eliminated from the analysis the impact of contextual, environmental and geographical factors. However, the relevance of the geographical area approach for comparative politics is not as straightforward as it may appear at first sight. Similarity is not necessarily linked to contiguity. Geographical proximity in itself is not always meaningful for comparing. There are nations or political systems that belong to the same region, are contiguous, and yet are very different—such is the case, for instance, for the South East Asian countries. As John Martz has shown, students of Latin American politics remain enduringly frustrated by the problems of diversity: none of the broad theories that were applied to Latin America as a whole, such as the dependencia (Cardoso and Faletto 1978), bureaucratic-authoritarianism (O’Donnell 1973) or the transition theory (O’Donnell et al., 1986; Linz and Stepan 1978), allowed a general conceptualization covering all

24

mattei dogan

countries of the region and region-wide comparative studies. There are indeed certain risks of confinement that the regional studies conceal when they put the emphasis too heavily on the specificities of the region. A comparison between “relatively similar” countries sets out to neutralize certain differences in order to permit a better analysis of others. This strategy is at the heart of the comparative method. As John Stuart Mill once stated, it is by reducing, insofar as possible, the number of interacting variables, that one has the means to observe the influence of factors one wishes to study. It is easier to test the weight of certain institutional rules on political behaviour by choosing democracies that have common creeds than by incorporating authoritarian regimes or pseudo democracies into the analysis. We know that the comparativist, unlike the chemist, can never eliminate the impact of environment. No two nations in the world would enable the researcher to measure the influence of the Protestant religion or certain rules of ownership “all things being equal in other respects”. What the researcher can do is to increase the pertinence of his conclusions by carefully choosing the political and social entities to be compared. For the researcher who studies political systems, analogies are to be sought either in the sociocultural environment of those systems or in their structures and features of operation. The homogeneity will be more a cultural one if, for example, Anglo-Saxon countries are chosen, and more a structural one if the researcher decides to study single-party regimes. The strategy of comparing similar countries has been criticized by Adam Przeworski: I do not know one single study which has successfully applied Mills’ cannon of only differences [“most similar systems design” in the PrzeworskiTeune 1970, terminology]. I continue to be persuaded, indeed, that the “most similar systems design” is just a bad idea. The assumption is that we can find a pair (or more) of countries which it will differ in all but two characteristics and that we will be able to confirm a hypothesis, that X is a cause of Y under a natural experiment in which a ceteris paribus holds. There are no two countries in the world, however, which differ in only two characteristic and in practice there are always numerous competing hypotheses. (Przeworski 1987)

Such a criticism is not justified because the similar countries are not chosen on simple characteristics, but on the criteria of basic analogies, such as the sociological context or the socio-economic level.

strategies in comparative sociology

25

5. Comparing Contrasting Countries by Functional Equivalence A comparison between two series of contrasting countries implies that the contrasts are of broad significance and delineate areas defined by systemic features. The notion of functional equivalence plays an important role in comparisons between contrasting countries. This approach has generated a great number of new concepts and terms. Here is an extract from Gabriel Almond’s “manifesto”: Thus, instead of the concept of “State”, limited as it is by legal and institutional meanings, we prefer “political system”; instead of “powers”, which again is a legal concept in connotation, we are beginning to prefer “functions”; instead of “offices” (legal again), we prefer “roles”; instead of “institutions”, which again directs us toward formal norms, “structures”; instead of “public opinion” and “citizenship training”, formal and rational in meaning, we prefer “political culture” and “political socialization”. We are not setting aside public law and philosophy as disciplines, but simply telling them to move over to make room for a growth in political theory that has been long overdue. (Almond 1960:4)

In the contrasting comparison, the researcher eliminates from the analysis the secondary differences and the similarities which may persist in spite of the profound contrasts. This strategy may generate new comparisons, also contrasting, but more refined. Comparativists initially contrasted democracy and totalitarianism, and then they focused on the differences between nazism and Stalinism. Just like the mountaineer who reaches the summit, he discovers another summit. The history of comparisons by contrasts looks like a chain of high mountains. 6. Conceptual Homogenization of a Heterogeneous Domain The ingenious comparativist chooses carefully the countries to be compared. The criterion of choice is not immediately obvious. In all cases, however, the choice must be made according to a clear concept. The comparativist can select countries from the four corners of the world, and find common characteristics to countries apparently dissimilar. It is he who creates a concept capable of homogenizing a heterogeneous series of countries. Some examples will illustrate how this conceptual homogenization is achieved. The consociational democracy was not understood or recognized until the concept was forged, describing a society segmented into

26

mattei dogan

religious, ethnic or cultural communities, but where consensus was institutionalized at the summit. In the same way, the concept of neocorporatism has allowed meaningful comparisons of some European countries. Another example is the concept of the “one-dominant partysystem”. In these examples, the explanatory hypothesis pre-existed the selection of countries to be compared. A recent example of this strategy is offered by a new interpretation of the concept of “presidentialism”, by Fred Riggs (1994). Many authors have compared presidential regimes and parliamentary regimes. We have learned a lot from this kind of comparison. But Riggs presents an innovative approach by comparing presidential systems to one another. The integrating concept—presidentialism—gives coherence to a heteroclite universe: Brazil, the United States, South Korea, Chile; in total more than 30 countries. This apparently disparate aggregation is conceptually coherent. Riggs finds a contrast, within this constructed universe, between the United States and all other presidential systems. He is the first scholar to try to explain the success of the American presidential system in the light of the failures of such systems in more than 30 countries in Latin America, Africa and Asia: “We must not reject comparisons between the US and other presidentialist regimes because of the failures of the latter—rather, they provide the information we need in order to explain the relative success of the United States”. He makes a clear distinction between formal constitutional rules and parapolitical constitutional practices. He arrives at a paradoxical conclusion: “The more democratic a presidentialist regime, the more likely it is to be overthrown and replaced by authoritarianism”. So for him the United States is an exceptional case: it is the only successful presidential system, if we exclude the particular case of the French system (which is at times super-presidential without countervailing powers, at times simply parliamentarian, but never truly presidential; it is unofficially called semi-presidential). Riggs uses two strategies of comparative research. First, conceptual homogenization; secondly, the identification of an exceptional case. The conceptual framework, more than anything else, helps the scientific knowledge of social or political phenomena to advance. The historian Paul Veyne defends this strategy, arguing that facts are emphasised by their place in an intellectual construction. “The spatiotemporal continuum is only a didactic framework that perpetuates the lazily narrative tradition. Historical facts are not organized by period or

strategies in comparative sociology

27

people, but by notion; they do not need to be replaced in their time, but grouped under their concepts. History does not study man in time; it studies human materials subsumed under concepts” (Veyne 1976:49). 7. Worldwide Statistical Comparisons “The principal problems facing the comparative method can be succinctly stated as many variables, small number of cases”. “There is, consequently, no clear dividing line between the statistical and comparative methods; the difference depends entirely on the number of cases” (Lijphart in 1971). More than three decades later, with the hindsight of the progress made in comparative politics, such a statement remains convincing only for certain types of comparisons. With 200 independent nations (in 2000), the number of existing cases does not seem to be that small. In the last two decades, many insignificant variables have been abandoned and other indicators, because of their interchangeability, have been combined in indices. Worldwide analysis, called by some scholars “holonational” (adapted from the anthropological term hologeistic) consists of the study of whole societies, counts each country as one case, computes formal mathematical measures of relationships among variables and uses these measures to test general theories (Naroll 1972:212–3). The larger the number of countries included in the comparison, the greater the need for quantitative data. Worldwide correlational analysis has experienced a period of stagnation and is out of breath today, an overused form of research. The main reason for this decline is the discrepancy between the quality of statistical data for the advanced countries and for the developing ones. Scholars became aware that in comparing the two sets of countries they were dealing with material of unequal accuracy. It became clear that the lower the level of development, the lower is also the validity of quantitative data. The difficulties encountered in worldwide correlational analyses mark one of the limits to statistical approaches in comparative politics. The weakness of worldwide statistical comparison can also be explained by the fact that it is based on national averages, neglecting the within-nation diversity.

28

mattei dogan 8. Cross-national Comparison of Intra-national Diversities

With very few exceptions, cross-national comparisons use national averages. But we all know that when on the Gauss curve the distance between average, mean and mode is great, an average is not a significant statistical value. In a distribution the average does not reflect skewness. In other words, the skewness of a distribution differently affects the mean, mode and average. The assumption is that the internal diversity of countries is less significant than the differences between them. But in reality most countries are characterized by an important internal diversity, either regional, or vertical in terms of social strata. Some of the most significant characteristics are distributed unevenly. Internal diversities can be ethnic, linguistic, religious, social, economic. Almost all countries could be ranked according to their degree of homogeneity-heterogeneity. In some matters, such as pluralism, internal diversity is an essential dimension. The internal diversity of countries is not necessarily related to their size. Some small countries are very heterogeneous and some large countries relatively homogeneous. Regional diversities are visible in all European countries except, perhaps, Denmark. There are three Belgiums, four Italys, eight Spains. In France there are old regional contrasts. Yugoslavia has exploded into six pieces. Instead of a single national average for the entire Soviet colossus, there are today 15 independent nation-states and as many national averages. Geographical diversity may be expressed in survey research by the notion of social context. When these contexts are taken into consideration, the risk of the “individualistic fallacy” (Scheuch 1966) is seriously reduced, particularly in ethnically diverse countries. “Cross-national comparison may be more fruitful when based upon within-nation comparison” (Verba 1971:309). For the analysis of intra-national diversities, statisticians and geographers long ago elaborated adequate indices, such as the Gini index of inequality, translated into Lorenz curves and coefficients of dispersion. We have the appropriate tools but the standardized statistical data on internal diversity were, until recently, scarce. An important indicator of internal diversity is the degree of linguistic homogeneity, which has been quantified for a large number of countries. Many political phenomena cannot be explained by national averages. Take, for instance, the level of poverty. People do not revolt against poverty as such, they revolt against injustice; they do not revolt against

strategies in comparative sociology

29

the national average of poverty. In statistical terms social inequalities may be expressed in standard deviations. In some countries, governments have been reluctant to collect and publish data on regional, ethnic or social inequalities. Nevertheless, the World Bank has published data on income inequality for many countries, and so has the OECD for fifteen Western countries. Regional disparities have been studied in many fields, including voting behaviour. Disparities among social strata and their changes have so far received little systematic comparative attention, except for Western Europe. Today we can do better. We have more data on many more countries and we know much more about the diversity within these countries. It is very likely that in the future more attention will be given to intra-national disparities because, for many significant variables, within-nation differences are larger than between-country differences. In this way it will be possible to explain a larger part of the variance. 9. Longitudinal, Diachronic and Asynchronic Comparisons Most international comparisons concern societies which exist at a given time. They are synchronic analyses. But societies can be compared over time, even at a distance of centuries. Distinctions should be made between longitudinal, diachronic and asynchronic comparisons. The longitudinal comparison throws light on continuous, gradual evolution, long-term change, irrespective of accidents of history. It is often based on statistical series. For instance, Peter Flora has made a comparative analysis of the development of mass democracy and the redistribution of national income in some fifteen European countries over a period of more than a century (Flora 1983). Another example is the work of Tatu Vanhanen covering some 150 countries over many decades (Vanhanen 1979). But the use of statistics is not indispensable. The longitudinal comparison is different from the analysis of social change, which emphasizes short-term changes or accelerations of history or historical turning points, and not what persists, or changes only slowly. The longitudinal comparison divides time up into periods, more or less mechanically, whereas the comparison of social change is based on successions of events, often dwelling on the concept of generation. The longitudinal comparison shows what would not appear in a factual narration.

30

mattei dogan

A method that relies on successive synchronic comparisons within a chronological framework is called diachronic (Thrupp 1970). A diachronic analysis compares two or several countries at different times, leaving aside the intermediary periods; for instance a comparison of social inequalities in Europe and the United States in 1900, 1950 and 2000. The asynchronic comparison is a twofold comparison in time and space. It involves a comparison of two or more countries or cultural areas at different moments; for instance a comparison between population growth in Europe in the mid-nineteenth century and the population growth in India and Brazil in the second half of the twentieth century without checking the intermediate trend. Such a comparison would vindicate Malthus by demonstrating that he was not a false prophet. The same kind of comparison can be made of the role played by mandarins in imperial China and contemporary Japan and France. The comparison between junkers and samurais made by Bendix comes under this asynchronic strategy. Another good example is Joshua Forrest’s comparison between the weak state in post-colonial Africa and Europe in the Middle Ages. The objective of the author is to show the similarities between the two worlds. He does so on the basis of an impressive amount of empirical evidence from contemporary Africa and mediaeval Europe, concerning the following features that characterize weak states: inadequate administrative capacity; low level of state penetration because of strong local powers; the dominance of informal politics, involving personal rule, unbounded power struggles, multiplicity of factions, use of force, military involvement and coups d’état, over formal political institutions. Forrest’s asynchronic comparative analysis illuminates certain problems of contemporary political systems in Africa. But the author remains very cautious about the possibility of extending the Africa-mediaeval Europe comparison to the study of future trends in African politics. On the contrary, he maintains that there is no certainty whatsoever that the historical paths of African politics will be similar to those of the post-mediaeval states of Europe. Forrest’s contribution is first of all an example of the asynchronic comparative method. But it also is an illustration of the strategy of conceptual homogenization of a heterogeneous field, since his analysis is built on the concept of “weak state”, bringing together countries from two continents separated by seven to ten centuries (Forrest 1994).

strategies in comparative sociology

31

10. Comparison of Causal Relationships Staggered over Time The time dimension is important for understanding political processes and effects. Rates of change are essential for the analysis of political development. Rapid changes may have different effects from slow ones. Comparisons of rates of change may reveal important differences. Time lags are crucial in understanding causality or probabilistic influence. Everything in politics takes time, and so do all changes in society. No social change is instantaneous. Even if communications take place with electronic speed, the social impact of political decisions takes time. Even revolutions need time to engender social consequences. Nevertheless, most comparative research over the last quarter century has used synchronic data, often because they seemed to be the only ones available. For a long time most survey data were synchronic; only recently have comparative time-series become available. Synchronic political analysis was an important step, but often it could only explain a fraction of the variance. This is one reason why many findings reach only minimal results, and often are not even published. A method for dealing with time is the use of lagged variables. If we assume for theoretical reasons, or from experience, that a change in variable A will have an impact on variable B, we must still ask how much later this impact will take place and have observable results. We must compare variables A and B not at the same time, but variable A at a certain moment and B at some later time. This delay may be quite long. The introduction of compulsory primary education in several Western countries around the 1860s was followed by the rise of the “yellow press” in the 1890s. The historian Daniel Vernet has demonstrated that in France, during the eighteenth century, revolutionary ideas and behaviour spread in the countryside two decades after the rise of radical ideas in the main cities. Other time lags may be short, depending on the scale of the processes involved, but some lag is always to be expected. For instance, the attainment of power by social democratic or similarly welfare-oriented parties—often in the form of coalitions—has been linked by several authors to the enactment of additional social welfare legislation and to an actual rise in welfare benefits. Many of these studies, however, have not given enough weight to time lags, and hence have underestimated the actual impact that occurred. The time lags involved include the time between the formation of the government, the enactment of

32

mattei dogan

specific legislation, its promulgation, its effective implementation at the administrative level, and the time it takes the public to learn to make full use of the opportunities under the new laws. The rise in the number of social security beneficiaries partly illustrates this process. In all Western democracies social expenditures have changed slowly, by an incremental trend. Many comparativists have tried to ascertain the importance of social democratic parties in the growth of government, but because they neglected the time dimension and the delayed, incremental social consequences of the participation of social democrats in power, they have succeeded in explaining only a small part of the variance (Dogan 2000:93–114). The vexed question of economic development and the prerequisites for the establishment of stable democratic regimes also involves considerable time lags too often neglected. Causal relationships in contemporary demographic trends in the Third World would emerge more clearly if urbanization and literacy were considered at a certain moment and birth rates and infant mortality one generation later. Such staggering does not require sophisticated statistical techniques. The neglect of the temporal dimension has long limited the explanation of variance. Its inclusion in research designs could enhance the potential for comparative quantitative analysis. 11. Comparison by Composite Indices Single isolated indicators are often misleading. When a researcher relies on only one or two indicators to measure a complex phenomenon, these are likely to be invalid measures. An example: some still use the number of radios per 1000 population as an indicator of the development of the entire communications network of a nation. While such extrapolation may have been valid several decades ago for many nations, there are today cases where this indicator is invalid. A relatively poor country could rank in radios per 1000 inhabitants as high as a relatively rich country. At the same moment the rich country could rank very high on television sets and daily newspaper circulation per 1000. Except for comparisons between the 50 or 60 poorest countries the indicator “radios per 1000” could today be abandoned. The same problem is evident in many other areas where there are complementary items, as in the transportation network. Cars, trains, buses, boats and aircraft all fulfil similar functions. The relative frequency

strategies in comparative sociology

33

in the use of one or more of these modes of transport is influenced by geography, average distances, cost and cultural preferences. In Europe the rail system is more developed than in the US, there being shorter distances to cover and higher population densities. The train is not seen as a lowly form of transport in Europe, as it is in the US. It would be misleading, then, to use air traffic as an indicator of the development of the transportation system. While many social scientists have assumed that the number of cars per thousand inhabitants is a valid indicator of development, they may not have recognized the importance of the fact that there are alternatives available. Energy consumption per capita is another variable which needs an index to help integrate various energy data. The consumption of energy can reflect many social indicators: industrialization, mechanization and even mass communication. Forms of energy include oil, electricity, coal, gasoline and nuclear energy. For purposes of international standardization, the index of energy expresses data in coal equivalents to oil, natural gas and electrical energy. Another aspect of the relevance of indicators is whether certain variables can meaningfully be quantified. It is not enough to assign numbers to events. The second edition of the World Handbook (Taylor and Hudson 1972) contains quantified data on indicators of political protest. Aside from problems of accuracy, these data are of questionable validity: do they really measure unrest in a society? Even if we grant that demonstrations, riots, armed attacks, deaths from domestic violence and governmental sanctions can be quantified accurately, it is still questionable whether we can assume that these categories represent the true level of unrest in a society. Discontent may not appear without a spark to bring it into the open. Even more fundamentally, the indicators of unrest fail to acknowledge the role of suppression in affecting the statistics. Dictatorial governments around the world suppress the expression of unrest. The existence of this underlying level of unrest was demonstrated by the crises in East Germany, Czechoslovakia, Poland and Hungary in 1989–90. By compounding various indicators in an index, the sociological significance of statistical data could be enhanced. Too often, isolated indicators are still treated by complex methods, even when a simple statistical treatment of indices would be sufficient. But in some cases the components of a composite indicator may obscure more than they illuminate. We now possess quantified indicators difficult or impossible to obtain in the 1970s for a large number of countries; for instance, for life

34

mattei dogan

expectancy, access to safe water, number of people per hospital bed, and school enrolment at age 10–12. By combining isolated indicators into indices, quantitative comparative analysis would be facilitated, because the number of variables would be reduced and their explanatory power enhanced. Certain indicators do not need to be combined into indices, because their explanatory power is sufficient, as attested by numerous empirical analyses. Among these privileged indicators is infant mortality. One does not need sophisticated factor analysis to understand why, sociologically, infant mortality is one of the best indicators in comparative research (Vanhanen 1989). 12. Comparison by Scoring and Scaling as a Substitute for Formal Statistics Many of the most significant aspects of political life cannot be treated in statistical terms. The alternative is scaling by experts. The recourse to judgemental rankings and to scoring finds a justification in a statement by the mathematician Tukey: “Far better an approximate answer to a right question, than an exact answer to the wrong question, which can always be made precise” (cited by Banks and Textor 1963:7). The translation of qualitative aspects into measurable variables requires scaling by judges. The involvement of judges raises the question of coder reliability: how likely are two or several judges to rate the same situation in the same manner? If an expert says that country A is more democratic than country B, and this last more than country C, he must admit also that A is more democratic than C. The reliability of an expert can be tested by the consistency of his rankings. To show the potential of scoring and judgemental rankings, I have selected four examples from the literature. The first one is from Phillips Cutright’s “National political development: its measurement and social correlates”. This article is one of the most cited in the literature on comparative politics and one of the few still relevant today of those published three decades ago. With the help of experts, Cutright constructed an index of political development. He allocated for each country two points for each year in which a parliament existed and where the minority party had at least 30 percent of the seats. He allocated only one point when the minority party was weaker, and no points for each year when no parliament existed. He did the same scoring for the executive branch. Over a period of 22

strategies in comparative sociology

35

years a country could accumulate 66 points. Cutright used a simple but pertinent index. The validity of his scoring can be tested retrospectively. He found for 1963 an imbalance for Chile, the Philippines, Indonesia, Nicaragua and Guatemala: political development was higher than socio-economic development. In the following years the regimes in these countries collapsed. The opposite was “predicted” for Spain, Portugal, Czechoslovakia and Poland. These countries were supposedly ripe for democracy. Cutright’s analysis based on scores and a simple statistical model should be compared with many other articles published at roughly the same time which disappeared from the literature despite the mountains of statistics on which they were built. Cutright’s method of scores could be applied retrospectively to Eastern Europe: the implosion in 1989–90 can be explained by the gap between the relatively high socio-economic level (education, health, urbanization, industrialization) and the low level of political development, before the implosion of the Soviet Union. A second example of scoring as a substitute for formal statistics is the voluminous book by Banks and Textor, A Cross-polity Survey (1963). They proposed a series of 57 dichotomized variables, most of which were directly political: interest articulation and aggregation, leadership charisma, freedom of group opposition, freedom of the press, role of the police, character of the bureaucracy, personalismo, westernization and others. The authors preferred significant aspects of political life to quantified but unimportant variables, even if their dichotomization was uncertain. They gave approximate answers to good questions. Another codification of variables which are not directly quantifiable was adopted by Irma Adelman and Cynthia Taft Morris in their Society, Politics and Economic Development: a quantitative approach (1967). This book has been severely criticized by some scholars (Kingsley Davis, among others) and appreciated by others. These contrasting evaluations can be explained by the fact that it consists of two parts. The first (pp. 1–129) contains an interesting discussion of 41 variables, most of which were and remain not directly quantifiable. The second part consists of a confusing factor analysis. I mention this book for its first part. I use the second part to try to vaccinate comparativists against the temptation to engage in factor analysis. Because of frequent malpractice in the use of this statistical tool, mass immunization is needed. In a series of volumes, Freedom in the World, Raymond D. Gastil (1979–90) has ranked countries with the help of experts according to two basic dimensions: political rights and civil liberties. The rating is

36

mattei dogan

on a seven-point scale by univocal ranking. Published annually since 1979, this series has become a rich source of documentation for comparative politics. After decades of progress in comparative politics we still face this dilemma: whether to have recourse to judgemental variables or to neglect some of the most important aspects of political life. 13. Comparing Ecological Environments Because of exaggerations of some older sociological schools of thought, particularly that of Ellsworth Huntington, who overstressed geographical determinism, as a reaction (see Sorokin), geographical conditions were neglected by sociologists for more than a generation; but most recently the evolution of the ecological sciences has greatly increased the possibilities for analysing the environment. Some economists, such as Andrew M. Kamarck of the World Bank, have spoken of tropical societies as distinct from those in the temperate zones. Three-quarters of Africa is in the tropics. The well-known fact that the vast majority of the world’s poor people live in the tropical or semitropical zones is highlighted by the “North-South” categorization. Human behaviour depends not only on temperature and humidity, but also on the rarity or prevalence of morbidity and debilitation. The frequency of infection by parasites and chronic malnutrition not only reinforce each other, but they also interact in feedback cycles with economic productivity and growth, speed or slowness of behaviour, human energy and capacity to work, and the gap between thoughts and feelings on one side and effective action on the other. Large cohorts of tropical populations are not sick enough to die but sick enough to remain poor. Epidemics or widespread tropical diseases may not destroy governments but the population may lack the energy to wipe them out. Malaria and sleeping sickness have been driven back but not eliminated. Hookworm, bilharzia and trachoma still blight the lives of hundreds of millions. Chronic malnutrition has been estimated to account for about two-thirds of morbidity and child mortality in Africa, South Asia and tropical Latin America. Trypanosomiasis kills horses and cattle and makes it difficult to get to the interior from the coast using animal transport. “The transport obstacle alone was quite sufficient to postpone for centuries any appreciable economic development in tropical Africa” (Kamarck 1976:19).

strategies in comparative sociology

37

Comparativists have not yet asked this difficult question: How far is the low level of development in most of Africa and in some Asian countries to be explained by their tropical environments? Such an interrogation is completely absent even in the recent books on Africa. In the southern regions of the United States and in Northern Australia similar conditions of heat and humidity prevail but the economic handicaps have been overcome. Before the coming of modern hygiene in the eighteenth and nineteenth centuries, major ecological handicaps also prevailed among the poorer strata in the temperate zones. Rats and lice spread plague; epidemics of cholera, typhus and tuberculosis were frequent and often endemic, and so were rickets and other diseases due to deficient nutrition. The mass availability of industrially produced soap and cotton underwear increased as early as the eighteenth century. Clean drinking water, free of epidemic germs, is not available everywhere. The spread of tea in parts of Asia, but not in Africa, meant that drinking water was boiled. Disinfectants came later, used first in hospitals and later in homes. Malaria was wiped out in the south of Italy only about 50 years ago. The experience of Western countries, temperate or hot, suggests that social and economic conditions can contribute substantially to reducing morbidity and mortality. It is the same for the highest strata in tropical countries: the Latin American upper classes have for centuries been healthier than the poor. A set of additional quantified new indicators, highlighting these conditions, could lead to significant revisions of many received theories of economic and political underdevelopment. They might even lead to a revision of the received and often ethnocentric notions about easy self-help for tropical nations. In the advanced countries the ecological problem is reappearing at a higher level in concerns for the “quality of life”. Access to green spaces, to woods and meadows, is becoming rarer and more difficult. Water and air are polluted, less often by germs and parasites and more often by industrial effluents. Smog burdens eyes and lungs. Along with such conditions new political movements and parties have arisen in European countries. Indicators for these ecological problems are available at least for urban areas in many countries, but they are difficult to integrate as international statistical series. This is why comparative researchers have been slow to use them and to relate them systematically to social and political issues. But these problems will not go away; they will grow. And social scientists will have to catch up with them. The ecological dimension may require nominal indicators as well as quantified ones.

38

mattei dogan 14. Comparing Mini-states and Mega-cities

In 1992, there were 214 countries and territories, 187 of them independent. Of these 214, only 132 have a population of over one million and only 122 over two million. One-quarter of the countries represented at the United Nations have together a population equivalent to that of Colombia, which ranks thirtieth among nations in demographic terms. At the same time, half the world’s population lives in four countries: China, India, the USA and Russia. For many comparative purposes, such disparity creates no difficulties. In a typology of political systems the size of the country does not matter. One can compare social mobility in a series of countries without taking their size into consideration. The political systems of Denmark and Costa Rica can be compared with those of India and Nigeria. It is appropriate to compare the presidential system of France and of Sri Lanka even if one country is six times larger than the other. But under some circumstances, size may have an impact on the functioning of a democratic regime. Size always has an impact in international relations: we cannot evaluate the role of Ghana and of Brazil in the international arena if we ignore their sizes. When the analytical approach is basically statistical, the number of cases and the diversity of their size can be an essential dimension. When it is remembered that Norway is not much more populous than Connecticut, there is a feeling of unease about a comparison of electoral behaviour in the United States and in Norway. A sample of 2000 individuals in each country might be statistically sufficient, but one cannot avoid certain doubts about the choice of this pair of countries. One remedy would be to weight the countries according to their demographic size. This is already done in some comparisons of European countries and in studies which consider the continent as a whole. In such analyses, France counts sixteen times more than Finland. The problem of the size of nations is aggravated when we contrast small states with giant cities. Considering only independent countries (leaving aside the territories), one of every two had in 2001 less than four million inhabitants. At the same moment there were eighty megacities of over four million people, many in middle-sized countries. During the 1950s, urbanists defined cities as agglomerations of 5,000 and, for some world regions, of 10,000 inhabitants. Later they adopted a criterion of 20,000 people, and still later of 100,000 people

strategies in comparative sociology

39

per agglomeration. Comparative political researchers followed these definitions, since they depended on the data made available to them. These changes in definition seemed to reflect reality, since urban centres evidently grew. But in part the new definitions were adopted for reasons of convenience rather than of insight. Giant cities of more than one million are a new category. They are of crucial importance in the politics of the countries in which they exist. In 1950, there were about 50 such cities; in 1982 there were 278; in 1992, about 330; in 2001 about 400. Statistical data on giant cities are not easily rendered comparable. Some include only the population in the city itself, administratively defined; others include the suburbs or the entire urbanized area gravitating to the central city. The United Nations has made a serious effort to standardize these criteria, but in many cases it is still necessary to evaluate rather than to count. These giant cities require separate treatment. Their number in a country makes a difference to its political system. If there is only one, it is apt to dominate the country and make it “monocephalic” (singleheaded), usually with a star-shaped system of internal communication, as in France, Britain, Austria, Peru, or the Republic of Korea. Some 30 or more countries, from Hungary to Mexico, Argentina and Thailand, are in this condition. Other countries are “polycephalic”: they have several giant cities, none dominating the others, with a grid-shaped system of transport and communications. Here we find some of the largest countries in the world—China, India, Russia and the United States—but also middle-sized countries, such as Germany, Italy, Canada, Spain, Australia, Poland, Morocco, and also a number of small countries such as Switzerland, The Netherlands and Belgium. Some countries are bicephalic or double-headed, such as Turkey, Syria and Vietnam. For certain comparisons of European countries it is useful to take into consideration the system of cities. All else being equal, the fact that France, Austria, Denmark, Ireland and Finland are “macrocephalic” countries, and, on the contrary, that Germany, Italy and The Netherlands are “polycephalic”, makes an important difference in many political domains. If Yugoslavia had a single powerful mega-city instead of six important regional cities, the dismemberment of the federation would probably have taken a different course. The link between a network of old major cities, born of history, and federalism is obvious.

40

mattei dogan

Comparing large American and European cities, one should take into consideration the public transportation system, particularly the underground infrastructure. The metros in Paris, London, Moscow or Tokyo represent an investment that Los Angeles would need more than 20 or 30 years to build. The cost of the Parisian metro is perhaps equivalent to the cost of the entire production of automobiles in the United States over two or three years. Such a comparison cannot remain at the statistical level. The metropolitan areas of Mexico City, Buenos Aires, São Paolo, Cairo, Bombay, Calcutta, Seoul and other mega-cities had in 1990 a larger population than each of the 120 smaller independent nations. In some countries, the primary city accommodates a significant part of the population (Athens and Santiago about 40 percent, Montevideo almost half, Beirut about three-quarters), and includes the lion’s share of economic, financial, cultural, educational, scientific, artistic and political activities of the country. A World of Giant Cities (Dogan and Kasarda 1988) is replacing progressively the world of territorial nations. In his Political Order in Changing Societies, Samuel Huntington asked in 1968, “What groups are most likely to be revolutionary in the city?”. In the last two decades it has been necessary to explain why the urban lumpenproletariat did not revolt despite the continuing growth of shantytowns, favelas (Brazil), poblaciones (Chile), barriadas (Lima), ciudades esperidida (Mexico), Kutcha (Calcutta) and other slums and bidonvilles at the peripheries of mega-cities from Casablanca to Bogota and from Bombay to Lagos. It may be that tomorrow many comparativists will have to give priority to political unrest in the giant cities of the Third World; they will then need new indicators to replace older ones. Lerner’s model (urbanization ∅ literacy ∅ communication ∅ participation) had a nice run in comparative studies but, for the study of primary cities in Asia, Latin America and Africa and their lumpenproletariat and troglodytes, it appears obsolete today. Some 40 quantified indicators are available for a large number of cities, not all standardized. There is an important monographic literature on mega-cities, but very few comparative studies. As the number of giant metropolises will inevitably continue to grow, there is a need for systematic comparison not only of metropolises, but also between small countries and giant cities; for instance, the budget of the municipality of New York or of the Metropolitan District of Mexico is higher than the national budget of dozens of small countries.

strategies in comparative sociology

41

15. Anomaly, Deviance, Exceptionalism and Uniqueness in Comparative Perspective Even in the most imaginative comparative research there remain, in the final analysis, certain irreducible phenomena which refl ect the originality of each country. History is the greatest generator of national configurations. The older a country, the more it has been shaped by its history. Two or several countries may have many features in common, but they are never identical, because the attributes are combined differently for each country. We are always facing unique realities that we call China, Switzerland, Egypt, Russia, India or Spain. In international comparisons we may distinguish anomalies, deviant cases and exceptional cases. It is a matter of degree. An anomaly is an unexpected position in a ranking, curve or diagram. Statistical eccentricities can be discovered by crossing variables in a scattergram. For instance, life expectancy in Bulgaria is higher than we would expect considering the other correlates of this country. The number of students per thousand inhabitants in India appears “abnormal” for a country with such a low standard of living. An anomaly can be revealed only in a comparative light. In comparative sociology it may play the same role as the clinical case in biology or medicine. The deviant case is less frequent than the anomaly, but it is more significant because it is an entire sector of the society or the political system which appears unusual, abnormal. Deviance can be defined in relation to a set of expectations drawn from a series of countries similar in many ways. The exceptional case is a multiple deviant case, an extreme case from many points of view. India, according to certain theories, should not be democratic; it is an enormous exception. The difference between the deviant case and the exceptional case is a difference of degree. Exceptionalism refers to an accumulation of several deviances in systemic or contextual characteristics, forming a configuration, a Gestalt. It is in this sense that S.M. Lipset considers Japan and the United States as exceptional cases. When we decompose the configuration into variables, the distinctiveness tends to be obscured, because we are extracting the variables from their contextual significance. Japan and the United States are exceptional as wholes, but when we “segment” these configurations, when we isolate variables and indicators, the differences between the two countries become differences of degree. Among the three dozen partial comparisons and sectoral analyses, Lipset always finds differences

42

mattei dogan

of degree; it is never zero percent in one case and 100 percent in the other. The exceptionalism of each of these two countries resides in their national configuration. The search for deviant and exceptional cases is an alternative strategy to the statistical approach. Instead of high correlations, we are looking for a meaningful clinical test. Exceptional cases do not limit the potential of international comparisons. On the contrary, the search for exceptions can be a sui generis strategy of comparative research, for only by comparing can one say that a country is or is not abnormal, deviant, exceptional. Concluding Comments Everyone agrees that the microscope and telescope serve different purposes. In the same way, some of these fifteen strategies could be appropriate for certain problems, but not necessarily for others. Not every strategy is equally useful in studying every kind of problem. The strategies set out here analytically as concurrent, are not exclusive in practice. Some of them may appear complementary. The adoption of one strategy rather than another depends on the nature of the phenomenon to be studied. But for many questions, several strategies can be combined at different stages of the comparative research. The hypotheses validated for a given country in a case study can be tested in a second country by a binary comparison. At a later stage, the analysis can be extended to two series of countries, each one relatively homogeneous, by a comparison of contrasting countries. The comparativist’s freedom of choice of countries is great, unless the problem studied is too closely linked to a determined context. Post-industrial democracies, traditional Islamic societies or totalitarian regimes are social contexts with profound characteristics which condition the choice of countries. But even then it is possible to adopt another strategy by choosing an exterior pole of reference such as emerging pseudo-democracies, secularized societies, or oriental despotic societies. The comparativist establishes causal relationships and observes the interaction of various factors by dividing social reality into specific sectors. Before comparing, it is necessary to segment, choosing at the same time the appropriate countries to be included in the comparative research. It is by such segmentation and choice that during the last three decades a new comparative social science came to be established.

strategies in comparative sociology

43

References Adelman, J., and C.T. Morris. 1971. Society, Politics and Economic Development: A Quantitative Approach. Baltimore: Johns Hopkins Press. Almond, Gabriel. 1960. “A Functional Approach to Comparative Politics.” In The Politics of Developing Areas, edited by G. Almond and J.S. Coleman. Banks, A., and R.B. Textor. 1963. A Cross-Polity Survey. Cambridge, MA: M.I.T. Press. Bebler, Anton, and Jim Seroka (eds.). 1990. Political Systems: Classifications and Typologies. Boulders, Co. Coleman, James. 1960. “The Political Systems of the Developing Areas.” In The Politics of Developing Areas. Princeton University Press. Cutright, Philips. 1963. “National Political Development: Measurement and Analysis.” American Sociological Review 28 (April). Dahl, R., and E. Tufte. 1973. Size and Democracy. Stanford: Stanford University Press. Dogan, Mattei. 2000. “Class, Religion, Party, Triple Decline of Electoral Cleavages in Western Europe.” In Party Systems and Voter Alignments Revisited, edited by L. Karnoven and S. Kuhnle. London: Routledge. ——. 2002. “Are there Paradigms in the Social Sciences?” In International Encyclopedia of the Social and Behavioral Sciences. Oxford: Elsevier. ——, and Dominique Pellassy. 1990. How to Compare Nations (Second edition), New York: Chatham House. ——, and John Kasarda. 1988. A World of Giant Cities, 2 vol., London: Sage. ——, and Peter Campbell. 1957. “Le Personnel Ministériel en France et en Grande Bretagne.” Revue Française de Science Politique 7(2). Flora, Peter (ed.). 1983. State, Economy and Society in Western Europe 1815–1975. Frankfurt: Campus Verlag. Forrest, Joshua, B. 1994. “Weak States in Post-Colonial Africa and Mediaeval Europe.” In Comparing Nations, edited by M. Dogan and A. Kazancigil. Oxford: Blackwell. Grosser, Alfred. 1972. L’Explication Politique, Introduction à l’Analyse Comparative. Paris, Colin. Holtz, R.T., and J.E. Turner. 1970. The Methodology of Comparative Research. New York: Free Press. Huntington, Samuel. 1968. Political Order in Changing Societies. New Haven: Yale University Press. Kamarck, Andrew M. 1976. The Tropics and Economic Development. New York: The World Band Publications. Lijphart, Arendt. 1971. “Comparative Politics and the Comparative Method.” American Political Science, Review 65. Linz Juan. 1975. “Totalitarian and Authoritarian Regimes.” In Handbook of Political Science, edited by F. Greenstein and N.W. Polsby. Reading: Addison-Wesley. Lipset, Seymour M. 1994. “American Exceptionalism-Japanese Uniqueness.” Pp. 153– 212 in Comparing Nations, edited by M. Dogan and A. Kazancigil. Oxford: Blackwell. Martz, John D. 1994. “Problems of Conceptualization and Comparability in Latin America.” In Comparing Nations, edited by M. Dogan and A. Kazancigil. Oxford: Blackwell. Mayer, Lawrence C. 1993. “Practicing what we Preach: Comparative Politics in the 1980s.” Comparative Political Studies 16, 2 ( July):173–194. Naroll, Raoul. 1972. “A Holonational Bibliography.” Comparative Political Studies 16, 2 ( July):5–2. Nowak, Stefan. 1977. “The Strategy of Cross-National Survey Research for the Development of Social Theory.” In Cross-National Comparative Survey Research, edited by A. Szalai and R. Petrella. Oxford: Pergamon Press.

44

mattei dogan

Peter, B. Guy. 1998. Comparative Politics, Theory and Methods. New York University Press. Przeworski, Adam. 1987. “Methods of Cross-National Research: an Overview.” In Comparative Policy Research: Learning from Experience, edited by M. Dierkes, H. Weeler, and A. Berthoin Antal. Berlin: Wissenschaft Zentrum. ——, and Henry Teune. 1970. The Logic of Comparative Social Inquiry. New York: Wiley. Pye, Lucian. 1958. “The Non-Western Political Process.” Journal of Politics 20(3). Ragin, Charles C. 1987. The Comparative Method. Berkeley: University of California Press. Riggs, Fred W. 1994. “Presidentialism in Comparative Perspective.” Pp. 72–152 in Comparing Nations, edited by M. Dogan and A. Kazancigil. Oxford, Blackwell. Sartori, Giovanni. 1994. “Comparing Miscomparing and the Comparative Method.” In Comparing Nations, edited by M. Dogan and A. Kazancigil. Oxford: Blackwell. Scheuch, Erwin. 1966. “Cross-National Comparisons Using Agreggate Data: Some Methodological Problems.” In Comparing Nations, edited by R.L. Merritt and S. Rokkan. New Haven: Yale University Press. Sigelman, L., and G.H. Gadbois. 1983. “Contemporary Comparative Politics: An Inventory and Assessment.” Comparative Political Studies 16, 3 (October):275–306. Taylor, Ch.L., and M.C. Hudson. 1972. World Handbook of Political and Social Indicators. New Haven: Yale University Press. Teune, Henry. 1990. “Comparing Countries: Lessons Learned.” In Comparative Methodology, edited by E. Oyen. London: Sage. Thrupp, Sylvia L. 1970. “Diachronic Methods in Comparative Politics.” In The Methodology of Comparative Research, edited by R.T. Holt and J.E. Turner. New York: The Free Press. Vanhanen, Tatu. 1979. Power and the Means of Power, A Study of 119 States 1850–1975, Ann Arbor, MI: University Microfilm International. ——. 1989. “The Level of Democratization Related to Socio-economic Variables in 147 States 1980–85.” Scandinavian Political Studies 12(2):95–127. Verba, Sidney. 1967. “Some Dilemmas in Comparative Research.” World Politics 20(1). ——. 1971. “Cross-National Survey Research: the Problem of Credibility.” In Comparative Methods in Sociology, edited by Ivan Vallier. Berkeley: University of California Press. Veyne, Paul. 1976. L’Inventaire des Différences. Paris, Seuil. Ward, R.E., and D.A. Rustow. 1964. Political Modernization in Japan and Turkey. Princeton University Press.

METHODS FOR ASSESSING AND CALIBRATING RESPONSE SCALES ACROSS COUNTRIES AND LANGUAGES Tom W. Smith, Peter Ph. Mohler, Janet Harkness, and Noriko Onodera Introduction Scientific research rests on the reliable and consistent measurement of phenomenon. In cross-national or cross-cultural survey research between countries or social groups that speak different languages, the goal of replicative measurement is greatly complicated by the necessity of designing and administering questionnaires in two or more languages. Only by assuring that the items in all languages and questionnaires are equivalent both in meaning and response scales can comparable measurement be obtained and valid inferences drawn. But the complexity of both survey measurement and of languages makes the goal of equivalency an extremely difficult challenge (Glick et al. 2004; Kumata and Schramm 1956; Ommundsen et al. 2002; Scheuch 1989; Smith 1988, 2002, 2004; Van de Vijver and Leung 1997). Each question has two parts: 1) the point of the inquiry or substance of what is being asked about and 2) the implicit or explicit categories in which the response is requested. When the question is open-ended, the requested response is unstructured (e.g. “What is the most important problem facing the country today?” and “Why did you vote for George W. Bush for President?”). But most survey questions are closed-ended with an explicit set of response categories or some type of response mechanism described (e.g. “If you were to consider your life in general these days, how happy or unhappy would you say you are on the whole: Completely happy, Very happy, Fairly happy, Not very happy, or Not at all happy?” and “Do you favor or oppose the death penalty for people convicted of murder?”). While there are effectively an unlimited number of subjects that questions ask about (and an wide variety of ways of asking about each subject), survey researchers tend to use a much smaller number of response categories in their questions. As Davis’ review (1993) of

46

Tom

w. smith et al.

301 questions on the 1985–1993 International Social Survey Program (ISSP) modules showed, several response scales were repeatedly used. For example . . . Scale

# of Items

Agree strongly/Agree/Neither agree nor disagree/ Disagree/Disagree strongly/Can’t choose Essential/Very important/Fairly important/ Not very important/Not important at all/Can’t choose Definitely allowed/Probably allowed/Probably not allowed/Definitely not allowed/Can’t choose Strongly in favor of/In favor of/Neither in favor of nor against/Strongly against Very important/Important/Neither important nor unimportant/Not important/Not important at all/ Can’t choose

92 26 22 11 9

Not only are the same scales utilized again and again, but certain terms tend to be repeated across scales. Note, for example, the use of “very” and “important” in the second and fifth examples above, of “strongly” in the first and fourth examples, and of “can’t choose” in all but the fourth example. Thus, by focusing on the response-scale part of questions, one deals with a set of measurement and translation issues that have widespread application across questions and surveys. In addition, most survey response scales seek to arrange responses along a underlying continuum such as agreement/disagreement, importance, allowance, being in favor of/against, etc.1 By assessing the position of each response category on the underlying continuum, the intensity of the response is determined. If this is done for items in two languages, it becomes possible to determine the equivalency of the individual response categories and ultimately of the response scale as a whole. The task then becomes developing a method for assessing where categories fall on a response continuum. This paper will examine 1) how response categories influence the reported distribution of results, 2) how to measure the intensity of response categories, 3) results from American and German pilot studies

1

Nominal scales do not do this, but these are rare in attitudinal scales.

methods for assessing and calibrating response scales

47

of response scales and a Japanese replication, 4) the use of alternative response scales, and 5) the implications of these results for cross-national research. Response Scales and Reported Distributions Reported distributions are a function of a) the true distribution of attitudes in the population and b) measurement properties of the response scale.2 How much of an underlying distribution is captured by a given term/category is a function of a) the underlying distribution and b) the number, intensity, positioning, and intervals between the scale points utilized. In general, a) the more points used, less of the distribution will be captured by a particular point, b) the closer two points are in intensity, the less of the distribution will be captured by each individual point, c) broader terms may capture more of the distribution than narrower terms (i.e. it is not only the mean intensity of a term, but its range, that determines how much of the distribution will be covered), and d) adding a new, more intense point to a scale can change how the previous end point was understood and alter (and typically increase) the share of the distribution captured by the displaced endpoint. The effect seems to be that some people avoid “extreme” categories where extremity is based on a category representing the end or extreme position on a scale, rather than on the extremity of the term actually used to express the scale point. To illustrate these points, let us start with the simplest case of a dichotomy: agree/disagree. Given the hypothetical distribution of attitudes in Figure 1, the reported distribution would be about 65% agree and 35% disagree. Now suppose a third category, “neither agree nor disagree,” was added and that half of the people closest to the midpoint (4.5) were attracted to this mid-category. The revised distribution would be agree 55%, neither agree nor disagree 17.5%, and disagree 27.5%. Next, suppose that “agree” was replaced with two categories “completely agree” and “somewhat agree”. If “completely agree” was at point 0 and “somewhat agree” at point 3, then the new distribution might be 15% completely agree and 40% somewhat agree. But suppose

2 This paper does not cover other reasons that people in various cultures may respond differently to the offered response options such as response effects due to social desirability or an extremity bias (Smith 2002; Javeline 1999).

48

Tom

w. smith et al.

the two new categories were “completely agree” and “strongly agree” with the former at point 0 and the latter at 2. The distribution would be something like 10% completely agree and 45% strongly agree. But if “strongly agree” was added as a third new category on the agreement side, then the distribution might become completely agree 10%, strongly agree 20%, and somewhat agree 25%. But if “completely agree” was then dropped, then the distribution might become 30% strongly agree and 25% somewhat agree.

X XX – – Agree 0 1 X = 5% of total

X X X – 2

X X X – 3

X X X X – 4

X X X – 5

X XXX0 – – – – 6 7 8 9

Di sagree

Figure 1. Hypothetical distribution.

Alternatively, assume that “slightly agree” was added and represented at 4. It might not only take over much of the somewhat agree cases, but draw in some of the distribution from neither agree or disagree. Along with a matching “slightly disagree” category these might bring back in say half of the distribution lost to the middle category above leaving 59.375% in the combined agree categories, 8.75% in neither agree nor disagree, and 31.875% in disagree. Thus, assuming a fixed true distribution, seven different response scales ranging from two to seven categories, and the simplest of rules for allocating cases, there is considerable variation in distributions reported. In this hypothetical example the % agreeing varies from 55% to 65% and strongly agreeing from 20% to 45%. Now consider what the impact might be of using two scales to measure two different populations with the same true distribution of an attitude as in Figure 1. Suppose that in population A the completely agree/somewhat agree/neither . . . scale was used with the resulting distribution of 15%, 40%, 17.5%, 25%, and 2.5%. In population B the strongly agree/somewhat agree/neither . . . scale was employed and the distribution was 30%, 25%, 17.5%, 17.5%, and 10%. Now assume that population B was interviewed in another language and the researcher was told that the second (non-English) scale was a translation of and equivalent to the first (English) scale. Comparing these two scales using

methods for assessing and calibrating response scales

49

the typical values of 1–5 for the five categories one would conclude that there was more agreement in population B than in population A (means respectively of 2.5 vs. 2.6) and that there was much more extremity in population B than in A (1+5 = 40% vs. 17.5%). Neither conclusion would be correct, but merely the artifact of mistranslations and/or misinterpretations of scales in two languages. Measuring the Intensity of Response Categories There are several ways to measure the strength of response categories along an underlying response scale. One approach is to have respondents rate the strength of terms defining each point on the scale. There are three standard variants of this approach. First, one can rank the terms from weaker to stronger (or from less to more or along any similar continuum) (Spector 1976). This, of course, only indicates their relative position and not the absolute strength or distance between terms. Second, one can rate each term on a numerical scale (usually with 10 to 21 points) (Wildt and Mazis 1978; Worcester and Burns 1975; Myers and Warner 1968; Cliff 1959; Jones and Thurstone 1955; Mosier 1941; Vidali 1975; Mittelstaedt 1971; Bartram and Yelding 1973; Traenkle 1987). This allows the absolute strength or distance between each term to be known and thus facilitates the creation of equal interval scales. Alternatively, it is also possible to use an alphabetical scale or unlabeled spaces, rungs, or boxes as in a semantic differential scale (Osgood et al. 1957). The letters or spaces are then transformed into their numerical equivalents. Finally, magnitude measurement techniques can be used to place each term on a ratio scale (Lodge et al. 1975, 1976, 1979, 1981, 1982; Hougland et al. 1992; Osinski and Bruno 1998). The magnitude measure techniques gives an arbitrary value to a reference term and has respondents rate other terms as ratios to this base term. This allows more precision than the numerical scale approach (since the terms are not constrained by the artificial limits of the bounded number scale). Of these three variants the middle seems most useful. On the one hand, the ranking method fails to provide the numerical precision that is necessary to calibrate terms across languages. On the other hand, the magnitude measurement technique is much more difficult to administer and much harder for respondents to work with (about 10–15% seem

50

Tom

w. smith et al.

unable to master the procedure). In addition, the extra precision that the magnitude measurement procedure can provide over that achievable using a 21-point scale approach does not appear to be needed. The direct rating approach has been used to rate words along various dimensions. Of most interest to us are those that either rate terms along a general good/bad or positive/negative dimension or which rate the intensity of modifiers (Wildt and Mazis 1978; Worcester and Burns 1975; Myers and Warner 1968; Cliff 1959; Jones and Thurstone 1955; Mosier 1941; Vidali 1975; Mittelstaedt 1971; Bartram and Yelding 1973; Lodge et al. 1975, 1976, 1979, 1981, 1982; Hougland et al. 1992; Bullinger 1995; Szabo et al. 1997; Skevington and Tucker 1999; Skevington 2002). Similarly, other studies have rated probability statements (Wallsten et al. 1986; Lichtenstein and Newman 1967); frequency terms (Spector 1976; Schaeffer 1991; O’Muircheartaigh et al. 1993; Strahan and Gerbasi 1973; Bradburn and Sudman 1979; Schriesheim and Schriesheim 1974; Hakel 1968; Simpson 1944); and terms used in reports to describe percentages from public opinion surveys (Crespi 1981 and “RAC . . .,” 1984). The studies generally show that a) people (usually college students) can perform the required ratings tasks,3 b) ratings and rankings are highly similar across different studies and populations, c) there is high test/retest reliability, and d) several different treatments or variations in rating procedures yield comparable results. Thus, the general technique seems robust and reliable.4 A second approach for assessing the intensity of scale terms and response categories is to measure the distributions generated by using different response scales (Smith 1979; Laumann et al. 1984; Michael and Michaels 1994; Hougland et al. 1992; Orren 1978; Sigelman 1990). In an experimental, across subjects design, one random group is asked to evaluate an object (e.g. presidential popularity or one’s personal happiness) with one set of response categories and a second random group

3 While this is reassuring, other studies show that various measurement artifacts can influence responses to numerical scales (Wilcox et al. 1989; Smith 1993; Schwarz and Hippler 1995; Schwarz et al. 1985; and Schwarz et al. 1991). See also, O’Muircheartaigh et al. 1993; Wright et al. 1997. 4 An exception is that vague frequency terms correspond to different absolute values depending on the commonness or rarity of the specified event or behavior. Thus, people who “usually” vote may vote once a year, but people who “usually” dine out dine out more than once a week (Schaeffer 1991; Bradburn and Sudman 1979).

methods for assessing and calibrating response scales

51

evaluates the same object with another set of response categories. Since the stimulus is constant and the sub-group assignment is random, the number of people attracted to each category will depend on the absolute location of each response category on the underlying continuum and the relative position of each of the scale points adopted. With some modeling around what the two observed distributions suggest are the underlying distribution, it is possible to estimate at what point each term is cutting the underlying scale (Clogg 1982, 1984). The alternative version uses a within subjects design in which people are asked the same question (i.e. presented with the same stimulus) two or more times with different response categories being used (Orren 1978). This differs from a test/retest reliability design in that a) the measurement instrument is not constant (since the response categories differ) and b) the two administrations are essentially consecutive without any intervening time and/or buffer tasks. This provides additional information since it allows the direct comparison of responses, but the initial evaluations may artificially influence responses to the later scales (e.g. a person may feel constrained to choose the same response in terms of position or term used on the first administration on a subsequent administration). The advantage of the distributional approaches is that they ask respondents only to do what they are normally required to do—to answer substantive questions with a simple set of response categories. The disadvantages are that a) it is harder to access a large number of response terms and thus is better suited for assessing a discrete response scale already adopted than for evaluating a large number of terms that might be utilized in possible response scales,5 b) results will depend on the precise underlying distribution and the modeling procedures adopted, and c) it creates more work for the analysts, since the strength of terms must be indirectly estimated from the distributions rather than directly calculated from respondent ratings. A final approach uses anchoring vignettes to establish comparability across measures (Banks et al. 2004; d’Uva et al. 2006; King et al. 2004;

It would be possible to evaluate more terms using more random sub-groups, but in order to maintain the same level of precision this would mean increasing the sample size. Similarly, the same people could be asked many repetitions of a question with different response scales, but this would soon become tedious and later repetitions would probably be distorted by the previous administrations. 5

52

Tom

w. smith et al.

Salomon et al. 2004). Short vignettes describing a person’s situation regarding the construct of interest are devised and then respondents evaluate the person’s situation and rate it. For example, the vignettes may describe a person’s health status and then ask respondents to rate that person’s health as “excellent, very good, good, fair, or poor.” Given that the vignette’s person’s objective, health-related conditions are fixed and identical across respondents, differences in ratings are deemed to reflect how the scale is understood and utilized by respondents. When comparing two groups such as respondents from two countries, the mean differences in responses to the vignettes can be used to anchor people’s ratings of their own health and this make those ratings more comparable across groups. As with the response-scale, calibration approach, the anchor-vignette approach does not have to be asked of all respondents on every survey, but can be used to generate general adjustment factors that can be applied whenever the tested construct and response scale are used. This approach rests on several assumptions. First, response consistency assumes that respondents use scales to rate people in vignettes in the same way that they use scales to rate their own situation. Second, vignette equivalence assumes the objective situations in vignettes are perceived by people across groups in the same way. While not implausible, neither of these assumptions have been seriously tested. Because the direct rating approach provides the quantified intensity scores needed in the most straight-forward manner, this was adopted as the main technique in this study. In addition, there may be context effects in the rating of the intensity of terms. For example “very” may be rated more intensely if it was the first strong term presented than if it followed other stronger terms (e.g. completely, extremely). Context effects have generally not been searched for in this line of research, but the randomization of order in several studies has tried to average out any such effects. This latter approach is generally utilized here, but an ordered vs. not ordered experiment is also included. American and German Pilot Studies and Japanese Replication Pilot studies were carried out in the United States and Germany to use the above approach to evaluate the translation and equivalency of response scales. The American pilot study was carried out on a quasi-representative sample of adults living in households. Ten sample

methods for assessing and calibrating response scales

53

points were selected to represent all four Census regions (West, South, Midwest, and Northeast) and three size of place strata (central cities, metropolitan areas outside of central cities, and non-metropolitan areas). Interviewers had quotas to fill based on gender, age, and employment status. They proceeded through neighborhoods in the selected communities until the quotas were completed. The study was designed and carried out by the National Opinion Research Center at the University of Chicago. Besides representing the adult population of United States on the stratification and quota variables (region, size of place, gender, age, and employment status), the sample is also representative on race and marital status. The sample does underrepresent the less educated segment of the population (less than a high school degree: pilot study 6%, General Social Survey 17%). Interviews were conducted in July/ August, 1995. A total of 119 interviews were collected, but two were lost in the mail for a final total of 117. The German pilot study stratified the country by states (Bundeslaender) and city size (cities over 100,000 vs. else). Within these areas interviewers filled quotas based on gender, age, and education. The study was designed and supervised by the Zentrum fuer Umfragen, Methoden, und Analysen, Mannheim, and interviewing was conducted by Infratest— Burke Sozialforschung, Munich. The sample closely matches German Census figures on gender, age, and education. Fieldwork was carried out in September, 1995. A total of 221 interviews were conducted. In order to see how the results between two linguistically-similar languages (and two societies with close cultural and historical ties as well) compared to findings from a country with a very dissimilar language (and more remote culture and history in general), the study was later replicated in Japan. The Japanese study was carried out by the NHK Broadcasting Culture Research Institute on a national sample of 405 in March, 2001 (Onodera 2002). American Results In the pilot study attempts were made both to assess the intensity that people assigned to particular terms and therefore response categories and to evaluate the meaning of the underlying continuum on which intensity was being measured. First, people were asked to rate the intensity of 27 phrases on a 21-point agree/disagree scale (See Qs A3 and B3 in the Appendix). Item order was randomized by sorting cards

54

Tom

w. smith et al.

Table 1. Mean scores on agree/disagree terms Term

Mean

Standard Deviation

Completely agree Definitely agree Strongly agree Very much agree Agree a lot Agree Basically agree Probably agree Tend to agree Moderately agree Somewhat agree Agree a little In the middle Neither agree nor disagree Can’t choose Undecided Disagree a little Somewhat disagree Moderately disagree Tend to disagree Probably disagree Disagree Not agree Disagree a lot Strongly disagree Very much disagree Definitely disagree Completely disagree

19.4 19.0 18.8 18.5 17.2 16.1 13.8 13.6 13.5 13.3 12.9 12.1 10.1 9.9 9.8 9.6 7.1 6.6 6.4 6.4 6.2 3.5 3.5 3.0 1.5 1.4 1.0 0.8

1.6 1.5 1.3 2.2 2.8 2.9 3.1 2.9 2.8 2.3 2.4 2.6 0.7 1.3 2.7 1.8 2.2 2.1 2.3 2.7 3.1 2.9 3.1 3.6 2.2 1.6 1.3 2.3 N = 97–101

containing the phrases, except for “basically” which was the first term rated by each respondent. Table 1 gives the means and standard deviations for the terms.6 In terms of magnitude and relative position the

6 On a scale-by-scale basis cases were excluded from the analysis that failed to carry out the ratings adequately. This excluded respondents who refused to do items, those with high item non-response, those who could not consistently associate terms with the proper pole, and those showing peculiar response patterns. People were not excluded for a few unusual responses, but for incomplete and erratic responses to the scale as a whole. There were 14 exclusions for Q. 3 (agree/disagree), 5 for Q. 4 (important/unimportant), 10 for Q. 5 (in favor of/against), and 12 for Q. 6 (ranges

methods for assessing and calibrating response scales

55

terms array themselves almost exactly as one would expect.7 The 11 agree terms run from “agree a little” at 12.1 to “completely agree” at 19.4. The four mid-point or uncertain terms are from 10.1 to 9.6. The 11 disagree terms range from “disagree a little” at 7.1 to “completely disagree” at 0.8. In addition, “not agree” exactly matches “disagree” at 3.5. Standard deviations follow a wave pattern. They are small near the extremes, increase as intensity moderates, and then decrease to their lowest level for the two mid-point categories (in the middle and neither agree nor disagree). The lower range for the categories near the extremes (strongly agree/disagree) is only partly a function of floor and ceiling effects resulting from respondents rating the terms at or near the end-points. The unbounded end of the range is usually a little smaller than that for broader and more moderate terms. For example, the average upper range for strongly disagree is +1.6 compared to +2.2 for disagree, while the average lower range for strongly agree is –1.9 compared to –2.2 for agree. Thus these terms appear to have more precise and limited meanings not only because of floors and ceilings, but also because their greater intensity also narrows people’s understanding of their meaning. The standard deviations narrow for the middle categories because people have a clear and consistent understanding of on agree/disagree). Overall, there were 15 respondents who were excluded for two or more individual scales. Exclusions were significantly associated with interviewer assessments that respondents misunderstood the word rating tasks and that these tasks were difficult. (The interviewer evaluations questions were “How was the respondent’s understanding of the word rating tasks? Completely understood/Mostly understood/ Mostly misunderstood/Completely understood” and “How hard were the word rating tasks for the respondent? Very difficult/Somewhat difficult/Somewhat easy/Very easy.”) Exclusions were also higher among the less educated, although only the association with Q. 5 was statistically significant. 7 On the agree/disagree rating scale the questionnaire was handed back to respondents after the question was completed and they were told by the interviewer “Please look over your answers. If you want to change any of your responses, indicate in the right-hand column, the one headed “CHANGES,” what number you now want to give a phrase.” Respondents to later questions were not given a chance to review their responses, but at any point while a question was being administered a respondent could change a response. Changes were fairly rare. 62.4% made no changes, 17.0% 1–2 change, 14.6% 3–6 changes, and 6.0% 7+ changes. On average 1.7 changes were made among the 28 phrases rated. Two type of changes were common. First, there were minor upward or downward adjustments to have responses better fit in with other phrases being rated. Second, there were pole corrections when respondents realized they had oriented their response to the wrong end of the scale. These usually resulted in large changes (e.g. from 2 to 18). In almost all cases, the changes moved answers towards the modal response.

56

Tom

w. smith et al.

what the mid-point of a scale is. The uncertain terms “can’t choose” and “undecided” are also placed near the middle, but the standard deviations are a bit higher because some people wanted to rate them as off-scale and gave some different responses such as 0 to try to convey this idea. (In addition, a few more people than for the other terms did not rate these terms for the same reason.) Second, a similar exercise was carried out on two important/unimportant scales (Qs. A4 and B4 in Appendix). As in the case of the agree/disagree scale, order of presentation was randomized by sorting. Table 2 reports the means and standard deviations. On one half of the sample people rated terms on an unipolar scale measuring degree of importance and on the other half on a bipolar scale of important/unimportant. The unipolar scale ran from 19.4 for “extremely important” to 1.4 for “not at all important.” The bipolar scale extended from 19.4 “extremely important” to 0.8 for “extremely unimportant.” On this scale middle terms were placed very near the mid-point (“in between” = 10.0; “neither important nor unimportant” = 9.5). There were 15 “important” terms that were rated on both scales. In 13 of these cases the terms were rated somewhat higher on the bipolar scale than on the unipolar scale. It appears that on the important/not important scale people adjust terms down towards the not important end of the scale. For example, “neither important nor unimportant” is scored at 9.0 instead of the mid-point of 10.0. This suggests that “unimportant” defines a more extreme position than the lack of importance does. The latter is seen by at least some as indicating the absence of importance rather than the presence of unimportance. Standard deviations are smaller for high terms on both the unipolar and bipolar scales, but the pattern is less clear at the lower end of these scales. “Not important” terms have the largest standard deviations of all terms on the bipolar scale and on the unipolar scale they have among the largest values. Some negative phrases tend to confuse people in general (Smith 1995) and especially on the bipolar scale people were less sure where to rate these terms vis-a-vis the “unimportant” terms. Third, Table 3 rates another set of terms, “against/in favor of,” and also carries out an order experiment (Qs. A5 & B5). In terms of the means and standard deviations both orders are similar to each other and to the pattern shown with “agree/disagree” in Table 1. In particular, the means have magnitudes and relative positions as one would expect and the standard deviations show the same wave pattern of going from small for extreme terms to larger for more moderate, general terms and then smaller for the middle term.

methods for assessing and calibrating response scales

57

Table 2. Mean scores on important/unimportant terms Term

Important: Extremely Very, very Exceptionally Completely Definitely Highly Very Quite IMPORTANT Pretty Probably Fairly Somewhat Slightly A little bit Neither imp. nor unimp. Not too Not very Not Not at all Unimportant: In between Slightly A little bit Somewhat Probably Fairly Very Pretty UNIMPORTANT Definitely Completely Extremely N

Important Only Important/ Combined List List Unimportant List Mean SD Mean SD Mean SD 19.4 19.0 18.9 18.6 18.5 18.2 18.2 16.8 15.1 15.0 13.0 13.4 12.2 10.8 10.1 9.0

1.2 2.4 2.4 2.9 2.0 1.9 1.5 2.8 3.6 3.1 3.4 3.5 3.5 3.4 4.2 3.6

19.4 – – 19.1 18.4 – 18.3 – 16.3 15.6 14.0 13.9 13.2 12.0 12.2 9.5

0.9 – – 1.8 2.0 – 2.6 – 2.8 2.6 3.1 2.2 2.5 2.7 2.4 2.4

19.4 – – 18.8 18.4 – 18.2 – 15.5 15.3 13.5 13.6 12.7 11.3 11.1 9.3

1.0 – – 2.5 2.0 – 2.1 – 3.5 2.9 3.3 3.0 3.1 3.2 3.7 3.1

6.8 4.7 2.4 1.4

3.6 3.4 3.4 3.3

– 5.5 4.1 3.0

– 4.2 4.4 4.2

– 5.1 3.2 2.2

– 3.8 4.0 3.8

– – – – – – – – – – – –

– – – – – – – – – – – –

10.0 8.0 7.9 6.6 6.1 5.8 5.1 4.7 3.6 1.8 1.3 0.8

2.2 2.7 3.0 2.7 3.8 3.2 6.7a 3.1 3.9 3.3 3.6 1.8

– – – – – – – – – – – –

– – – – – – – – – – – –

56–58

51–54

109–112

This item has a small number of cases coded near the high end of the scale (16–20). These cases create the large standard deviation (6.7) and also make the mean (5.1) much higher than the median (2). In all other cases the mean and median are very close (almost always within +/– 1). Inspection of the cases to see why a high number of pole reversals (i.e. errors of reference) occurred on this item did not reveal any special cause.

a

58

Tom

w. smith et al.

Table 3. Ratings of In favor of and Against Ascending Order Mean StdDev

Mixed Order Mean StdDev

Strongly against Against Slightly against Neither against nor in favor of Slightly in favor of In favor of Strongly in favor of

1.6 3.9 6.9 9.6 12.2 15.9 18.9

2.0 2.2 2.4 1.7 2.3 2.4 1.3

1.3 3.5 7.2 9.8 12.1 15.6 18.6

1.9 3.4 1.9 1.4 1.9 2.8 1.6

% with all items rated in ascending order

71.7

–

37.0

–

N

53

52–54

The order experiment did however reveal a decided difference in terms of the consistency of ratings. On the version that arranged terms in ascending order from “strongly against” to “strongly in favor of ” as they would be presented as part of a response scale, 72% of people rated all seven terms in ascending order without any inconsistency. On the version that presented the terms in a fixed, unordered sequence only 37% of respondents rated all seven terms in ascending order. This indicates that presenting the terms in ascending order, as they are presented as actual response scales, provides people with additional information and constrains how people perceive and evaluate the terms. When terms are organized as a scale, people are more likely to perceive and treat them as such. Fourth, Table 4 shows that the values assigned to terms at both ends of the scales for agree/disagree, important/unimportant, and against/in favor of (Table 1–3) are highly symmetrical. The first column gives the mean rating for each term when associated with the positive/top end of the scale. The second column gives the rating when used in conjunction with the lower end of the continuum. The third column reverses the numbers in the second column to show what they equal if rated at the opposite end. Comparing the first and third columns shows how similar and symmetrical the ratings are. With one exception, all terms rated at the positive end practically match how they are rated at the negative pole. This indicates that people assign these terms a consistent value regardless of their positive or negative orientation.

methods for assessing and calibrating response scales

59

Table 4. Symmetry in ratings A. Ratings of Agree/Disagree (Samples A and B) Agree Disagree Completely Definitely Strongly Very much A lot AGREE/DISAGREE Not agree Probably Tend to Moderately Somewhat A little

19.4 19.0 18.8 18.5 17.2 16.1 – 13.6 13.5 13.3 12.9 12.1

0.8 1.0 1.5 1.4 3.0 3.5 3.5 6.2 6.4 6.4 6.6 7.1

20 – Disagree 19.2 19.0 18.5 18.6 17.0 16.5 16.5 13.8 13.6 13.6 13.4 12.9

B. Ratings of Importance/Unimportance (Sample B) Important Unimportant 20 – Unimp. Extremely Completely Definitely Very IMPORTANT/UNIMPORTANT Pretty Probably Fairly Somewhat A little bit Slightly

19.4 19.1 18.4 18.3 16.3 15.6 14.0 13.9 13.2 12.2 12.0

0.8 1.3 1.8 5.1 3.6 4.7 6.1 5.8 6.6 7.9 8.0

C. Ratings of In favor of/Against (Samples A and B) In favor of Against Strongly In favor of/Against Slightly a

See note “a” in Table 2.

18.7 15.8 12.2

1.5 3.7 7.1

19.2 18.7 18.2 14.9 16.4 15.3 13.9 14.2 13.4 12.1 12.0 20 – Against 18.5 16.3 12.9

60

Tom

w. smith et al.

Table 5. Comparisons across rating scales A. Agree/Disagree and Important/Unimportant (Sample B) Agree Important Disagree Completely Definitely Very much/very BASE WORD Probably Somewhat A little/a little bit

19.4 18.9 18.2 15.6 13.5 12.5 12.1

19.1 18.4 18.3 16.6 14.0 13.2 12.2

0.8 1.2 1.4 3.7 6.5 6.6 7.1

B. Agree/Disagree and In favor of/Against (Samples A and B) Agree In favor of Disagree Strongly BASE WORD Neither Agree/In favor or . . . a

18.8 16.1 9.9

18.7 15.8 9.7

1.5 3.5 –

Unimportant 1.3 1.8 5.1a 3.6 6.1 6.6 7.9 Against 1.5 3.7 –

See note “a” in Table 2.

Fifth, Table 5 shows that terms are also rated in a highly similar manner when the underlying continuum varies. Part A indicates that terms rated on the agree/disagree and important/unimportant scales have highly consistent values. Part B reveals that terms rated on agree/ disagree and in favor of/against are also quite similar. Along with the results from Table 4, this indicates that ratings are robust and that terms probably have similar intensities across various scales. Sixth, the rating of scales are also quite stable across sub-groups. Subgroup differences were examined for all items rated in the agree/ disagree and important/unimportant scales. Differences by gender, age, education, and race were examined. While a few statistically significant results emerged, there were no consistent differences either across samples or demographics. Education showed the most significant differences (6 of 43) Pearson’s correlations, but only one significant one-way analysis of variance. The education effects that do appear seem to be related to the greater difficulty of less educated respondents in carrying out the rating task, rather than to systematic differences in the meaning of terms. Finally, Table 6 examines intra- and inter-respondent variability in the rating of terms. Intra-respondent variability was measured by selecting eight terms rated on the agree/disagree scale and reminding people

methods for assessing and calibrating response scales

61

Table 6. Range of acceptable values Mean Rangea

StdDevb

2.6 4.0 4.1 1.6 1.6 4.1 3.9 2.6

1.3 3.1 2.9 1.3 2.7 3.5 3.1 2.1

Strongly agree Basically agree Agree Neither agree nor disagree Can’t choose Disagree Not agree Strongly disagree a b

Difference between high and low limits in Q. 6 Standard deviation of items in Q. 3

what score they had assigned to the terms. Next, people were asked what was the lowest value they would accept for the term and what was the highest (Qs. A6 & B6). If they thought that no variation from their earlier assignment was acceptable, then that same value was entered as the minimal and maximum score for the term and the acceptable range was 0. The first column shows the mean interval between the top and bottom values. First, these values follow the wave pattern described earlier for the standard deviations (which are presented in the second column for comparison). Acceptable ranges are narrow at the extremes and at the middle and widest between the middle and extremes. Second, the ranges are almost perfectly symmetrical with strongly agree/strongly disagree and agree/disagree showing the same means. Third, most people see these term as somewhat malleable. They do not believe that the terms have only a precise and invariant value (like agree = 16.1), but see terms covering a range of values (e.g. 14–18). Next, assessments were made of the meaning of the underlying dimension on which the above terms were arrayed. First, the similarity between different pair of words were examined. In Table 7 five pair of words were compared with the pair “agree/disagree” (Qs. A8 & B8). People evaluated how similar the “agree/disagree” pair was to each of the other pairs. “For/against” and “favor/oppose” were considered to be the most similar, “positive/negative” the next closest, and “like/dislike” and “important/unimportant” the least alike. This indicates that “agree/disagree,” and “for/against” and “favor/oppose” come closest to tapping a similar underlying dimension, while the other pairs define more distinct continuums.

62

Tom

w. smith et al.

Table 7. Closeness of the meaning of various pairs of words to Agree/Disagree Meana

% Very much the same

2.4 3.2 2.9 2.3 2.9

34.7 8.9 8.8 29.7 14.9

For/against Important/unimportant Like/dislike Favor/Oppose Positive/negative

n = 101

Response scale ran from 1 = Very much the same to 4 = Very much different. Lower number indicates pairs are closer.

a

Table 8. Terms used in the definition of agree Accept, acceptance Accord, accordance Against (not) Agree, agreeable, agreeing, agreement Alike Approve Congenial Consensus, consent Disagree (not) Favor For Harmony In line with Like, liking Mutual OK Same Similar Support True Valid

6 2 1 13 1 3 1 4 1 6 6 2 2 3 1 2 16 1 2 2 1

Then, the similarity of other terms to those used in the “agree/disagree” (“agree,” “neither agree nor disagree,” and “disagree”) and “important/unimportant” (“important” and “unimportant”) dimensions were assessed by an open-ended item that asks people to define these terms.8 8 The definition tasks were found to be fairly hard by many people. A number of interviewers noted in the evaluation section that particular people had problems expressing themselves and often used the word itself as part of the definition.

methods for assessing and calibrating response scales

63

Table 8 lists the terms offered to define “agree.” The list basically includes synonyms along with repetitions of “agree” itself. The use of this list will be discussed in the comparative section below. America and Germany Compared The preceding analysis indicates how useful the evaluations of the response terms are for understanding response scales in general. Here the use of this information for comparing scales in two countries and languages is considered. Tables 9 and 10 show that overall there is a high correspondence between the agree/disagree and important/unimportant scales in the United States and their counterparts in Germany. Table 9 presents the mean ratings for the agree/disagree and the two German counterparts stimme zu/lehne ab and stimme zu/stimme nicht zu. The American scores correlate almost as highly with both German scales (respectively r = 0.993 and 0.986) as the two German scales associate with each other (r = 0.995) and most means are close and not statistically different from one another (Mohler et al. 1997). Despite this extremely high correlation and the general correspondence in scale scores, there are some important differences in the mean values. First, the base words (e.g. agree, stimme zu, disagree, lehne ab, etc.) have more extreme meanings in German than in English. For example agree is 16.1 in English and stimme zu 17.4–17.5 in German. Semantically “lehne ab (from “ablehnen”) is more like “to reject (e.g. an idea)” than disagree; German lacks a verb directly corresponding to “disagree”. Second, “definitely” is a stronger term in English than “bestimmt” is in German. Third, while “strongly” is a weaker term in English than either “completely” or “definitely,” this does not appear to be the case in German where “voll und ganz” shows up as the strongest German term. But this disagreement is really in using “voll un ganz” to stand for “strongly”. It means literally “fully and wholly” and as such it is not unexpected that it is rated more highly. In retrospect, perhaps the German phrase “stimme stark zu” should have been used to match “strongly agree”. Fourth, while “a lot” is an intensifier in English both in terms of semantics and its performance in this study, “ziemlich” in German is more complicated. In terms of semantics “ziemlich” is a middle to high intensifier depending on context. “Ziemlich” X can mean X to a considerable, but not extreme, degree or as very X. There is a direct and literal way of expressing very X in German (“sehr” X). In selecting “ziemlich” X with the intention of meaning very X, one

64

Tom

w. smith et al.

Table 9. American/German scores on Agree/Disagree, Stimme zu/Lehne ab, and Stimme zu/Stimme nicht zu English/German

Agree/Stimme zu: Completely/Voellig Definitely/Bestimmt Strongly/Voll und ganz Very much/Sehr A Lot/Ziemlich AGREE/STIMME ZU Basically/Im grunde Probably/Wahrscheinlich Tend to/Eher Moderately/Maessig Somewhat/Teilweise A little/Ein bisschen

America

Germany Stimme zu/ Stimme zu/ Lehne ab nicht Means SD Means SD

Means

SD

19.4 19.0 18.8 18.5 17.2 16.1 13.8 13.6 13.5 13.3 12.9 12.1

1.2 1.5 1.3 2.2 2.8 2.9 3.1 2.9 2.8 2.3 2.4 2.6

19.3 17.9 19.7 17.6 16.0 17.5 14.4 13.8 13.8 12.3 13.3 12.5

1.8 2.5 1.4 2.8 2.2 2.5 2.9 2.8 2.5 2.4 2.5 2.5

19.3 17.6 19.8 18.3 16.4 17.4 14.6 14.0 13.8 10.4 12.8 11.7

1.4 3.1 0.7 2.1 2.6 2.8 3.5 3.1 3.1 3.9 2.5 3.5

0.7 1.3

10.0 9.7

1.3 1.5

9.9 9.6

1.3 2.4

2.7

9.5

2.5

8.5

3.7

1.8

10.0

1.1

10.0

0.5

2.2 2.1

6.7 6.8

2.4 2.5

– 7.6

– 3.1

2.3 2.7 3.1 2.9 3.1

6.6 5.9 6.1 2.9 3.5

3.1 2.4 2.6 2.6 3.6

– 6.0 4.9 1.2 1.8

– 3.1 2.5 2.4 2.2

3.6

4.1

2.6

4.4

3.1

2.2

1.7

2.7

0.4

0.9

1.6

2.0

2.6

1.2

2.9

1.3 2.3

2.7 1.1

3.0 2.6

1.7 0.6

2.8 1.6

Middle/Mitte: In the middle/In der mitte 10.1 Neither agree nor disagree/ 9.9 Stimme weder zu noch lehne ab/Stimme weder zu noch nicht zu Can’t Choose/Kann ich 9.8 nicht sagen Undecided/Unentschieden 9.6 Diasagree/Lehne ab/Stimme nicht zu: A little/Ein bisschen 7.1 Somewhat/Teilweise/ 6.6 Zum teil nicht Moderately/Maessig 6.4 Tend to/Eher 6.4 Probably/Wahrscheinlich 6.2 DISAGREE/LEHNE AB 3.5 Not agree/STIMME 3.5 NICHT ZU A lot/Ziemlich/ 3.0 Ueberwiegend nicht Strongly/Stark/ 1.5 Ueberhaupt nicht Very much/Sehr/ 1.4 Entschieden nicht Definitely/Bestimmt 1.0 Completely/Voellig/ 0.8 Ganz und gar nicht

methods for assessing and calibrating response scales

65

Table 10. American/German scores on Important/Unimportant and Wichtig/Unwichtig English/German Important/Wichtig: Extremely/Aeusserst Completely/Voellig Definitely/Bestimmt Very/Ganz IMPORTANT/WICHTIG Pretty/Schon ziemlich Probably/Wahrscheinlich Fairly/Einigermassen Somewhat/Teilweise A little bit/Ein wenig Not very/Nicht sehr Not/Nicht Not at all/Ueberhaupt nicht Middle: Neither important nor unimportant/Weder wichtig noch unwichtig In between/Dazwischen Unimportant/Unwichtig: A little bit/Ein wenig Somewhat/Teilweise Probably/Wahrscheinlich Fairly/Einigermassen Very/Sehr Pretty/Schon ziemlich UNIMPORTANT/ UNWICHTIG Definitely/Bestimmt Completely/Voellig Extremely/Aeusserst

America Means SD

Germany Means SD

19.4 18.8 18.4 18.2 15.5 15.3 13.5 13.6 12.7 11.1 5.1 3.2 2.2

1.2 2.5 2.0 2.1 3.5 2.9 3.3 3.0 3.1 3.7 3.8 4.0 3.8

18.6 18.8 16.5 18.4 16.4 15.7 13.0 11.9 12.9 11.0 5.8 2.6 1.5

3.0 3.0 2.6 2.9 2.8 2.7 3.5 3.9 2.8 3.5 3.7 3.4 2.8

9.3

3.1

9.7

2.3

10.0

2.2

10.1

1.9

7.9 6.6 6.1 5.8 5.1 4.7 3.6

3.0 2.7 3.8 3.2 6.7 3.1 3.9

7.0 6.6 5.3 5.2 1.5 4.8 2.2

3.2 2.6 3.1 3.1 3.7 4.6 3.4

1.8 1.3 0.8

3.3 3.6 1.8

2.6 1.1 0.9

3.1 3.0 2.8

66

Tom

w. smith et al.

opts for stylistic understatement. In literal terms however “ziemlich” remains less intense than “very” or “a lot”. As a result, one can understand it as meaning less than “very” and thus more moderate. Thus, if “stimme zu” is taken as an absolute, “Stimme ziemlich zu” can be understood as less than the absolute, meaning something like a good bit of agreement, but not all. This apparently is what many German respondents did, since empirically it acted as a deintensifier. In the ZUMA survey 58–85% of respondents rated it less strongly than they rated the base words (stimme zu, lehne ab, stimme nicht zu). It maybe that “ziemlich” is not an appropriate translation of “a lot.” Table 10 shows the mean scores for important/unimportant and wichtig/unwichtig. As before, the cross-national, scale scores correlate very strongly (r = 0.987) and most means are quite close. But again there are some notable differences. First, as in previous comparisons, the base words are stronger in German than in English (e.g. unimportant = 3.6 and unwichtig = 2.2). Second, “definitely” again shows up as stronger than “bestimmt.” The difference in the intensity of base terms may be a general difference between English and German. The pattern appears not only for agree/disagree and important/unimportant (See Tables 9 & 10), but also for in favor of/against. Next, Table 11 shows the frequency of English terms used to define “agree” and German terms used to define “stimme zu.” The next step in the analysis is to take each English term and translate it into German and each German term and translate it into English. “Agree” and “stimme zu” will be judged to mean the same thing to the extent that a) the German terms offered as meaning “stimme zu” match the German terms translated from English terms used to define “agree” and b) the relative frequency of these terms are similar. Perfect correspondence would involve only matched terms appearing and in the same proportions. While the detailed analysis has not been carried out, the terms in Table 11 clearly show both much overlap and some distinctions. For example, “bejahe,” etc. means “accept,” “give consent” both of which appear among the English terms and translations of “approve,” etc. include “zustimmung” which is among the German terms offered. But “positiv” in German is equivalent to “positive” in English and this term is not mentioned in the American survey.

methods for assessing and calibrating response scales Table 11. Terms used in the definition of Agree and Stimme zu A. Agree Accept, acceptance Accord, accordance Against (not) Agree, agreeable, agreeing, agreement Alike Approve Congenial Consensus, consent Disagree (not) Favor For Harmony In line with Like, liking Mutual OK Same Similar Support True Valid

6 2 1 13 1 3 1 4 1 6 6 2 2 3 1 2 16 1 2 2 1

B. Stimme zu Akzeptieren, akzeptabel, akzeptiere Anerkannen Befueworte, befueworten, befuewortung Bejahe, bejahen, bejahung Dafuer Einverstanden, Einverstandnis, einverstandniserkaerung Gleiche, gleichen, gleicher Grosse Grund, grunde Gut Identisch Positiv, positive Richtig, richtige, richtigkeit Selbe, selben, selber Voll, volle, volles Uebereinstimmen, uebereinstimmung Ueberzeugt Ueberzeugung Zustimme, zustimmen, zustimmung Zutreffend Zuveriaessig

3 1 4 8 42 55 23 1 4 8 1 7 9 6 11 7 9 3 14 1 1

67

68

Tom

w. smith et al.

America and Japan compared The Japanese instrument was based on a translation of the American instrument. The German questionnaire was not consulted. The larger differences between Japanese and English compared to English and German and the greater challenge of coming up with equivalent terms for testing were apparent from the start. Two bilingual translators, one a native Japanese speaker and the other a native English speaker, independently translated the 50 some agree/disagree and importance terms to be rated. They came up with only one translation that was identical (although many were similar). Moreover, of the 28 terms used for agree/disagree, the final terms used in the study agreed with either of the initial translations in only 11 of 28 cases (Onodera 2002).9 Overall, there was a high agreement on the ranking of terms between the English and Japanese on both the agree/disagree and importance scale (Tables 12 and 13). There was also usually close agreement on the absolute scores given. However, absolute differences were somewhat larger than for the English-German comparisons. On agree/disagree 9 of 28 terms differed by one point or more for English and Japanese, but only 6 of 28 differed by that much for English and German. Similarly, on importance 12 of 26 terms were separated by a point or more for English/Japanese, while only 4 of 25 varied by this much for English/ German. The main differences were that 1) the more intense terms on agreedisagree were always rated more towards the endpoints in English than the Japanese terms were (this was also true, but to a lesser extent, for the importance terms), 2) there was closer ratings toward the high end of the importance scale and less towards the low end, and 3) while there was close agreement on the mid-point location of “dochira to mo ieru/in the middle” and “dochira to mo ienai/neither agree nor disagree”, there was less agreement on “wakarani/can’t choose”.10 Given the linguistic gulf between English and Japanese, the matching of terms was notable. But where differentiation occurred, further work on comparable translation and scaling are indicated. 9 Also, there is the matter as to how closely the base terms agreed. The Japanese terms used for agree/disagree are not literal equivalents, but come closer to the notion of “thinking X is the case/thinking X is not the case.” 10 On “wakarani/can’t choose” 15–19% of Japanese respondents gave no rating (left blank) apparently thinking of the items as off-scale and a number of people gave ratings of zero—perhaps trying to convey the same idea (Onodera 2002).

methods for assessing and calibrating response scales

69

Table 12. American/Japanese scores on Agree/Disagree and Sou Omou/ Sou Omowani English/Japanese Agree/Sou Omou: Completely/Mattaku Definitely/Zettai Strongly/Hijou ni Strongly/Tsuyoku Very much/Ooi ni A lot/Kanari AGREE/SOU OMOU Basically/Kihonteki ni wa Probably/Osoraku sou da to omou Tend to/Dochira ka to ieba Moderately/Aru teido Somewhat/Tashou A little/Yaya Middle/Dochira to mo Ieru: In the middle/Dochira to mo ieru Neither agree nor disagree/ Dochira to mo ienai Can’t Choose/Wakaranai Undecided/Kimerarenai Disagree/Sou Omowanai: A little/Yaya Somewhat/Tashou Moderately/Amari Tend to/Dochira ka to ieba Probably/Osoraku sou de wa nai to omou DISAGREE/SOU OMOWANAI Not agree/Sou Omowanai A lot/Kanari Strongly/Kesshite sou wa omowanai Very much/Zenzen Definitely/Zettai Completely/Mattaku

English

(Means) Japanese

19.4 19.0 18.8 18.8 18.5 17.2 16.1 13.8 13.6 13.5 13.3 12.9 12.1

17.7 18.8 17.7 17.7 17.1 16.4 15.7 14.7 14.4 13.5 13.8 12.8 13.2

10.1 9.9

10.4 10.0

9.8 9.6

8.3 9.2

7.1 6.6 6.4 6.4 6.2

6.2 6.7 6.0 6.4 5.8

3.5 3.5 3.0 1.5 1.4 1.0 0.8

3.7 3.7 3.7 2.7 2.1 1.4 2.1

70

Tom

w. smith et al.

Table 13. American/Japanese scores on Important/Unimportant and Juuyou Dearu/Juuyou De Wa Nai (Means) English Japanese Important/Juuyou Dearu: Extremely/Kiwamete Completely/Mattaku Definitely/Zettai Very/Hijou ni IMPORTANT/JUUYOU DEARU Pretty/Kanari Probably/Dochira ka to ieba Fairly/Aru teido Somewhat/Tashou wa Slightly/Yaya A little bit/Maa Middle/Dochira To Mo Ieru: Neither important nor unimportant/Dochira to mo ienai In between/Dochira to mo ieru Unimportant/Juuyou De Wa Nai: Slightly/Yaya A little bit/Maa Somewhat/Sorehodo Probably/Dochira ka to ieba Fairly/Am ari Not very important/Taishite Very/Sukoshi mo Pretty/Hotondo Not Important/JUUYOU DE WA NAI UNIMPORTANT/JUUYOU DE WA NAI Not at all/Zenzen Definitely/Zettai Completely/Mattaku Extremely/Kesshite

19.4 19.1 18.4 18.3 16.3 15.6

18.1 18.0 18.9 18.4 16.8 17.3

14.0 13.9 13.2 12.0 12.2

13.2 13.9 13.4 13.4 13.6

10.0

9.9

9.5

10.0

8.0 7.9 6.6 6.1 5.8 5.5 5.1 4.7 4.1 3.6 3.0 1.8 1.3 0.8

6.3 5.4 5.9 6.5 5.0 5.0 3.0 2.8 2.5 2.5 1.6 1.3 1.5 2.4

Note: Except as indicated the American means are from the important/unimportant column in Table 2. As Table 2 shows, the English tried two formulations important and unimportant, and important and not important. Japanese does not have two ways of expressing the presence/absence of importance comparable to the English. Slightly different results emerge if the other English scale or the combined scales are used.

methods for assessing and calibrating response scales

71

Using Pilot Results to Formulate Response Scales In an ideal world, a five-point, response scale would mark the lowest value, the exact midpoint, the highest value and two intermediate values with equal intervals between the points. In the case of the this study, the highest agreement value should be somewhat near 20, the intermediate agreement value near 15, the midpoint at 10, the intermediate disagreement value near 5 and the highest disagreement value at 0. Moreover, this should be the case for all countries and cultures (i.e. full agreement would represent in all cultures ‘20’, intermediate agreement ‘15’ and so on). But as these pilot studies indicate, real world response scales are not ideally calibrated in such a way. The studies tested response categories in an out-of-survey context, placing them in the unusual context of modifier comparison. As suggested later, respondents answering real survey questions may well fix the unbalanced scales and ‘transform’ them into well-calibrated scales. But this is something that will require further research. In the meanwhile, the pilot studies suggest what the next steps may be. As one can see from Table 14, response scale terms used in America, Germany, and Japan do not form an ideally balanced scale. The midpoint is the only answer category fitting optimal response scale properties across all three countries. There is a difference in the highest agreement as well as the expected compression of the Japanese scale, for example. On the other hand, one could be satisfied with the results because the end-points in all three countries are at a marked distance from the midpoint. At the same time, the intermediate points raise some questions. The American results indicate that “disagree” (3.5) category is nearer to “strongly disagree” (1.5) than ideal, leaving a substantial gap between “disagree” and “Neither nor . . .” (9.9). Similar observations can be made for Germany and Japan. Thus the pilot studies suggest that respondents in each country rate expressions differently that are assumed at face value to “match” across languages. As an alternative, one could assemble a different set of answer categories from the pilot lists as illustrated in Table 15. The categories presented here spread more equally over the 0–20 possible values. They use the highest and lowest categories in each country (thus reducing the difference between the highest agreement in America and Japan from .9 to only .2). They are closer at the 15 and 5 intermediate points, and use the lowest agreement level available (with still a notable difference between Germany and America on one hand and Japan on the other).

72

Tom

w. smith et al.

Table 14. Means of response categories commonly used in ISSP surveys Item IDs D/US A20/v A16/b A4/p A3/j A5/w A9/e

German Expressions

American Expressions

Japanese Expressions

Stimme voll und ganz zu Stimme zu Stimme weder zu noch lehne ab Lehne ab Lehne stark ab Kann ich nicht sagen

Strongly agree Agree Neither agree nor disagree Disagree Strongly disagree Can’t choose

Hijouni sou omou Sou Omou Dochira tomo ieanai

Mean Mean Mean Germany USA Japan

Sou omowanai Kesshite sou wa omowanai Wakaranai

19,87

18,80

17,70

19,05 9,77

16,00 9,90

15,70 10,00

2,41 1,21

3,50 1,50

3,70 2,70

9,42

9,80

8,30

Table 15. Means of best matched response categories from pilot studies Item IDs D/US A17/h A1/a A4/p A21/o A2/i

German Expressions Stimme bestimmt zu Stimme im Grunde zu Stimme weder zu noch lehne ab Lehne maessig ab Lehne stark ab

American Expressions Definitely agree Basically Agree Neither agree nor disagree Moderately Disagree Definitely disagree

Japanese Expressions

Mean Mean Mean Germany USA Japan

Zettai

19,22

19,00 18,80

Kihonteki ni wa Dochira tomo ieanai

14.93

13.80

14.70

9.77

9.90

10,00

Amari

6.63

6.40

6.00

Zettai/Zenzen/ Mattaku

1.21

1.00

2.10

Whether such a response scale would achieve better measurement than standard response scales would need to be tested. Implications and Future Research First, in general the comparison of American and German (and to a lesser extent American and Japanese) results on the agree/disagree and important/unimportant scales indicate a close, but not perfect, correspondence between the scale terms in general and in particular for terms used in prior ISSP scales (e.g. the five-point, agree/disagree scale). Some scale disparities do exist and the above rating scores could be used to suggest the use of alternative terms in future response scale

methods for assessing and calibrating response scales

73

Table 16. Mean correlations using raw and adjusted response values for Agree/Disagree ISSP questions A. Religion Mean Inter-item Correlations Mean Correlation with Gender Age Years of Education Highest Educational Degree Church Attendance B. Environment Mean Inter-item Correlations Mean Correlation with Gender Age Years of Education Highest Educational Degree Church Attendance

(Pearson’s r) Raw .169

Adjusted .160

.069 .127 .153 .141 .219

.071 .123 .147 .136 .215

Raw .148

Adjusted .130

.088 .094 .181 .177 .076

.072 .089 .162 .156 .073

Source: ISSP, 1991 and 1993.

or the adjustment of past scales according to their position on the underlying continuum. Regarding the latter, attitudinal scales are often used in analysis as if they were interval scales with equal distances between each response. For example, a five-point agree/disagree scale will be used in analysis with the response points assigned values of one to five. But the above analysis indicates that the response points do not have equal intervals between them. For example, the scores on the American five-point agree/disagree scale are 18.8, 16.0, 9.9, 3.5, and 1.5 and the intervals are 2.8, 6.2, 6.4, and 2.0. To estimate the impact of these miscalibrations, 16 agree/disagree items on the 1991 ISSP religion module and 18 agree/disagree items on the 1993 ISSP environment module were inter-correlated with themselves and five demographics (gender, age, years of education, highest educational degree, and frequency of church attendance) using both the raw 1–5 scale and the 18.8–1.5 scale. Overall, there was little difference in the raw or adjusted correlations (Table 16). What impact there is, is for the adjusted correlations to decrease. This may mean that the raw scale scores apply more regularity

74

Tom

w. smith et al.

to attitudes than really prevails so that the adjusted figures show the marginally lower and truer associations. Alternatively, when presented with the terms as a response scale in the context of a survey, people may assign them equal distances and shift from scale-independent evaluations of the response terms to more ordered, scale-dependent assessments. This would mean that the scale adjustment would be less than optimal since respondents had already self-adjusted their responses. Even if the placement of terms in a scale tends to more firmly establish order and distance than when the individual terms are used independently, the utilization of terms that naturally represented the proper interval should facilitate creation of an optimal response scale. Second, these results offer some tentative ideas about what kinds of scales might produce more equivalent, cross-national comparisons. Symmetrical, bipolar scales with an explicit middle point are probably best for cross-national scales. First, people have a very clear understanding of what is the mid-point. It provides people with a third anchor point (in addition to the end points). Second, the division into two sides means that even if sub-categories within the two sides do not match that summing the categories within each side should produce comparable recoded categories. Third, modifiers generally appear to be balanced. For example, strongly agree and strongly disagree have reciprocal values. Of course it is important that bipolar pairs exist in each language. This will not always be the case (Harkness 2003, 2005). Unipolar scales without an explicit mid-point that ask about the amount of some quantity are likely to be more problematic. First, setting aside the translation of specific terms, it would be harder to match categories across languages since on these scales the mid-point is either not clearly defined or subsumed into some broad middle category. Second, the terms used tend to be asymmetrical which makes the matching across languages harder to achieve. Third, research indicates that on unipolar scales people confound terms and position (Klockars and Yamagishi 1988). Without the mid-point clearly defined, people will often assume that the middle category represents the middle even when the term used (e.g. good or bad) is clearly towards the positive or negative end. Fourth, people may not consistently understand what the low end of a pure unipolar scale mean. For example, if people are rating values as from high to low on “conservativism,” does a low conservativism score mean the value is very liberal or merely that it is not conservative and perhaps moderate. Fourth, without a clear mid-

methods for assessing and calibrating response scales

75

point it is possible for unipolar scales to “slide over,” so that categories are unintentionally tilted towards the upper or lower end.11 Finally, more research is needed. It would be extremely useful both to cover more languages and have larger samples. Specific issues that need further study are: 1. How common is it that German terms are stronger (more near the extremes) than corresponding English and what can be done to compensate for this? 2. How common is it that extreme English terms are rated more moderately in Japanese? 3. Would numerical response scales with only the endpoint labeled be more equivalent across languages than labeled scales? How would a scale with only the ends and mid-point labeled perform? 4. Does presenting terms in response scales change the intensity that people associate with them (and the distance between categories)? In particular, to what extent does the use of terms in as response scales lead to them taking on the formal attributes of scale items (e.g. ordered and equally spaced)? 5. When people hold an attitude that lies between two response scale points, how do they decided to choose the category that is higher or lower than their precise position? Do they select the nearest category or are other decision rules used? 6. Do well-constructed response scales reduce respondent burden by reducing respondent effort to understand and “fix” less wellconstructed response scales? 7. Are the positions of terms/categories independent of the substance of the scale (i.e. its subject matter and the nature of public attitudes toward it)? Would people rate “completely agree” about a moderate statement about the economy the same as they would in reference to an extreme statement about religion?

11 On unipolar and bipolar scale in general see Ostrom 1987. A suspect scale is the Eurobarometer life satisfaction scale (“On the whole are you very satisfied, fairly satisfied, not very satisfied, or not at all satisfied with the life you lead?) (European Commission 1996). Year-to-year changes across the European Union are minor and inter-country difference are large and pretty stable. It is suspected that the large intercountry differences are in part due to differences on the intensity of terms used in the scale, variations in translation of the underlying dimension itself, or both.

76

Tom

w. smith et al.

Achieving equivalence in cross-cultural, multiple-language surveys is a challenge. The numerical scaling of response options is one technique that can be used to further this goal. In combination with the optimal, general translation procedures (Harkness 2003; Harkness et al. 2004; Smith 2004), it can notably assist in that endeavor. References Banks, James et al. 2004. “International Comparisons of Work Disability.” Discussion Paper IZA DP No. 1118. Institute for the Study of Labor. Bartram, Peter, and David Yelding. 1973. “The Development of an Empirical Method of Selecting Phrases Used in Verbal Rating Scales: A Report on a Recent Experiment.” Journal of the Market Research Society 15 ( July):151–156. Bradburn, Norman M., and Seymour Sudman. 1979. Improving Interview Method and Questionnaire Design. San Francisco: Jossey-Bass. Bullinger, Monika. 1995. “German Translation and Psychometric Testing of the SF-36 Health Survey: Preliminary Results from the IQOLA Project.” Social Science Medicine 41:1359–1366. Cliff, Norman. 1959. “Adverbs as Multipliers.” Psychological Review 66 ( January):27–44. Clogg, Clifford C. 1982. “Using Association Models in Sociological Research: Some Examples.” American Journal of Sociology 88:114–134. ——. 1984. “Some Statistical Models for Analyzing Why Surveys Disagree.” In Surveying Subjective Phenomena, edited by Charles F. Turner and Elizabeth Martin. Volume 2. New York: Russell Sage. Crespi, Leo P. 1981. “Semantic Guidelines to Better Survey Reportage,” Office of Research, International Communication Agency, Memorandum, August 11. D’Uva, Teresa Bago et al. 2006. “Does Reporting Heterogeneity Bias the Measurement of Health Disparities.” Tingergen Institute Discussion Paper, TI 2006–033/3. Davis, James A. 1993. “[Memorandum to] ISSP Methodology Group,” September. European Commission. 1996. Eurobarometer: Public Opinion in the European Union. No. 45. Brussels: European Commission. Glick, Peter et al. 2004. “Bad but Bold: Ambivalent Attitudes towards Men Predict Gender Inequality in 16 Nations.” Journal of Personality and Social Psychology 86 (May):713–728. Hakel, Milton D. 1968. “How Often is Often?” American Psychologist 23 ( July):533–534. Harkness, Janet. 2003. “Questionnaire Translation.” In Cross-Cultural Survey Methods, edited by Janet A. Harkness, Fons J.R. Van de Vijver, and Peter Philip Mohler. New York: John Wiley & Sons. ——. 2005. Report to the ISSP General Assembly on Behalf of the Translation Group. Mannheim: ZUMA. ——, Beth-Ellen Pennell, and Alisu Schoua-Glusberg. 2004. “Survey Questionnaire Translation and Assessment.” In Methods for Testing and Evaluating Survey Questionnaires, edited by Stanley Presser et al. New York: John Wiley & Sons. ——, Peter Ph. Mohler, Tom W. Smith, and James A. Davis. 1997. Final Report of the Project on ‘Research into the Methodology of Inter Cultural Surveys’ (MINTS). Transcoop Research Reports for ZUMA and NORC. Mannheim: ZUMA. Hougland, James G., Timothy P. Johnson, and James G. Wolf. 1992. “A Fairly Common Ambiguity: Comparing Rating and Approval Measures of Public Opinion.” Sociological Focus 25 (August):257–271. Javeline, Debra. 1999. “Response Effects in Polite Cultures: A Test of Acquiescence in Kazakhstan.” Public Opinion Quarterly 63:1–28.

methods for assessing and calibrating response scales

77

Jones, Lyle V., and L.L. Thurstone. 1955. “The Psychophysics of Semantics: An Experimental Investigation.” Journal of Applied Psychology 39 (February):31–36. King, Gary, Christopher J.L. Murray, Joshua A. Salomon, and Ajay Tandon. 2004. “Enhancing the Validity and Cross-Cultural Comparability of Measurement in Survey Research.” American Political Science Review 98 (February):191–207. Klockars, Alan J., and Midori Yamagishi. 1988. “The Influence of Labels and Positions in Rating Scales.” Journal of Educational Measurement 25 (Summer):85–96. Kumata, Hideya, and Wilbur Schramm. 1956. “A Pilot Study of Cross-Cultural Meaning.” Public Opinion Quarterly 20:574–584. Laumann, Edward O., John H. Gagnon, Robert T. Michael, and Stuart Michaels. 1994. The Social Organization of Sexuality: Sexual Practices in the United States. Chicago: University of Chicago Press. Lichtenstein, Sarah, and J. Robert Newman. 1967. “Empirical Scaling of Common Verbal Phrases Associated with Numerical Probabilities.” Psychon. Sci. 9:563–564. Lodge, Milton. 1981. Magnitude Scaling: Quantitative Measurement of Opinions. Beverly Hills: Sage Publications. ——, and Bernard Tursky. 1979. “Comparisons between Category and Magnitude Scaling of Political Opinion Employing SRC/CPS Items.” American Political Science Review 73:50–66. ——, and Bernard Tursky. 1981. “On the Magnitude Scaling of Political Opinion in Survey Research.” American Journal of Political Science 25 (May):376–419. ——, and Bernard Tursky. 1982. “The Social-Psychological Scaling of Political Opinion.” In Social Attitudes and Psychophysical Measurement, edited by Bernd Wegener. Hillsdale, NJ: Lawrence Erlbaum Associates. ——, David Cross, Bernard Tursky, Joseph Tanenhaus, and Richard Reeder. 1976. “The Psychophysical Scaling of Political Support in the ‘Real World’,” Political Methodology 3:159–182. ——, David V. Cross, Bernard Tursky, and Joseph Tanenhaus. 1975. “The Psychological Scaling and Validation of a Political Support Scale.” American Journal of Political Science 19 (November):611–649. ——, Joseph Tanenhaus, David Cross, Bernard Tursky, Mary Ann Foley, and Hugh Foley. 1976. “The Calibration and Cross-Modal Validation of Ratio Scales of Political Opinion in Survey Research.” Social Science Research 5:325–347. MacKuen, Michael B., and Charles F. Turner. 1984. “The Popularity of Presidents, 1963–1980.” In Surveying Subjective Phenomena, edited by Charles F. Turner and Elizabeth Martin. Volume 2. New York: Russell Sage. Mittelstaedt, Robert A. 1971. “Semantic Properties of Selected Evaluative Adjectives: Other Evidence.” Journal of Marketing Research 8 (May):236–237. Mohler, Peter Ph., Janet Harkness, Tom W. Smith, and James A. Davis. 1996. “Calibrating Response Scales Across Two Languages and Cultures.” Paper presented to the International Sociological Association Conference, Colchester, July. Mosier, Charles. 1941. “A Psychometric Study of Meaning.” Journal of Social Psychology 13 (February):123–140. Myers, James H., and W. Gregory Warner. 1968. “Semantic Properties of Selected Evaluation Adjectives.” Journal of Marketing Research 5 (November):409–412. O’Muircheartaigh, Colm A., George D. Gaskell, and Daniel B. Wright. 1993. “The Impact of Intensifiers.” Public Opinion Quarterly 57 (Winter):552–565. Ommundsen, Reidar, Sven Moerch, Tony Hak, Knud S. Larsen, and Kees Van Der Veer. 2002. “Attitudes towards Illegal Immigration: A Cross-national Methodological Comparison.” Journal of Psychology 136:103–110. Onodera, Noriko. 2002. “Notes on Broadcast Research: Answers Depend on Adverbial Terms Used in Questions—A Review of Expressions of Degree in Choices Used in International Comparative Studies.” The NHK Monthly Report on Broadcast Research ( January):62–75.

78

Tom

w. smith et al.

Orren, Gary R. 1978. “Presidential Popularity Ratings: Another View.” Public Opinion 1 (May/June):35. Osgood, Charles E., George J. Suci, and Percy H. Tannenbaum. 1957. The Measurement of Meaning. Urbana, IL: University of Illinois Press. Osinski, I.C., and A.S. Bruno. 1998. “Response Categories in Likert-scale.” Psicothema 10 (November):623–631. Ostrom, Thomas M. 1984. “RAC Quantifies the Vast Majority.” The Sampler from Response Analyses 31 (March):2. ——. 1987. “Bipolar Survey Items: An Information Processing Perspective.” In Social Information Processing and Survey Methodology, edited by Hans-J. Hippler, Norbert Schwarz, and Seymour Sudman. New York: Springer-Verlag. Salomon, Joshua, Ajay Tandon, and Christopher J.L. Murray. 2004. “Comparability of Self-Rated Health: Cross-Sectional Multi-country Survey Using Anchoring Vignettes.” British Medical Journal 358 ( January 31):258–264. Schaeffer, Nora Cate. 1991. “Hardly Ever or Constantly? Group Comparisons Using Vague Quantifiers.” Public Opinion Quarterly 55 (Fall):395–423. Scheuch, Erwin. 1989. “Theoretical Implications of Comparative Survey Research: Why the Wheel of Cross-Cultural Methodology Keeps on Being Reinvented.” International Journal of Sociology 4 ( June):147–167. Schriesheim, Chester, and Janet Schriesheim. 1974. “Development and Empirical Verification of New Response Categories to Increase the Validity of Multiple Response Alternatives Questionnaires.” Educational and Psychological Measurement 34 (Winter):877–884. Schwarz, Norbert, and Hans-J. Hippler. 1987. “What Response Scales May Tell Your Respondents: Informative Functions of Response Alternatives.” In Social Information Processing and Survey Methodology, edited by Hans-J. Hippler, Norbert Schwarz, and Seymour Sudman. New York: Springer-Verlag. ——. 1995. “The Numeric Values of Rating Scales: A Comparison of their Impact in Mail Surveys and Telephone Interviews.” International Journal of Public Opinion Research 7 (Spring):72–74. Schwarz, Norbert, Baerbel Knaeuper, Hans-J. Hippler, Elisabeth Noelle-Neumann, and Leslie Clark. 1991. “Rating Scales: Numeric Values May Change the Meaning of Scale Labels.” Public Opinion Quarterly 55:570–582. Schwarz, Norbert, Hans-J. Hippler, Brigltte Deutsch, and Fritz Strack. 1985. “Response Scales: Effects of Category Range on Reported Behavior and Comparative Judgments.” Public Opinion Quarterly 49 (Fall):388–395. Sigelman, Lee. 1990. “Answering the 1,000,000-Person Question: The Measurement and Meaning of Presidential Popularity.” Research in Mircropolitics 3:209–226. Simpson, Ray H. 1944. “The Specific Meanings of Certain Terms Indicating Differing Degrees of Frequency.” Quarterly Journal of Speech 30 (October):328–330. Skevington, Suzanne M. 2002. “Advancing Cross-cultural Research on Quality of Life: Observations Drawn from the WHOQOL Development.” Quality of Life Research 11 (March):135–144. ——, and Christine Tucker. 1999. “Designing Response Scales for Cross-cultural Use in Health Care: Data from the Development of the UK WHOQOL.” British Journal of Medical Psychology 72 (March):51–61. Smith, Tom W. 1979. “Happiness: Time Trends, Seasonal Variations, Intersurvey Differences, and Other Mysteries.” Social Psychology Quarterly 42:18–30. ——. 1993. “An Analysis of Response Patterns to the Ten-Point Scalometer.” American Statistical Association 1993 Proceedings. Alexandria, VA: ASA. ——. 1995. “The Holocaust Denial Controversy.” Public Opinion Quarterly 59 (Summer): 269–295. ——. 2002. “Developing Comparable Questions in Cross-National Surveys.” In CrossCultural Survey Methods, edited by Janet Harkness, Fons van de Vijver, and Peter Ph. Mohler. London: Wiley Europe.

methods for assessing and calibrating response scales

79

——. 2004. “Developing and Evaluating in Cross-National Survey Instruments.” In Methods for Testing and Evaluating Survey Questionnaires, edited by Stanley Presser et al. New York: John Wiley & Sons. Spector, Paul E. 1976. “Choosing Response Categories for Summated Rating Scales.” Journal of Applied Psychology 61 ( June):374–375. Strahan, Robert, and Kathleen Carrese Gerbasi. 1973. “Semantic Style Variance in Personality Questionnaires.” Journal of Psychology 85 (September):109–118. Szabo, Silvija, John Orley, Shekhar Saxena, and Alison Harper. 1997. “An Approach to Response Scale Development for Cross-Cultural Questionnaires.” European Psychologist 2 (September):270–276. Traenkle, Ulrich. 1987. “Auswirkungen der Gestaltung der Antworskala suf Quantitative Urteile.” Zeitschrift fuer Sozial Psychologie 18:88–99. Van de Vijver, Fons, and Kwok Leung. 1997. “Methods and Data Analysis of Comparative Research.” In Handbook of Cross-Cultural Personality, edited by J.W. Berry, Y.H. Poortinga, and J. Pandey. Boston: Allyn and Bacon. Vidali, Joseph J. 1975. “Context Effects on Scales Evaluatory Adjective Meaning.” Journal of the Market Research Society 17 ( January):21–25. Wallsten, Thomas S., David V. Budescu, Amnon Rapoport, Rami Zwick, and Barbara Forsyth. 1986. “Measuring the Vague Meanings of Probability Terms.” Journal of Experimental Psychology 115 (December):348–365. Wilcox, Clyde, Lee Sigelman, and Elizabeth Cook. 1989. “Some Like it Hot: Individual Differences in Responses to Group Feeling Thermometers.” Public Opinion Quarterly 53 (Summer):246–257. Wildt, Albert R., and Michael B. Mazis. 1978. “Determinants of Scale Response: Label Versus Position.” Journal of Marketing Research 15 (May):261–267. Worcester, Robert M., and Timothy R. Burns. 1975. “A Statistical Examination of the Relative Precision of Verbal Scales.” Journal of the Market Research Society 17 ( July):181–197. Wright, D.B., G.D. Gaskell, and Colm O’Muircheartaigh. 1995. “How Response Alternatives Affect Different Kinds of Behavioural Frequency Questions.” British Journal of Social Psychology 36 (December):443–456.

80

Tom

w. smith et al.

Appendix: American Questionnaires Sample A: Time: _____________

Section A: Translation 1. If you were to consider your life in general these days, how happy or unhappy would you say you are on the whole . . . Very happy Fairly happy Not very happy Not at all happy

1 2 3 4

2. What is your opinion of the following statement? It is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes. Do you . . . Agree strongly Agree Neither agree nor disagree Disagree Disagree strongly

1 2 3 4 5

CAN’T CHOOSE

8

3. In order to help us write better and more understandable questions, we need to know how people like you use certain words. Here is a scale that goes from 0 to 20. The zero (0) point means you totally and completely disagree with an idea and 20 means you totally and completely agree with an idea. I’m going to read you some terms and I’d like you to tell me what number best represents how much agreement or disagreement the word or phrase means. A. What score between 0 and 20 would you give to . . .

methods for assessing and calibrating response scales

81

HAND CARD Q. 3

First Response Changes a. basically agree _____ _____ SHUFFLE CARDS AND ASK REST OF PHRASES b. agree _____ _____ c. agree a little _____ _____ d. agree a lot _____ _____ e. can’t choose _____ _____ f. completely agree _____ _____ g. completely disagree _____ _____ h. definitely agree _____ _____ i. definitely disagree _____ _____ j. disagree _____ _____ k. disagree a little _____ _____ l. disagree a lot _____ _____ m. in the middle _____ _____ n. moderately agree _____ _____ o. moderately disagree _____ _____ p. neither agree nor disagree _____ _____ q. not agree _____ _____ r. probably agree _____ _____ s. probably disagree _____ _____ t. somewhat agree _____ _____ u. somewhat disagree _____ _____ v. strongly agree _____ _____ w. strongly disagree _____ _____ x. tend to agree _____ _____ y. tend to disagree _____ _____ z. undecided _____ _____ aa. very much agree _____ _____ bb. very much disagree _____ _____ Code 95 = verbatim; 96 = can’t rate term; 98 = Don’t know term; 99 = missing/no answer/unreadable. Code 0.5 for value between the 21 numbered scale points. If range given, code 95 and indicate range specified. B. HAND RESPONSE SHEET WITH ANSWERS RECODED TO RESPONDENT AND SAY: Please look over your answers. If you want to change any of your responses, indicate in the right hand column, the one headed “CHANGES,” what number you now want to give a phrase. 4. Now we’re going to use a similar scale that goes from 0 to 20 to rate some additional phrases. On this scale 0 indicates something of the lowest importance possible, something last and least in importance, and 20 indicates the highest importance possible, something that is first and foremost in importance. As I read you each phrase, please tell me what number best represents how much importance the phrase indicates.

82

Tom

w. smith et al.

HAND CARD Q. A4 SHUFFLE CARDS AND ASK IN THAT ORDER a. Pretty important _____ b. Definitely important _____ c. Not too important _____ d. Extremely important _____ e. Not very important _____ f. Fairly important _____ g. Highly important _____ h. Probably important _____ i. Not at all important _____ j. Exceptionally important _____ k. Not important _____ l. Very important _____ m. Somewhat important _____ n. Important _____ o. Neither important nor unimportant _____ p. Quite important _____ q. Very, very important _____ r. A little bit important _____ s. Slightly important _____ t. Completely important _____ 5. And now consider a similar scale going from 0 to 20. Point 0 indicates that someone is totally and completely against an idea and point 20 means that someone is totally and completely in favor of the idea. I’m going to read you some terms and I’d like you to tell me what number best represents how much someone is either against or in favor of an idea. HAND CARD Q. A5 a. b. c. d. e. f. g.

Slightly against Strongly in favor of Against Strongly against Neither against nor in favor of Slightly in favor of In favor of

_____ _____ _____ _____ _____ _____ _____

GO BACK TO Q. 3, LOOK UP THE RATINGS GIVEN TO TERMS USED IN 6A-H, AND ENTER IN MIDDLE COLUMN, “PREVIOUSLY GIVEN,” BELOW: 6. Now let’s consider again a few of the terms you rated about disagreement and agreement. Here again is the scale that goes from 0 to 20. The zero (0) point means you totally and completely disagree with an idea and 20 means you totally and completely agree with an idea. You gave “basically

methods for assessing and calibrating response scales

83

agree” a score of [mention number given to Q. 3]. Now I want you to think about what is the lowest score that you feel would still represent the phrase “basically agree” and what would be the highest score that would still mean “basically agree,” that is what numbers would represent the range from high to low that would describe where “basically agree” fits on our scale from 0 to 20. First, what would be the lowest number for “basically agree?” And what would be the highest number? REPEAT FOR 6B-H. HAND CARD Q. 6 Lowest a. b. c. d. e. f. g. h.

Basically agree Strongly agree Neither agree nor disagree Disagree Can’t Choose Not agree Strongly Agree Agree

_____ _____ _____ _____ _____ _____ _____ _____

INTERVIEWER: FILL-IN FROM Q. 3 Previously Given _____ _____ _____ _____ _____ _____ _____ _____

Highest _____ _____ _____ _____ _____ _____ _____ _____

NOTE: IF ANY OF THE TERMS WERE NOT RATED 0–20 IN Q. 3, THEN SKIP AND DO NOT ASK IN Q. 6. 7. Now, I’m going to ask you about some of words we’ve just been discussing. What does the word “agree” mean? What does it involve? How about “disagree”? What does it mean or involve? And WHAT does the phrase “neither agree nor disagree” mean? What does it involve? And what about “important”? And how about “unimportant”? A. Agree B. Disagree C. Neither agree nor disagree D. Important E. Unimportant 8. I’m going to read several pairs of words and I would like you to compare and contrast these with the pair “agree/disagree”. I want you to tell me whether you think they mean very much the same as “agree/disagree”, somewhat the same as “agree/disagree”, somewhat different from “agree/disagree” or very much different from “agree/disagree.”

84

Tom

w. smith et al.

First, does the phrase “for/against” mean very much the same as “agree/ disagree”, somewhat the same as “agree/disagree”, somewhat different from “agree/disagree” or very much different from “agree/disagree”? REPEAT FOR 8B-E.

a. b. c. d. e.

Very Much the Same for/against 1 important/ unimportant 1 like/ dislike 1 favor/ oppose 1 positive/ negative 1

Somewhat the Same 2

Somewhat Different 3

Very Much Different 4

DK 8

2

3

4

8

2

3

4

8

2

3

4

8

2

3

4

8

9. What language did you mainly speak at home when you were a child? _______________ 10. What languages do you now speak at home? First Mentioned: _______________ Second Mentioned: _______________ Q. 3

-

0 – Totally and Completely Disagree 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 – Totally and Completely Agree

methods for assessing and calibrating response scales Q. A4

-

0 – Lowest Possible Importance/Last and Least in Importance 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 – Highest Possible Importance/First and Foremost in Importance

Q. A5 -

85

0 – Totally and Completely Against 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 – Totally and Completely in Favor of

86 Q. 6

Tom

-

w. smith et al.

0 – Totally and Completely Disagree 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 – Totally and Completely Agree

Q. 7 Agree/Disagree Very Much the Same Somewhat the Same Somewhat Different Very Much Different

1 2 3 4

Sample B: Time: _____________

Section A: Translation (B) 1. If you were to consider your life in general these days, how happy or unhappy would you say you are on the whole . . . Completely happy Very happy Fairly happy Not very happy Not at all happy

1 2 3 4 5

methods for assessing and calibrating response scales

87

2. What is your opinion of the following statement? It is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes. Do you . . . Completely agree Somewhat agree Neutral Somewhat disagree Completely disagree

1 2 3 4 5

CAN’T CHOOSE

8

3. In order to help us write better and more understandable questions, we need to know how people like you use certain words. Here is a scale that goes from 0 to 20. The zero (0) point means you totally and completely disagree with an idea and 20 means you totally and completely agree with an idea. I’m going to read you some terms and I’d like you to tell me what number best represents how much agreement or disagreement the word or phrase means. A. What score between 0 and 20 would you give to . . . HAND CARD Q. 3

First Response Changes a. basically agree _____ _____ SHUFFLE CARDS AND ASK REST OF PHRASES b. agree _____ _____ c. agree a little _____ _____ d. agree a lot _____ _____ e. can’t choose _____ _____ f. completely agree _____ _____ g. completely disagree _____ _____ h. definitely agree _____ _____ i. definitely disagree _____ _____ j. disagree _____ _____ k. disagree a little _____ _____ l. disagree a lot _____ _____ m. in the middle _____ _____ n. moderately agree _____ _____ o. moderately disagree _____ _____ p. neither agree nor disagree _____ _____ q. not agree _____ _____ r. probably agree _____ _____ s. probably disagree _____ _____ t. somewhat agree _____ _____

88 u. v. w. x. y. z. aa. bb.

Tom

somewhat disagree strongly agree strongly disagree tend to agree tend to disagree undecided very much agree very much disagree

w. smith et al. _____ _____ _____ _____ _____ _____ _____ _____

_____ _____ _____ _____ _____ _____ _____ _____

Code 95 = verbatim; 96 = can’t rate term; 98 = Don’t know term; 99 = missing/no answer/unreadable. Code 0.5 for value between the 21 numbered scale points. If range given, code 95 and indicate range specified. B. HAND RESPONSE SHEET WITH ANSWERS RECODED TO RESPONDENT AND SAY: Please look over your answers. If you want to change any of your responses, indicate in the right hand column, the one headed “CHANGES,” what number you now want to give a phrase. 4. Now we’re going to use a similar scale that goes from 0 to 20 to rate some additional words. On this scale 0 indicates something that is completely and totally unimportant and 20 indicates something that is completely and totally important. As I read you various terms, please tell me what number best represents how much importance the word or phrase means. HAND CARD Q. B4 SHUFFLE CARDS AND ASK IN THAT ORDER a. a. b. c. d. e. f. g. h. i. j. k. l. m. n. o. p. q.

Pretty important Pretty important Definitely important Probably unimportant Extremely important Not very important Fairly important Very unimportant Probably important Not at all important Fairly unimportant Not important Very important Somewhat important Unimportant Completely important Neither important nor unimportant Somewhat unimportant

____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____

methods for assessing and calibrating response scales r. s. t. u. v. w. x. y. z. aa.

Important Pretty unimportant A little bit unimportant Slightly unimportant A little bit important Definitely unimportant Extremely unimportant Slightly important Completely unimportant In between

89

____ ____ ____ ____ ____ ____ ____ ____ ____ ____

5. And now consider a similar scale going from 0 to 20. Point 0 indicates that someone is totally and completely against an idea and point 20 means that someone is totally and completely in favor of the idea. I’m going to read you some terms and I’d like you to tell me what number best represents how much someone is either against or in favor of an idea. HAND CARD Q. B5 a. b. c. d. e. f. g.

Slightly against Against Strongly against Neither against nor in favor of Slightly in favor of In favor of Strongly in favor of

_____ _____ _____ _____ _____ _____ _____

GO BACK TO Q. 3, LOOK UP THE RATINGS GIVEN TO TERMS USED IN 6A-H, AND ENTER IN MIDDLE COLUMN, “PREVIOUSLY GIVEN,” BELOW: 6. Now let’s consider again a few of the terms you rated about disagreement and agreement. Here again is the scale that goes from 0 to 20. The zero (0) point means you totally and completely disagree with an idea and 20 means you totally and completely agree with an idea. You gave “basically agree” a score of [mention number given to Q. 3]. Now I want you to think about what is the lowest score that you feel would still represent the phrase “basically agree” and what would be the highest score that would still mean “basically agree,” that is what numbers would represent the range from high to low that would describe where “basically agree” fits on our scale from 0 to 20. First, what would be the lowest number for “basically agree?” And what would be the highest number?

90

Tom

w. smith et al.

REPEAT FOR 6B-H. HAND CARD Q. 6 Lowest a. b. c. d. e. f. g. h.

Basically agree Strongly agree Neither agree nor disagree Disagree Can’t Choose Not agree Strongly Agree Agree

_____ _____ _____ _____ _____ _____ _____ _____

INTERVIEWER: FILL-IN FROM Q. 3 Previously Given _____ _____ _____ _____ _____ _____ _____ _____

Highest _____ _____ _____ _____ _____ _____ _____ _____

NOTE: IF ANY OF THE TERMS WERE NOT RATED 0–20 IN Q. 3, THEN SKIP AND DO NOT ASK IN Q. 6. 7. Now, I’m going to ask you about some of words we’ve just been discussing. What does the word “agree” mean? What does it involve? How about “disagree”? What does it mean or involve? And WHAT does the phrase “neither agree nor disagree” mean? What does it involve? And what about “important”? And how about “unimportant”? A. Agree B. Disagree C. Neither agree nor disagree D. Important E. Unimportant 8. I’m going to read several pairs of words and I would like you to compare and contrast these with the pair “agree/disagree”. I want you to tell me whether you think they mean very much the same as “agree/disagree”, somewhat the same as “agree/disagree”, somewhat different from “agree/ disagree” or very much different from “agree/disagree.” First, does the phrase “for/against” mean very much the same as “agree/disagree”, somewhat the same as “agree/disagree”, somewhat different from “agree/ disagree” or very much different from “agree/disagree”?

methods for assessing and calibrating response scales

91

REPEAT FOR 8B-E.

a. b. c. d. e.

Very Much the Same for/against 1 important/ unimportant 1 like/ dislike 1 favor/ oppose 1 positive/ negative 1

Somewhat the Same 2

Somewhat Different 3

Very Much Different 4

DK 8

2

3

4

8

2

3

4

8

2

3

4

8

2

3

4

8

9. What language did you mainly speak at home when you were a child? _______________ 10. What languages do you now speak at home? First Mentioned: _______________ Second Mentioned: _______________ Q. 3 -

0 – Totally and Completely Disagree 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 – Totally and Completely Agree

92

Tom

w. smith et al.

Q. B4 -

0 – Completely and Totally Unimportant 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 – Completely and Totally Important

Q. B5 Strongly against Against Slightly against Neither against nor in favor of Slightly in favor of In favor of Strongly in favor of

methods for assessing and calibrating response scales -

0 – Totally and Completely Against 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 – Totally and Completely in Favor of

-

0 – Totally and Completely Disagree 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 – Totally and Completely Agree

Q. 6

93

94

Tom

w. smith et al.

Q. 7 Agree/Disagree Very Much the Same Somewhat the Same Somewhat Different Very Much Different

1 2 3 4

PART TWO

RADICAL SOCIAL CHANGE

THE TRANSITION TO CAPITALISM IN CHINA AND RUSSIA Erich Weede 1. Introduction For 1979, the World Bank (1981:134–135) provided numbers which imply a Soviet per capita product about 16 times as high as the Chinese product. For 1991, when Russia became the (most important) successor state of the USSR, the World Bank (1993:238–239) reported a Russian per capita product about 8.7 times as high as the Chinese product. Since the Chinese economy grew nearly 8% per year in the 1980s, most of the narrowing of the gap—from 16:1 to about 9:1—had been due to Chinese success rather than Soviet failure. If one had used purchase power parity data (World Bank 1993:296–297) instead, then Russia per capita income in 1991 was only 4.1 times as high as Chinese income. But the decline of the Soviet Union was only the beginning.1 According to The Economist (1997:5) and the World Bank (1996:26), after the dissolution of the Soviet Union the Russian economy lost about half of its size until the mid-1990s.2 This implies worse economic losses than those suffered by the Russians during World War II. Simultaneously, China’s economy continued to grow. At the turn of the millennium, in 2000, Russia’s gross national income per capita was about twice as high as China’s (World Bank 2002:232–233). Looking at purchase power parity data hardly affects this conclusion. Thereafter, high prices for Russia’s natural resources, in particular oil and gas, generated some recovery for Russia’s fortune. In 2005, Russian gross national income per capita in dollar terms was 2.6 times as high as Chinese income. For living standards, however, the ratio for per capita incomes in purchase power parity terms, which was only 1.6, might be more relevant 1 Trenin (2002:2) has summarized the impact of the collapse of the USSR in the following terms. Russia kept 50% of its population, 60% of its industry, and 70% of its territory. 2 According to Silverman and Yanowitch (2000:151), in 1998 agricultural production and GDP had fallen to 60% of production in 1991. Industrial production had fallen to 50% in the same period.

98

erich weede

(World Bank 2007:288–289). In this respect at least, China is still catching up with Russia. How could this happen? 2. The Communist Heritage of China and Russia Both countries still suffer the consequences of Communist rule or misrule. Both societies lost millions of people because of the cruelty and incompetence of their rulers which resulted in terrible losses of life, frequently because of starvation. Together, the number of victims of Soviet and Chinese Communism may be on the order of magnitude of a hundred million, although different researchers disagree about whether Soviet or Chinese Communism was more lethal (Courtois et al. 1998; Rummel 1994). If one trusts the results from cross-national analyses of growth rates (Barro and Sala-i-Martin 1995; Levine and Renelt 1992; Sala-i-Martin, Doppelhofer, and Miller 2004; World Bank 1993a:51), then Russia or the Soviet Union and China enjoyed favourable prospects for growth.3 Compared to the developed countries of the West and Japan, both countries were and still are poor and lagging behind, which promises potential advantages of backwardness. They should not yet suffer from diminishing returns to investment. They still have some potential to reallocate labour from agriculture to more productive employment elsewhere. As a matter of fact, of course, both the Soviet Union and China did not permit labour to go where the rewards were highest. Instead mobility was controlled.4 Probably most important of all, they could borrow technologies from more advanced economies and thereby benefit from the fruits of the earlier establishment of economic freedom and capitalism in the West (Weede 2006). Moreover, their poverty protected both of them from post-materialism and an erosion

3 As Easterly (2001:200ff.) has pointed out, presumed determinants of growth (including level of development, investment and human capital) tend to be much more stable than growth rates. Therefore, “luck” or unknown determinants of growth must be important too. 4 In contemporary China the surplus rural workforce might have been as high as 150 million at the turn of the millennium. Up to 100 million people might live and work in places where they should not be, according to official regulations. Only in 2001 was lifting the restrictions at least considered and only for a later date (The Economist 2001b). Russia’s restrictive internal registration system also survived the demise of Communism (The Economist 2001a:8).

the transition to capitalism in china and russia

99

of achievement motivation by affluence (Inglehart 1997; but see also Mehlkop 2000 for criticism). Although the availability and quality of data for socialist countries has always been poor (Winiecki 1988), there is reason to believe that both countries, the USSR as well as China, invested5 a lot, although not always efficiently.6 Compared with nations at the same level of development, both countries did a lot for human capital formation. Investment and human capital formation should have permitted both of them to realize the potential advantages of backwardness. But this happened to an ever lesser degree in the Soviet Union after the 1960s, and it really happened in China only in the era of Deng Xiaoping. Under Mao, the Chinese economy grew more slowly than the global economy (Maddison 1998:15–16, 97). Looking at the second half of the 20th century and comparing growth rates with either the losers of World War II, Germany and Japan, or with the Asian tigers, especially Taiwan or South Korea, the economic performance of none of the socialist great powers looks very impressive. The reasons for the growth deficits of both of them have been known for centuries, or at least decades. According to Adam Smith (1776–1976), there are few incentives to produce rather than to shirk without property. According to Mises (1920), private property in the means of production is a prerequisite of scarcity prices in factor markets and therefore of a rational allocation of resources.7 One may point to a link between state-owned enterprises on the one hand and comparative-advantage-denying strategies of development on the other hand (Lin, Cai and Li 2003). The nationalization of the means of production facilitates value-subtracting production, i.e., turning useful raw materials which can be sold on free markets into unwanted and less valuable goods which can no longer be sold in free markets, but only to consumers in socialist economies who have no choice but to accept

5 Actually, the impact of investment on growth is disputed. According to Blomström, Lipsey and Zejan (1996), there is much better evidence for growth affecting later investment than for earlier investment affecting growth. In recent work investment is increasingly seen as endogenous to growth (Bleany and Nishiyama 2002:44). 6 Only after the dissolution of the Soviet Union did investment in Russia collapse. In the late 1990s investment was only about one-fifth of what it still was in 1990 (Lynch 2001:19). This lack of investment and maintenance in the Yeltsin era had resulted in seriously deteriorating infrastructure. 7 For a modern treatment of similar ideas and an identical conclusion, see Michael Keren (2002).

100

erich weede

shoddy products. According to Hayek (1945), central planning by itself is incompatible with the mobilization of knowledge which is necessarily dispersed over millions of heads. The mobilization and the expansion of knowledge by innovation is dependent on an economy where a large number of actors exert free choices, where they command resources (labour, knowledge, etc.) of their own and where there are incentives to put them to good use. Under central planning, however, knowledge unavailable to the leadership exerts no beneficial impact. More than forty years ago, on the occasion of the ‘great leap forward’, Mao Zedong acted contrary to this Hayekian insight with the most tragic results. Then, comparatively small agricultural collectives encompassing villages sometimes consisting of only a few extended families or clans were combined into so-called people’s communes. Thereby, property rights became ever more attenuated. One’s living standard thereafter depended largely on the effort of a multitude of others whom one did not even know. Nor did most peasants share feelings of solidarity with most other commune members. Worst of all, the leadership of the commune decided what should be done, when, and on which field. Whereas peasants usually know from experience what grows best on which ones of their fields, an education in Marxism or Mao Zedong thought provides a poor substitute. Nowadays, estimates of the numbers of starvation victims exceed 30 million (Fu 1993:235, 304; Kristof and WuDunn 1994:66). A crucial task of all ex-socialist societies remains the re-establishment of respect for property rights. As Olson (2000) recognized, one may regard socialism and nationalized property as a kind of educational program to make people not respect property rights. In societies with private property rights, most of the costs of protecting property are borne by private owners, not by the state. If these private property owners no longer exist, then there is nobody to cover these costs. Securing property rights is much easier, if owners cooperate with the police and the courts. They install locks, they pay private guardians, they provide the police and the authorities with information after a theft or robbery has occurred. Without their assistance, fighting crimes against property becomes much more difficult. Where there is private property, the state can rely on the self-interests of owners to assist it in safeguarding property rights. The situation is quite different in a society with little private property, widespread scarcity, and prices which do not reflect supply and demand.

the transition to capitalism in china and russia

101

There producers and potential customers face strong incentives for illegal exchange. A typical case concerns the illegal sale of public property stolen at the place of work. This benefits the worker and thief. The customer benefits, too, because the stolen goods might not be available legally. Since such exchanges benefit all of those involved, nobody will cooperate with the police. If the police somehow learn about the exchange, there is an incentive to bribe it to buy its silence. Therefore, regulation and central planning boost crime and corruption.8 Although both countries, China and Russia, suffer from this heritage of socialism, the mis-education in not respecting property rights lasted much longer in Russia than in China, especially in rural China where it was essentially overcome after three decades. Moreover, it should be easier to re-establish respect for private property rights in a countryside where people know each other, where reputation is easily established and lost, than in big cities where people do not even know their neighbours. Finally, political instability shortens the time horizon of the authorities and therefore increases the kleptocratic inclinations of the powerful (Heilmann 2000:128). Compared to China after 1979, Russia looked much less stable, at least before the Putin presidency. 3. The Transition to Capitalism in Comparative Perspective Despite some attempts, neither China nor Russia succeeded in establishing the rule of law (see chapters IVb and Xc in Weede 2000; Blankenhagel 2000; Pei 1998, 2006; Zakaria 2003:89ff.; Yavlinsky 1998), which is a prerequisite for safe property rights, economic freedom and incentives to work productively. Compared to Russia, however, China is a fairly safe country. Whereas more than 40% of all surveyed Russians claim to have been victims of crime, less than 20% of all Chinese claim to have suffered from it. Whereas only about 10% of all Russians asked are satisfied with the police, about 80% of all Chinese were so at the beginning of the millennium (Newman 2002:28). Because of being vulnerable to ethnic separatism, Russia under Yeltsin succeeded in continuing neither the tsarist and Communist tradition of strong central authority, nor an orderly and voluntary devolution of administrative controls and tasks to local and regional levels. Only under

8

See the econometric study by Peter Graeff and Guido Mehlkop (2003) for evidence.

102

erich weede

Putin was central authority restored. By contrast, China could afford to shift many economic policy responsibilities from the central state to provinces, counties, cities and townships. According to Montinola, Qian and Weingast (1995), this shift of much decision-making from central to lower levels of authority contributed to the rise of “market-preserving” federalism in China. In my view, the Chinese thereby established a functional equivalent of the rule of law. In contrast to the central authorities, local and regional authorities have to compete more vigorously for the favour of investors, including overseas Chinese investors. Therefore, they have to act as if they desire to respect private property rights and provide some infrastructure. Whoever does worst in these respects, whoever is more corrupt than neighbouring units, whoever engages in more arbitrary and confiscatory taxation than others will drive investors, capital and even qualified workers elsewhere.9 From this perspective, the task of the central government is to guarantee a common market and to prevent the rise of local or regional protectionism. By contrast to China, the post-Communist Russian state has been less capable of containing regional protectionism, at least during the Yeltsin era (Heilmann 2000:210–211). Legally, one may argue that Russia is a federal state, whereas China is not. Administratively, both political systems are similar in consisting of five levels of government. What seems to matter, however, are neither similarities in administrative arrangements, nor legal fictions, but the actual independence of various levels of government from each other.10 Without a reasonable degree of fiscal independence of various layers of government, federalism is unlikely to be market-preserving. As Zhuravskaya (2000:134) has demonstrated, Russian cities enjoy little fiscal independence and face few incentives to improve the local economies because additional local revenues are almost entirely taxed away by regional authorities. Instead, Russian cities waste a lot of resources

9 Although population movements within the People’s Republic of China are still controlled, foreigners, overseas Chinese or Taiwanese enjoy more choices. About a million people from Taiwan live on the Chinese mainland. There even have been a quarter million cross-strait marriages (Ross 2005:82). Such people can discriminate in their choice of workplace and residence in favor of the better governed parts of China. 10 Feld and Voigt (2003) provide a beautiful econometric demonstration of the relevance of ‘de facto’ rule of law and the irrelevance of ‘de jure’ rule of law in accounting for economic performance. According to Blankart (2007), the constitutional ideal is “institutional congruence”, which implies that decision-making happens at the same level of government where benefits and costs occur.

the transition to capitalism in china and russia

103

on subsidizing loss-making enterprises. Whereas Chinese federalism may look weaker than Russian federalism from a legalistic perspective, the former may still be market-preserving, whereas the latter might be “market-hampering”. In Zhuravskaya’s (2000:148) view, “there is quite a lot of evidence that Russian local governments conduct predatory policies toward business, while Chinese governments make efforts to promote entrepreneurial activities in their communities”. In Treisman’s (2000:66) view, revenue sharing between various levels of government in Russia leads to “overgrazing of the common tax and bribe basis”.11 Moreover, differential treatment of regional governments in Russia seems to be politically rather than economically motivated. The main objective has been appeasement of potential troublemakers, especially in ethnic republics (Shleifer and Treisman 2000). By contrast to Russia, the central government of China kept tighter control of senior provincial-level appointments and frequently rotated them from one province to another (World Bank 2002:115). Only under Putin has the primacy of the central government been vigorously reasserted in Russia. Fiscal federalism in China suffers from some weaknesses too (Li 2000). From the late 1970s to the late 1990s, budgetary revenue as a proportion of GDP has been falling. More and more local and provincial revenue has been shifted off-budget. Farmers carry a heavy and frequently arbitrarily imposed burden (Wen 2000). The tax share of the central government had fallen so low that there was a major reform in 1994, which reinforced the role of the central government in tax collection. The fiscal system has not been successful in transferring resources from richer to poorer provinces. The reverse side of this latter shortcoming, however, might be that tax collection in China does not undermine local and regional autonomy. Therefore, it does not weaken

11 According to Easterly (2001:248ff.) “centralized corruption is less damaging than decentralized corruption”. Within Russia corruption has not only grown in magnitude, but under Yeltsin it has become more decentralized, thereby reinforcing the tendency to overgraze the bribe basis. Moreover, the combination of fiscal federalism and electoral pressure in Russia has made regional governments expand public employment beyond fiscal capabilities, resulting in wage arrears, strikes und ultimately bailouts from the central government in the 1990s (Gimpelson and Treisman 2000). Of course, there is a lot of decentralized corruption in China too. Pei (2006:132 and 167) even refers to “local mafia states” and argues: “The combination of lagging political reforms, entrenchment of rent-seeking groups, and decentralization of state predation is a recipe for deteriorating governance.” Although Pei is very good at listing all the weaknesses of China, he is less persuasive in reconciling his criticism with China’s stellar economic performance.

104

erich weede

market-preserving federalism and incentives to work and to make profits too much. In China, fiscal decentralization actually curtailed government size (Zhu and Krug 2007). Both countries suffered from state-asset depletion and self-enrichment by politically well-connected persons (Shleifer and Treisman 2000; Xu 2000). This resulted in declining government revenue as a proportion of GDP, especially at the central level of government. In Russia, insider privatization, the loans-for-shares deal and bank profits from short-term treasury bills are the most important examples of self-enrichment for a few at the expense of the many. In China, asset transfers from stateowned enterprises (SOEs) to semi-public companies, SOE-attached collectives, joint ventures, or even purely private enterprises siphoned off investment capital supplied by the state-owned banks, profitable divisions and trademarks from SOEs. The SOEs were left with money-losing divisions and burdens, like social service obligations and loans, whereas the ‘new’ enterprises could prosper. This process of asset-stripping may be called ‘insider privatization with Chinese characteristics’. In both countries the mode of privatization is likely to have improved the incentives for economic performance (Shleifer and Treisman 2000:33, 38; Xu 2000:87) at the expense of social justice. A potential long-run consequence of this process might be undermining the legitimacy of the transition from socialism to capitalism and the market. Since the Chinese economy works much better than the Russian one, Russia is more likely than China to be at risk sooner rather than later. The reimposition of state control on many previously privatized enterprises in Putin’s Russia is compatible with this view. One of the reasons why China did so much better than the Soviet Union or Russia after 1979 is that responsibility for agriculture was transferred to ever smaller work units until the level of households had been reached. Moreover, the rent contracts of peasants became more extended in time. So, incentives to work and opportunities for the application of knowledge significantly improved in the Chinese countryside, although there is still no private property in land.12 By contrast, Russia had not yet overcome the legacy of collectivization at the beginning of the 21st century. According to Ryback (2000), only

12 Comparing property rights in land across Chinese provinces, the World Bank, (2002:35) concluded “that higher levels of transferability were positively correlated with higher levels of farm investment.”

the transition to capitalism in china and russia

105

6% of the agricultural land was worked by private farmers and another 3% consisted of garden plots at the turn of the century. But these private plots are much more productive than other land. At the end of the 20th century, they accounted for 90% of the potatoes, 75% of the vegetables, and 55% of the meat produced in Russia (Silverman and Yanowitch 2000:157). Since China remained more rural than Russia into the early 21st century, agricultural reforms by themselves could never have had as beneficial an impact in Russia as they had in China in the early 1980s. Moreover the Chinese economy was never as centralized as the Russian one was under Soviet rule. Heilmann (2000:60) estimates that the state commanded about 90% of the Soviet economy, whereas the Chinese government commanded a mere 20% after the de facto privatization of agriculture. Since the 1950s there have always been waves of centralization and decentralization in China (Qian 2000). During periods of decentralization, village, township, county and provincial administrations got valuable experience in running local or regional economies. This experience was put to good use in later township village enterprises (TVEs).13 Their most important characteristic is that they have to compete with each other. In spite of collective ownership they have to behave as if they were capitalist enterprises. Actually, some of them have been quite close to private enterprises which preferred to take cover under some collective. Moreover, there are rising numbers of truly private enterprises in China, some of which are owned by overseas Chinese. Admittedly, the government wasted a lot of time instead of privatizing state-owned enterprises. Although many of them were bankrupt according to Western standards and nevertheless continued to operate for years, even though China could not afford to subsidize them endlessly (Lardy 1998; Weede 2000, chapter IVb), growth elsewhere at least reduced the economic weight of state-owned enterprises.14 13 According to Che and Qian (1998:490–491), TVEs are a second best solution to the ownership problem under insecure property rights. In their view “local government-owned enterprises have more secure property rights than private enterprises have because the national government expects them to better serve its interests . . . they provide more revenues to the national government; and they also spend more on local public goods . . .” 14 According to The Economist (2000:93–97), SOEs constituted about 28% of the Chinese economy at the turn of the century, but accounted for 44% of urban employment, 70% of government revenue and 80% of bank loans. From 1998 to 2000 SOEs released about 21 million workers (Taube 2001:135), most of whom did not benefit from a social security net. According to Pei (2006:2–3), the share of SOEs in industrial output fell

106

erich weede

Whereas Russian small enterprises are weak, Chinese small enterprises are comparatively strong. Why this difference is important has been persuasively argued by Pejovich (2001:28): “The small enterprises are the breeding ground for entrepreneurs, a work ethic, and a capitalist exchange culture. They educate ordinary people to appreciate a way of life that rewards performance, promotes individual liberties, and places high value on self-responsibility and self-determination”. Because of its head-start in nurturing small enterprises, capitalism stands a chance of growing much deeper roots in China than in Russia. Another reason for the differential success of China and Russia is the differential degree of export orientation of both economies. Econometric studies (Bleaney and Nishiyama 2002; Dollar 1992; Edwards 1998; Greenaway and Nam 1988; World Bank 1993a) demonstrate that open or export-oriented economies do grow more rapidly than others. Whereas China integrated itself into the global economy, the Soviet Union and its Russian successor largely neglected exports and the benefits to be derived therefrom. According to Taylor and Jodice (1983:226–228), exports and imports constituted little more than 10% of the Soviet economy during the 1960s and 1970s. A 6% export ratio for China in 1979 (World Bank 1981:142) demonstrates much similarity between the Soviet Union and pre-reform China. By 1991 the Chinese export ratio had leapt to 20% (World Bank 1993:254). Because of the dissolution of the Soviet Union and the resulting disorder, we don’t know the comparable number for Russia. In the long run, however, the dissolution of the USSR should have increased the importance of foreign trade for the Russian economy. By and large, economic size and trade dependence are negatively correlated. Whereas China achieved export growth of 13% in the 1990s, the Russians achieved a meagre 2.3% (World Bank 2000–2001:294–295).15 Because of Hong Kong’s entrepot role, Chinese numbers are more likely to be understated than overstated. Whereas Russians sell commodities, oil and gas, the Chinese

from 78 to 41% from 1978 to 2002, whereas the share of the private sector (including foreign invested enterprises) rose from 0.2 to 41% of industrial output. 15 According to Shleifer and Treisman (2000:103), foreign trade dependence helps to reduce corruption. Obviously, the civilizing impact of trade must be weaker in Russia than in China. As Colombatto (2001) has argued, Western advice and Western policies have not been helpful in promoting a free and open Russian economy.

the transition to capitalism in china and russia

107

sell labour-intensive products.16 As Sachs and Warner (1995) or Bleaney and Nishiyama (2002) have demonstrated, high ratios of natural resource exports to GDP are generally associated with low growth rates. Because of currently high prices of oil and other natural resources, Russian exports in 2005 were about one-third of the value of Chinese exports, or a quarter if one includes Hong Kong’s exports with China’s. Given China’s huge preponderance in population size, these numbers look fairly good for Russia. But export quality remained what it was. Manufactured exports constituted 91% of Chinese exports, but merely 21% of Russian exports. The gap in high technology exports is also large and in favour of China. Its share in Chinese exports was 30% in China, but 9% in Russia (World Bank 2007:296–297). For China globalization worked. According to The Economist (2007:5) “with a trade-to-GDP ratio around 70% and a sea of foreign investment, China is one of the world’s most open economies”. In the first decade of the 21st century China had already overtaken Japan as an exporter and remained behind only Germany and the United States (The Economist 2007:8). In 2008 China is likely to become the biggest exporter in the world. Differences in economic openness also manifest themselves in differential attractiveness for foreign investment. In 1998, China received about 43 billion US dollars, whereas Russia got less than 3 billion (World Bank 2000–2001:314–315). According to The Economist (2001a:5), the gap was similar in 2000. In 2004, however, Russia’s attractiveness to foreign capital looked much better. The Russians received nearly a quarter of what Mainland China got, or about one-seventh, if one includes Hong Kong with China (World Bank 2007:296–297). As foreign direct investment tends to be more productive than domestic investment, as foreign investment does not crowd out domestic investment (de Soysa and Oneal 1999), as foreign takeovers of enterprises frequently lead to performance improvements in transition economies 16 Rodrik (2006) notes that China’s export basket is more sophisticated than one would expect from its level of economic development. Consumer electronics and auto parts are examples of such exports. Although no government knows in advance where investment in non-traditional exports is likely to succeed, experimentation with subsidies or (in the case of auto parts) local content requirements may help to upgrade a country’s export basket and thereby improve its growth rate. In Rodrik’s (2006) account, this is what China succeeded in. From a political economy perspective, the main requirement for this type of industrial policy is to stop subsidizing losers. Picking winners only is impossible, but giving up losers is not. Although Rodrik does not even raise the question, one may speculate about the differential feasibility of giving up losers in autocracies and democracies.

108

erich weede

(World Bank 2002:66), the long-standing preference of foreign capital owners for China over Russia looks like a self-fulfilling prophecy. Moreover, Chinese and Russian capitalists feel like foreigners toward their home countries. Capital flight in Russia exceeded inward investment for some time.17 In China it is the other way round. Much of Chinese flight capital re-enters under the foreign capital label via the backdoor (Heilmann 2000:235). Although Russia benefited from high and rising prices of oil during the first decade of the 21st century, which stabilized the currency and boosted economic growth, the middle-run prospects of Russia look much poorer than the prospects of China. In the early 1990s, Russia succeeded in privatizing a lot of its state-owned enterprises (Aslund 1995:223ff.; Layard and Parker 1996:125), but the preferential treatment of insiders in assigning private property rights, the dubious links between the new tycoons and the authorities, corruption and the absence of the rule of law together prevented the establishment of Russian industries (except for the extraction of natural resources) which can compete in global markets. Foreign investors acquired little property and control in Russia. But they might have contributed to healthier government finances in the 1990s in Russia, to more efficient management and to access to foreign technology and markets (Heilmann 2000:176–177). Therefore, the story of Russian privatization left Russian industry with serious handicaps.18 It cannot balance the Russian disadvantages compared to China—of a still largely collectivized agriculture, of establishing too few young enterprises, of inward orientation of its civilian manufacturing industries and too little foreign investment in them, of especially weak property rights. It is likely that China will build on its advantages and that Russia will fall further and further behind China. Russia does not owe this modest position to a poor endowment with natural resources—quite to the contrary—but to institutional deficits

17 The Economist (2001a:5) reported a monthly capital flight of $1.5 billion to be compared with yearly foreign direct investment of $2.7 billion. Russian capital flight is probably related to its frequently illegitimate acquisition. In 1998, 65% of the respondents in a survey favored confiscation of illegally acquired wealth (Silverman and Yanowitch 2000:151). Under such circumstances, capital flight looks rational. 18 Shleifer and Treisman (2000) argue that it was impossible to privatize Russian industry more efficiently. In their view, stakeholders or veto players had to be bought off. Insider privatization was a way of substituting an inefficient economic arrangement for an even less efficient prior arrangement. They cite restructuring and better productivity after privatization as support for their view.

the transition to capitalism in china and russia

109

and policy errors.19 One should not conclude therefrom that Russia could easily change and become as successful as China has been since the 1980s. Russia has neither established the rule of law, nor decentralized and developed market-preserving federalism, nor established private property rights in agricultural land, nor even started the route to export-oriented development. Worse still, political obstacles arising out of Russia’s ethnic heterogeneity might have prevented even a most enlightened Russian government from enacting more efficient reforms (see Shleifer and Treisman 2000). Except for the flat income tax of 13% which resulted in an 80% increase of government revenue within a year after its introduction, too many Russian reforms remained mere facades (Siegl 2001; World Bank 2002:64). Putin restored the Russian tradition of administration by subordination without recognizing its cost. According to Shevtsova (2003:97), “the regime of subordination went against Putin’s goal of building an efficient market economy, which demands freedoms and initiative”. Although both countries have made great progress on the road to economic freedom, albeit from a dismal starting point, China always scored better than Russia on the Fraser Index of economic freedom in the late 20th century. Only in 2004 did Russia at last come close to China’s score (Gwartney, Lawson, with Easterly 2006:19–21).20 The comparative performance of Russia and China fits with the view that economic freedom is productive (Doucouliagos and Ulubasoglu 2006; Gwartney et al. 2006; Liu 2007; Weede 2006). The Russians quickly established only a single trait of capitalist economies, i.e., inequality. In Russia, the top 10% received about 39% of the income—in the United States and China it was between 30% and 31%—at the end of the 20th century. In Russia the top 20% obtained nearly 54% of the income; in China and the United States it was between 46% and 47% (World Bank 2000–2001:282–283). 19 It has been argued (Lal 1998:3; de Soysa 2000), however, that rich resource endowments might reinforce predatory behavior and thereby make institutional and economic development more difficult. According to Ross (1999:297), “there is now strong evidence that states with abundant resource wealth perform less well than their resource poor counterparts, but there is little agreement on why this occurs”. For further evidence on the negative effects of the resource curse, see Weiffen (2004) and Collier (2007). 20 O’Driscoll, Holmes and Kirkpatrick (2001), however, disagree. In their view, economic freedom in China and Russia had been quite similar in the late 1990s, but Russia pulled ahead of China in 2000. For 2007, however, Kane, Holmes, and O’Grady (2007) place China and Russia very close to each other on their economic freedom scale.

110

erich weede

Admittedly, China is one of those globalizing economies where the size distribution of income deteriorated in the 1980s and 1990s. Because of strong economic performance, incomes of the poorest quintile in China nevertheless improved by 3.8% per year (Dollar and Kraay 2001:40). But the Chinese distribution of income has become more unequal than the Russian distribution according to the most recent data (World Bank 2006:280–281). Although inequality of income is essential in order to elicit effort and hard work and in order to guide producers to doing what consumers actually want, not all observable income inequality everywhere is justifiable in this way. Corruption, theft, robbery and rent-seeking may also produce income inequality. Most Russians do not explain inequality in their country by differences in effort, merit or hard work. Instead they believe that “swindlers and manipulators” stand the best chance to become rich in contemporary Russia (Silverman and Yanowitch 2000:33). The trademark of capitalism is not inequality of income, but competition among producers and satisfaction of the wants of consumers. In these respects, Russia is lagging. Male life expectancy provides another cue that Chinese society works better than Russian society. In poorer China it was about 63 years; in comparatively richer Russia it was only 58 years in the mid-1990s (Heilmann 2000:16). In 2004, life expectancy at birth was 70 for males and 73 for females in China, but still merely 59 for males and 72 for females in Russia (World Bank 2007:288–289). Russia might lose about 750,000 people per year, or up to 22 million in the first fifteen years of the 21st century (Legvold 2001:63). Establishing good government has always been easier in China than in Russia, because Chinese society is ethnically more homogeneous than Russian society. In addition to ethnic heterogeneity, there might be a second obstacle to good government and growth-promoting policies, i.e., income inequality and a potential for distributional struggles (Easterly 2001, chapter 13). When Communism was still practiced in Russia and China, both countries were fairly egalitarian at low levels of prosperity. Economic reforms and the transition to capitalism increased income inequality in both societies. Russia faces two obstacles to stable, effective and growth-promoting government, i.e., ethnic heterogeneity and inequality, and China merely one of them. Whereas China—like Japan before it (Weede 2004)—has succeeded in making nationalism an engine of growth, Russian nationalism always focused on the acquisition of territory and great power rhetoric (Trenin 2002), but never put the

the transition to capitalism in china and russia

111

idea of becoming a rich country and a global economic powerhouse high on its political agenda. Greenfeld (2001:218) argues that nationalism and “the view of the economy as a battlefield in the struggle for national supremacy” provides much of the motivation for economic growth. As in Japan more than one hundred years ago, nationalism in China might succeed in legitimizing entrepreneurship, private property rights, and capitalism and thereby overcome the traditional contempt and lack of respect from which merchants suffered in the Confucian societies of East Asia. 4. Conclusion In comparative terms, the Chinese transition to capitalism has been a success, whereas the Russian transition so far would have been a failure, had it not been rescued by high and rising energy prices during the Putin presidency. China enjoyed a head-start of about a dozen years. Whereas the Russian economy suffered negative growth rates at the beginning of this transition, China grew vigorously and persistently since the transition began. The size of the Chinese economy grew eightfold; per capita incomes sevenfold (Pei 2006:2). Although the protection of private property rights, rule of law, and economic freedom leave much to be desired in both countries, China overcame the legacy of socialism sooner and to a greater degree than Russia. Only China, but not yet Russia, benefits from the existence and competition of a multitude of small and dynamic enterprises. Whereas China seems to have established market-preserving federalism, Russia suffers from market-hampering federalism. Moreover, China has opened its doors more vigorously for foreign trade and investment than Russia did. The one head-start of Russia over China, i.e., early privatization of state-owned enterprises, did not result in beneficial consequences, because the preferential treatment of insiders and the weakness of the rule of law in Russia neutralized much of this potential Russian advantage. Moreover, the reassertion of state ownership under Putin has squandered some of the impressive productivity gains achieved by the previous, but transient privatization (Aron 2006). If Chinese administrative and economic policies were more efficient than Russia’s, this raises the question whether regime differences might account for divergent economic outcomes. To put it simply, in Russia democratization lead to capitalist reforms, whereas in China we have

112

erich weede

seen vigorous economic reforms under the autocratic guidance of a nominally still Communist party. Does this establish an inherent economic superiority of autocracies like China over more democratic, albeit still illiberal regimes like Russia? The economic history of China itself tells a different story. An autocracy is capable of getting its economic decisions disastrously wrong, as the Chinese did under Mao Zedong during the great leap forward resulting in mass starvation. It is also capable of promoting an economic miracle, as the Chinese did under Deng Xiaoping’s leadership and his successors. Less autocratic leadership is less likely to result in either extreme. As has been argued elsewhere (Weede 1996; Quinn and Wooley 2001), autocracies differ from democracies not in their average performance, but in their variation of performance. After some decades of very poor performance under autocratic policy-makers the Chinese benefited from a string of better results recently. Neither China nor Russia have yet established limited government and the rule of law. Whoever does so makes enlightened leadership (and luck in getting it) less important than they still are in most transitional societies. Like Russia, China is destined to gray without having become a rich country. Although both countries are aging the hard way (Eberstadt 2006), this may well be worse for a resource-poor country like China, which has to work its way towards prosperity, than for Russia with its abundant natural resources. By 2020, China’s population will stagnate. By 2015, 120 million Chinese, or about 9% of the population, will be older than 65 (Eberstadt 1998, 2006; England 2005:17). Since the mandatory retirement age for male employees is still 60 (for females it is 55), this lower age threshold might be more relevant. In 1990, about 9% of the Chinese population was over age 60; in 2030 it is likely to be about 22%, i.e., on the order of magnitude of 300 million people (Williamson and Shen 2004:3). Another decade later the percentage might be 26 and the number of old people about 400 million (England 2005:23). Then China’s share of old people might be higher than the Chinese share of the global population. Currently, the Chinese state has promised pension benefits to only about 10% of these hundreds of millions of old people. Whether paid for privately by their children or publicly by the state, support for the elderly will soon become a significant burden on the Chinese economy.21 Although China cannot 21 This number is only a crude guess. It has been ‘confirmed’ at the International Institute for Sociology Conference in Beijing in July 2004 by John B. Williamson who

the transition to capitalism in china and russia

113

continue to grow by increasing inputs, whether capital or labour, as it did so massively and successfully in the past, the size of its labour force will continue to grow for another twenty years and then start to decline gently. Moreover, internal migration from the countryside to the cities will contribute to the avoidance of a European-type labour shortage and contribute to growth (England 2005:118). Nevertheless, given the already high level, major increases in capital investment are inconceivable. The workers for increasing the input of labour simply will no longer be available. Worse still, on top of the aging problem China faces a major imbalance between men and women. Since there might be 16 to 20% more men than women, about every sixth Chinese man will not be able to find a wife (Eberstadt 1998:63; Poston 2004). Small and rich countries might close this gap by inviting in foreign women. The world’s most populous country, and still a comparatively poor one, cannot solve its home-grown problems in this way. Conceivably, China’s gender imbalance could even contribute to future political instability (Hudson and Boer 2002; Poston 2004). It has been demonstrated in a cross-national econometric study (Bloom and Williamson 1998) that economic growth rates are affected by differential growth rates of dependent and working-age populations. In the future, China’s growth rate will be reduced by the stronger growth rate of its dependent population, whereas Russia’s growth rate might remain dependent on the prices of its natural resources, especially its oil and gas exports. It is conceivable that exports of natural resources could keep a greying Russia solvent. According to West (2005:125), “oil and gas revenues and taxes are as much as 50 percent of government revenues, generate most of the country’s foreign exchange and subsidize domestic industry and agriculture”. It is also conceivable that the reassertion of state control under Putin could increase wastefulness, reduce the quality of management, decrease investment and prevent a massive flow of Western capital and technology to the Russian energy sector (Aron 2006; West 2005). Certainly, an aging Russia is not prepared for a sharp drop in oil and gas prices. The faster China’s has done research on old-age security in China together with Chinese social scientists. Most rural Chinese cannot look forward to public promises for their old age. Since about one-third of them do not have sons who are traditionally responsible for supporting their parents, their prospects are bleak. If they have to work during old age, they suffer the consequences of little formal schooling and no work experience beyond the low-income agricultural sector (Eberstadt 2006).

114

erich weede

economy grows, the higher China’s demand for natural resources and foreign commodities, the better the Russian terms of trade and Russian growth prospects might become.22 Concerning institutional and political development, however, Russia’s ‘resource curse’ (Collier 2007; Ross 1999; Weiffen 2004) generates more problems for Russia’s future than for China’s future.23 References Aron, Leon. 2006. “What Does Putin Want?” Commentary 122(5):19–24. Aslund, Anders. 1995. How Russia Became a Market Economy. Washington, DC: Brookings. Barro, Robert J., and Xavier Sala-I-Martin. 1995. Economic Growth. New York: McGraw-Hill. Blankart, Charles B. 2007. Föderalismus in Deutschland und Europa. Baden-Baden: Nomos. Blankenhagel, Alexander. 2000. “Legal Reforms in Russia.” Journal of Institutional and Theoretical Economics 156(1):99–119. Bleany, Michael, and Akira Nishiyama. 2002. “Explaining Growth: A Contest Between Models.” Journal of Economic Growth 7(1):43–56. Blomström, Magnus, Robert E. Lipsey, and Mario Zejan. 1996. “Is Fixed Investment the Key to Growth?” Quarterly Journal of Economics 111(1):269–276. Bloom, David E., and Jeffrey G. Williamson. 1998. “Demographic Transitions and Economic Miracles in Emerging Asia.” The World Bank Economic Review 12(3):419–455. Che Jiahua, and Yingyi Qian. 1998. “Insecure Property Rights and Government Ownership of Firms.” Quarterly Journal of Economics 113:467–496. Collier, Paul. 2007. The Bottom Billion. Why the Poorest Countries Are Failing and What Can Be Done About It. Oxford: Oxford University Press. Colombatto, Enrico. 2001. “Was Transition About Free-Market Economics?” Journal des Economistes et des Etudes Humaines XI(1):63–76. Courtois, Stephane. 1998. “Die Verbrechen des Kommunismus.” Pp. 11–43 in Das Schwarzbuch des Kommunismus, by Stephane Courtois et al. München: Piper. De Soysa, Indra. 2000. “The Resource Curse: Are Civil Wars Driven by Rapacity or Paucity?” Pp. 113–135 in Greed and Grievance: Economic Agendas in Civil Wars, edited by Mats Berdal and David M. Malone. Boulder, CO: Lynne Rienner. ——, and John R. Oneal (1999). “Boon or Bane? Reassessing the Productivity of Foreign Direct Investment.” American Sociological Review 64(5):766–782.

22 For a similar argument on the relationship between China’s growth and export success on the one hand and deteriorating terms of trade on the other hand, see The Economist (2007:10). 23 Environmental problems have not been considered in this paper. One may argue that air and water pollution—and water shortages in the North—pose a serious threat to China’s future (Harding 2007). Since Russia supports a much smaller population on a bigger territory than China, the consequences of environmental degradation promise to be more severe in China than in Russia.

the transition to capitalism in china and russia

115

Dollar, David. 1992. “Outward Oriented Developing Economies Really Do Grow More Rapidly.” Economic Development and Cultural Change 40(3):523–544. ——, and Aart Kraay. 2001. Trade, Growth, and Poverty. Washington, DC: World Bank (Working Paper). Doucouliagos, Chris, and Mehmet Ali Ulubasoglu. 2006. “Economic Freedom and Economic Growth.” European Journal of Political Economy 22(1):60–81. Easterly, William. 2001. The Elusive Quest for Growth: Economists’ Adventures and Misadventures in the Tropics. Cambridge, MA: MIT Press. Eberstadt, Nicholas. 1998. “Asia Tomorrow, Gray and Male.” The National Interest 53:56–65. ——. 2006. “Growing Old the Hard Way: China, Russia, India.” Policy Review 136:15–39. Economist, The. 1997. “Survey: Russia.” The Economist 344, 8025 ( July 12th). ——. 2000. “China’s State Owned Enterprises.” The Economist 356, 8190 (September 30th):93–97. ——. 2001a. “Survey: Russia. Putin’s Choice.” The Economist 360, 8231 ( July 21st). ——. 2001b. “Mobility in China. Off to the City.” The Economist 360, 8237 (September 1st):48. ——. 2007. “Reaching for a renaissance. A special report on China and its region.” The Economist 382, 8522 (March 31st). Edwards, Sebastian. 1998. “Openness, Productivity and Growth: What Do We Know?” Economic Journal 108:383–398. England, Robert Stowe. 2005. Aging China. The Demographic Challenge to China’s Economic Prospects. Westport, CT: Praeger (for the Center for Strategic an International Studies, Washington, DC). Feld, Lars P., and Stefan Voigt. 2003. “Economic Growth and Judicial Independence: Cross-Country Evidence Using a New Set of Indicators.” European Journal of Political Economy 19(3):497–527. Fu Zhengyuan. 1993. Autocratic Tradition and Chinese Politics. Cambridge: Cambridge University Press. Gimpelson, Vladimir, and Daniel Treisman. 2002. “Fiscal Games and Public Employment: A Theory with Evidence from Russia.” World Politics 54(2):145–183. Graeff, Peter, and Guido Mehlkop. 2003. “The Impact of Economic Freedom on Corruption: Different Patterns for Rich and Poor Countries.” European Journal of Political Economy 19(3):605–620. Greenaway, David, and Chong Hyun Nam. 1988. “Industrialization and Macroeconomic Performance in Developing Countries under Alternative Trade Strategies.” Kyklos 41:419–435. Greenfeld, Liah. 2001. The Spirit of Capitalism. Nationalism and Economic Growth. Cambridge, MA: Harvard University Press. Gwartney, James D., Randall D. Holcombe, and Robert A. Lawson. 2006. Institutions and the Impact of Investment on Growth.” Kyklos 59:255–273. Gwartney, James, Robert Lawson, with William Easterly. 2006. Economic Freedom of the World. Annual Report 2006. Vancouver, BC: Fraser Institute, and Potsdam: Liberales Institut. Harding, Harry. 2007. “Think Again. China.” Foreign Policy 159:26–32. Hayek, Friedrich August Von. 1945. “The Use of Knowledge in Society.” American Economic Review 35(4):519–530. Heilmann, Sebastian. 2000. Die Politik der Wirtschaftsreformen in China und Rußland. Hamburg: Mitteilungen des Instituts für Asienkunde, Band 317. Hudson, Valerie M., and Andrea Den Boer. 2002. “A Surplus of Men, A Deficit of Peace: Security and Sex Ratios in Asia’s Largest States.” International Security 26(4):5–38. Inglehart, Ronald. 1997. Modernization and Postmodernization. Cultural, Economic and Political Change in 43 Societies. Princeton: Princeton University Press.

116

erich weede

Kane, Tim, Kim R. Holmes, and Mary Anastasia O’Grady. 2007. 2007 Index of Economic Freedom. Washington, DC: Heritage Foundation, and New York: Wall Street Journal. Keren, Michael. 2002. “Socialism and Stalinism: Never the Twain Shall Part? Or Why Can’t we have Liberal Socialism?” Paper presented at the European Public Choice Meeting, Belgirate (Italy), April 4–7. Kristof, Nicholas D., and Sheryl Wudunn. 1994. China Wakes. New York: Random House. Lal, Deepak. 1998. Unintended Consequences: The Impact of Factor Endowments, Culture, and Politics on Long-Run Economic Performance. Cambridge, MA: MIT Press. Lardy, Nicholas R. 1998. China’s Unfinished Economic Revolution. Washington, DC: Brookings. Layard, Richard, and John Parker. 1996. The Coming Russian Boom. New York: Free Press. Legvold, Robert. 2001. “Russia’s Unreformed Foreign Policy.” Foreign Affairs 80(5):62–75. Levine, Ross, and David Renelt. 1992. “A Sensitivity Analysis of Cross-Country Growth Regressions.” American Economic Review 82:942–963. Li Shi. 2000. “Efficiency and Redistribution in China’s Revenue Sharing System.” Pp. 103–122 in Governance, Decentralization and Reform in China, India and Russia, edited by Jean-Jacques Dethier. Dordrecht and Boston: Kluwer Academic Publishers. Lin Justin Yifu, Fang Cai, and Zhou Li. 2003. The China Miracle: Development Strategy and Economic Reform. Hong Kong: The Chinese University Press. Liu Lirong. 2007. Wirtschaftliche Freiheit und Wachstum. Münster: Lit-Verlag. Lynch, Allen C. 2001. “Einen Schritt vor, zwei Schritte zurück. Wurzeln des wirtschaftlichen Dilemmas in Ruland.” Internationale Politik 56(10):17–26. Maddison, Angus. 1998. Chinese Economic Performance in the Long-Run. Paris: OECD. Mehlkop, Guido. 2000. “Methodische Probleme bei der Analyse von Wertvorstellungen und Wirtschaftswachstum.” Zeitschrift für Soziologie 29(3):217–226. Mises, Ludwig Von. 1920. “Die Wirtschaftsrechnung im sozialistischen Gemeinwesen.” Archiv für Sozialwissenschaft und Sozialpolitik 47(1):86–121. Montinoala, Gabriella, Yingyi Qian, and Barry Weingast. 195. “Federalism Chinese Style. The Political Basis for Economic Success in China.” World Politics 48(1):50–81. Newman, Graeme. 2002. “Crimes and Punishments.” Foreign Policy 127:28–29. O’Driscoll, Gerald P., Kim R. Holmes. and Melanie Kirkpatrick. 2001. Index of Economic Freedom. New York: Wall Street Journal, and Washington, DC: Heritage Foundation. Olson, Mancur. 2000. Power and Prosperity: Outgrowing Communist and Capitalist Dictatorships. New York: Basic Books. Pei Minxin. 1998. “Is China Democratizing?” Foreign Affairs 77(1):68–82. ——. 2006. China’s Trapped Transition. The Limits of Developmental Autocracy. Cambridge, MA: Harvard University Press. Pejovich, Svetozar. 2001. “After Socialism. Where Hope for Individual Liberty Lies.” Journal des Economistes et des Etudes Humaines XI(1):9–30. Poston, Dudley L. 2004. “The Demographic Destiny of China, South Korea and Taiwan: Changes and Implications for the Family.” Paper presented at the 36th World Congress of the International Institute of Sociology, Beijing, July 7–11. Qian Yingyi. 2000. “The Process of China’s Market Transition (1978–1998).” Journal of Theoretical and Institutional Economics 156(1):151–171. Quinn, Dennis P., and John T. Wooley. 2001. “Democracy and National Economic Performance: The Preference for Stability.” American Journal of Political Science 45(3):634–657. Rodrik, Dani. 2006. What is so Special About China’s Exports? Cambridge, MA: Harvard University, Kennedy School of Government, Faculty Research Paper RWP 06–001.

the transition to capitalism in china and russia

117

Ross, Michael L. 1999. “The Political Economy of the Resource Curse.” World Politics 51(2):297–322. Ross, Robert S. 2005. “Assessing the China Threat.” The National Interest 81:81–87. Rummel, Rudolph J. 1994. Death by Government. New Brunswick, NJ: Transaction. Ryback, Andrzej. 2000. “Putin steuert bei der Bodenreform Zick-Zack-Kurs.” Financial Times Deutschland, March 20:14. Sachs, Jeffrey D., and Andrew M. Warner. 1995. “Natural Resource Abundance and Economic Growth.” Cambridge, MA: NBER Working Paper W 5398. Sala-I-Martin, Xavier, Gernot Doppelhofer, and Ronald I. Miller. 2004. “Determinants of Long-Term Growth.” American Economic Review 94(4):813–835. Shevtsova, Lilia. 2003. Putin’s Russia. Washington, DC: Carnegie Endowment for International Peace. Shleiffer, Andrei, and Daniel Treisman. 2000. Without a Map. Political Tactics and Economic Reform in Russia. Cambridge, MA: MIT Press. Siegl, Elfie. 2001. “Rußland baut Reformfassaden.” Frankfurter Allgemeine Zeitung 205 (September 4):17. Silverman, Bertram, and Murray Yanowitch. 2000. New Rich, New Poor, New Russia. Armonk, NY: Sharpe. Smith, Adam. 1776–1976. An Inquiry into the Nature and Causes of the Wealth of Nations. Oxford: Oxford University Press. Taube, Markus. 2001. “Ostasien—neues Gravitationszentrum der Weltwirtschaft im 21. Jahrhundert?” Pp. 119–139 in Jahrbuch für internationale Sicherheitspolitik, edited by Erich Reiter. Hamburg: Mittler. Taylor, Charles Lewis, and David A. Jodice. 1983. World Handbook of Political and Social Indicators. 3rd ed. New Haven, CT: Yale University Press. Treisman, Daniel. 2000. “Russia’s Federal System of Public Finance: Trends, Politics, and Pressing Issues.” Pp. 65–68 in Governance, Decentralization and Reform in China, India and Russia, edited by Jean-Jacques Dethier. Dordrecht and Boston: Kluwer Academic Publishers. Trenin, Dmitri. 2002. The End of Eurasia. Washington, DC: Carnegie Endowment for International Peace. Weede, Erich. 1996. “Political Regime Type and Variation in Economic Growth Rates.” Constitutional Political Economy 7(3):167–176. ——. 2000. Asien und der Westen. Baden-Baden: Nomos. ——. 2004. “Comparative Economic Development in China and Japan.” Japanese Journal of Political Science 5(1):69–90. ——. 2006. “Economic Freedom and Development.” CATO Journal 26(3):511–524. Weiffen, Brigitte. 2004. “The Cultural-Economic Syndrome: Impediments to Democracy in the Middle East.” Comparative Sociology 3(3–4):353–375. Wen Tie Jun. 2000. “The Reform of the Agricultural Tax System.” Pp. 279–296 in Governance, Decentralization and Reform in China, India and Russia, edited by Jean-Jacques Dethier. Dordrecht and Boston: Kluwer Academic Publishers. West, J. Robinson. 2005. “The Future of Russian Energy.” The National Interest 80:125–127. Williamson, John B., and Ce Shen. 2004. “Do Notional Defined Contribution Accounts Make Sense as Part of the Old-Age Security Mix for China?” Paper presented at the 36th World Congress of the International Institute of Sociology, Beijing, July 7–11. Winiecki, Jan. 1988. The Distorted World of Soviet-Type Economies. London: Routledge. World Bank. 1981. 1993. 1996. 2000–2001. 2002. 2006. 2007. World Development Reports. New York: Oxford University Press. ——. 1993a. The East Asian Miracle. New York: Oxford University Press. Xu Yi-Chong. 2000. “State Assets Depletion and Economic Reform in China.” Studies in Comparative International Development 35(1):73–100. Yavlinsky, Grigory. 1998. “Russia’s Phony Capitalism.” Foreign Affairs 77(3):67–79.

118

erich weede

Zakaria, Fareed. 2003. The Future of Freedom. New York: Norton. Zhu Ze, and Barbara Krug. 2007. Is China a Leviathan? Rotterdam: Erasmus University, Rotterdam School of Management, Manuscript. Zhuravskaya, Ekaterina V. 2000. “Market-Hampering Federalism: Local Incentives for Reform in Russia.” Pp. 127–170 in Governance, Decentralization and Reform in China, India and Russia, edited by Jean-Jacques Dethier. Dordrecht and Boston: Kluwer Academic Publishers.

SOCIAL STRUCTURE AND PERSONALITY DURING THE PROCESS OF RADICAL SOCIAL CHANGE: A STUDY OF UKRAINE IN TRANSITION Melvin L. Kohn, Valeriy Khmelko, Vladimir I. Paniotto and Ho-fung Hung The theoretical question that motivates this inquiry is whether the relationships between social structure and personality previously found in both Western and non-Western, capitalist and socialist societies (Kohn et al. 1990; Kohn and Slomczynski 1990) during periods of apparent social stability continue to obtain even during periods of radical social change. Following Williams (1970), we define social change as change in the structure of the society, not merely as an eventful or dramatic period in the life of that society: “Change occurs when there is a shift in pattern, when new relationships emerge . . . ” (Williams 1970:620–621). By radical social change, we refer not to the pace of change but to the nature of the change—the transformation of one political and economic system into a quite different system. Our exemplar of radical social change is the transformation of the countries of Eastern Europe and the former Soviet Union from socialism to nascent capitalism. The question of whether the relationship of social structure and personality continues to hold under conditions of radical social change has been provisionally answered by comparative analyses of Poland and Ukraine, based on cross-sectional surveys of the adult populations of the urban areas of those countries in 1992–93 (Kohn et al. 1997). In all those respects in which socialist Poland had shown a pattern of relationships of social class and of social stratification with personality similar to that found in studies of the capitalist United States and Japan, it continued to do so after the advent of nascent capitalism: Under conditions of radical social change, just as under conditions of social stability, people of more advantaged class position, and of higher social-stratificational level, enjoyed much greater opportunity to be self-directed in their work—that is, to do more substantively complex work, to be less closely supervised, and to work under less routinized conditions—than did people of less advantaged social-structural position. Occupational self direction, in turn, continued to be conducive

120

melvin l. kohn et al.

to more self-directed orientations to self and society and to greater intellectual flexibility. Where, however, socialist Poland had differed from the United States and Japan (notably, in that people of more privileged position in the capitalist countries had a stronger sense of well-being, and people of less privileged position were more distressed, while nearly the opposite obtained in then-socialist Poland), Poland in transition now fully exemplified the capitalist pattern. Ukraine seemed to be following a similar trajectory, albeit at a slower pace: Ukraine showed the same pattern of relationships between social structure and personality as did Poland, but all the relationships were weaker in magnitude, with those for distress not even statistically significant.1 The evidence of the cross-sectional analyses thus demonstrates that the radical social change attendant on the transformation of the social and economic structures of Poland and Ukraine had not fundamentally affected the relationships between social structure and personality, at least for the employed segments of the population, except insofar as the social structures of these countries had become more like those of capitalist countries. Yet, the comparative analyses of Poland and Ukraine tell us little about the dynamics of the ongoing process. Not only were the analyses necessarily cross-sectional, but the transformation of Poland had occurred so rapidly that, by the fall and early winter of 1992–93, the relationships of social class and social stratification with personality already exemplified the capitalist pattern of the United States and Japan (Kohn et al. 1990; 1997). The Ukrainian transformation had not advanced nearly so far, and the relationships between social structure and personality, while similar to those for Poland, were not nearly so sharply pronounced; Ukraine was still very much in process of transformation to nascent capitalism. In terms of movement away from the long-term domination of the economy by a centralized system of command, though, the Ukrainian transformation was even more profound than the Polish. For sixty years prior to the beginning of the transformation, Ukrainians had had 1 Kohn et al. (1997) speculated about whether the weaker relationships for Ukraine than for Poland of job conditions and personality, and thus also of class and stratification with personality, were more likely a carryover of Ukraine’s history as part of the Soviet Union or the result of the extreme conditions of uncertainty that Ukraine was experiencing at the time of the 1992–93 survey. Without pertinent data from Ukraine while it was part of the Soviet Union, there was no way to be certain which explanation was valid.

a study of ukraine in transition

121

no experience with private enterprise: even small private enterprises were forbidden. In the rural areas of Ukraine, where more than 80 percent of the population lived, private ownership was eradicated by draconian measures, beginning in the 1930’s. In Poland, by contrast, small private business was never forbidden, and agriculture was never socialized. Moreover, the economy of Ukraine was an integral part of the centralized economy of the USSR. With the disintegration of the USSR, the industrial connections of the enterprises of Ukraine with tens of thousands of enterprises in fourteen newly independent countries of the former Soviet Union were abruptly broken. The result was a much sharper decline in production and in the standard of living in Ukraine than in Poland. Thus, in terms of the depth of the changes that were occurring, the early years of the transformation in Ukraine were extremely radical. The process was hardly complete in 1992–93, and was still ongoing for some years to come. Herein lies the impetus and the opportunity for the present study. With the realization that Ukraine had been at a very early stage of a very radical transformation at the time of the cross-sectional surveys in 1992–93, the Ukrainian collaborators in the present research grasped the unique opportunity to secure the data that would make possible longitudinal analyses of the dynamics of change during the ongoing process of radical social change. In the spring and summer of 1996 they re-interviewed all those men and women in the original sample who had been in the labor force at the time of the initial interview. This made possible the conversion of a cross-sectional survey conducted at a time when the transformation of Ukraine had barely begun into a longitudinal data-set extending three to three and a half years into the ongoing transformation. For this study, even more than for most studies of social structure and personality, context is crucial. Khmelko’s (2002) analysis of macrosocial change in the first decade of Ukrainian independence documents that by 1996 (and even later), although Ukraine had left its former socialist economy far behind, it had not moved decisively to a capitalist social and economic structure. This paper, then, is not a study of Ukraine before and after its transformation from socialism to capitalism, nor of Ukraine during and after the transformation, but of Ukraine during the early stages of an ongoing transformation whose eventual outcome was still uncertain. The strategic value of a longitudinal study of Ukraine during these years is that it enables us to study the dynamics of the relationships between

122

melvin l. kohn et al.

social structure and personality during the ongoing process of radical social change, our analytic lens being an examination of what happens to these relationships under such uncertain, changing conditions. This inquiry thus provides an extreme test of whether the relationships of social structure and personality found in studies conducted under conditions of apparent social stability obtain even during the ongoing process of radical social change. As will be apparent in the analyses to be presented shortly, the test is not only longitudinal, and not only conducted during the very process of radical social change, but extremely severe for a reason that we did not anticipate and that contrasts sharply with the findings of many studies of personality conducted during times of social stability: the over-time correlations of the dimensions of personality we study are astonishingly low. The central questions we shall pursue, then, are (1) whether the relationships of social structure and personality are meaningful, non-trivial in magnitude, and consistent over time even during the very process of radical social change and even in the face of instability of personality during this period of time; and (2) if they are meaningful, non-trivial, and consistent under these extraordinary circumstances, what makes this possible. Sample and Methods of Data-Collection The Baseline Cross-sectional Survey of 1992–93 The initial, cross-sectional survey of Ukraine, which we now treat as the baseline for our longitudinal analyses, was conducted in the winter of 1992–93. It was based on face-to-face interviews with representative samples of all men and women living in urban areas of the country. The sample was drawn by the Ukrainian members of our research team, who designed a method to overcome the limitations of past procedures for selecting samples in the former Soviet Union and the poor quality of official statistics in Ukraine. Their method is based on multi-stage random sampling: the first stage being to sample from seven hundred districts, then to successively sample post offices, streets, buildings, and apartments, and finally residents aged 18 or older living in the selected dwellings. The survey was carried out by the Kiev International Institute of Sociology, a Research Center that the Ukrainian investigators had created in 1990. Since sociologists in the former Soviet Union had had little

a study of ukraine in transition

123

experience in conducting surveys based on face-to-face interviews (see Kohn 1993), they had to develop their survey research center almost from scratch. Fortunately, they had the expert assistance of Michael Haney of the Research Institute of Radio Liberty, who conducted intensive interviewer training sessions (in Russian, which he speaks fluently), in preparation for surveys that the Center carried out for Radio Liberty, with further training in the conduct of academic surveys by the uniquely knowledgeable sociologist and Sovietologist, Michael Swafford, again in Russian, which he too spoke fluently. By the time the Institute carried out the cross-sectional survey of 1992–93, it had a trained and experienced field staff and a good system for ensuring that their interviews were of high quality. The investigators successfully interviewed 81% of their designated respondents, interviewing 2322 people (966 men and 1356 women). The apparent over-representation of women in this sample reflects the demographic composition of the country. (For further information about sampling and methods of data collection, and for information about the methods used for pretesting the interview schedule and for insuring comparability of meaning and measurement, not only between the Russian and Ukrainian versions of the interview schedules, but also with past studies of other countries in other languages, see Appendix A of Kohn et al. 1997.) The Follow-up Survey With limited resources for fieldwork, the investigators restricted the follow-up survey to those respondents in the original survey who had at that time been in the labor force—defined in Ukraine, as in the United States, as either gainfully employed or not employed and looking for work. This was a strategic subsample for studying movement into and out of the ranks of the employed and for studying the psychological concomitants of interclass mobility. Limiting the follow-up survey to people in the labor force, though, had the corresponding disadvantage of precluding longitudinal analysis of housewives and pensioners. Still, we can compensate for much of this loss by juxtaposing the cross-sectional data about people who in 1992–93 were housewives or pensioners to the follow-up data about people who in 1992–93 were in the labor force but by 1996 had become housewives or pensioners. Securing an adequate completion rate in the 1996 follow-up survey proved to be even more difficult than in the baseline survey of 1992–93,

124

melvin l. kohn et al.

in part because many Ukrainians had become disillusioned with the formal institutions of their society; and in part, too, because economic conditions were so difficult that many people—even employed urbanites—spent the time when they were not at their jobs doing what amounts to subsistence farming, in small plots in or near the cities or towns in which they live, and so were not available to be interviewed.2 With great persistence, the Ukrainian investigators did secure interviews with approximately 75% of their intended sample—admittedly, no longer fully representative of the population to which we would like to generalize, but—interpreted cautiously—useful for the study of the dynamics of change. The Over-Time Stability of Personality Authoritarian Conservatism An obvious question with which to begin our analysis is how stable was personality during this period of radical change. To answer this question, and to provide crucial indices of personality for analyses to follow, we developed longitudinal measurement models of the same dimensions of personality as had been studied in the cross-sectional analyses of Poland and Ukraine for 1992–93 (see Kohn et al. 1997, Appendix Table A-1), and had earlier been studied for the United States (Kohn and Schooler 1983, Chapter 6 and Appendices C and D), Poland when it was socialist (Kohn and Slomczynski 1990, Chapter 4), and Japan (Kohn et al. 1990). We began with authoritarian conservatism vs. open-mindedness, deliberately selected as a well-measured dimension of orientation to self and society, one that had been shown to be highly stable in the longitudinal analyses of U.S. men. An initial model showed the over-time correlation (which we shall call its stability) of this dimension of orientation to be astonishingly low (at 0.18 for men and 0.37 for women), particularly for the relatively short interval

2 A survey of a representative sample of 4500 Ukrainian households carried out by the Kiev International Institute of Sociology in the summer of 1996 found that approximately 62 percent of the urban population were engaged in subsistence agriculture, spending on the average 24 hours per week on such activities. The urban residents who engaged in these agricultural activities were not limited to manual workers, but included the self-employed and small-scale employers.

a study of ukraine in transition

125

of three to three-and-a-half years, even considering the tumultuous times that Ukraine was then experiencing. By contrast, for U.S. men over a 10-year period of much greater social stability, from 1964 to 1974, the over-time correlation had been 0.78 (Kohn and Schooler 1983:328). What may be a more apt comparison, even though based on a small sample: Bogdan Mach’s analysis of subsamples of 99 men and 98 women representative of the southern half of Poland during approximately the same span of time as the Ukrainian analyses,3 albeit during a more advanced stage of transformation, yielded over-time correlations for both men and women about as high as that for U.S. men: 0.78 for men and 0.76 for women. We were so astonished at the extraordinarily low stability of authoritarian conservatism for Ukraine that we thought it unwise to pursue the analysis until we were confident that the finding was not an artifact of some flaw in the fieldwork, or in matching baseline respondents and follow-up respondents, or in data-processing.4 We therefore retraced our steps, beginning with the selection and locating of respondents for the follow-up survey, not only reviewing field notes but also conducting brief re-interviews of respondents in the follow-up survey, to be certain that they really were the same people as the original respondents. We also checked our procedures for merging the baseline and follow-up data-files, to be certain that we had not mismatched any 1992–93 and 1996 respondents. Since we had earlier found that the information provided in four respondents’ initial and follow-up interviews about such identifying personal characteristics as age, gender, educational attainment, and marital and parental status had been so inconsistent as to suggest that the wrong person had been re-interviewed or even that one or the other interview had been fraudulent, we did a systematic analysis of the consistency of such information for all people in the longitudinal sample. We also refined the criteria that had been employed for including respondents in the follow-up sample.

3 The baseline survey for this analysis (and for other Polish analyses to be discussed below) consisted of face-to-face interviews with a representative sample of all men and women living in the urban areas of Poland (see Kohn et al. 1997). The follow-up survey was conducted with a small but representative subsample of those members of the original sample living in the southern half of the country. 4 We here follow the strategy that, before embarking on substantive interpretations of cross-national differences, one should always attempt to rule out the possibility of these differences being a methodological artifact (Kohn 1987:719–721).

126

melvin l. kohn et al.

We found no inconsistencies that required excluding other respondents from the sample (although we did find a few inconsistencies that seem to have resulted from miscoding rather than mismatching, which we were able to correct). We did, however, find that 28 respondents who had said they were “working” at the time of the initial interview did not provide convincing evidence that they were actually employed, or were on leave from their jobs, or were actively seeking employment. We removed them from the sample as not really being in the labor force. Neither their removal from the sample nor the correction of inconsistent information made the slightest difference in the measurement model of authoritarian conservatism. In particular, the stability of authoritarian conservatism was unchanged, both for men and for women (for the final measurement model, see Table 1). We count this as important evidence against the possibility that the low over-time correlations were somehow the result of methodological artifact. There is much more evidence to come. Other Dimensions of Orientation The logically next question is whether the low stability of authoritarian conservatism is indicative of a more general pattern of instability of orientation to self and society or is somehow unique to this particular dimension. To answer this question entails assessing the stability of all seven dimensions of orientation that we use in our analyses—anxiety, authoritarian conservatism, receptiveness or resistance to change, personally responsible standards of morality, self-confidence, self-deprecation, and trustfulness. We developed satisfactory longitudinal measurement models for all these dimensions of orientation—satisfactory, as judged by the models’ fitting the data well (see the first column of Table 2) and the parameters being consistent with those of cross-sectional models for Ukraine and other countries and with longitudinal measurement models for the United States and Poland. From these models, we learn that, although authoritarian conservatism is one of the two least stable of the seven dimensions of orientation for men, it is of intermediate rank for women, and in any case (even for men) is not unique in having a low stability: several dimensions of orientation had only small-to-modest stabilities for one or both genders (see the second and third columns of Table 2). We also learn that the magnitudes of over-time correlation

a study of ukraine in transition

127

Table 1. Longitudinal measurement model of authoritarian conservatism, Ukraine (1992–93 and 1996) Concept and Indicators

Standardized Paths: Concept to Indicators Men

Women

1992–93 1996

1992–93 1996

Authoritarian Conservatism The most important thing to teach children is absolute obedience to their parents. It’s wrong to do things differently than past generations did. Any good manager should be demanding and strict with people under him in order to gain their respect. In this complicated world, the only way to know what to do is to rely on specialists. No decent man can respect a woman who has engaged in sexual relations before marriage. One should always show respect to those in authority. You should obey your superiors whether or not you think they’re right. Young people should not be allowed to read books that are likely to confuse them. Over-time correlation of the concept Ratio of chi-square to degrees of freedom Root mean square error of approximation Number of cases

.50*

.40*

.64*

.53*

.67*

.49*

.62*

.47*

.44*

.56*

.44*

.42*

.41*

.46*

.47*

.31*

.42*

.31*

.52*

.54*

.48*

.31*

.41*

.42*

.30*

.52*

.36*

.53*

.30*

.40*

.49*

.42*

.18*

.37* 2.11 0.05

380

460

*p <.05. Note: A high score on the indicator signifies strong agreement with the statement; a low score, strong disagreement.

128

melvin l. kohn et al.

Table 2. Over-time correlations of first-order dimensions of orientation, Ukrainian men and women 1992–93 to 1996, U.S. men 1964 to 1974, and Polish men and women 1992 to 1996 Over-time Correlation (Stability) Ukraine 1992–93 to 1996 (RMSEA) Men Women Dimension of Orientation: Anxiety Authoritarian Conservatism Receptiveness to Change Personally Responsible Standards of Morality Self-confidence Self-deprecation Trustfulness Number of Cases

United States Poland 1964 to 1974 1992 to 1996 Men

Men Women

(0.05) (0.05)

.26* .18*

.28* .37*

.53* .78*

.42* .78*

.62* .76*

(0.06)

.36*

.16

–

.56

.74*

(0.04)

.56*

.38*

.65*

.52*

.63*

(0.05)

.31* .17 .42*

.27* .18* .59*

.52* .55* .81*

.32 .55* –

.29* .56* –

452

687

99

98

(0.04)

379

Notes: 1. RMSEA is an acronym for Root Mean Square Error of Approximation. 2. Self-confidence and self-deprecation are estimated in a single measurement model, hence there is only one measure of goodness of fit for the combined model. 3. The data for U.S. men are taken from Kohn and Schooler 1983: Table C-1, p. 328. 4. The data for Poland were provided in measurement models developed by Bogdan Mach. 5. No longitudinal measurement models were created for U.S. men for receptiveness to change or for Polish men or women for trustfulness.

vary considerably, from 0.17 to 0.56 for men and from 0.16 to 0.59 for women.5 There is, moreover, a discernible pattern to the relative stabilities of the several dimensions of orientation. For both men and women, personally responsible standards of morality and trustfulness are the most stable of these dimensions of orientation, while self-

5 That there is considerable variability in the stability of the various dimensions of orientation, and even substantial differences between men and women in the stability of several dimensions (with men’s stability being higher for receptiveness to change and personally responsible standards of morality, and women’s being higher for authoritarian conservatism and trustfulness) is further evidence that low stability of personality cannot be attributed to some general methodological artifact.

a study of ukraine in transition

129

deprecation, self-confidence, and anxiety are among the least stable. In terms of a long-standing (albeit admittedly ad hoc) classification (Kohn 1969:78–84), self-conception is generally less stable than is social orientation, at least during this period of ongoing social change. By comparison, the magnitudes of over-time correlation were consistently much higher for U.S. men over the ten-year period from 1964 to 1974—the range of correlations in those relatively stable times being 0.53 to 0.81 (see column 4 of Table 2). Still, aside from authoritarian conservatism, the pattern of which dimensions of orientation are more stable, and which are less stable, is much the same for U.S. men as for Ukrainian men and women. The magnitudes of the over-time correlations for Polish men and women during approximately the same timeperiod as the Ukrainian study (Table 2, columns 5 and 6) were generally (albeit not uniformly) substantially higher than those for Ukrainians of the same gender—some of the correlations being as high for one or both genders as those for U.S. men. Mach’s analyses show that, for Poland just as for the United States, the pattern of relatively higher and relatively lower stabilities was much the same as that for Ukrainian men and women, the exception for Poland, just as for the United States, being the high stability of authoritarian conservatism. In short, with the notable exception of Ukraine’s low stability of authoritarian conservatism, particularly for men, there is nothing unusual about the Ukrainian pattern of which are the more stable dimensions of orientation, and which are the less stable. What is remarkable about Ukraine during this period of radical social change is not pattern but magnitude of stability: most dimensions of orientation were much less stable for Ukraine than for the U.S. (during a much less tumultuous decade in the United States) and, perhaps by a somewhat smaller margin, than during the same time-period (but at a later stage of transformation) in Poland. A “Second-order” Measurement Model of Orientation Despite the instability of these dimensions of orientation, we hypothesized that a “second-order” confirmatory factor analysis6 would yield 6 A second-order confirmatory factor analysis is essentially a factor analysis of a factor analysis, in which one asks what are the fundamental dimensions underlying a set of factors inferred in analyses of measured indicators. Since we had seven first-order dimensions of orientation, measured at two times, based on a set of 76 measured indicators, a true (full-information) second-order analysis that simultaneously estimated both the paths from first-order factors to their indicators and the paths from

130

melvin l. kohn et al.

the same two underlying dimensions—self-directedness of orientation versus conformity to external authority, and a sense of distress versus a sense of well-being—as had been found in earlier analyses of the cross-sectional data of the baseline Ukrainian survey and the comparable survey of Poland (Kohn et al. 1997, Table 1), and in studies of the United States (Kohn and Schooler 1983, Fig. 6.3, p. 147; Miller et al. 1986, Table 2); Poland when it was socialist (Kohn and Slomczynski 1990, Fig. 4.3, p. 87), and Japan (Kohn et al. 1990). Indeed, it did (see Table 3). Self-directedness of orientation is reflected primarily in being receptive to change and in being open-minded (i.e., in not having authoritarian-conservative beliefs), and secondarily in having personally responsible standards of morality and being trustful of others—all of which is in accord with our theoretical premises and with past analyses. Distress is reflected primarily in being self-deprecatory and anxious, secondarily in lacking self-confidence—which certainly appears to be face-valid, and which is consistent with past analyses. The model, which fits the data well, is quite similar for men and for women and for the baseline and follow-up surveys. Of pivotal interest for our analysis, the over-time stabilities of these underlying dimensions of orientation are, as would now be expected, very low: 0.28 and 0.22 for self-directedness of orientation and distress for men, 0.18 and 0.32 for self-directedness of orientation and distress for women. We take these figures to be as close to reality as we are able to come (see the latter half of note 6). For U.S. men, the corresponding figures are 0.78 and 0.59 (Kohn and Schooler 1983, Fig. 6.3, p. 147). Thus, a major finding of the research is that there has been great instability in these underlying dimensions of personality, over a second-order factors to first-order factors seemed too formidable to test all at once, particularly in a two-population model for the two genders. We therefore used the correlations of factor scores based on the first-order factors as the input to a confirmatory factor analysis that treated these factor scores as if they were measured variables. (We therefore put quotation marks around “second-order” in Table 3, to signify that this is not a true second-order model but an approximation thereto.) Since correlations of factor scores invariably underestimate true correlations, we became concerned lest the approximate “second-order” model, based as it was on correlations of factor scores for the first-order dimensions, might underestimate the over-time correlations of the second-order concepts. To be certain that this had not occurred, we replicated the model (separately for the two genders), as full-information models, simultaneously estimating the first-order and second-order portions of the model. The new estimates of the over-time correlations of self-directedness of orientation and distress were very close to those of the approximate model, assuring us that we had not under-estimated those correlations.

a study of ukraine in transition

131

Table 3. “Second-order” longitudinal measurement model of orientation, Ukraine (1992–93 and 1996) Second-order and First-order Concepts

Standardized Paths, Second-order Concepts to First-order Concepts Men

Women

1992–93

1996

1992–93

1996

–.40* .25*

–.48* .40*

–.58* .27*

–.47* .37*

.36* .78*

.38* .55*

.19* .56*

.34* .62*

–.43* .85* .46*

–.45* .65* .51*

–.25* .70* .58*

–.38* .72* .54*

–.15*

–.29*

–.40*

–.36*

Self-directedness of Orientation Authoritarian Conservatism Personally Responsible Standards of Morality Trustfulness Receptiveness to Change Distress Self-confidence Self-deprecation Anxiety Correlation between Self-directedness of Orientation and Distress Over-time Correlation of: Self-directedness of Orientation Distress

.28* .22*

Ratio of Chi-square to Degrees of Freedom Root Mean Square Error of Approximation Number of Cases

.18* .32* 2.33 0.06

379

452

*p <.05.

period of three to three and a half years of ongoing social change. This finding flies in the face of an accumulating body of research and writing in psychology (of which Costa et al. 2000 is prototypic) that finds the stability of personality to be so invariably high that some investigators (e.g., McCrae et al. 2000; McGue et al. 1993) infer that there must be a predominantly biological (mainly genetic) basis to personality. The pertinent studies, though, were not only based on distinctly different facets of personality from those studied in research on social structure and personality, and on samples weighted to more

132

melvin l. kohn et al.

advantaged segments of the population, but also—and crucially—even though some tangential evidence is based on studies in countries that were experiencing or had experienced radical social change, the longitudinal analyses were invariably carried out under conditions of much greater social stability than those experienced in Ukraine during the time of our research. Prior research does not contradict our finding of great instability in underlying dimensions of personality because, so far as we know, there had been no prior studies of the stability of these or similar dimensions of personality under conditions of radical social change, only research that generalized far beyond what was justified by the empirical evidence. Intellectual Flexibility There is one more dimension of personality that is essential, not only to our assessment of the stability of personality, but also to our analysis of the continuing relationship of social structure and personality: intellectual flexibility, which we measure by assessing the respondent’s intellectual performance in the interview situation itself. Our measurement model of intellectual flexibility (see Table 4) is based on our appraisals of the adequacy of the respondents’ answers to two seemingly simple but highly revealing cognitive problems; their propensity to “agree” when asked agree-disagree questions; and (for 1996) the interviewer’s assessment of the respondent’s intelligence, following a long session that required a great deal of thought and reflection (a rating that had inadvertently been left out of the initial interview). The full rationale for this approach, the wording of the questions asked, and cross-sectional measurement models for Poland and Ukraine are presented in Kohn et al. (2000:203–205). We do not assume that any of the indicators of intellectual flexibility is unbiased; but we do assume that all the indicators reflect, to some substantial degree, people’s effectiveness in coping with the intellectual demands of complex tasks. We rely on confirmatory factor analysis to assure that the biases of any indicator are not reflected in our measure of the underlying concept, provided that these biases are not shared by all of the other indicators. The model fits the data well, its ratio of chi-square to degrees of freedom being 1.76 and its root mean square error of approximation being 0.03. The over-time correlations for intellectual flexibility (at 0.52 for men and 0.55 for women) are considerably more robust than those

a study of ukraine in transition

133

Table 4. Longitudinal measurement model of intellectual flexibility, Ukraine (1992–93 and 1996) Indicators

Standardized Paths: Concept to Indicators Men

“Commercials” Question “Kiosk” Question “Agree” Score Interviewer’s Appraisal of Respondent’s Intelligence

1992–93

1996

1992–93

1996

.44* .46* –.07 –

.23* .23* –.26* .63*

.45* .36* –.22* –

.49* .22* –.46* .51*

Ratio of Chi-Square to Degrees of Freedom Root Mean Square Error of Approximation Over-time Correlation of the Concept Number of Cases

Women

1.76 0.03 .52*

.55*

379

459

*p <.05.

for either self-directedness of orientation or distress, but much lower than that for the intellectual flexibility of U.S. men, which was 0.93. We have to take into account, though, that the Ukrainian model may underestimate the true over-time correlations, because of the baseline survey’s leaving out the interviewer’s assessment of the respondent’s intelligence. In an analysis of the cross-sectional Polish data, comparing the full fourindicator model to a model using only the three indicators that were available for Ukraine, Kohn et al. (2000:204) found that the absence of the fourth indicator results in underestimates of the correlations of intellectual flexibility with social-structural variables by approximately one-fifth. By extrapolation, we infer that our longitudinal measurement model may underestimate the over-time correlation of intellectual flexibility for Ukraine to approximately the same degree, which would imply that the true over-time stability for Ukraine would be on the order of 0.60–0.65. An over-time stability of this magnitude would still be decidedly smaller than that for U.S. men during a more stable ten-year period. In all probability, intellectual flexibility was much less stable for Ukraine during this period of ongoing radical social change than for the United States during more stable times, but nevertheless much more stable than was either self-directedness of orientation or distress.

134

melvin l. kohn et al. Social Structure and Personality

That self-directedness of orientation and distress were so unstable during this period of radical social change in Ukraine—and that even intellectual flexibility was in all probability decidedly less stable than it had been during a decade of apparent social stability in the United States—poses the central question of this paper in extreme form: Is a stable relationship between social structure and personality possible in the face of instability of personality? Conceptualization and Index of Structural Location In assessing the relationships between social structure and personality during the process of social and economic transformation, we conceptualize social structure in a way that encompasses not only the employed segments of the population, but also those who were not gainfully employed. To do this, we incorporate a classification of social class that Kohn et al. (1997) developed for the employed segments of the Polish and Ukrainian populations during the transitional period, adding appropriate categories for the non-employed segments of those populations. We call the resultant classification structural location. By “social classes” we mean groups defined in terms of their relationship to ownership and control over the means of production, and of their control over the labor power of others. Social classes are distinct groups (“nominal” categories), not a continuum or even a set of categories that can be ranked as higher or lower along some single underlying dimension. Our classification of social class had its origins in Erik Wright’s (1978) conceptualization of social class for advanced capitalist societies, as modified for the United States by Kohn and Schoenbach (1983), and then further modified for socialist Poland by Kohn and Slomczynski (1990). For Poland and Ukraine in transition, Kohn et al. (1997) dropped the distinction between manual workers employed in the centralized sectors of Poland’s socialist economy and manual workers employed in ancillary sectors of that economy as no longer relevant to economies in transition to capitalism, and added a distinction between those owners who employed a substantial number of non-family employees (the employers) and those owners who did not (the self-employed), a distinction that was becoming more relevant to Poland and Ukraine in transition than it had been for socialist Poland. They also added a distinction between experts and other non-manual

a study of ukraine in transition

135

employees, which seemed to be particularly pertinent to transitional economies. The resultant set of social classes is depicted in the first seven categories of the classification below. During the period of transformation, we must also take account of the non-employed segments of the population: not only those who were not gainfully employed and were seeking employment (the unemployed), but also other non-employed people who occupy distinctive locations in the socio-economic structure, notably housewives and pensioners. This is necessary in the present inquiry despite the follow-up sample being limited to people who were in the work force at the time of the initial survey, because in the ensuing three or three and a half years many people had become housewives or pensioners. Our resulting classification of structural location thus combines the classification of social class that Kohn et al. (1997) developed for the employed with the categories enumerated by Kohn et al. (2000 and 2002) for the principal segments of the nonemployed portion of the adult population, as follows: (1) Employers: Owners of the means of production who have at least three nonfamily employees. (2) Self-employed: Owners of the means of production who have no more than two nonfamily employees, and members of the owners’ families who are employed in such enterprises. (3) Managers: Employees who direct and control the operation of a firm, organization, or major governmental unit or large subdivision thereof, as well as other employees in appropriate occupational categories who directly or indirectly supervise the activities of more than 50 people, some of whom themselves are supervisors. (4) Supervisors: All other employees who supervise at least two people. (5) Experts: Nonsupervisory employees who work in professional occupations that usually require university or polytechnical institute education. (6) Nonmanual workers: All nonsupervisory employees whose work includes a substantial nonmanual component, other than those classified as experts. (7) Manual workers: Nonsupervisory employees whose work is predominantly manual in character. (8) Unemployed: Those who are not gainfully employed at least 15 hours per week and are seeking employment.

136

melvin l. kohn et al.

(9) Pensioners: People who have retired on pension, whether they retired for reasons of age, had willingly or otherwise taken early retirement, or had suffered some disability. (10) Housewives: Women who were not gainfully employed for 15 or more hours per week and considered themselves to be housewives, even if they were seeking paid employment. (11) That small portion of the adult population who are full-time students—a category excluded from the present analysis, because none of the people who were in the labor force at the time of the initial survey went back to being full-time students. For pertinent analyses we shall also distinguish women who were employed but were on maternity or child-care leave from their jobs; and also people employed less than 15 hours per week (but who were not actively seeking full-time employment). Structural Location and Personality in 1992–93 and in 1996 Since the sub-sample of 1992–93 respondents who were re-interviewed in 1996 does not include people who were housewives or pensioners at the time of the baseline survey, but some people who were in the labor force at the earlier time had moved into the ranks of the non-employed by the time of the 1996 survey, the most appropriate baseline against which to compare the 1996 subsample is the entire cross-sectional sample (other than students) who were interviewed in 1992–93 (see Table 5).7

7 Our treatment of some of the categories in the classification of structural location differs somewhat from earlier analyses of the 1992–93 surveys. The Kohn et al. (2002) analyses of the 1992–93 Ukrainian sample excluded respondents older than 65, to make the analyses comparable to those for Poland, where the sample had an age cut-off of 65; but, for an analysis limited to Ukraine, no such age cut-off is necessary, and for comparisons of the 1992–93 sample to the 1996 respondents it is informative to include older respondents, particularly pensioners. The earlier analyses also included adult respondents, generally younger adults, who were full-time students. Those students not having been included in the follow-up survey, and none of the respondents in that survey having in the meantime become full-time students, it would make for non-comparability of the 1992–93 and 1996 respondents to include those respondents who were then students in the 1992–93 analyses, so we excluded them. We also combined two pairs of categories. For both men and women, we combined managers and employers, there being few respondents in either category, both categories entailing control over the labor power of others, and the respondents in the two categories being similar in our measures of personality. We also combined the relatively small number of women on maternity or child-care leave from their jobs with the much larger number of housewives. Here, too, the respondents in the two categories differ little in our

a study of ukraine in transition

137

The 1996 sample is not fully representative of the urban adult population of Ukraine in that year, not only because follow-up samples never do completely match cross-sectional samples, but also because the pensioners and housewives in our 1996 sample are limited to previously employed women and men who became pensioners or housewives in the ensuring few years. This limitation on the representativeness of the 1996 sample entails a small loss but also a strategic gain, for it focuses our attention on people who became nonemployed during the period of radical social change. The pattern of relationships between structural location and personality is much the same for 1996 as it was for 1992–93: Men and women in more advantaged structural locations were more self-directed in their orientations, more intellectually flexible, and less distressed than were men and women in less advantaged structural locations. Specifically, as was true for men in 1992–93 (there being too few women who were either managers or employers at that time to draw firm conclusions about them), the managers and (the few) employers, and the experts, were at one extreme in 1996, the pensioners at the other. It is notable that the pensioners of 1996 (who, as a result of our criteria for sample selection, were newly pensioned) were as extreme in their lack of selfdirection and intellectual flexibility, and in their high degree of distress, as were the pensioners of 1992–93 (who, as a result of our criteria for sample selection, were not included in the follow-up sample). Stability in the relationship of structural location and personality co-existed with instability of personality. Insofar as there were changes in the relationship of structural location and personality from 1992–93 to 1996, the figures for Ukraine in 1996 are closer to those for the Poland of 1992–93 than for the Ukraine of that time—Poland having been further along in its transformation. (This conclusion is based on comparisons of the present Table 5 with Table 1 of Kohn et al. 2002.) In particular, in 1992–93 there was a decided difference between the Polish and Ukrainian unemployed, with unemployed Poles being somewhat less self-directed and less intellectually flexible than most other people of their gender, particularly the employed, while

measures of personality. And we add the few people who had lost their jobs and had given up active efforts to find another to those who were still actively looking for paid employment, to create a combined category, “seeking work or discouraged.” None of these modifications of our basic classification distorts the findings of Table 5, but they do make for greater comparability between 1992–93 and 1996.

138

melvin l. kohn et al.

Table 5. Structural location and personality, Ukraine (1992–93 and 1996) Standardized Differences from the Mean for all People of the Particular Gender in the Pertinent Sample Number of Cases

Self-directedness of Orientation

Distress

Intellectual Flexibility

(1992–93) (1996) 1992–93 1996 1992–93 1996 1992–93 1996 A. MEN Structural Location Managers & Employers Self-employed Supervisors Experts Non-manual Workers Manual Workers Seeking Work or Discouraged Pensioners Employed < 15 Hours/Week

(27) (15) (116) (31) (19) (319) (37)

(17) (21) (69) (20) (5) (165) (23)

.69 .24 .16 .77 .45 .03 .27

.59 .00 .12 .46 – –.10 –.04

–.26 –.12 –.32 –.43 –.20 –.09 .31

–.42 .12 –.02 –.33 – –.08 .35

.60 –.14 .11 .39 .39 .00 .21

1.05 .04 .21 .93 – –.27 –.21

(217) (6)

(15) (14)

–.43 –

–.52 –.02

.36 –

.44 .14

–.24 –

–.31 –.07

.32* .15

.31* .16

Correlation (eta) Correlation (eta) Limited to the Employed

.38* .26*

.32* .30*

.27* .24*

.60* .62*

B. WOMEN Structural Location Managers & Employers Self-employed Supervisors Experts Non-manual Workers Manual Workers Seeking Work or Discouraged Pensioners Employed < 15 Hours/Week Housewives or on Maternal or Child-care Leave Correlation (eta) Correlation (eta) Limited to the Employed *p <.05.

(3) (10) (102) (74) (130)

(10) (15) (72) (48) (88)

– .31 .27 .67 .19

.63 –.11 .08 .24 –.10

– –.13 –.26 –.21 –.22

–.74 .26 –.18 –.11 –.13

– –.03 .17 .60 .18

.64 –.16 .32 .51 .08

(234)

(121)

.00

–.02

–.07

.02

–.04

–.37

(31) (356) (5)

(22) (18) (22)

.19 –.42 –

–.08 –.10 .19

–.02 .35 –

.35 .28 .39

.43 –.34 –

.17 –.60 –.11

(194)

(13)

.20

–.48

–.20

.16

.17

–.15

.24* .18

.33* .26*

.35* .25*

.37* .33*

.31* .14

.47* .46*

a study of ukraine in transition

139

unemployed Ukrainians were somewhat more self-directed in their orientations and intellectually flexible than most other people of their gender, as much or more so than the average for all of the employed. Kohn et al. (2002) attributed the relatively higher degree of self-directedness of orientation of the Ukrainian than of the Polish unemployed of 1992–93 to the initial brunt of Ukrainian unemployment striking the younger and better educated. By 1996, unemployed Ukrainians (particularly the men) were relatively less self-directed and intellectually flexible than they had been three years earlier. And unemployed Ukrainian women, who in 1992–93 had not been much more distressed than the mean for their gender, had joined unemployed Ukrainian men—and unemployed Poles—in being decidedly more distressed than other Ukrainians of their gender. Housewives and women on maternal or child-care leave had also become relatively more distressed—more like the involuntary Polish housewives of 1992 than like the Ukrainian housewives of that time, who were averse to full-time employment. Ukraine was past the initial brunt of unemployment. As for class differences among the employed: the pattern of class differences for Ukraine in 1992–93 was very similar to that for Poland (see Kohn et al. 1997, Table 2), although the magnitudes of relationship of class and personality were greater for Poland. The cross-national differences (between Poland and Ukraine) in the patterns of relationship of structural location and personality were entirely among the non-employed. The magnitudes of relationship between structural location and personality (as measured by eta, a correlational statistic appropriate to a nominal variable such as structural location) are roughly similar for the Ukraine of 1996 to what they had been in 1992–93 for both selfdirectedness of orientation and distress,8 and had probably increased

8 We attempt to assess the “true” magnitudes of correlation (etas), corrected for attenuation in factor scores or latent-variable scores. The 1992–93 figures are based on factor scores produced by the FSCORE program, which provides the exact correlation between true scores and factor scores for all three dimensions of personality; with that information we are able to correct the etas to take the unreliability of factor scores precisely into account. The 1996 figures are based on latent-variable scores produced by the LISREL program, which provides no information about the relationship between true scores and latent-variable scores. We have estimated the approximate magnitudes of the relationships between latent-variable scores and true scores for self-directedness of orientation and distress by systematically comparing magnitudes of correlation of the latent-variable scores with other ordinal variables to the true correlations produced in corresponding full-information models. For intellectual flexibility, where we have created both factor scores and latent-variable scores—which correlate perfectly with

140

melvin l. kohn et al.

for intellectual flexibility (even taking into account that the figure for 1992–93 may be an under-estimate9). Certainly, the relationships between structural location and personality did not appreciably weaken during those years. Further insight into the relationship of social structure and personality during the ongoing transformation of Ukraine is provided by examining the magnitudes of correlation (the etas), not for all segments of the urban population, but only for the employed—i.e., for social class. Table 5 shows that the etas for the relationship of social class with intellectual flexibility are approximately as large as are those for structural location—thus, class differences among the employed are as sizeable as are the differences between the employed and the non-employed. For self-directedness of orientation, some of the etas for class are substantially smaller than are those for structural location. For distress, the etas for social class are consistently much smaller than those for structural location and are not even statistically significant; it is the pensioners (for both years for both men and women) and the unemployed and partially employed (both years for men, only in 1996 for women, then joined by the housewives) who make the overall relationship of structural location and distress robust. Perhaps symptomatic of Ukraine’s equivocal transition from socialism to capitalism, the manual workers—who in socialist Poland had enjoyed a strong sense of well-being but who in (nascent) capitalist Poland had become notably distressed—in Ukraine, both in 1992–93 and three to three and a half years later in 1996, hovered close to the mean for all urban adults of their gender, employed or non-employed. The really sharp contrast when it comes to a sense of

each other—we use the correlations of true scores and factor scores as our estimate of the correlation of true scores and latent-variable scores. If we had made no effort to adjust the etas for unreliability in the factor scores or latent-variable scores for the three dimensions of personality, all the estimates of magnitude would be smaller, but the general conclusion, that magnitudes of relationship were generally as high or higher in 1996 as in 1992–93, would still hold true. 9 The issue, once again, is that the 1992–93 index of intellectual flexibility lacks a crucial indicator, the interviewer’s assessment of the respondent’s intelligence. As noted in the text, the index may underestimate the correlations of social-structural position with intellectual flexibility, thus measured, by approximately 20%. Even if we increased our estimate of the eta for the relationship of structural location with intellectual flexibility in 1992–93 by 20%, though, the eta for 1992–93 would still be considerably smaller than that for 1996, when the crucial indicator was included in the measurement model.

a study of ukraine in transition

141

well-being or distress is not of manual workers with the more advantaged social classes, but of the employed with the non-employed. We must therefore keep in mind that the relationships of social structure with self-directedness of orientation and intellectual flexibility are substantial in magnitude and statistically significant whether or not the non-employed are included in the analysis, while the relationships of social structure with distress are statistically significant and substantial in magnitude only when the non-employed are taken into account. We must also keep in mind that the relationships of social class with all three dimensions of personality were weaker for Ukraine in 1992–93 than for Poland at about the same time (Kohn et al. 1997), so intranational consistency over time for Ukraine means continued inter-national inconsistency between a Poland that had moved rapidly toward fully developed capitalism and a Ukraine that had not. The ongoing process of radical social change in Ukraine did not weaken the relationships of social structure and personality, but it may have prevented the relationships of class and stratification with personality from becoming as strong as they were in Poland. Still, whatever accounts for the different patterns of relationship of social structure with distress in contrast to self-directedness of orientation and intellectual flexibility (an important issue, to which we shall return), it is noteworthy that all relationships of social structure and personality remain consistent over time. This is true both for social class and for structural location, and for all three dimensions of personality. Despite the deep structural changes in the economy and the society during the three years intervening between our surveys, and despite great instability of personality during that period, the pattern and even the magnitudes of relationship of social class and of structural location to personality remained remarkably consistent. One major interpretive task is to explain the stability of these relationships despite change in all their constituent elements. Social Stratification and its Component Dimensions For a finer-grained analysis of the relationship of social structure to personality among employed men and women, we utilize an alternative conceptualization of socio-economic structure—social stratification. We define social stratification as the hierarchical ordering of society, a single dimension inferred from the covariation of educational attainment, occupational status, and job income. To index social-stratificational

142

melvin l. kohn et al.

position, we developed a (true) second-order confirmatory factor-analytic model (see Table 6), with each of the first-order component factors having two indicators. This model closely follows the cross-sectional model developed for the Ukraine of 1992–93 (Kohn et al. 1997, Appendix Table A-2), now extended to be longitudinal. The 1992–93 portion of the longitudinal model shows that educational attainment reflects social-stratificational position most strongly, and occupational status next most strongly (they were roughly equal in the cross-sectional model), with income of decidedly lesser importance, especially for men (as was also true in the cross-sectional model). By 1996, income had become a stronger indicator of social-stratificational position than in 1992–93, both for men and for women, although still not nearly as strong as educational attainment or occupational status. In other respects the 1992–93 and 1996 portions of the model are quite similar, as they are for men and for women. The longitudinal measurement model of social-stratificational position fits the data well, the ratio of chi-square to degrees of freedom being 1.72 and the root mean square error of approximation being 0.04. The most striking thing about the longitudinal model is the extremely high stability of employed men’s and women’s social-stratificational positions—the over-time correlations being 0.98 for men and 0.97 for women. At first blush, such high over-time consistency in socialstratificational position would seem to be incompatible with the considerable amount of inter-class mobility that had occurred in Ukraine during that period (as will be shown below). The explanation lies in the component dimensions of social stratification. Educational attainment is necessarily highly stable: no one could have declined in educational attainment, nor could employed people have greatly increased their educational attainment in three or three and a half years of part-time schooling. Occupational status and income, especially the latter, were more open to change. In fact, the over-time correlations of the three component dimensions of social stratification (as assessed in a firstorder analogue to the second-order measurement model presented in Table 6) are in accord with our expectation. The over-time correlations of educational attainment were very high (0.87 for men and 0.85 for women), those for occupational status somewhat lower (0.70 for men and 0.77 for women), and those for income by far the lowest of all (0.25 for men and 0.33 for women). Social-stratificational position being inferred from the covariance of these three dimensions, it can be highly stable even when there is considerable inter-class mobility.

a study of ukraine in transition

143

Table 6. Longitudinal measurement model of social-stratificational position, Ukraine (1992–93 and 1996) Concepts and Indicators

Men 1992–93

Educational Attainment Years of Schooling Educational level Income Job income (log) Family income (log) Occupational status Treiman (International) scale Ukranian Prestige Scale

Social-Stratificational Position Educational attainment Income Occupational status Over-time correlation of socialstratificational position Ratio of chi-square to degrees of freedom Root mean square error of approximation Number of cases

Women 1996

1992–93

1996

Standardized paths, concepts to indicators .97* 1.00*

.95* .99*

.98* 1.00*

.97* .96*

.70* .94*

.90* .89*

.83* .48*

.85* .74*

.99* .85*

.98* .89*

.97* .94*

.96* .95*

Standardized paths, social stratificational position to first-order concepts .94* .12* .70*

.95* .30* .66*

.96* .26* .74*

.98*

.92* .55* .78* .97*

1.72 .041 384

463

*p <.05.

Social Stratification and Personality in 1992–93 and 1996 10 The extremely high stability of social-stratificational position does not necessarily mean that its relationships to personality would necessarily

10 Since social stratification, self-directedness of orientation, and distress are ordinal variables, we are able to precisely assess the magnitudes of their inter-correlations by creating a combined measurement model of all three concepts, fixing the paths from concepts to indicators (the lambdas) at the unstandardized values achieved in the separate measurement models, allowing the same correlations of residuals, and leaving the unexplained residuals free. We limit the analyses of social stratification and personality to men and women who were employed both in 1992–93 and in 1996, thus eliminating the possibility that any differences we might find in the patterns of relationship for the two years might result from movement into or out of the ranks of the employed.

144

melvin l. kohn et al.

be similar in 1996 to what they were in 1992–93. On the contrary, the very instability of personality might suggest that social-stratificational position must have lost some or even most of its relevance for personality during those years, else high stability of social position would have resulted in correspondingly high stability of personality. We find, though (in Table 7), that social stratification was of greater importance for the self-directedness of orientation and intellectual flexibility of employed men in 1996 than it was in 1992–93, and was of at least as great importance for the self-directedness of orientation and intellectual flexibility of employed women in 1996 as it was in 1992–93. For distress, though, the safest conclusion we can draw is that social stratification was only weakly and generally not significantly related to distress for either men or women in either year. As we learned from the analyses of the relationships of structural location with distress, what mattered most for distress during these times of radical social change was not social-structural variations among the employed, but whether or not one was employed at all. Explaining the Consistent Relationships of Social Structure and Personality Despite Instability of Personality We have answered the question of whether social structure can have a consistent and meaningful relationship to personality even in the face of instability of personality: it certainly did in Ukraine during a prolonged period of extremely radical social change. We have yet to answer the more fundamental question, what accounts for stability in the relationship of social structure and personality despite instability of personality? Nor have we broached the related question, what accounts for the instability of personality? A sociologically appealing hypothesis that, if valid, would simultaneously answer both questions is that the considerable social mobility that must have resulted from the transformation of the Ukrainian economy would have led to changes in the personalities of those who were mobile, consonant with their new positions in the social structure. Such changes in personality could be described at an individual level as instability of personality and at a social level as sustaining a consistent relationship of social-structural position and personality. We do not have the data to trace every step of the implied processes, but we certainly can assess whether social mobility has been meaningfully related to changes in personality.

a study of ukraine in transition

145

Table 7. Social-stratificational position and personality, Ukraine, 1992–93 and 1996 Zero-order Correlations of Social-Stratificational Position with Personality, for Men or Women Employed Both in 1992–93 and in 1996 Employed Men

Correlations of SocialStratificational Position with: Self-directedness of Orientation Distress Intellectual Flexibility Number of Cases

Employed Women

1992–93

1996

.23*

.43*

.36*

.35*

–.15* .47*

–.10 .87*

–.03 .57*

–.14 .74*

352

1992–93

1996

412

*p <.05.

Social Mobility and Changes in Personality Are the consistent relationships of social structure and personality that we have found meaningfully related to changes in social-structural position? Certainly, there was a great deal of social mobility in Ukraine from 1992–93 to 1996. Only 198 of the 337 men, and only 197 of the 459 women, in our longitudinal sample were in the same structural location in 1996 as in 1992–93. Movement had taken place in both directions—from more to less advantaged structural locations and from less to more advantaged locations. For example (to take a crucial segment, and one that is clearly less advantaged than most), only two of the 21 men who were not gainfully employed and were looking for work in 1992–93 were still (or again) unemployed in 1996. And only one had dropped out of the labor force and stopped looking for employment. All 18 others were employed at least 15 hours per week, most of them as manual workers, but some as non-manual workers, supervisors, or self-employed. Conversely, of the 21 men who were not gainfully employed and were looking for work in 1996, all but four had been working at least 15 hours per week in 1992–93. The figures are similar for women, with only four of the 20 women who were unemployed in 1992–93 still (or again) unemployed in 1996. All this is consistent with (but by no means proof of) the possibility that changes in structural location attendant on the transformation of the Ukrainian economy

146

melvin l. kohn et al.

might provide the key to understanding consistent relationships between social-structural position and personality despite widespread change in personality. If people’s personalities changed consonantly with changes in their social-structural positions, then instability in personality would be fully in accord with stability in the relationships of social structure and personality. Movement to and from the ranks of the employed. We look first at changes in self-directedness of orientation and in distress associated with movement from the ranks of the full-time employed in 1992–93 to the various categories of non-employment in 1996, and from unemployment in 1992–93 to employment in 1996 (see Table 8).11 The N’s are small, but the changes are dramatic, and despite the small N’s, three of the four most crucial differences are statistically significant. With one intriguing exception, the findings are consistent for men and for women. Movement from unemployment to employment was associated with an increase in self-directedness of orientation and a decrease in distress—both for men and for women, although the magnitudes of change in orientation were more substantial for men. Conversely, movement from employment to unemployment was associated with a decrease in self-directedness of orientation and an increase in distress—this too both for men and for women. The one cross-gender inconsistency is in what happens when previously employed people become pensioned: Those men who changed from being employed to becoming pensioned, as best we can infer from an N of only 12, behaved similarly to men who changed from being employed to becoming unemployed—they showed a decrease in self-directedness of orientation and an increase in distress. Those women who changed from being employed to becoming pensioned (a slightly larger N of 17) behaved in precisely the opposite way: their level of self-directedness increased, their level of distress decreased. The cross-gender difference in distress, although based on 11 For the analyses of change in personality, we make use of a new component of LISREL, which produces a type of factor score (which Jöreskog and colleagues call latent-variable scores) that correlate perfectly with ordinary factor scores but have a feature that is extremely valuable for our purposes: The intercorrelations of latent-variable scores for factors derived from the same measurement model perfectly reproduce the correlations among the latent variables in that measurement model ( Jöreskog 2000). This is crucial for our analyses, for correlations of ordinary factor scores derived from a longitudinal confirmatory factor-analytic measurement model underestimate the correlations of time-1 and time-2 measures of the same concept, obviating one of the principal purposes of the longitudinal analyses—to assess the relationships between change in social-structural position and change in personality.

+.57 –.05 0.03

–.31

–.36 +.02 0.05

(19)

(12) (283)

+.39

–

–

(6)

–.44

+.50

Distress

(18)

Self-directedness of Orientation

0.32

–.46 +.04

+.01

–

+.28

Intellectual Flexibility

Note: Employed is defined as gainfully employed for an average of 15 or more hours per week.

Unemployed in 1992–93, employed in 1996 Not employed at the time of either interview Employed in 1992–93, and: (a) not employed (either looking for work or discouraged) in 1996 (b) a pensioner in 1996 Employed in 1992–93 and also in 1996 P (difference between those who changed from unemployed to employed and those who changed from employed to not employed)

(N)

Men

(17) (308)

(20)

(4)

(16)

(N)

Women

0.20

+.35 –.02

–.16

–

+.26

0.01

–.16 .04

+.42

–

–.17

0.34

–.41 .00

+.16

–

–.14

Self-directedness Distress Intellectual of Orientation Flexibility

Change from 1992–93 to 1996

Table 8. Movement to and from the ranks of the employed and change in personality, Ukraine, from 1992–93 to 1996

a study of ukraine in transition 147

148

melvin l. kohn et al.

small N’s, is statistically significant. Becoming pensioned was a letdown for men, a boon for women. As for intellectual flexibility: movement into and out of the ranks of the unemployed was not significantly related to change in intellectual flexibility for either men or women. But movement from employment in 1992–93 to becoming a pensioner by 1996 was associated with a decrease in intellectual flexibility for both genders. Hence, at the extreme—movement into and out of the ranks of the employed—change in structural location was significantly, substantially, and meaningfully related to changes in self-directedness of orientation and distress, albeit not to changes in intellectual flexibility, except in the case of the newly pensioned. But, since only a small portion of the population moved from employment to non-employment, or the reverse, this is at most only a small part of the explanation of the consistency of the relationships of social structure and personality despite changes in personality. To pursue the hypothesis that mobility from one structural location to another contributed substantially to the consistent relationships of social structure and personality that we have found for the sample as a whole requires that we look also at mobility among that vast majority of men and women who were employed both in 1992–93 and in 1996. Inter-class mobility. We begin this portion of the analysis by employing a theoretically derived classification of inter-class mobility, the crux of which is that (in our conceptualization) social class represents the intersection of three dimensions of inequality: ownership vs. nonownership; supervisory authority over others or the lack thereof; and, among non-supervisory employees, their employment situation (as experts, non-manual workers, or manual workers). We classified movement from more advantaged to less advantaged class position, and the reverse, along each of the three dimensions. We also combined the three types of movement from more to less advantaged class positions into an aggregate category, movement to a class position that is more advantaged along any of these dimensions. We similarly aggregated the three types of movement from less to more advantaged class positions into a single category. The hypothesis, of course, was that movement from a more advantaged to a less advantaged class position (along any or all of the dimensions of inequality) would result in an increase in distress and a decrease in self-directedness of orientation and perhaps also in intellectual flexibility; and that movement in the reverse

a study of ukraine in transition

149

direction would have opposite effects. No such thing. Neither the people who moved to a more advantaged class position nor those who moved to a less advantaged class position changed their levels of self-directedness of orientation, distress, or intellectual flexibility to a substantial (or statistically significant) degree. For those men and women who were employed both in 1992–93 and in 1996, we find no evidence that movement to or from a more advantaged class position was significantly linked to change in any of the three fundamental dimensions of personality. The theoretically derived approach not having yielded any evidence that inter-class mobility along the more advantaged-less advantaged axes was linked to changes in personality among people employed in both years, we resorted to an empiricist approach: We examined a complete cross-classification of all class positions in 1992–93 and 1996, to see whether there were any movements from any social class to any other social class that were associated with substantial change in any of the three principal dimensions of personality. Even where the N’s were small, we searched to find whether we could discern any patterns of change. We could not discern any consistent pattern. Our conclusion is that inter-class mobility was not substantially linked to changes in personality. There certainly was considerable class mobility, but—unlike the strong evidence for movement from the ranks of the employed to the unemployed and the reverse—we found no evidence at all that inter-class movement among those who were employed both in 1992–93 and in 1996 was linked to substantial changes in personality. Change in social-stratificational position (and its component dimensions) and change in personality. The very strong over-time correlations of socialstratificational position for both men and women mean that change in the social-stratificational positions of those men and women who were employed at the times of both interviews could not account for more than a small portion of the change in personality, because there was little change in their social-stratificational positions. Still, it is possible that change in one or another of the three component dimensions of social stratification—educational attainment, occupational status, and particularly income, which was not nearly so stable—might bear a more substantial relationship to change in personality. To assess these possibilities, we created change scores for social stratificational position and each of its component dimensions, and calculated their correlations with change in self-directedness of orientation,

150

melvin l. kohn et al.

distress, and intellectual flexibility. Change in none of the component dimensions of social stratification (nor in the overall index of socialstratificational position) is significantly related to change in either selfdirectedness of orientation or intellectual flexibility, either for men or for women. Change in social-stratificational position or in one or more of its components is significantly, albeit only modestly, related to change in distress—with an increase in educational level being significantly related to a decrease in distress for men and change in relative income significantly related to change in distress for women. Even these relationships are only modest in magnitude. The findings for change in social-stratificational position are thus entirely consistent with those for inter-class mobility. The main lesson of our analyses of both inter-class mobility and change in socialstratificational position (and in its component dimensions) is that—for the employed—social mobility does little to explain the consistent relationships between social structure and personality in the face of extreme social change and instability of personality. Only for movement to and from the ranks of the non-employed does social mobility help explain the consistency of relationships between structural location and personality. By the same token, only for movement from the ranks of the employed to the nonemployed, and vice versa, does social mobility help to explain instability of personality. In fact, instability of personality is just as pronounced for the continuously employed portion of the Ukrainian population as for those who were unemployed at the time of one or the other of our surveys. What, then, does account for instability of personality? And what accounts for the over-time consistency in the relationships of social structure and personality for that vast majority of the adult Ukrainian population who were employed at the times of both the initial and follow-up interviews? The hypothesis that would have answered both questions had it been valid having failed, we offer a pair of related hypotheses, both of them based on extrapolations from longitudinal studies done under conditions of social stability. Stability of Social Conditions and Instability of Personality A central finding of longitudinal analyses of U.S. men, going back to a prototypic analysis of the reciprocal effects of the substantive complexity of work and intellectual flexibility (Kohn and Schooler 1978; 1983, Chapter 5), and greatly extended in analyses of occupational

a study of ukraine in transition

151

structure and personality (Kohn and Schooler 1982; 1983, Chapter 6) is that stability of social conditions helps account not only for the small amount of change in personality found in those studies, but also for the much greater amount of stability in personality (see the extended discussion with regard to cognitive functioning in Kohn 1996). Our hypothesis is that the obverse is also true: that under conditions of radical social change, unstable social conditions are conducive to instability of personality. If so, our quest to find linkages between social mobility and changes in personality was doomed to failure, precisely because instability of personality is not linked to social mobility, nor to social structure, but to unstable social conditions. Absent any contrary evidence (and we have searched assiduously for such evidence), we infer that conditions of change and uncertainty must have been so widespread in Ukraine during the years of our study as to produce a great deal of instability in people’s personalities regardless of their positions in the social structure and regardless of whether they were socially mobile or immobile. The extreme uncertainty about what would be the future of a Ukraine that was no longer socialist and no longer part of the Soviet Union, but was not firmly established as a capitalist, fully independent, and democratic society must have affected nearly everyone in the country, regardless of class position or social status. Although this pervasive uncertainty must have contributed, perhaps decisively, to the instability of personality, it must have been so widely experienced that it did not much affect the relationships between social structure and personality—except for that (small) proportion of the populace who experienced radical change in the most direct and compelling way, by moving to or from the ranks of the employed. This, we readily admit, is only inference—entirely consonant with our data but by no means demonstrated by our analyses. All we can say with certainty on this issue is that we have unequivocally shown a degree of instability of personality undreamed of in past analyses—our own or anyone else’s—and, despite that, overtime consistency in the relationships of social structure and personality during this period of extreme social change in Ukraine. Moreover—and crucially—the relationships of social structure and personality are consistent with those found for other countries during periods of apparent social stability.

152

melvin l. kohn et al.

The Role of Proximate Conditions of Life in Explaining the Over-Time Consistency in the Relationships of Social Structure and Personality for the Employed Segments of the Population Another central finding of longitudinal analyses of U.S. men, and also of other studies conducted in diverse countries during periods of social stability, is that position in the larger social structure is linked to personality primarily through the proximate conditions of life, job conditions in particular, attendant on social-structural position. The cross-sectional study of Poland and Ukraine in transition (Kohn et al. 1997) suggests that, even at a time of radical social change, position in the larger social structure is closely linked to the same proximate conditions of life that studies conducted during times of greater social stability have consistently shown to be the principal mechanisms by which socialstructural position affects personality. Our hypothesis is that stability in crucial proximate conditions of life largely accounts for continuity in the relationship of social structure and personality even during the ongoing process of radical social change. One proximate condition—the substantive complexity of work—has proved to be particularly pertinent for explaining the relationships of both social class and social stratification with personality, not only during times of apparent social stability in socialist Poland and in the capitalist United States and Japan (Kohn et al. 1990), but also in transitional Poland and, to a lesser extent, in Ukraine at a very early stage of transition (Kohn et al. 1997). By the substantive complexity of work, we mean the degree to which performance of the work requires thought and independent judgment. Substantively complex work by its very nature requires making many decisions that must take into account ill-defined or apparently conflicting contingencies. We hypothesize that the substantive complexity of work has continued to play a key role in the relationships of class and stratification with intellectual flexibility and self-directedness of orientation during the process of radical social change. To play such a role, the substantive complexity of work would have had to continue to be linked both to position in the social structure and to these dimensions of personality, at least as strongly in 1996 as in 1992–93. Moreover, to help account for the consistency in the relationships of class and stratification with self-directedness of orientation and intellectual flexibility, the substantive complexity of work would itself have had to be highly stable, whatever changes were occurring in the economic structure of the country or in general social conditions.

a study of ukraine in transition

153

Table 9. Longitudinal measurement model of the substantive complexity of work in paid employment, Ukraine (1992–93 and 1996) Concept and Indicators

Standardized Paths: Concept to Indicators Men 1992–93

1996

1992–93

1996

.87* .35* .81* .58* –.72* .33* .79*

.89* .31* .86* .66* –.47* .37* .83*

.90* .45* .75* .56* –.59* .25* .85*

.90* .52* .88* .54* –.47* .21* .95*

Substantive Complexity of Work Complexity of work with data Complexity of work with things Complexity of work with people Hours of work with data Hours of work with things Hours of work with people Overall complexity Over-time Correlation of the Concept

.83*

Ratio of Chi-Square to Degrees of Freedom Root Mean Square Error of Approximation Number of Cases

Women

.81* 2.80 .07

313

363

*p <.05.

Our measurement model of the substantive complexity of work in paid employment (see Table 9) is a longitudinal extension of the cross-sectional model developed for transitional Ukraine by Kohn et al. (1997, Table A-3). Our information about the substantive complexity of work is based on detailed questioning of the respondents about their work with data, things, and people. These questions provide the basis for seven ratings of each person’s job: our appraisals of the complexity of that person’s work with data, with things, and with people; our appraisal of the most complex work that the respondent ordinarily does, regardless of whether that work is in the domain of work with data, with people, or with things; and the respondent’s estimates of the time spent working at each of the three types of activity. We treat the seven ratings as indicators of the underlying but not directly measured concept, the substantive complexity of that job. The model fits the data reasonably well, with a ratio of chi-square to degrees of freedom of 2.80 and a root mean square error of approximation of 0.07.

154

melvin l. kohn et al.

This measurement model demonstrates one key fact essential for the hypothesis to be valid—that the over-time correlations of the substantive complexity of work were very high (0.83 for men and 0.81 for women). Another key set of conditions essential for the hypothesis to be valid is that the substantive complexity of work be strongly and meaningfully correlated with both social class and social stratification, both in 1992–93 and in 1996. These conditions are unequivocally met (see Table 10). The etas for the relationships of social class with the substantive complexity of work for employed men and for employed women hover close to 0.80 both for 1992–93 and for 1996. The social classes whose members rank highest in the substantive complexity of their work are the (combined category of) managers and employers, and the experts; those who rank lowest are the manual workers. (Certainly, no surprise.) The correlations of social-stratificational position with the substantive complexity of work for employed men and women are of approximately the same magnitudes as those for social class. One could not imagine more consistent figures—consistent across both time and gender, and highly consistent with comparable figures for other countries during times of apparent social stability.12 Here, then, is a likely source of stability in the relationships of class and stratification with self-directedness of orientation and intellectual flexibility, assuming of course that the substantive complexity of work continued to bear a similar relationship to these dimensions of personality for Ukraine in 1996 as in 1992–93—as it, in fact, did. The definitive test of the hypothesis is to statistically control the substantive complexity of work, to assess the degree to which the correlations (etas for social class, Pearsonian r’s for social stratification) are thereby reduced. Insofar as they are, to that extent are the relationships of class and stratification with these dimensions of personality attributable to the impact of social-structural position on the substantive complexity of work, which in turn affects these dimensions of personality. For the hypothesis to be supported, the percentage reductions in the relationships of both class and stratification with self-directedness

12 The high over-time stabilities of both social-stratification position and the substantive complexity of work provide definitive evidence that the much lower over-time stabilities of our indices of personality cannot be an artifact of some general methodological artifact. Moreover, the strong correlations of both social class and social stratification with the substantive complexity of work belie any thought that the weaker correlations of social structure with personality are somehow an artifact of random measurement error.

*p <.05.

Correlation (eta) Correlation, socialstratificational position and substantive complexity of work

Managers & Employers Self-employed Supervisors Experts Non-manual Workers Manual Workers

Social Class

(16) (7) (66) (16) (9) (167)

(1992–93) (15) (16) (66) (19) (5) (161)

(1996)

Number of Cases

.79* .78*

1.53 – .73 1.31 – –.57

1992–93

.82* .76*

1.63 –.24 .71 1.44 – –.60

1996

Substantive Complexity of Work

Men

(2) (6) (62) (45) (82) (117)

(1992–93) (11) (9) (72) (41) (74) (113)

(1996)

Number of Cases

.82* .81*

– – .70 1.00 .28 –.94

1992–93

.81* .81*

.77 – .70 .92 .31 –.98

1996

Substantive Complexity of Work

Women

Standardized Differences from the Mean for all People of the Particular Gender in the Pertinent Sample

Table 10. Social structure and the substantive complexity of work in paid employment for Ukrainian men and women employed both in 1992–93 and in 1996

a study of ukraine in transition 155

156

melvin l. kohn et al.

of orientation and intellectual flexibility must be at least as great for 1996 as for 1992–93. If so, then combined with the very high stability of the substantive complexity of work, it would be a fair inference that the substantive complexity of work contributes substantially to the stability of the relationships of social structure and these dimensions of personality (for the employed) even under continuing conditions of radical social change and despite massive instability of personality. Given our findings to this point, we have no similar hypothesis for distress, only puzzlement as to why class and stratification were not significantly related to distress either in 1992–93 or in 1996, as they were for Poland in 1992. Our findings for self-directedness of orientation and intellectual flexibility are generally consonant with the hypothesis (see Table 11). The percentage reductions in magnitude of correlation for 1992–93 range from a modest 19% to a substantial 51%, with most of the percentage reductions in the 20’s and 30’s. The percentage reductions generally increase by 1996, some of them (particularly those for intellectual flexibility) to the 60’s. A puzzling exception is that the percentage reduction for the correlation of social class with self-directedness of orientation for women drops from 51% in 1992–93 to 15% in 1996; but there is no such diminution in the relationship of social stratification with self-directedness of orientation for women, nor is there a diminution of percentage reduction in the relationships of either class or stratification with self-directedness of orientation for men. Notwithstanding this single exception, we infer that the consistency of relationships of class and stratification with self-directedness of orientation and intellectual flexibility is partly attributable to the continuing role of the substantive complexity of work in bridging people’s positions in the larger social structure and these dimensions of their personalities.13 We believe that other proximate conditions of life associated with social-structural position must similarly contribute to the consistency of relationships of class and stratification with self-directedness of orientation. There is some evidence as well that the degree to which the relationships of class and stratification with intellectual flexibility are attributable to the substantive complexity of work was increasing from 1992–93 to 1996, 13 Unfortunately, we do not have the data to assess whether the finding for Poland (in Kohn et al. 2002, Table 6) that the relatively low level of complexity of the activities of the unemployed and the pensioners contributed substantially to explaining their relatively low levels of self-directedness of orientation and intellectual flexibility holds also for the Ukrainian unemployed and pensioners.

a study of ukraine in transition

157

Table 11. The degree to which the relationships of social class and social stratification with personality are attributable to the substantive complexity of work, Ukraine (1992–93 and 1996) Self-directedness of Orientation

A. MEN Correlation (eta) with social-class position Percent reduction in eta when the substantive complexity of work is statistically controlled Correlation with social stratificational position Percent reduction in the correlation when the substantive complexity of work is statistically controlled B. WOMEN Correlation (eta) with social-class position Percent reduction in eta when the substantive complexity of work is statistically controlled Correlation with social stratificational position Percent reduction in the correlation when the substantive complexity of work is statistically controlled *p <.05.

Distress

Intellectual Flexibility

1992–93

1996

1992–93 1996 1992–93 1996

.26*

.30*

.15

.16

.24*

.62*

19%

60%

46%

5%

24%

61%

.23*

.43*

–.15*

–.10

.51*

.84*

28%

49%

66%

11%

32%

34%

.25*

.33*

.14

.18

.26*

.46*

51%

15%

27%

21%

40%

67%

.36*

.35*

–.03

–.14

.51*

.59*

20%

42%

–

91%

38%

64%

158

melvin l. kohn et al.

which helps explain the increasing magnitudes of relationship of social structure and intellectual flexibility. The preceding argument is, of course, predicated on the assumption that the substantive complexity of work is not merely correlated with, but has a continuing effect on, intellectual flexibility and self-directedness of orientation even during a prolonged period of radical social change. The ultimate test of this assumption is to replicate for Ukraine the prototypic model of the reciprocal effects of social structure and personality of the longitudinal U.S. analyses—Kohn and Schooler's causal model of the reciprocal effects of the substantive complexity of work and intellectual flexibility (Kohn and Schooler 1978; 1983, Chapter 5). We do not have the instrumentation to separately assess the crosslagged and contemporaneous reciprocal effects of the substantive complexity of work and intellectual flexibility for Ukraine, but we can do the next best thing, which is to assess their total effects on each other, whether those effects be contemporaneous, lagged, or a combination of the two. To be able to make precise comparisons of the U.S. and Ukrainian models, we modify the original U.S. model to do the same,14

14 The prototypic U.S. model (Kohn and Schooler 1978, Figures 4 and 5; Kohn and Schooler 1983, Figures 5.4 and 5.5) allowed both cross-lagged effects (of the substantive complexity of work at time-1 on intellectual flexibility at time-2, and of intellectual flexibility at time-1 on the substantive complexity of work at time-2) and contemporaneous effects (the effects of each at time-2 on the other at the same time). This model enabled us not only to assess the magnitudes of the effects of the substantive complexity of work and intellectual flexibility on each other, but also to depict the timing of those effects. Although we do not have the instrumentation to separate crosslagged from contemporaneous effects in the Ukrainian data, we can assess the total effects by not allowing the cross-lagged effects, thereby using the cross-lagged effects of the substantive complexity of work at time-1 on intellectual flexibility at time-2 and of intellectual flexibility at time-1 on the substantive complexity of work at time-2 as instruments to identify the contemporaneous effects, which are thereby converted into total effects. By thus modifying the U.S. model, as we do in Figure 1, we have converted the strong cross-lagged effect of intellectual flexibility at time-1 on the substantive complexity of work at time-2, and the strong contemporaneous effect of the substantive complexity of work at time-2 on intellectual flexibility at that same time, as depicted in the original model, into strong total effects of each on the other. We lose the ability to show that the effects of intellectual flexibility on the substantive complexity of work are predominantly lagged, and the effects of the substantive complexity of work on intellectual flexibility are predominantly contemporaneous, but we preserve the essential finding that the total effects of each on the other are moderately strong and statistically significant. (For a more extended discussion of the logic of using cross-lagged effects as instruments to identify total effects, see Kohn and Slomczynski 1990:127–131; see also the brief discussion in Chapter 8 of Kohn 2006.)

a study of ukraine in transition

159

and use the modified U.S. model as a baseline against which to assess the models for Ukrainian men and women.15 Our hypothesis in this comparison stems directly from the preceding analyses: that the relationships between the substantive complexity of work and intellectual flexibility (for both men and women) would be similar to that in the U.S. model (for men), except for a much lower stability of intellectual flexibility: Thus, we hypothesized a high stability of the substantive complexity of work, just as in the U.S. model; a much lower stability of intellectual flexibility in the Ukrainian models than in the U.S. model; and, just as in the U.S. model, strong reciprocal effects of the substantive complexity of work and of intellectual flexibility on each other. The first half of this hypothesis, about the stabilities of the substantive complexity of work and of intellectual flexibility, is precisely what the models do show (see Figure 1). The second half of this hypothesis falls short of what the models clearly demonstrate: Contrary to our expectation, the Ukrainian models do not show strong reciprocal effects of the substantive complexity of work and intellectual flexibility, but strong effects of the substantive complexity of work on intellectual flexibility, and only weak, statistically non-significant effects of intellectual flexibility on the substantive complexity of work. Our hypothesis had built on the evidence of stability in the substantive complexity of work, but had not fully taken into account that unstable facets of personality could not greatly affect job conditions. It is not that the substantive complexity of work and intellectual flexibility continue to re-enforce each other even during a period of intense social change, but that the substantive complexity of work continues to affect intellectual flexibility, and thereby to link social-structural position to intellectual flexibility.

15 The U.S. model is only for men, since Kohn and Schooler did not have the data to estimate a comparable model for women. The Ukrainian data being based on representative samples of men and women, we are able to estimate models for both genders. The U.S. model also included as exogenous variables the substantive complexity of work at an earlier time than the baseline interview (based on retrospective accounts of earlier jobs), as well as several social characteristics, some of which were used as instruments. In the Ukrainian models, we lack the requisite information to measure the substantive complexity of work at a time prior to the baseline interview (which is the primary reason why we cannot separately identify cross-lagged and contemporaneous effects). We do include some of the same social characteristics as in the U.S. model (e.g., age, religious background, mother’s and father’s educational attainment) but substitute other social characteristics more appropriate to Ukraine (such as whether the respondent’s native tongue is Ukrainian or Russian) for some of the social characteristics that were used in the model for the United States (such as region of the country in which the respondent was raised).

160

melvin l. kohn et al. Substantive Complexity 1964 .36*

.76*

Education .52* Intellectual Flexibility 1964

.43* .03

.68* .35*

Education .42* Intellectual Flexibility 1992–93

.25*

–.01 .68*

.60* .20

.50*

.18

UKRAINIAN MEN RMSEA = 0.053 N = 384 Substantive Complexity 1992–93 .65* .51*

Education .33* Intellectual Flexibility 1992–93

.59* .13

.39*

.25

UKRAINIAN WOMEN RMSEA = 0.043 N = 320

–.08 u

u

.19

Intellectual Flexibility 1996 * = P < .05

u

(Correlation of substantive complexity of work and intellectual flexibility in 1996 = .75*)

Substantive Complexity 1996

.24

u

.09

Intellectual Flexibility 1996 * = P < .05

–.06*

(Correlation of substantive complexity of work and intellectual flexibility in 1974 = .80*)

Substantive Complexity 1996

.30*

u

.46*

Intellectual Flexibility 1974 * = P < .05

U.S. MEN RMSEA = 0.053 N = 687 Substantive Complexity 1992–93

Substantive Complexity 1974

–.03 u

(Correlation of substantive complexity of work and intellectual flexibility in 1996 = .72*)

Figure 1. The Reciprocal Effects of the Substantive Complexity of Work and Intellectual Flexibility: U.S. Men, Ukrainian Men, and Ukrainian Women.

That the substantive complexity of work affects (and hardly, if at all, reflects) intellectual flexibility well into a period of radical social change confirms the essential premise on which this entire analysis of radical social change has been based, although not quite in the way that we had expected. The continuing relationships between the substantive complexity of work and intellectual flexibility are built on the base of employed men and women continuing to hold jobs of essentially similar

a study of ukraine in transition

161

levels of complexity even during a time of considerable change in the socio-economic structure of the country. This is clearly a key to the continuing relationship of social structure and intellectual flexibility even in the face of instability of what under other social conditions is an exceptionally stable facet of personality. The model thus shows how the continued stability of the substantive complexity of work sustains the relationship between social structure and personality even when personality is itself unstable.16 Perceived Economic Well-being What, then, explains the (consistent) lack of relationship of class and stratification with distress for the employed? Table 11 sheds little light on this question, for the issue is not what explains the relationships of class and stratification with distress, but why the relationships of class and stratification with distress are so weak. A partial answer may be given in the zero-order r’s of the substantive complexity of work with distress, which are too weak to provide much of a linkage from social structure to distress. Presumably, if the correlations of the substantive complexity of work and distress were stronger, so too would be the correlations of class and stratification with distress. But this partial explanation does not explain why the non-employed segments of the population were more distressed than the employed, thus making the relationship of structural location and distress more powerful than that between social class and distress.

16 One last comment on the comparability of the models for the U.S. men, Ukrainian men, and Ukrainian women: As Figure 1 shows, the correlation of the residuals of time2 substantive complexity and time-2 intellectual flexibility is small (–0.06), negative, but statistically significant. When a correlation of residuals of reciprocally-related variables is statistically significant, that correlation should certainly be a parameter in the model, and not fixed at zero. The comparable correlations for Ukrainian men and women are also small and negative, but not statistically significant. It is our understanding that retaining non-significant correlations of residuals (rather than fixing them at zero) is a judgment call, but also that negative correlations should be retained if at all possible. Both because we wanted to make the Ukrainian models as precisely comparable to the U.S. model as possible, and because those correlations are negative, we think it desirable to keep those parameters open. If, arguendo, they were fixed at zero the models would yield the same conclusions as above, albeit with the paths from intellectual flexibility to the substantive complexity of work even weaker than when the model includes that correlation of residuals (at 0.14 for women and –0.06 for men, the latter both theoretically implausible and statistically anomalous because it is of opposite sign to the reciprocal path from the substantive complexity of work to intellectual flexibility).

162

melvin l. kohn et al.

On the basis of the findings from the cross-sectional analyses of the nonemployed (Kohn et al. 2002), we speculated that, during this period of economic turmoil in Ukraine, the effect of the substantive complexity of work on feelings of well-being or distress might have been overshadowed by those of another proximate condition of life, a sense of economic adversity. Our index is based on a confirmatory factor-analytic measurement model of perceived economic well-being (or lack thereof ), which we think of as the subjective face of economic adversity (see Table 12). Our longitudinal measurement model is an extension of the crosssectional model developed for Ukraine in Kohn et al. (2002, Table 3). It has four indicators: the log of per capita household income (which ties perceptions to economic reality); the respondents’ reports about whether their households had experienced difficulties during the past year in buying food, paying bills, providing for entertainment, or having funds for taking a vacation; their assessments of whether their financial situations had improved or gotten worse during the past three years; and their degree of satisfaction or dissatisfaction with their financial situations. The model fits the data well, the ratio of chi-square to degrees of freedom being 0.65 and the root mean square error of approximation being 0.00. The over-time correlation of the concept, as one might expect in economically precarious times, is only about half of that for the substantive complexity of work: 0.43 for men and 0.40 for women. Using factor scores based on this measurement model, we find that the correlations of perceived economic well-being or duress with distress were stronger than those of the substantive complexity of work with distress (see Table 13). These analyses also show that, although perceived economic well-being was meaningfully related to social-class position, with managers and employers having the strongest sense of economic well-being and (among the employed) manual workers having the greatest sense of economic duress, perceived economic well-being was not nearly so strongly correlated with class position and stratificational level as was the substantive complexity of work. Thus, instead of linking class and stratification with distress, the sense of economic well-being or adversity contributed little to those relationships and may even have weakened them, by displacing the substantive complexity of work in influencing people’s feelings of well-being or distress. Perceived economic well-being did, however, serve to link structural location to distress, because some non-employed or partially employed segments of the population who in 1996 suffered a strong sense of economic duress—in particular, the unemployed and the partially employed (who were employed

a study of ukraine in transition

163

Table 12. Longitudinal measurement model of perceived economic well-being, Ukraine (1992–93 and 1996) Indicators

Standardized Paths: Concept to Indicators Men 1992–93

Women 1996

1992–93

1996

Per capita household income (logged)

.35*

.49*

.29*

.44*

Respondent’s reports on whether his (or her) household experienced difficulties in past year in buying food, paying bills, providing for entertainment, or taking a vacation

–.63*

–.66*

–.60*

–.54*

Respondent’s assessment of whether his (or her) financial situation had improved or gotten worse during the past 3 years

.62*

.41*

.55*

.54*

Respondent’s satisfaction with his (or her) financial situation

.53*

.63*

.56*

.58*

Over-time correlation of the concept

.43*

Ratio of chi-square to degrees of freedom Root mean square error of approximation Number of cases

.40* 0.65 0.00

368

456

*p <.05.

Table 13. Correlations of the substantive complexity of work and of perceived economic well-being or duress with feelings of distress, Ukraine (1992–93 and 1996) Men

Women

1992–93

1996

1992–93

1996

–.16*

–.05

–.15*

–.18*

–.26*

–.28*

–.33*

–.27*

Correlations with Distress of: Substantive Complexity of Work Perceived Economic Wellbeing or Duress *p <.05.

164

melvin l. kohn et al.

less than 15 hours per week)—were more distressed than were most people of their gender. They thus stood in sharp contrast to the most economically advantaged segments of the employed population—in particular, the employers and managers, and the male experts—who had a strong sense of economic well-being and were also among the least distressed of their gender. Conclusions Limitations of the Analysis This study was initiated at a time when it had become apparent that Ukraine was still very much in process of transformation but it seemed that a follow-up study had better be undertaken as quickly as possible if the opportunity to do such a study were not to be lost. There was no time to apply for additional funding for fieldwork, and so fieldwork was supported mainly out of the revenues of the Ukrainian investigators’ survey organization. This precluded attempting to re-interview everyone interviewed in the initial survey. The investigators decided to limit the follow-up survey to the respondents who were in the labor force at the time of the 1992–93 survey, a strategically useful decision but one that precludes our being able to do longitudinal analyses of the non-employed. We do not think this imperils the validity of our analyses of those people who became non-employed in the interval between the initial and follow-up surveys, but it prevents us from assessing the full impact of the transformation on the non-employed. The follow-up survey having been planned before the completion of the cross-sectional analyses of the complexity of activities of the non-employed (Kohn et al. 2000; 2002), the evidence was not yet in place that the complexity of non-work activities is as strongly related to personality for the non-employed as is the substantive complexity of work for the employed. The follow-up survey might otherwise have secured information about the complexity of non-work activities for those people who became unemployed, housewives, or pensioners during the interval between surveys. Lacking this information, we are unable to assess the role of complexity of activities of the non-employed in explaining the relationships between structural location and personality, as we did the role of the complexity of work in paid employment in explaining the relationships of social class and social stratification with personality for the employed. Our interpretation is thus less encompassing than it might have been.

a study of ukraine in transition

165

Summary and Interpretation Our analyses have yielded two principal findings. One is that during a period of three to three and a half years of ongoing radical social change attendant on the transformation of Ukraine, there was extreme instability in urban men’s and women’s self-directedness of orientation and sense of well-being or distress, and (as best we can judge) lesser but still substantial instability even in what is ordinarily the most stable facet of personality—intellectual flexibility. So unexpected was this instability of personality that we devoted considerable effort to verifying that the finding was not merely an artifact of some defect in method. We are convinced, and have presented evidence (in the text and in notes 5 and 12), that it was not. Our other principal finding is that, despite radical social change and despite instability of personality, the relationships of social class and social stratification with personality, which at the time of the baseline survey of 1992–93 were generally similar in pattern albeit somewhat weaker in magnitude than those for Poland in transition at that same time, or than those for Poland when it was socialist, or than those for the capitalist United States and Japan, remained similar in pattern after three to three-and-a-half tumultuous years, and some may have increased in magnitude. The one striking exception was that the relationships of class and stratification with distress, although consistent over time, were consistently weak. But, when the non-employed segments of the population are included in the analysis, there is a substantial and significant relationship between structural location and distress, consistent for 1992–93 and for 1996—because the non-employed were consistently more distressed than the employed. Much of the analysis of this paper has been an attempt to reconcile what, on its face, would seem to be an incompatibility in these findings. If personality was so unstable, how could the relationships of social structure and personality be consistent over time? We initially hypothesized that the extraordinary instability of personality that we had found might be linked to social mobility—that people who moved from more advantaged to less advantaged structural locations might have experienced increased distress, a decline in self-directedness of orientation, perhaps even a decline in intellectual flexibility; and that people who moved from less advantaged to more advantaged structural locations might have experienced an increased sense of well-being, greater self-directedness of orientation, and increased intellectual flexibility. At the extreme—the extreme being mobility into and out of the ranks of the employed—we did find evidence in support of this

166

melvin l. kohn et al.

hypothesis, in that such movement was substantially and significantly related to consonant change in self-directedness of orientation and distress, albeit not to any significant change in intellectual flexibility. But, for the great majority of urban adults who were employed both in 1992–93 and in 1996, there was little evidence of any substantial linkage between social mobility and change in personality. Try as we might, we could not find any substantial or statistically significant relationship between inter-class mobility and change in personality. As for change in social-stratificational position, the key fact is that, notwithstanding great change in the social and economic structure of the country, and notwithstanding a substantial amount of inter-class mobility, there was very little change in overall social-stratificational position. The most that one can conclude is that change in social-stratificational position or in its component dimensions is significantly but only modestly related to change in feelings of well-being or distress, albeit not to change in self-directedness of orientation or in intellectual flexibility. Moreover, different components of social stratification are significantly related to change in distress for men and for women. One could hardly conclude from such findings that we have shown instability of personality to be closely or consistently linked to social mobility. What, then, explains the consistently significant relationships of social class and social stratification with self-directedness of orientation and intellectual flexibility during a period of three to three and a half years of radical social change and despite great instability of personality? Our hypothesis was that the same proximate conditions of life that have been shown to link position in the larger social structure with personality in prior studies of capitalist and socialist, Western and non-Western, societies and—especially pertinent here—in studies conducted during periods of apparent social stability and of radical social change, continued to play this bridging role throughout this period of Ukraine’s transformation. The underlying premise of this hypothesis, of course, is that massive social change as experienced during this period of time in Ukraine did not affect the linkages of structural location to people’s immediately impinging conditions of life. In support of this premise, we found that class and stratification were as closely linked to the substantive complexity of work during the extended period of radical social change in Ukraine as they have been shown to be in other societies during times of apparent social stability. And the substantive complexity of work was linked to self-directedness of orientation and to intellectual flexibility just as in other counties at more stable times, although perhaps not quite as strongly.

a study of ukraine in transition

167

Moreover—a crucial ingredient in explaining the stability of relationships of social structure and personality despite social change and instability of personality—we hypothesized that the crucial proximate conditions of life were themselves highly stable. Our study examined one key condition, one that has been shown in study after study to be important for explaining the relationships of social class and social stratification to personality—the substantive complexity of work in paid employment. Its stability even during the three to three-and-a-half years of radical social change and instability of personality encompassed in our study proved to be very high—both for men and for women. And, since we also find that the substantive complexity of work accounts for a modest-to-substantial portion of the relationships of both social class and social stratification with self-directedness of orientation and, to a greater extent, with intellectual flexibility, we believe that we have found an important component of the explanation for consistency in the relationship of social structure and personality even when personality is itself unstable. It lies in the stability of the proximate conditions of people’s lives, which remain strongly linked to position in the larger social structure, and which continue to be the bridge from position in the larger social structure to self-directedness of orientation and intellectual flexibility, albeit not to distress. The ultimate test, as far as the substantive complexity of work is concerned, was to replicate the causal model of the substantive complexity of work and intellectual flexibility that Kohn and Schooler had estimated for U.S. men as the prototypic assessment of the reciprocal effects of job conditions and personality. If the above rationale is valid, then the substantive complexity of work should affect the intellectual flexibility of Ukrainian men and women during this period of radical social change much as it had affected the intellectual flexibility of U.S. men during an extended period of social stability. Our hypothesis was also premised on the assumption that the stability of the substantive complexity of work, even in a model that statistically controls the effects of intellectual flexibility, educational attainment, and pertinent social characteristics would be strong for Ukrainian men and women, while that of intellectual flexibility would necessarily be much weaker than in the U.S. model. Thus far, the findings were precisely as hypothesized. What we did not expect, but what in retrospect seems altogether plausible, was that the reciprocal effect of intellectual flexibility on the substantive complexity of work would be small and statistically non-significant. But how could it be otherwise? We put the question this way not only because the mathematics of the equations so dictate, but also because

168

melvin l. kohn et al.

the social psychology of the situation makes the outcome inevitable. If personality is unstable, how can personality (in this instance, the dimension of personality we call intellectual flexibility) affect the individual’s ability either to modify his or her job conditions or to affect processes of promotion, retention, and job selection? Even taking the equivocal relationship of social structure and distress fully into account, the major lesson we learn from the longitudinal analyses of Ukraine—a lesson that could never have been gleaned from any cross-sectional study – is one of stability despite change: stability in the relationships of social structure and personality despite ongoing massive change in the social and economic structure of the country, and despite massive instability in the personalities of its adult populace. Cross-national comparativists though we be, we face the very same problem of generalizability of our findings that every case study faces: From the study of a single country undergoing an exceptionally uncertain and rather prolonged period of radical social change, we can hardly conclude that the relationships of social structure and personality would necessarily be the same in other countries experiencing radical social change, particularly countries experiencing other types of transformation. Might a study of a country experiencing radical social change under different political or cultural or economic conditions—say, a study of China—yield different findings? Certainly, it might. Our findings challenge authors and readers alike to design and carry out such studies. We can safely conclude from our case study of Ukraine, though, that the general pattern of relationships between social structure and fundamental dimensions of personality that has now been found in diverse countries—capitalist and socialist, Western and non-Western, during times of apparent social stability and of radical social change—has been found again even in the midst of an extended period of profound social change and despite extreme instability of personality. Moreover, the explanation for this consistency of relationship is much the same under conditions of radical social change as during times of apparent social stability. Even a single instance of a country evidencing such great consistency in the relationships of social structure and personality under such extreme conditions is enough to demonstrate that these relationships are remarkably robust.

a study of ukraine in transition

169

Acknowledgments Our research was supported by the State Committee for Science and Technology of Ukraine, the National Science Foundation [Grants #SES–9107584 and #SRB–9728374], and the National Council for Soviet and East European Research (now the National Council for Eurasian and East European Research). We are indebted to Leonid Finkel for help in devising the sampling frame for Ukraine; to Andrzej Wejland for teaching the Ukrainian investigators his methods of intensive pretesting; to Andrzej Wejland and Roman Lentchovskii for help in translating interview questions from Polish to Russian; to Michael Haney and Michael Swafford for training the Ukrainian interviewers; to our Polish collaborators—Kazimierz M. Slomczynski, Wojciech Zaborowski, Krystyna Janicka, and Bogdan Mach—for conducting the surveys on which the Polish measurement models in Table 2 are based, and to Bogdan Mach for developing those models; to Ronald Schoenberg for sleuthing problems in our creation of factor scores; and to Mesfin Mulatu for guidance in creating the latent-variable scores. We are also indebted to Carmi Schooler and Mesfin Mulatu for their incisive critiques of an earlier draft of this paper, and to the faculty and graduate students of the Department of Sociology of the Johns Hopkins University for their stimulating discussion of the theoretical implications of this paper at a departmental seminar. References Costa Jr., Paul T., Jeffrey H. Herbst, Robert R. McCrae, and Ilene C. Siegler. 2000. “Personality at Midlife: Stability, Intrinsic Maturation, and Response to Life Events.” Assessment 7:365–378. Jöreskog, Karl G. 2000. “Latent Variable Scores and their Uses.” Accessible on the LISREL Web site, http://www.ssicentral.com/lisrel/column6.htm. Khmelko, Valeriy. 2002. “Macrosocial Change in Ukraine: The Years of Independence.” Sisyphus: Sociological Studies XVI:125–136. Kohn, Melvin L. 1969. Class and Conformity: A Study in Values. Homewood, IL: Dorsey (2nd edition, University of Chicago Press, 1977). ——. 1987. “Cross-National Research as an Analytic Strategy: American Sociological Association, 1987 Presidential Address.” American Sociological Review 52:713–731. ——. 1993. “Doing Social Research Under Conditions of Radical Social Change: The Biography of an Ongoing Research Project.” (1992 Cooley-Mead Award Lecture, American Sociological Association.) Social Psychology Quarterly 56:4–20. ——. 1996. “Review Essay: The ‘Bell Curve’ from the Perspective of Research on Social Structure and Personality.” 1996. Sociological Forum 11:395–411. ——. 2006. Change and Stability: A Cross-National Analysis of Social Structure and Personality. Boulder, CO: Paradigm Publishers.

170

melvin l. kohn et al.

Kohn, Melvin L., and Carmi Schooler. 1978. “The Reciprocal Effects of the Substantive Complexity of Work and Intellectual Flexibility: A Longitudinal Assessment.” American Journal of Sociology 84( July):24–52. ——. 1982. “Job Conditions and Personality: A Longitudinal Assessment of their Reciprocal Effects.” American Journal of Sociology 87(May):1257–1286. ——. 1983. “With the collaboration of Joanne Miller, Karen A. Miller, Carrie Schoenbach, and Ronald Schoenberg.” Work and Personality: An Inquiry into the Impact of Social Stratification. Norwood, NJ: Ablex. Kohn, Melvin L., and Carrie Schoenbach. 1983. “Class, Stratification, and Psychological Functioning.” Pp. 154–189 in Work and Personality: An Inquiry into the Impact of Social Stratification, by Melvin L. Kohn, and Carmi Schooler. Norwood, NJ: Ablex. Kohn, Melvin L., and Kazimierz M. Slomczynski, with the Collaboration of Carrie Schoenbach. 1990. Social Structure and Self-Direction: A Comparative Analysis of the United States and Poland. Oxford, England: Basil Blackwell. Kohn, Melvin L., Atsushi Naoi, Carrie Schoenbach, Carmi Schooler, and Kazimierz M. Slomczynski. 1990. “Position in the Class Structure and Psychological Functioning in the United States, Japan, and Poland.” American Journal of Sociology 95:964–1008. Kohn, Melvin L., Kazimierz M. Slomczynski, Krystyna Janicka, Valeri Khmelko, Bogdan W. Mach, Vladimir Paniotto, Wojciech Zaborowski, Roberto Gutierrez, and Cory Heyman. 1997. “Social Structure and Personality under Conditions of Radical Social Change: A Comparative Analysis of Poland and Ukraine.” American Sociological Review 62:614–638. Kohn, Melvin L., Wojciech Zaborowski, Krystyna Janicka, Bogdan W. Mach, Valeriy Khmelko, Kazimierz M. Slomczynski, Cory Heyman, and Bruce Podobnik. 2000. “Complexity of Activities and Personality Under Conditions of Radical Social Change: A Comparative Analysis of Poland and Ukraine.” Social Psychology Quarterly 63:187–208. Kohn, Melvin L., Wojciech Zaborowski, Krystyna Janicka, Valeriy Khmelko, Bogdan W. Mach, Vladimir Paniotto, Kazimierz M. Slomczynski, Cory Heyman, and Bruce Podobnik. 2002. “Structural Location and Personality during the Transformation of Poland and Ukraine.” Social Psychology Quarterly 65:364–385. McCrae, Robert R., Paul T. Costa, Jr., Fritz Ostendorf, Alois Angleitner, Martina Hrebickova, Maria D. Avia, Jesus Sanz, Maria L. Sanchez-Bernardos, M. Ersin Kusdil, Ruth Woodfield, Peter R. Saunders, and Peter B. Smith. 2000. “Nature Over Nurture: Temperament, Personality and Life Span Development.” Journal of Personality and Social Psychology 78:173–186. McGue, Matt, Steven Bacon, and David T. Lykken. 1993. “Personality Stability and Change in Early Adulthood: A Behavioral Genetic Analysis.” Developmental Psychology 29:96–109. Miller, Karen A., Melvin L. Kohn, and Carmi Schooler. 1986. “Educational SelfDirection and Personality.” American Sociological Review 51:372–390. Williams Jr., Robin M. 1970. American Society: A Sociological Interpretation. 3rd ed. New York: Alfred A. Knopf. Wright, Erik Olin. 1978. Class, Crisis and the State. London: New Left Books.

PART THREE

VALUES, CULTURE AND DEMOCRACY

A THEORY OF CULTURAL VALUE ORIENTATIONS: EXPLICATION AND APPLICATIONS Shalom H. Schwartz Recent years have seen the emergence of several theories and instruments for mapping and comparing national cultures. Interestingly, the main theories have emerged from different disciplines. Hofstede developed his theory of work values (e.g., 1980, 2001) to make sense of data gathered for purposes of management by IBM. The four and later five dimensions he derived to compare country cultures have been widely applied in the fields of business and management. Inglehart developed his theory of materialism-postmaterialism (e.g., 1977, 1990), which he later refined to include two dimensions (e.g., 1997, Inglehart and Baker 2000), in order to address issues in political science and sociology about the effects of modernization. My own theory emerged later than these two (Schwartz 1994b, 1999, 2004) out of my studies of individual differences in value priorities and their effects on attitudes and behavior, a sub-field of social psychology. This article presents my theory of seven cultural value orientations that form three cultural value dimensions. At the cost of greater complexity than the other theories, this theory permits more finely tuned characterization of cultures. To validate the theory, I present analyses of data from 73 countries, using two different instruments. Conceptual and empirical comparisons of the cultural value orientations with Inglehart’s two dimensions clarify their similarities and differences. Using the seven validated cultural orientations, I generate a worldwide map of national cultures that identifies distinctive cultural regions. The article then addresses the question of the antecedents that give rise to national differences on the cultural value dimensions. Finally, it examines some consequences of prevailing cultural value orientations on attitudes and behavior within countries. The approach presented here is distinctive in deriving the seven cultural orientations from a priori theorizing and then testing the fit of these orientations to empirical data. Moreover, the a priori theorizing specified a coherent, integrated system of relations among the orientations, which was then tested.

174

shalom h. schwartz Cultural Orientations—Basic Assumptions

I view culture as the rich complex of meanings, beliefs, practices, symbols, norms, and values prevalent among people in a society. The prevailing value emphases in a society may be the most central feature of culture (Hofstede 1980; Inglehart 1977; Schwartz 1999; Weber 1958; Williams 1958). These value emphases express shared conceptions of what is good and desirable in the culture, the cultural ideals. Cultural value emphases shape and justify individual and group beliefs, actions, and goals. Institutional arrangements and policies, norms, and everyday practices express underlying cultural value emphases in societies. For example, a cultural value emphasis on success and ambition may be reflected in and promote highly competitive economic systems, confrontational legal systems, and child-rearing practices that pressure children to achieve. The preference element in cultural value orientations—values as ideals—promotes coherence among the various aspects of culture. Because prevailing cultural value orientations represent ideals, aspects of culture that are incompatible with them are likely to generate tension and to elicit criticism and pressure to change. In a society whose cultural value orientations emphasize collective responsibility, for example, a firm that fires long-term employees in the interests of profitability is likely to elicit widespread criticism and pressure to change policies. Of course, cultures are not fully coherent. In addition to a dominant culture, subgroups within societies espouse conflicting value emphases. The dominant cultural orientation changes in response to shifting power relations among these subgroups. But change is slow. Another important feature of cultural value orientations is that they are relatively stable (Hofstede 2001; Schwartz et al. 2000). Some researchers argue that elements of culture persist over hundreds of years (e.g., Kohn and Schooler 1983; Putnam 1993). Yet, cultural value orientations do change gradually. Societal adaptation to epidemics, technological advances, increasing wealth, contact with other cultures, and other exogenous factors leads to changes in cultural value emphases. Culture joins with social structure, history, demography, and ecology in complex reciprocal relations that influence every aspect of how we live. But culture is difficult to measure. To reveal the cultural orientations in a society, we could look at the themes of children’s stories, at the systems of law, at the ways economic exchange is organized, or at socialization practices. These indirect indexes of underlying orientations

a theory of cultural value orientations

175

in the prevailing culture each describe a narrow aspect of the culture. When researchers try to identify culture by studying the literature of a society or its legal, economic, family, or governance systems, what they seek, implicitly or explicitly, are underlying value emphases (Weber 1958; Williams 1968). Therefore, studying value emphases directly is an especially efficient way to capture and characterize cultures. A Theory of Cultural Value Orientations Cultural value orientations evolve as societies confront basic issues or problems in regulating human activity. People must recognize these problems, plan responses to them, and motivate one another to cope with them. The ways that societies respond to these basic issues or problems can be used to identify dimensions on which cultures may differ from one another. The cultural value orientations at the poles of these dimensions are Weberian ideal-types. I derived value dimensions for comparing cultures by considering three of the critical issues that confront all societies. The first issue is the nature of the relation or the boundaries between the person and the group: To what extent are people autonomous vs. embedded in their groups? I label the polar locations on this cultural dimension autonomy versus embeddedness. In autonomy cultures, people are viewed as autonomous, bounded entities. They should cultivate and express their own preferences, feelings, ideas, and abilities, and find meaning in their own uniqueness. There are two types of autonomy: Intellectual autonomy encourages individuals to pursue their own ideas and intellectual directions independently. Examples of important values in such cultures include broadmindedness, curiosity, and creativity. Affective autonomy encourages individuals to pursue affectively positive experience for themselves. Important values include pleasure, exciting life, and varied life. In cultures with an emphasis on embeddedness, people are viewed as entities embedded in the collectivity. Meaning in life comes largely through social relationships, through identifying with the group, participating in its shared way of life, and striving toward its shared goals. Embedded cultures emphasize maintaining the status quo and restraining actions that might disrupt in-group solidarity or the traditional order. Important values in such cultures are social order, respect for tradition, security, obedience, and wisdom.

176

shalom h. schwartz

The second societal problem is to guarantee that people behave in a responsible manner that preserves the social fabric. That is, people must engage in the productive work necessary to maintain society rather than compete destructively or withhold their efforts. People must be induced to consider the welfare of others, to coordinate with them, and thereby manage their unavoidable interdependencies. The polar solution labeled cultural egalitarianism seeks to induce people to recognize one another as moral equals who share basic interests as human beings. People are socialized to internalize a commitment to cooperate and to feel concern for everyone’s welfare. They are expected to act for the benefit of others as a matter of choice. Important values in such cultures include equality, social justice, responsibility, help, and honesty. The polar alternative labeled cultural hierarchy relies on hierarchical systems of ascribed roles to insure responsible, productive behavior. It defines the unequal distribution of power, roles, and resources as legitimate. People are socialized to take the hierarchical distribution of roles for granted and to comply with the obligations and rules attached to their roles. Values like social power, authority, humility, and wealth are highly important in hierarchical cultures. The third societal problem is to regulate how people manage their relations to the natural and social world. The cultural response to this problem labeled harmony emphasizes fitting into the world as it is, trying to understand and appreciate rather than to change, direct, or to exploit. Important values in harmony cultures include world at peace, unity with nature, and protecting the environment. Mastery is the polar cultural response to this problem. It encourages active self-assertion in order to master, direct, and change the natural and social environment to attain group or personal goals. Values such as ambition, success, daring, and competence are especially important in mastery cultures. In sum, the theory specifies three bipolar dimensions of culture that represent alternative resolutions to each of three problems that confront all societies: embeddedness versus autonomy, hierarchy versus egalitarianism, and mastery versus harmony (see Figure 1). A societal emphasis on the cultural type at one pole of a dimension typically accompanies a de-emphasis on the polar type, with which it tends to conflict. Thus, as we will see below, American culture tends to emphasize mastery and affective autonomy and to give little emphasis to harmony. And the culture in Singapore emphasizes hierarchy but not egalitarianism and intellectual autonomy.

a theory of cultural value orientations

177

HARMONY Unity With Nature World at Peace

EGALITARIANISM Social Justice Equality

EMBEDDEDNESS Social Order, Obedience Respect for Tradition

INTELLECTUAL AUTONOMY

HIERARCHY

Broadmindedness Curiosity

Authority Humble

MASTERY AFFECTIVE AUTONOMY

Ambition Daring

Pleasure

Figure 1. Cultural dimensions: prototypical structure.

The cultural value orientations are also interrelated based on compatibility among them. That is, because certain orientations share assumptions, it is easier to affirm and act on them simultaneously in a culture. For example, egalitarianism and intellectual autonomy share the assumption that people can and should take individual responsibility for their actions and make decisions based on their own personal understanding of situations. And high egalitarianism and intellectual autonomy usually appear together, as in Western Europe. Embeddedness and hierarchy share the assumption that a person’s roles in and obligations to collectivities are more important than her unique ideas and aspirations. And embeddedness and hierarchy are both high in the Southeast Asian cultures I have studied. The shared and opposing assumptions inherent in cultural values yield a coherent circular structure of relations among them. The structure reflects the cultural orientations that are compatible (adjacent in the circle)

178

shalom h. schwartz

or incompatible (distant around the circle). This view of cultural dimensions as forming an integrated, non-orthogonal system, distinguishes my approach from others. Hofstede (1980, 2001) conceptualized his dimensions as independent. He assessed them as orthogonal factors. Inglehart (1997) derived his orthogonal dimensions empirically from a factor analysis of nation-level correlations among numerous attitudes and beliefs. Measuring Cultural Value Orientations I assume that the average value priorities of societal members point to the underlying cultural emphases to which they are exposed (Schwartz 2004). Like Hofstede (2001) and Inglehart (1997), I therefore infer the cultural value orientations that characterize societies by averaging the value priorities of individuals in matched samples from each society. My measurement of value priorities differs from that prevalent in survey research, however. I focus on basic values. Consensus regarding how to conceptualize basic values has emerged gradually since the 1950’s. It includes six main features (explicated more fully in Schwartz 2005a): (1) Values are beliefs that are linked inextricably to affect. (2) Values refer to desirable goals that motivate action. (3) Values transcend specific actions and situations (e.g., obedience and honesty are values that are relevant at work or in school, in sports, business, and politics, with family, friends, or strangers). This feature distinguishes values from narrower concepts like norms and attitudes that usually refer to specific actions, objects, or situations. (4) Values serve as standards or criteria that guide the selection or evaluation of actions, policies, people, and events. (5) Values are ordered by importance relative to one another to form a system of priorities. This hierarchical feature also distinguishes values from norms and attitudes. (6) The relative importance of values guides action. The tradeoff among relevant, competing values is what guides attitudes and behaviors (Schwartz 1992, 1996; Tetlock 1986). Implicitly, most survey researchers hold conceptions of values close to this one. Unlike the method I adopt below, however, many of the value items used in survey research are inconsistent with some of these features. The items often refer to specific situations or domains. They do not

a theory of cultural value orientations

179

measure ‘basic’ values in the sense of values that are relevant across virtually all situations. This affects value priorities. Consider the item ‘giving people more say in important government decisions’. Support for or opposition to the current government influences the importance respondents attribute to this goal (Israeli data from 1999). The meaning of such items depends on the interaction between people’s ‘basic’ values and the context and domain in which the items are measured. Researchers often combine responses to items from a number of specific domains in order to infer underlying, trans-situational values (e.g., materialism). But, because situation-specific items are sensitive to prevailing socio-political conditions, the choice of items may still substantially influence both group and individual-level priorities (e.g., Clarke et al. 1999). Contrary to features 5 and 6, many survey items do not measure values in terms of importance. Instead, they present attitude or opinion statements and employ agree-disagree, approve-disapprove, or other evaluative response scales. The researcher may then try to infer indirectly the importance of the values presumed to underlie these attitudes or opinions. But multiple values may underlie any given attitude or opinion. Hence, it is hazardous to infer basic value priorities from responses to specific attitude and opinion items. In order to discover basic values with this approach, one must ask numerous questions across many domains of content. One then searches for underlying consistencies of response that may or may not be present. Such an approach requires many items and may not discern clear sets of basic value priorities. Inglehart adopted this approach in deriving his two updated dimensions of culture. He describes the tradition/secular-rational dimension, for example, as centrally concerned with orientations toward authority (Inglehart and Baker 2000). He bases this on five items that load together in a factor analysis (importance of God, importance of obedience and religious faith for children, justifiability of abortion, sense of national pride, and attitude toward respect for authority). The secular/rational pole of this orientation is not measured directly. It is inferred from responses that reject these five items. The two items that load most strongly on this factor both concern religion. The broader meaning of this dimension is inferred from the correlations of the five-item index with various beliefs and attitudes. The meaning of such dimensions, derived by inference from correlations among diverse items rather than clearly defined and operationalized a priori, is necessarily loose.

180

shalom h. schwartz

To operationalize the value priorities of individuals, in one set of studies, I used the Schwartz Value Survey that includes 56 or 57 value items (SVS: Schwartz 1992; Schwartz and Boehnke 2004). These abstract items (e.g., social justice, humility, creativity, social order, pleasure, ambition) are each followed in parenthesis by a phrase that further specifies their meaning. Respondents rate the importance of each “as a guiding principle in MY life.” Respondents from cultural groups on every inhabited continent have completed the survey, anonymously, in their native language (N > 75,000).1 This survey is intended to include all the motivationally distinct values likely to be recognized across cultures, a claim for which there is growing evidence (Schwartz 2005a). Values whose meanings differ across cultures should not be used in cross-cultural comparison. Otherwise, group differences might reflect the fact that different concepts are measured in each group. Separate multi-dimensional scaling analyses of the value items within each of 66 countries established that 45 of the items have reasonably equivalent meanings in each country (Schwartz 1994a, 1999; Fontaine et al. 2005). To test the theory, I specified in advance a set of three to eight value items expected to represent each of the seven cultural orientations. Data from representative national samples in 20 countries, gathered as part of the European Social Survey (2002–3), provided a second test of the theory. This survey includes a 21-items short version of the Portrait Values Questionnaire (PVQ: Schwartz et al. 2001; Schwartz 2003, 2005b, 2006) designed to measure basic individual values. Each portrait describes a person’s goals, aspirations, or wishes that point implicitly to the importance of a value. For example, “Thinking up new ideas and being creative is important to her. She likes to do things in her own original way.” Regarding each portrait, respondents answer: “How much like you is this person?” Respondents’ own values are inferred from their self-reported similarity to people described implicitly in terms of particular values. Within-country analyses in each country confirmed reasonable meaning equivalence across countries for all 21 items. To test the theory, I specified in advance items expected to represent the cultural orientations.

1

I am indebted to over 100 collaborators for their aid in gathering the data.

a theory of cultural value orientations

181

Empirical Evaluation of the Theory of Cultural Value Orientations A first assessment of the validity of the seven cultural value orientations and the relations among them used SVS data gathered in 1988–2000. Participants were 80 samples of schoolteachers (k-12) from 58 national groups and 115 samples of college students from 64 national groups, together constituting 67 nations and 70 different cultural groups. Samples from ethnically heterogeneous nations came from the dominant, majority group. Most samples included between 180 and 280 respondents. For each sample, we computed the mean ratings of the 45 value items and then correlated items across samples. This treats the sample as the unit of analysis. The sample level correlations are statistically independent of the correlations across individuals within any sample. Thus, the analyses are at the sample (country) or culture level, not the individual level. Correlations between the sample means were used in a multidimensional scaling analysis (Borg and Lingoes 1987; Guttman 1968) to assess the presence of the seven cultural orientations and the relations among them. The 2-dimensional projection in Figure 2 portrays the pattern of intercorrelations among values, based on the sample means. Each value item is represented by a point such that the more positive the correlation between any pair of value items the closer they are in the space, and the less positive their correlation the more distant. Comparing Figure 2 with Figure 1 reveals that the observed content and structure of cultural value orientations fully support the theorized content and structure. This analysis clearly discriminates the seven orientations: The value items selected a priori to represent each value orientation are located within a unique wedge-shaped region of the space.2 Equally important, the regions representing each orientation form the integrated cultural system postulated by the theory: They emanate from the center of the circle, follow the expected order around the circle, and form the poles of the three broad cultural dimensions. The second assessment of the theory of cultural value orientations used the 21-item PVQ data from the representative national samples One item, ‘accepting my portion in life,’ emerged in the egalitarianism region rather than in the expected harmony region which is adjacent. Its correlations with other items also failed to support an interpretation as a harmony value. We therefore dropped it from the analyses. 2

182

shalom h. schwartz

of the ESS. To obtain a sufficient number of samples for a reliable Multidimensional Scaling Analysis, we split the 20 countries into 52 cultural groups, each with a minimum of 40 respondents.3 We computed group means for each item and correlated the items across the groups. Figure 3 presents the two dimensional projection of relations among the items. The observed content and structure of cultural value orientations in this figure again support the theorized content and structure fully. Because the ESS value scale has few items and was not designed to measure cultural orientations, only three or fewer items represent each orientation (except embeddedness with six). Nonetheless, there are seven distinguishable regions representing the seven orientations. The regions follow the expected order around the circle and form the three polar cultural dimensions. The score for each cultural value orientation in a country is the mean importance rating of the value items that represent it. Prior to computing these scores, we centered each individual respondent’s ratings of the value items on his/her mean rating of all of the items. This controls for individual as well as group biases in use of the response scales.4 In order to increase the reliability of country scores based on the SVS data, I combined the means of the teacher and student samples in the 52 countries in which both types of samples were available. In 21 countries, only either teacher or student data were available. For these countries, I estimated the missing sample means based on regression coefficients generated by regressing student and teacher means from the 52 countries where both were available on one another. For a lower-bound estimate of the adequacy of the combined teacher- and student-based SVS scores for measuring cultural differences among societies, I correlated them with the scores based on 25 representative national or sub-national samples from the ESS. I refer to this as a ‘lower-bound’ estimate because of a set of methodological factors that would weaken any associations: The ESS scores are each based on very few items, as noted above, the ESS data were gathered an average of seven years later, the ESS and SVS methods of measurement differ greatly and, most important, the ESS countries come only from Europe plus Israel, thereby substantially restricting the range of

My thanks to Ariel Knafo for deriving these ethnic group samples. Schwartz (1992, 2006) further explains how to perform the scale use correction and why it is necessary. 3 4

*CURIOUS

VARIED LIFE*

MASTERY

*DARING

HIERARCHY

*SOCIAL POWER

*WEALTH

Figure 2. Cultural level MDS-195 samples, 70 cultural groups (coefficient alienation .18).

AFFECTIVE AUTONOMY

INFLUENTIAL*

SOCIAL*RECOGNITION CHOOSING OWN GOALS*

*INDEPEDENT

*AUTHORITY

OBEDIENT* *MODERATE POLITENESS* *HONOR ELDERS NATIONAL CLEAN* *SECURITY *FAMILY *SELF DEVOUT* DISCIPLINE SECURITY *PROTECT MY *WISDOM PUBLIC IMAGE RECIPROCATION *OF FAVORS *HUMBLE

*SOCIAL ORDER

EMBEDDEDNESS *RESPECT TRADITION *FORGIVING

*CAPABLE SUCCESSFUL* AMBITIOUS*

*WORLD OF *WORLD BEAUTY AT PEACE

*PROTECT ENVIRONMENT *UNITY WITH NATURE

PLEASURE*

EXCITING LIFE

ENJOYING*LIFE

INTELLECTUAL AUTONOMY

*CREATIVITY

*BROADMINDEDNESS *FREEDOM

*EQUALITY

*LOYAL

*SOCIAL JUSTICE

HONEST*

RESPONSIBLE*

HELPFUL*

EGALITARIANISM

ACCEPT MY PORTION IN LIFE*

HARMONY

a theory of cultural value orientations 183

7

follow rules

5

wealth 2

13 4

show abilities

successful

risk 15 excitement

MASTERY

others

tell 17

secure surroundings

14

state protect

fun

21 pleasure

new 6 experiences

12

8

understand listen

loyal18 devoted

AFFECTIVE AUTONOMY

10

good time spoil self

free, own INTELLECTUAL 11decisions AUTONOMY creative 1 original

EGALITARIANISM

help others

equality

3 for all

HARMONY

Figure 3. Cultural level MDS-52 cultural groups from ESS (coefficient alienation .14).

HIERARCHY

20

tradition

16

behave properly

EMBEDDEDNESS

inconspicuous modest 9

care for 19 nature

184 shalom h. schwartz

a theory of cultural value orientations

185

scores on the cultural orientations. The observed Pearson correlations ranged from .45 (mastery) to .80 (intellectual autonomy), mean .63. In light of the methodological problems that weaken these correlations, this finding provides considerable support for the adequacy of the combined teacher- and student-based SVS scores for capturing cultural differences among whole societies Contrasting the Inglehart and Schwartz Dimensions Based on the European and World Value Surveys, Inglehart (e.g., Inglehart and Baker 2000) extended his earlier work on materialism and post-materialism to propose two value dimensions on which to compare national cultures, tradition vs. secular-rational and survival vs. self-expression. This section discusses conceptual and empirical relations between these dimensions and the cultural orientations presented above. Inglehart derived scores on the two dimensions for 72 countries, 63 of which overlap with the countries I have studied with the SVS. Correlations between the Inglehart and Schwartz dimensions across these countries provide an empirical basis for assessing their similarities and differences. To further assess relations among the Inglehart and Schwartz dimensions, I used the scores on my dimensions derived from the PVQ21 in the ESS countries. I computed separate scores for East and West Germany because Inglehart has separate scores for these regions. These 21 groups vary less than the 63 countries in the SVSbased analyses both culturally and socially. Should both sets of analyses yield similar results, despite diverse samples and different methods, we can have confidence in their robustness. The tradition/secular-rational dimension centrally concerns orientations toward authority. It contrasts societies in which religion, nation, and family are highly important with those in which they are not so important. In traditional societies, children’s first duty is to their parents and parents are expected to sacrifice themselves for their children. Male dominance is the norm, absolute standards of morality prevail, and national pride is high. “Societies with secular-rational values have the opposite preferences on all of these topics” (Inglehart and Baker 2000:25). The five-item index used to measure this dimension correlates highly with these beliefs and attitudes. The two items that load most strongly on the factor, however, both concern religious belief.

186

shalom h. schwartz

This dimension overlaps conceptually with my autonomy/embeddedness dimension. Both concern the degree to which the individual is submerged in all-encompassing structures of tight mutual obligations. Inglehart’s description implies that, in traditional societies, people’s ties to their religious, national, and family groups are the source of meaning in their lives—a core aspect of embeddedness. The weakening of encompassing structures and of absolute standards in secular-rational societies frees individuals to think, do, and feel more independently—a core aspect of autonomy. Now consider correlations between the dimensions across 63 countries with the SVS measure of my dimensions and, in parentheses, across the 21 ESS countries with the PVQ21 measure. The tradition/secularrational dimension correlates .60 (.51) with intellectual autonomy, .53 (.38) with affective autonomy, –.57 (–.33) with embeddedness, and .60 (.41) with the autonomy/embeddedness dimension. As the pairs of correlations indicate, both sets of analyses yield similar pictures. For a more in-depth examination, I elaborate on the 63-country analysis. The autonomy/embeddedness dimension shares considerable variance (36%) with the tradition/ secular-rational dimension, but they array nations somewhat differently. For example, East Germany ranks in the top 10% of nations on both dimensions and Zimbabwe in the bottom 15%. But Bulgaria, China, and Estonia rank in the top 10% on tradition/secular-rational but the bottom third on autonomy/embeddedness. The centrality of religion in the Inglehart index may explain this difference in country locations on the two dimensions. Their high secular-rational rankings may be due to a breakdown of religious faith and absolute standards of traditional morality during decades of communist rule. The autonomy/embeddedness dimension gives less weight to religious faith. It focuses more on how legitimate it is for individuals to cultivate unique ways of thinking, acting, and feeling vs. submerging the self in an encompassing collectivity. The culture in Bulgaria, China, and Estonia may have become quite secular. At the same time, the low scores on autonomy/embeddedness suggest that the culture still stresses finding meaning through ties to the in-group. The two dimensions apparently capture different aspects of the culture in these countries. The tradition/secular-rational dimension also has some conceptual overlap with the egalitarianism/hierarchy dimension. Both concern deference to authority. The emphasis on national pride in traditional societies also expresses a hierarchical orientation, and the preference

a theory of cultural value orientations

187

for male dominance is compatible with hierarchy and opposed to egalitarianism. The overlap is limited, however, because the Inglehart dimension does not relate to the primary focus of egalitarianism/hierarchy. Both egalitarianism and hierarchy seek to preserve the social fabric by promoting responsible behavior that considers the welfare of others. They differ in grounding such behavior in voluntary choice based on internalized commitments (egalitarianism) or in conformity to the obligations and expectations of ascribed roles (hierarchy). The empirical associations reflect little overlap. The tradition/secular-rational dimension correlated –.00 (.22) with egalitarianism, –.32 (–.21) with hierarchy, and .22 (.25) with the egalitarianism/hierarchy dimension. Unexpectedly, given no obvious conceptual overlap, the tradition/ secular-rational dimension correlated somewhat positively with the harmony/mastery dimension (.36/.34). Thus, more secular-rational societies are also societies that tend more to emphasize fitting into the natural and social world as it is, trying to understand and appreciate rather than to change or to exploit. The nations with cultures especially high on both harmony and secular-rational orientations are all in Western Europe. These nations have well-to-do, educated, and involved publics. Such publics responded to two devastating wars and to severe ecological crises by developing rationally-based relations of harmony among themselves and with the environment. Inglehart’s second dimension, survival/self-expression, contrasts societies in which people primarily focus on economic and physical security (survival) with societies in which security is high and qualityof-life issues are central (self-expression). In the latter, many people are well educated and work in the services. This demands of them more freedom of judgment, innovation, and autonomous decision-making and equips them with relevant communication and information-processing skills. Trust, tolerance, subjective well-being, political activism, and concern for the environment are high. At the survival pole, people feel threatened by and are intolerant of those who are different (e.g., ethnically or in sexual preference) or who seek cultural change (e.g., women’s movements). At the self-expression pole, difference and change are accepted and even seen as enriching, and out-groups are increasingly seen as meriting equal rights. This dimension also overlaps with the autonomy/embeddedness dimension. They both concern the degree to which individuals should be encouraged to express their uniqueness and independence in thought, action, and feelings. Empirically, they are substantially associated.

188

shalom h. schwartz

Survival/self-expression correlated –.66 (–.65) with embeddedness, .55 (.40) with affective autonomy, .57 (.65) with intellectual autonomy, and .64 (.63) with the autonomy/embeddedness dimension, with which it shares 41% of its variance. These two dimensions array nations quite similarly. For example, Sweden and Denmark rank very high on both dimensions and Uganda and Zimbabwe rank very low. But there are differences. Both Ghana and West Germany are moderate on survival/self-expression, but Ghana is very low on autonomy/embeddedness and West Germany very high. The egalitarianism/hierarchy dimension also overlaps conceptually with survival/self-expression. The latter pits trust, tolerance, and support for the equal rights of out-groups against low trust, intolerance, and rejection of out-groups as threatening. This closely parallels some aspects of high vs. low egalitarianism. Political activism and opposition to change in accepted roles also conflict with hierarchy, though these elements are less critical. The empirical correlations support these inferences. Survival/self-expression correlated .72 (.63) with egalitarianism, –.41 (–.25) with hierarchy, and .59 (.50) with the egalitarianism/hierarchy dimension with which it shares 35% of its variance. Sweden and Norway are very high on both dimensions and Bulgaria is very low on both. But many nations do not exhibit consistent ranks on the two cultural dimensions. Japan is very low on egalitarianism (vs. hierarchy) but moderately high on self-expression (vs. survival), for example. My dimension emphasizes the idea that Japanese culture organizes relations of interdependency in role-based hierarchical terms. The Inglehart dimension may reflect the consequences for culture of Japanese society’s wealth, high level of education, and advanced service economy. Thus, these two dimensions capture different, not necessarily contradictory, aspects of culture. The contrasting feelings of interpersonal threat vs. trust and the focus on material security vs. environmental protection of the survival/ self-expression dimension parallel a low vs. high harmony orientation. Empirically, however, there is no association either with harmony or with the harmony/mastery dimension. Harmony emphasizes fitting into and accepting the social and natural environments rather than changing or using them. Self-expression implies a more activist orientation to people and nature. This may explain the lack of association. In sum, there is substantial overlap between Inglehart’s tradition/ secular-rational dimension and my autonomy/embeddedness dimension and between Inglehart’s survival/self-expression dimension and both

a theory of cultural value orientations

189

my autonomy/embeddedness and egalitarianism/hierarchy dimensions. Given the differences in the way the dimensions were derived, in the scales used to measure them, and in the nature of the samples studied, this overlap is striking. It strongly supports the idea that these dimensions capture real, robust aspects of cultural difference. On the other hand, some nations exhibit substantially divergent rankings on the overlapping dimensions. This makes clear that each dimension also captures unique aspects of culture. Thus, for example, the level of conventional religious commitment in a country appears to influence tradition/secular-rational scores crucially but have much less impact on autonomy/embeddedness scores. In contrast, the cultural emphasis on extended in-group bonds vs. pursuit of individual uniqueness appears to influence autonomy-embeddedness scores more than tradition/secular rational-scores. Moreover, my harmony/mastery dimension apparently taps aspects of culture not measured by the Inglehart dimensions, as evidenced by its low correlations with them. As we will see, harmony/mastery is the only cultural dimension not strongly related to socio-economic development. Countries as a Cultural Unit Almost all large, comparative, cross-cultural studies treat countries as their cultural unit. Countries are rarely homogeneous societies with a unified culture. Inferences about national culture may depend on which subgroups are studied. The research on my cultural dimensions with the SVS used country scores from teacher and student samples rather than representative national samples. This makes it especially important to establish that scores derived from different types of samples order countries in the same way on the dimensions. If a meaningful general culture impacts upon varied groups within countries, the order of countries on cultural dimensions should be quite similar whether we measure culture using one type of subsample from the dominant group or another. The same countries should score higher and the same countries lower on each cultural orientation whether the set of samples consists, for example, of older or of younger respondents. I assessed consistency in the relative scores of countries on the seven cultural orientations measured with the SVS, using three types of subsamples. I first assessed whether younger and older subcultures yield similar relative national scores. I split the teacher samples into those 37 years

190

shalom h. schwartz

or younger and those older. There were 55 countries with at least 33 respondents in each age group. The relative national scores, based on these two subgroups, were very similar. Correlations ranged from .96 for embeddedness to .78 for mastery (mean .91). I also compared national scores using male vs. female student subgroups across 64 countries. This yielded similar results: Correlations ranged from .96 for embeddedness to .82 for egalitarianism (mean .90). Finally, I compared national scores based on the teacher vs. the student samples across 53 countries. Correlations ranged from .90 for egalitarianism to .57 for mastery (mean .81). Although still substantial, the somewhat weaker correlations in this last comparison reflect the fact that the samples differed in both age and occupation. This suggests that closely matching the characteristics of the samples from each country is critical when comparing national cultural orientations. These data demonstrate that the similarity of cultural value orientations within countries, when viewed against the background of cultural distance between countries, is considerable. Taken together, the findings support the view that countries are meaningful cultural units. In comparing national cultures, however, it is important to insure that the samples from different countries are matched on critical characteristics (e.g., all teacher samples, all student samples, or all properly drawn national samples). Cultural Distinctiveness of World Regions Both theoretical arguments and empirical analyses suggest that there are culturally distinct world regions (Hofstede 2001; Huntington 1993; Inglehart 1997; Schwartz 1999). This section examines the locations in cultural space of 76 cultural groups based on the combined teacher and student samples (73 countries, with Israel split into Arabs and Jews, Germany into East and West, and Canada into Anglo and Frenchspeaking national groups). For this purpose, I first standardized the mean importance of all seven cultural orientations within each group. Each group profile therefore reflects the relative importance of each cultural orientation within a national group. Unlike Inglehart’s mapping of national cultures on two orthogonal variables (his dimensions), I map them simultaneously on seven variables (my cultural orientations). Mapping locations of the national groups on the seven orientations in a two dimensional space necessarily entails some imprecision. The

a theory of cultural value orientations

191

richness gained in describing cultures largely compensates for the loss of precision because the degree of imprecision is limited (see below). The ‘co-plot’ multidimensional scaling technique (Raveh 2000) maps the cultural distances between groups. It computes a matrix of profile differences between all pairs of groups by summing the absolute differences between the groups on each of the seven value orientations. From this matrix it generates a two-dimensional spatial representation of the similarities and differences among groups (see Figure 4). It then calculates vectors (optimal regression lines) in the MDS space that show the direction of increasing scores for each of the seven orientations. In Figure 4, I drew the full vector for embeddedness from lower left to upper right and short arrows to indicate the angles of the vectors for the other orientations. These other vectors also extend through the center of gravity of the figure, just above Romania. The correlation between the actual scores of the cultural groups on an orientation and their locations along the vector that represents the orientation appear in parentheses. The substantial magnitude of these correlations (range .75 to .98) indicates that the locations of most samples provide quite an accurate picture. This is because countries usually exhibit a profile that reflects the coherence of the theoretical structure of cultural dimensions. If the culture of a country emphasizes one polar value orientation, it typically deemphasizes the opposing polar orientation. Moreover, the relative importance of adjacent cultural orientations is usually similar too. For example, Italian culture, compared to all the others, is very high both in egalitarianism and in the adjacent harmony orientation but very low in the opposing hierarchy and adjacent mastery orientations. Chinese culture shows the reverse profile.5 Locations of nations along these vectors relative to one another reveal, graphically, the specific ways in which national cultures resemble or differ from one another. For example, the farther a nation toward the upper right, the greater the cultural emphasis on embeddedness relative to other nations and the farther toward the lower left, the less the cultural 5 Japan presents a striking exception. Seven samples from around Japan reveal an unusual combination of cultural elements. The culture strongly emphasizes hierarchy and harmony but not embeddedness, which is adjacent to them, and it strongly emphasizes intellectual autonomy but not the adjacent egalitarianism. Thus, the location of Japan in the co-plot is necessarily misleading. This unusual combination would not surprise many scholars of Japanese culture (e.g., Benedict 1974; Matsumoto 2002). It points to a culture in tension and transition.

192

shalom h. schwartz

emphasis on embeddedness. To locate a nation on a cultural orientation, draw a perpendicular line from the position of the nation to the vector for that orientation. Perpendiculars drawn to the embeddedness vector in Figure 4 show that this orientation is especially emphasized in Yemen, less so in Macedonia, and very little in East Germany. Consider two examples of how Figure 4 represents the cultural profile of a country on all seven cultural orientations. Culture in Sweden (upper left) strongly emphasizes harmony, intellectual autonomy, and egalitarianism and moderately emphasizes affective autonomy. The cultural emphasis on embeddedness is low, and it is very low for mastery and hierarchy. In contrast, in Zimbabwe (lower right), mastery, embeddedness, and hierarchy are highly emphasized, affective autonomy moderately emphasized, and egalitarianism, intellectual autonomy, and harmony receive little cultural emphasis. The spatial map of the 76 national cultures reveals seven transnational cultural groupings: West European countries (clear circles) to the far left, English-speaking countries in the lower left center (dark circles), Latin American countries in the center (shaded circles), East European countries in the upper center and to the left of Latin America (lightly shaded circles), South Asian countries (shaded circles) in a band to the right, Confucian influenced countries below them to the right (clear circles), and African and Middle Eastern countries (dark circles) to the far right and above. Only nine cultures are located outside of their expected region. Four of these are from the culturally diverse Middle East (Turkey, Greek Cyprus, Israel Arabs, Israel Jews). The regions show striking parallels with the zones Huntington (1993) suggested and those Hofstede (1980) and Inglehart and Baker (2000) found. Schwartz and Ros (1995) and Schwartz and Bardi (1997) provide initial explanations for the emergence of the English-speaking, West European, and East European cultural profiles. Most regions reflect some geographical proximity. Hence, some of the cultural similarity within regions is doubtless due to diffusion of values, norms, practices, and institutions across national borders (Naroll 1973). But shared histories, language, religion, level of development, and other factors also play a part. To illustrate the sensitivity of the cultural orientations to such factors, consider the cultures that are not located in their expected regions. French Canadian culture is apparently closer to West European and particularly French culture than to English speaking Canadian culture, reflecting its historic and linguistic roots. East German culture is close

SPAIN

NORWAY

FINLAND

ITALY

AFFECTIVE AUTONOMY (.92)

CHILE

HUNGARY POLAND SINGAPORE

INDONESIA

GEORGIA FIJI BOSNIA HZ PHILIPPINES ESTONIA CYPRUS BOLIVIA Gr MALAYSIA

SLOVAKIA

MEXICO

CZECH REP

SLOVENIA

LATVIA

UGANDA

GHANA

NIGERIA

SENEGAL

EMBEDDEDNESS (.98)

S AFRICA

YEMEN

HIERARCHY (.87) CHINA

Figure 4. Co-plot map of 76 national groups on seven cultural orientations (coefficient alienation .11).

MASTERY (.88)

ROMANIA NEPAL VENEZUELA IRAN ARGENTINA YUGO- TURKEY MACEDONIA NAMIBIA ISRAEL BRAZIL SLAVIA RUSSIA PERU ARABS COSTA BULGARIA JORDAN PORTUGAL RICA UKRAINE TAIWAN NETHERLANDS ZIMBABWE AUSTRALIA GREECE FRANCE INDIA CROATIA HONG KONG IRELAND JAPAN S KOREA CANADA INTELLECTUAL NEW UK ENG AUTONOMY (.93) ZEALAND ISRAEL USA JEWS THAILAND

BELGIUM SWEDEN GERMANY DENMARK AUSTRIA W CANADA GERMANY E FR

SWITZERLAND FR

EGALITARIANISM (.75)

HARMONY (.79)

ETHIOPIA

EGYPT CAMEROON

a theory of cultural value orientations 193

194

shalom h. schwartz

to West German rather than part of the East European region. This probably reflects continued cultural ties from the pre-communist era. Turkish culture is higher on egalitarianism and autonomy and lower on hierarchy and embeddedness than its Middle Eastern Muslim neighbors are. This probably reflects its secular democracy, long history of East European influence, and recent struggles to join the West. Greek Cypriot culture is relatively high in embeddedness and low in autonomy. This may reflect its history of over 1000 years of rule by the Byzantine and Ottoman Empires and its Eastern Orthodox religion. Israeli Jewish culture is close to the English-speaking cultures and distant from the surrounding Middle East to which its Arab culture is close. Europeans founded Israel and it has strong political and economic links to the USA. Among the Latin American countries, the populations of Bolivia and Peru were least exposed to European culture and economically least developed. This probably explains why their cultures are much higher in hierarchy and embeddedness than those of their neighbors. For Japan, see footnote 5. Next, let us examine the cultural orientations that characterize each distinct cultural region. I base these characterizations on the actual cultural orientation scores because, as noted above, locations on seven variables in two dimensions cannot be perfect. Nonetheless, the locations of regions on the vectors in Figure 4 are quite accurate and highly informative. West Europe. Corresponding to its location on the left of Figure 4, West European culture emphasizes intellectual autonomy, egalitarianism, and harmony more than any other region. It is the lowest region on hierarchy and embeddedness. This profile holds even after controlling for national wealth (GDP per capita in 1985). Thus, though West Europe’s high economic level may influence its culture, other factors are apparently critical. This cultural profile is fitting for a region of democratic, welfare states where concern for the environment is especially high (cf. Ester et al. 1994). Although when compared with other world regions, West European countries share a broad culture, there is substantial cultural variation within the region too. Consider two examples. Greek culture is the least typical of Western Europe—higher on mastery and lower on intellectual autonomy and egalitarianism than the others are. French and Swiss French cultures display a relatively high hierarchy orientation for Western Europe, together with the usual high affective and intellectual autonomy. They apparently retain a somewhat hierarchical orientation despite their emphasis on autonomy. Detailed analysis of such

a theory of cultural value orientations

195

variations is beyond the scope of this article, but cultural differences within regions are meaningful. English-Speaking. The culture of the English-speaking region is especially high in affective autonomy and mastery and low in harmony and embeddedness, compared with the rest of the world. It is average in intellectual autonomy, hierarchy, and egalitarianism. The culture in America differs from that in other English-speaking countries by emphasizing mastery and hierarchy more and intellectual autonomy, harmony, and egalitarianism less. This profile points to a cultural orientation that encourages an assertive, pragmatic, entrepreneurial, and even exploitative orientation to the social and natural environment. With the exception of the USA, this region is particularly homogeneous. Cultural Differences in the ‘West’. There is a widespread view of Western culture as individualist. Hence, the differences within the West that the more complex conception of cultural orientations reveals deserve more detailed explication (see Schwartz and Ross 1995). Comparisons of 22 West European samples with six United States samples show large and significant differences on six of the seven culture orientations. Egalitarianism, intellectual autonomy, and harmony are higher in Western Europe; mastery, hierarchy, and embeddedness are higher in the United States. Using the term “individualist” to describe either of these cultures distorts the picture these analyses reveal. Cultural orientations in Western Europe are individualist in one sense: They emphasize intellectual and affective autonomy and de-emphasize hierarchy and embeddedness relative to other cultures in most of the world. But West European priorities contradict conventional views of individualism in another sense: They emphasize egalitarianism and harmony and de-emphasize mastery. That is, this culture calls for selfless concern for the welfare of others and fitting into the natural and social world rather than striving to change it through assertive action. This runs directly counter to what individualism is usually understood to mean. Cultural emphases in the United States show a different but equally complex pattern: The individualistic aspect of American value orientations is the emphasis on affective autonomy and mastery at the expense of harmony. This combination may be the source of the stereotypical view of American culture as justifying and encouraging egotistic self-advancement. But this is not prototypical individualism because intellectual autonomy is relatively unimportant. Moreover, both hierarchy and embeddedness, the orientations central to collectivism, are high compared with Western Europe. This fits the emphasis on

196

shalom h. schwartz

religion, conservative family values, and punitiveness toward deviance in America. The empirical profile is compatible with in-depth analyses of American culture (Bellah et al. 1986; Etzioni 1993). Confucian. The Confucian-influenced region also exhibits a pragmatic, entrepreneurial orientation. However, this orientation combines a heavy emphasis on hierarchy and mastery with a rejection of egalitarianism and harmony as compared with other regions. This region emphasizes embeddedness more than all the European and American cultures. This cultural profile is consonant with many analyses of Confucian culture (e.g., Bond 1996). Within-region differences are small except for Japan, which is substantially higher on harmony and intellectual autonomy and lower on embeddedness and hierarchy. Africa and the Middle East.6 The cultural groups from sub-Saharan and North Africa and the Muslim Middle East form a broad region that does not break down into clear sub-regions. These cultures are especially high in embeddedness and low in affective and intellectual autonomy. Thus, they emphasize finding meaning in life largely through social relationships and protecting group solidarity and the traditional order rather than cultivating individual uniqueness. This fits well with the conclusions of studies of the Middle East (e.g., Lewis 2003) and sub-Saharan Africa (e.g., Gyekye 1997). There is a great deal of variation within the region on all but embeddedness, egalitarianism, and intellectual autonomy. South Asia. The culture in the South Asian region is particularly high in hierarchy and embeddedness and low in autonomy and egalitarianism. This points to an emphasis on fulfilling one’s obligations in a hierarchical system—obeying expectations from those in roles of greater status or authority and expecting humility and obedience from those in inferior roles. As in Africa, here social relationships rather than autonomous pursuits are expected to give meaning to life. With the exception of India’s especially high rating on mastery, all the groups are culturally quite homogenous. The variety of dominant religions (Hinduism, Roman Catholicism, Islam, Buddhism, Methodist Protestantism) in this region does not produce cultural heterogeneity on the basic orientations.

6

I exclude Cyprus, Israeli Jews, and Turkey, which were discussed above.

a theory of cultural value orientations

197

East Europe. The East European cultures are low in embeddedness and hierarchy compared with Africa, Asia, and the Middle East, but higher in these cultural orientations than Western Europe and the Americas. Although the East European cultural groups do form a region in the spatial projection, the cultures in this region vary substantially on hierarchy, mastery, and harmony. A closer look at the map reveals that the Baltic and East-Central states form a sub-region toward the top center and the Balkan and more Eastern states form a sub-region to their right and below.7 The former are higher in harmony, intellectual autonomy, and egalitarianism and lower in mastery and hierarchy than the Balkan and more Eastern states. The Baltic and East-Central states have stronger historical and trade links to Western Europe, were less penetrated by totalitarian communist rule, and threw it off earlier. This may help account for why their profile is closer to that of Western Europe. In an earlier paper, my colleagues and I inferred from the profile of the East-Central states that their population had largely rejected both the rhetoric and the social organization of communist regimes, while insisting on their intellectual independence (Schwartz and Bardi 1997; Schwartz et al. 2000).8 Latin America. Finally, the culture of the Latin American region is close to the worldwide average in all seven orientations. Moreover, excepting Bolivia and Peru, whose populations have been least exposed to European culture, this region is particularly homogeneous culturally. Some researchers describe Latin American culture as collectivist (e.g., Hofstede 2001; Triandis 1995). Compared with Western Europe, this seems to be so. Latin America is higher in hierarchy and embeddedness, presumably the main components of collectivism, and lower in intellectual autonomy, presumably the main component of individualism. The opposite is the case, however, when we compare Latin America to Africa and the Middle East, South Asia, and Confucian-influenced cultures. This example highlights the importance of the frame of comparison. The culture of a group may look different when viewed in a worldwide perspective than when inferred from narrower comparisons.

Georgia and Bosnia-Herzegovina are exceptions that require further study. The finding that East European countries that experienced more invasive communist rule were lower in harmony and higher in mastery undermines our earlier interpretation of the high harmony scores as reflecting adaptation to life in totalitarian regimes by avoiding trouble and refraining from taking initiatives. 7 8

198

shalom h. schwartz Relations of Culture to Socioeconomic, Political, and Demographic Characteristics

Having seen that national cultural groups and regions of the world differ systematically on the cultural dimensions, two critical questions arise. How do these cultural differences arise? And do they matter? Socio-economic, political, and demographic factors all impact on culture and, I will argue, are themselves reciprocally influenced by culture. I present analyses for one key variable representing each of these factors—socio-economic development, level of democracy, and household/family size. Welzel, Inglehart, and Klingemann (2003) traced a causal sequence among socioeconomic development, values, and democratization. They argued that socioeconomic development increases individual resources and thereby gives people the means to make choices. Utilizing these means, people cultivate their presumed inherent human desire for choice, giving rise to “mass emancipative values.” Pursuit of these values leads to democratization—institutionalizing freedom rights that provide the legal guarantee of choice. Using a measure of emancipative values drawn from the World and European Value Surveys, Welzel et al. present analyses showing that values indeed mediate the relationship between socioeconomic development and democratization. Their causal argument contradicts views that values impact on socioeconomic development or that democratization influences values. I next examine how well this argument holds when we employ different measures of values and a set of 72 countries, 15 of which do not overlap those they studied. In order to simplify the empirical presentations, I will use the three polar value dimensions formed by the seven cultural orientations rather than the separate orientations. Two of the cultural value dimensions express what Welzel et al. call emancipative values—the dimensions of autonomy versus embeddedness values and of egalitarianism vs. hierarchy values. The first dimension emphasizes autonomous choice and cultivation of individuals’ unique ideas and preferences rather than following and preserving traditional and externally imposed ideas and preferences. The second dimension emphasizes voluntary regulation of behavior based on equality rather than regulation of behavior through submission to role expectations built into existing hierarchies.

a theory of cultural value orientations

199

Socioeconomic Development and Democratization The top panel of Table 1 presents correlations across 73 countries between four indexes of socioeconomic development and scores on the three value dimensions. The latter are computed by subtracting the score for the second pole from the first (e.g. harmony minus mastery). The values data are largely from the mid-90’s.9 The correlations portray associations with development about 10 and two years earlier, concurrently, and almost 10 years later. The second panel of Table 1 presents correlations with democratization scores from Freedom House (higher scores signify greater civil liberties and political rights) about 10 years earlier, concurrently, and seven years later (Freedom House, various years). As expected, all correlations with the autonomy and egalitarianism dimensions are substantial, whereas those with the harmony vs. mastery dimension are low. To assess causal relations, I carried out a path analysis in which earlier indexes of democratization predict later indexes and in which development might affect democratization either directly or through values. Following Welzel et al., I measured development with the Vanhanen (1997) ‘index of power resources’ for 1993. This index includes measures of material and intellectual resources and the complexity of the occupational system. This makes it better than a pure economic index for measuring individual resources. Because the 1995 index of democratization is included in Figure 5a to predict the 2002 index, paths from other predictors signify effects on change in the level of democratization. Figure 5a reveals that cultural emphases on autonomy and on egalitarianism values both predict increases in democratization. Earlier development affects these cultural values and they, in turn, fully mediate the effects of development on increasing democratization. Democratization in 1985 does not predict later cultural values over and above the effects of development.

9 Combining values data from somewhat different points in time poses little problem because value change at the national level on these dimensions is very slow (see Schwartz et al. 2000).

200

shalom h. schwartz

Table 1. Correlations of cultural value dimensions with socioeconomic development democratization, and household size N

Autonomy Egalitarianism Harmony minus minus minus Embeddedness Hierarchy Mastery

Socioeconomic Development 1985 GDPpc 1995 GDPpc 1993 Index of Power Resources 2004 GNIpc

73 73 72 73

.58** .73** .78** .75**

.42** .46** .58** .53**

.27* .19 .14 .20

Democratization 1985 Freedom House Index 1995 Freedom House Index 2002 Freedom House Index

73 73 73

.58** .71** .70**

.46** .52** .57**

.03 .36** .38**

Household Size 1985 Average Family Size 2001 Average Household Size

73 73

–.70** –.75**

–.61** –.36**

–.41** –.32**

**p<.01, *p<.05, 2-tailed Notes: GDPpc=Gross Domestic Product per Capita, from the World Bank; GNIpc=Gross National Income per Capita, from the World Bank; Average Household/Family Size=from the Encyclopaedia Britannica Almanac

The path analysis clearly supports the Welzel et al. causal argument. For them, development influences values which, in turn, influence change in democracy levels. Here, development influences change in democracy levels only insofar as it affects autonomy and egalitarian values, not directly. Support for this causal sequence derives from a different set of cultural values than those studied by Welzel et al. and from a somewhat different set of countries and of years. The current analysis further demonstrates that the prior level of democracy has no impact on cultural values, once development is controlled. However, this analysis does not test whether cultural values have a reciprocal influence on socioeconomic development. The path analysis in Figure 5b addresses this issue. Because no Vanhanen index of individual resources is available for the years after 1993, I chose the latest, best, available index of socioeconomic development for this analysis—gross national income per capita in 2004. Figure 5b reveals that neither democracy nor egalitarianism values in 1995 predict change in socioeconomic development from 1993 to 2004. However, autonomy values significantly predict the change. Thus, this dimension of cultural values reciprocally influences socioeconomic development. I cannot estimate the relative strength

.59**

.78**

.69**

Values 1995: Egalitarianism vs. Hierarchy

Values 1995: Autonomy vs. Embeddedness

Democracy 1995

.16**

.18**

.64** Democracy 2002

R2 = .75

.59**

Values 1995: Egalitarianism vs. Hierarchy

Values 1995: Autonomy vs. Embeddedness

Democracy 1995

.73**

.20**

Gross National Income per capita 2004

R2 = .78

Figure 5b. Causal model predicting change in socioeconomic level with democracy and cultural values across 73 countries.

Notes: Broken arrows indicate nonsignificant paths. ** p<.01, * p<.05.

Individual Resources 1993

.78**

.69**

Figure 5a. Causal model predicting change in democracy with socioeconomic level and cultural values across 73 countries.

Notes: Broken arrows indicate nonsignificant paths. ** p<.01, * p<.05.

Individual Resources 1993

Democracy 1985

a theory of cultural value orientations 201

202

shalom h. schwartz

of these reciprocal influences because we lack earlier measures of the cultural values. Nonetheless, with the caveat that the earlier and later measures of socio-economic development differed, this analysis provides evidence that contradicts the Welzel, et al. unidirectional view of relations between values and development. Family/Household Size The demographic characteristic I examine is average family/household size in a country. Researchers have largely overlooked the importance of this variable as a determinant and consequence of culture. The third panel in Table 1 reports the correlations of the cultural value dimensions with average family size in 1985 and with average household size in 2001.10 The negative correlations indicate that the larger the average family or household, the greater the cultural emphasis on embeddedness, hierarchy, and mastery values. At the country level, cultural value orientations relate strongly to family size. The size of people’s own family, however, does not relate to their own personal values. How does this happen? The key is societal norms for managing family relations. Societal norms reflect what is required and possible in order for the typical family to function smoothly. These norms reflect the prototypical size and composition of households. They specify how to organize families, raise children, and regulate interaction. Families largely conform to these norms, so variation in the size of particular families has less impact on its members. How might family/household size influence culture? Where the typical household is large, it is crucial for behavior to be predictable. This requires high levels of social control from above. Emphasizing obedience to authority, conformity to norms, and fulfilling role obligations unquestioningly is functional. If family members view themselves as inseparable parts of a family collectivity and identify with its interests, even large families can run smoothly. These family practices and norms foster cultural embeddedness and hierarchy in the society. Large families are incompatible with cultural autonomy and egalitarianism. The demands of coordination in large families preclude treating each member as a unique individual with equal rights. They discourage permitting each family member to make decisions autonomously and 10 Data are from the Encyclopaedia Britannica Almanac. The date (1985, 2001) is the median of about a 5 year period for which the data were reported.

a theory of cultural value orientations

203

to pursue his or her own ideas, interests, and desires. A greater need for pragmatic problem solving in larger families may account for the somewhat stronger emphasis on mastery values. The preceding explains the possible causal influence of family/household size on cultural value orientations. The influence of cultural values on family/household size is also likely to contribute to the correlations. Autonomy values, in particular, encourage having few children so that each can develop his or her unique abilities and interests. Autonomy and egalitarianism values encourage and justify women’s pursuit of meaningful non-family roles. This too reduces the number of children. Embeddedness values promote commitment to the in-group. They sanctify group continuity and, hence, having many children to promote it. Autonomy values sanctify individual choice. They justify weighing children against alternative paths for achieving personal meaning in life, such as careers. The path analysis in Figure 6 examines possible causal relations among culture, average household/family size, and socioeconomic level. For many countries, I found no statistics for average household size that antedated the mid-90’s measurement of culture or for average family size that post-dated this period. Examination of concurrent measures of average household and family size for 60 countries around 1983 reveals that they correlated highly (.87).11 I therefore used household and family size as proxies for one another in the analysis. Starting on the left in Figure 6, we see that greater socioeconomic resources (Individual Resources 1980) lead to smaller families (Family Size 1985), but family size does not appear to affect change in socioeconomic level (Individual Resources 1993). Next, the figure reveals a substantial influence of family size on all three cultural value dimensions, over and above the direct effects of socioeconomic resources. Larger families in a country induce less cultural emphasis on harmony, egalitarianism, and autonomy values, whereas greater socioeconomic resources lead to more emphasis on autonomy and egalitarianism but do not affect harmony vs. mastery. Moreover, family size apparently mediates some of the effect of socioeconomic resources on cultural values (indirect paths from Individual Resources 1980 through Family

11 Household size data are from Kurian (1984) and family size data from the Encyclopaedia Britannica Almanac 1985.

204

shalom h. schwartz

Size 1985).12 Clearly, family size is important in the development of culture. The right half of Figure 6 reveals that all three cultural value dimensions predict change in household size. Moreover, two dimensions, egalitarianism vs. hierarchy and autonomy vs. embeddedness, fully mediate relations of socioeconomic resources to change in household size. Thus, increasing socioeconomic level appears to promote a drop in the size of households only insofar as it leads to change in cultural values. The three cultural value dimensions also partially mediate effects of earlier family size on later household size. To the extent that smaller families and rising resources increase autonomy values, they promote a decrease in household size. To the extent that smaller families increase harmony values, they also promote decreasing household size, though this effect is weak. The above findings parallel the zero-order correlations; but the path from egalitarianism to household size reverses the correlation. Although smaller families promote a cultural emphasis on egalitarianism vs. hierarchy values, a cultural emphasis on egalitarianism values contributes to maintaining larger households over time. In other words, an emphasis on hierarchy values promotes a reduction in household size. This effect holds even when the other two value dimensions are not included as predictors. One speculative interpretation is that both formal and informal normative influence on families is greater in societies whose culture emphasizes hierarchy values. If governments seek to raise productivity through increasing women’s participation in the workforce, they are likely to generate societal norms that oppose large families. Such norms may reduce family size more effectively in cultures high in hierarchy. Congruent with this interpretation, the greatest reductions in household size have occurred in China, with its formal anti-natalist policies, and in the East Asian “Tigers” which have rapidly moved toward market economies and whose culture is highly hierarchical.

12 Direct paths from Individual Resources 1980 to the three cultural dimensions are left out of the figure to avoid clutter. None is significant.

Individual Resources 1993

–.36**

.33**

–.41**

.56**

–.42**

Values 1995: Autonomy vs. Embeddedness

Values 1995: Egalitarianism vs. Hierarchy

Values 1995: Harmony vs. Mastery

–.44**

.31**

–.15**

Household Size 2001

R2 = .73

Figure 6. Causal model predicting change in average household/family size with socioeconomic level and cultural values across 73 countries.

Notes: Broken arrows indicate nonsignificant paths. ** p<.01, * p<.05.

.97**

Individual Resources 1980

–.60**

Family Size 1985

.60**

a theory of cultural value orientations 205

206

shalom h. schwartz Consequences of Cultural Value Orientations

Finally, we cons ider some consequences of national differences in the cultural value orientations. We examine how culture relates to women’s position in society, to a selection of social attitudes held by societal members, and to important social behaviors. Women’s Equality The equality of women and their opportunities for autonomous decision-making is one domain in which cultural orientations are likely to influence practices. Women should have greater independence to develop their own capabilities and follow their own preferences if the culture emphasizes autonomy rather than embeddedness. Similarly, cultures that emphasize egalitarian rather than hierarchical, role-based regulation of interdependence and work are likely to promote greater equality. A cultural preference for harmonious relations in contrast to assertive mastery might also enhance women’s equality, because women around the world value benevolence more and power less than men (Schwartz and Rubel 2005). Women’s equality and autonomy are greater in wealthier and more economically developed countries (e.g., Apodaca 1998). Doubtless, some of this association is direct. Material and intellectual resources free individuals—men and especially women—from some of their dependence on the support of their families, enabling them to strike out on their own and to demand more equal opportunities. However, the prevailing cultural orientations may mediate the impact of increased individual resources that accompanies national wealth. Cultural orientations may legitimize and facilitate but also delegitimize and inhibit the pursuit of equality. This can occur through informal or formal sanctions experienced in everyday interaction and through encounters with the structures, practices, and regulations of societal institutions that are grounded in and justified by the cultural orientations. To assess possible mediation by culture, I first correlated both individual resources and the cultural value dimensions with several indicators of women’s equality. I then examined relations of individual resources with these indicators, controlling the effects of culture. This reveals whether cultural mediation substantially shrinks the association with resources. Ratings of women’s equality in 69 countries in four domains—social,

a theory of cultural value orientations

207

health, education, and employment—in 1988 (Population Crisis Committee 1988) served as one set of indicators. The Vanhanen country scores for power resources in 1993 indexed availability of resources to individuals. The data on the cultural dimensions and resources postdate the equality data by five to six years on average. However, this should have little effect on the associations: The cultural orientations demonstrate considerable stability (Schwartz et al. 2000), and analyses using gross domestic product per capita in 1985 to index country wealth yielded essentially the same results. The top panel of Table 2 shows the correlations of the three cultural value dimensions and individual resources with the five indicators of women’s equality. All three cultural dimensions as well as individual resources correlate significantly with the overall average and with most sub-categories of equality. Autonomy vs. embeddedness has the strongest associations, followed by individual resources, egalitarianism vs. hierarchy, and harmony vs. mastery. All correlations are in the expected direction. The last row of the panel reveals the effect of introducing the three cultural value dimensions as mediators. Clearly supporting cultural mediation, the variance in the overall index of women’s equality accounted for by individual resources shrinks 80%, when culture is controlled. Interestingly, though culture mediates resource effects in all four domains, mediation is smallest (50% reduction) for employment equality, the domain most directly concerned with producing wealth. The bottom panel of Table 2 examines cultural mediation of two other indicators of women’s equality and autonomy, the percent of ministerial positions filled by women in national parliaments in 1994–98 (United Nations Women Watch 1999) and the percent of married women aged 19–45 who used any method of contraception in 1990–97 (United Nations Population Division 1998). Cultural value dimensions and individual resources correlated substantially with both indicators, as expected. The more socio-economically developed the country and the more its culture emphasized autonomy, egalitarianism, and harmony values, the more political equality and sexual autonomy women enjoyed. Culture partially mediated the relation of resources to women’s attainment of ministerial positions, reducing the variance explained by 70%. It fully mediated the relation of resources to contraception use, reducing the variance explained by 97%. Here too, the main mediator was the cultural emphasis on autonomy vs. embeddedness values.

208

shalom h. schwartz

Table 2. Culture and women’s equality: correlations and mediation of the effects of country wealth

Autonomy vs. Embeddedness Egalitarianism vs. Hierarchy Harmony vs. Mastery Individual Resources 1993 Individual Resources 1995 controlling culture

Women’s Equality in 1988 Health Educational Employment

N

Social

66

.58**

.76**

.70**

.42**

.73**

66

.31*

.39**

.45**

.13

.39**

66

.35**

.23

.18

.25*

.31**

65

.48**

.66**

.67**

.50**

.68**

65

.09

.21

.27*

.35**

.30*

Autonomy vs. Embeddedness Egalitarianism vs. Hierarchy Harmony vs. Mastery Individual Resources 1993 Individual Resources 1995 controlling culture

Average

N % Ministers who are Women 1994–98

N

% Married Women 19–45 who use Contraception 1990–97

70

.51**

53

.76**

70

.54**

53

.26*

70

.26*

53

.11

70

.63**

52

.64**

70

.34**

52

.20

**p<.01, *p<.05, 2-tailed.

Social Attitudes To examine relations of cultural value emphases to the social attitudes of societal members, I drew on data from the 2000 wave of the World Value Survey (Inglehart et al. 2004). Table 3 presents correlations of the cultural value dimensions with selected attitudes concerning qualities important for children to learn, conventional morality, competition, and tolerance for out-group members. It also presents correlations of these variables with indexes of country levels of wealth and democratization from the same period. To assess the extent to which cultural values mediate effects of wealth and democracy on attitudes, it presents partial correlations controlling the cultural values in parentheses.

a theory of cultural value orientations

209

People in countries with cultures high in embeddedness and low in autonomy choose obedience and hard work, but reject imagination, as especially important for children. The chosen qualities promote conforming and contributing to the group in traditional ways, the rejected quality would threaten tradition by generating potentially disruptive, original ideas. Countries whose culture emphasizes hierarchy and mastery also view hard work—critical to insure fulfilling role obligations faithfully and struggling persistently to master reality—as especially important. Obedience does not relate significantly to the egalitarianism/ hierarchy dimension, but it does go with a cultural hierarchy orientation as one might expect. Only harmony/mastery, but neither national culture, nor wealth, nor democracy predicts the importance of unselfishness. A mastery orientation may legitimize selfishness because it justifies self-assertion in order to get ahead. People in countries whose culture emphasizes embeddedness, hierarchy, and mastery consider unconditional respect for parents and religion more important and they more strongly oppose behavior that threatens traditional family and sexual mores compared to people in countries where autonomy, egalitarianism, and harmony values prevail. These findings are congruent with the fact that the two attitudes express conformity to expectations of authorities and extended kin-groups but thwart innovative, unconventional behavior that expresses individual preferences, and they reject new realities. People are more likely to view competition as good if they live in countries with cultures that emphasize hierarchy and mastery. Both these cultural orientations justify the differential distribution of resources to which competition is directed, whereas their opposing poles, egalitarianism and harmony, call for cooperative regulation of interdependence. Readiness to accept immigrants, foreign workers, or people with criminal records as neighbors is lower in cultures that emphasize embeddedness and hierarchy as opposed to autonomy and egalitarianism. In such cultures, people are more likely to fear exposure to values, beliefs, norms, practices, traditions, etc., that differ from their own. Such exposure challenges and endangers the values, beliefs, etc. they cherish and might undermine the established hierarchical structure of roles they assume to be necessary for the smooth functioning of their society. A cultural emphasis on harmony values, which encourages tolerance, also correlates with acceptance of those with criminal records significantly and of immigrants weakly.

55 55

% who say would not like as neighbor: Immigrants/foreign workers People with a criminal record

A

–.71** –.75** –.58** –.63** –.21

59 59 60 60 52 –.45** –.44**

–.32* –.63** –.36** –.50** –.32*

–.57**

–.14 –.57** .24 –.05

Egalitarianism vs. Hierarchy

–.23 –.30*

–.42** –.39** –.41** –.42** –.29*

–.40**

–.21 –.27* –.04 –.30*

Harmony vs. Mastery

–.43** –.21

–.57** –.79** –.48** –.49** –.40**

–.64**

–.45** –.41** .29* –.05

(.00) (.27)

(–.13) (–.48**) (–.06) (.04) (–.30*)

(–.11)

(–.17)A (–.05) (–.05) (–.04)

Democracy 1995 Freedom House

–.40** –.57**

–.48** –.75** –.38** –.46** –.14

–.71**

–.41** –.62** .56** .11

(.08) (–.39**)

(.01) (.41**) (.07) (.08) (.05)

(–.32*)

(–.10) (–.37**) (.33*) (.11)

Economic Level GDPpc 1995

Correlations in parentheses are controlled for country scores on the three cultural value dimensions. **p<.01, *p<.05, 2-tailed.

–.56** –.43**

–.77**

–.52** –.53** .48** .03

57

58 58 58 58

% who select as an especially desirable quality for children to learn Obedience Hard Work Imagination Unselfishness

Autonomy vs. Embeddedness

% who agree that: Must respect parents regardless of qualities & faults Religion very important in life Homosexuality never justifiable Abortion never justifiable Divorce never justifiable Competition is good

N

Social Attitude

Table 3. Cultural Values and Social Attitudes: Correlations and Mediation of Democracy and Economic level Effects by Culture (Partialed)

210 shalom h. schwartz

a theory of cultural value orientations

211

Table 3 indicates that national level of democracy correlates significantly with 10 of the 12 social attitudes, as does national economic level. Does culture mediate these associations? I take a reduction of at least 60% in the variance that a structural variable explains to indicate substantial mediation. Using this criterion, the partial correlations in Table 3 reveal that culture substantially mediates the effects of each structural variable for nine of the 10 attitudes it predicts. Culture does not mediate the tendency to approve of competition more where democracy is lower, and it shrinks by only 47% the association of lower national wealth with rejection of neighbors who have a criminal record. These findings suggest that social structural variables may affect individuals’ attitudes largely through their effects on the cultural orientations that prevail in the society.13 Attitudes and Behavior in the European Social Survey Thus far, we have examined relations of cultural orientations to the attitudes and behavior of societal members using culture scores based on the SVS. As discussed above, the human values items in the European Social Survey (ESS) also provide scores for these orientations. These scores are based on few items and the heterogeneity and number of countries in the ESS data set is limited. However, studying relations of the cultural value orientations to attitudes and behavior in the representative national samples of the ESS can shed light on the robustness of the cultural value theory. Here, I examine relations of the cultural value orientations to one attitude (opposition to immigration) and two types of behavior (membership in voluntary organizations and political activism). I also present relations of social structural variables to the attitude and behaviors and assess the extent to which culture mediates these relations. In Schwartz (2006), I present more elaborate analyses of the relations of culture to these and other attitudes and behavior, using hierarchical linear

13 Although less plausible, one might suggest that cultural orientations influence individuals’ attitudes largely through their effects on the social structural variables. Applying the same criterion of a reduction of at least 60% in the explained variance, this time partialing culture on the structural variables, reveals that democracy and economic level substantially mediate the effects of autonomy vs. embeddedness for only two of the 10 attitudes it predicts, of egalitarianism vs. hierarchy for four of the nine attitudes it predicts, and of harmony vs. mastery for none of the nine attitudes it predicts (partial correlations available from the author).

212

shalom h. schwartz

modeling to take both individual level (e.g., age, education, personal values) and country level variables into account simultaneously. Table 4 presents correlations with the seven cultural orientations, with measures of country wealth (gross domestic product per capita in 1999) and average annual inflation between 1990 and 1999 from the ESS macro-data file, and with average household size in 2001. The index of democracy varies insufficiently across these countries to merit inclusion. Three ESS items measured opposition to accepting ‘other’ immigrants—those of a different race/ethnic group, from poorer European, and poorer non-European countries. A summary index of these items revealed great variation in levels of opposition across countries. 14 Column 1 in Table 4 reveals that opposition to ‘other’ immigrants correlates negatively with intellectual autonomy, egalitarianism, and country economic level, and positively with inflation and household size. The two cultural orientations emphasize openness to and tolerance for what is new and different as well as treating others as moral equals. As such, they encourage acceptance of immigrants and militate against opposing them. To assess mediation of structural effects by culture, I partialed on these two cultural orientations. The partial correlations in parentheses suggest full mediation of structural effects. Table 4. Correlations of cultural value orientations with selected attitudes and behavior across European Social Survey countries and mediation of structural effects Cultural Value Orientation

Opposition to ‘Other’ Immigrants N

Embeddedness Intellectual Autonomy Affective Autonomy Hierarchy Egalitarianism Harmony Mastery GDPpc 1999 Inflation 1990–99 Household Size 2001

15 .49 –.63* –.33 .33 –.68** –.23 .19 –.56* (.12)A .69** (.12) .45 (–.04)

Membership in Voluntary Organizations 18 –.76** .79** .42 –.26 .60** .06 .07 .54* (.31) –.56* (–.06) –.73** (–.39)

Political Activism 20 –.63** .84** .40 –.48* .78** .17 –.18 .53* (.39) –.68** (–.19) –.68** (–.20)

** p<.01, * p<.05, 2-tailed. Notes: ACorrelation partialed for intellectual autonomy and egalitarianism cultural orientations.

14 I included only the 15 West European countries in this analysis, because immigration to East Europe and to Israel has different meanings.

a theory of cultural value orientations

213

Consider next membership in voluntary organizations, a core aspect of social capital (e.g., Putnam 2000). The number of memberships in 12 types of organizations (e.g., sports, humanitarian, labor, religious) indexed this variable in the ESS. Column 2 of Table 4 reports correlations with membership. Three cultural orientations and all three structural variables predicted country differences. Greater intellectual autonomy, egalitarianism, and country wealth go with joining voluntary organizations. Cultural embeddedness, higher inflation, and larger households accompany lower membership. Intellectual autonomy and egalitarianism both entail a cultural view of individuals as independent actors with rights and responsibilities to express these interests through voluntary action. This would encourage joining voluntary organizations. In contrast, cultural embeddedness may discourage unnecessary involvement with people outside the broad in-group; it emphasizes loyalty and devotion to the in-group instead. Such a cultural atmosphere would not support membership in voluntary groups in the wider society. After partialing on intellectual autonomy and egalitarianism, the variance explained by GDPpc, by inflation, and by household size all shrink by at least 60%, suggesting substantial mediation by culture. As a final example, consider political activism. This was measured as the number of politically relevant, legal acts that respondents reported performing in the 12 past months out of nine (e.g., contacting a politician, participating in a public demonstration, boycotting a product). Intellectual autonomy and egalitarianism predicted greater political activism (column 3 of Table 4), whereas the opposing cultural orientations, embeddedness and hierarchy predicted less activism. Political activism necessitates taking initiatives, expressing views that may oppose conventional practices or expectations, often promoting causes that go beyond in-group self-interest (e.g., protecting the weak or the environment). Such action fits a cultural atmosphere that encourages autonomous thought, individual responsibility, and cooperative work, but it violates a cultural atmosphere focused on preserving the status quo and the authoritative social order. All three structural variables also correlated significantly with political activism. The cultural orientations of intellectual autonomy and egalitarianism mediated the strong negative effects of inflation and large households on political activism, but not the enhancing effect of country wealth. The three sets of findings in Table 4 suggest, with only one exception, that country wealth, inflation, and household size

214

shalom h. schwartz

influence these important attitudes and behaviors largely through their influence on the prevailing cultural value orientations.15 The findings with the ESS data together with those reported earlier for combined teacher and student samples across many nations demonstrate that the cultural orientations have robust effects even when measured with different instruments, in different types of samples, and across a wide variety of countries. It is often more convenient to work with the three cultural dimensions, as reported in Tables 1–3. However, the correlations of the single cultural orientations in Table 4 show that doing so may sometimes obscure meaningful information. The orientations that constitute the poles of each dimension correlate in opposing directions in almost every case, but the strength of the correlations can vary considerably. For all three examples in Table 4, egalitarianism has much stronger correlations than hierarchy, its opposing pole. Moreover, the correlations of intellectual autonomy are much stronger than those of affective autonomy, the orientation with which it combines to form the autonomy pole of the autonomy/embeddedness dimension. These patterns doubtless reflect the particular topics of study. It is therefore worthwhile, in each case, to examine whether combining cultural orientations to form dimensions loses important information. Conclusions This article presented a theory of seven cultural value orientations that form three cultural value dimensions. Although it is more complex than other dimensional theories of cultural variation, this theory permits more finely tuned characterizations of cultures. Analyses of data from 73 countries using one instrument and from 20 countries using another demonstrate the validity of the seven cultural orientations. Equally important, the analyses showed that these orientations form an integrated circular structure that captures their theorized compatibilities and oppositions. This yields three broad dimensions (Figure 1).

15 Here too, one might speculate that the two cultural orientations influence individuals’ attitudes and behavior largely through their effects on the social structural variables. Controlling for all three structural variables, the variance explained by the two key cultural values shrinks substantially in five of the six cases. However, the partial correlations for egalitarianism and intellectual autonomy are larger than those for the structural variables in 15 of 18 comparisons (2 values × 3 structural variables × 3 dependent variables).

a theory of cultural value orientations

215

The analyses demonstrated that all three cultural dimensions contribute uniquely to the explanation of important social phenomena. The autonomy/embeddedness and egalitarianism/hierarchy dimensions often showed a similar pattern of positive or negative associations. This reflects the positions of their component orientations, as adjacent or opposed, in the circular structure of cultural orientations. The harmony/mastery dimension exhibited a different pattern of associations. It correlated least strongly with indicators of socio-economic development. Its correlations with the attitudes and behavior studied here were also weaker, perhaps because this article focused on variables related to development. On the other hand, measurement of harmony and mastery may be problematic. The aspects of culture that the harmony/ mastery dimension captures may be especially distinctive. It added a unique element to our understanding of attitudes toward unselfishness and competition. The most striking finding when comparing the mapping of national cultures in the research based on the Hofstede, Inglehart, and Schwartz approaches is that they identify such similar cultural regions around the world. At least two of the three approaches, and usually all three, identify African, Confucian, East-Central European (ex-communist), English-Speaking, Latin American, South Asian, and West European regions. This is amazing, considering how different the approaches are. They differ in their basic cultural constructs, in their methods of measurement (work values and attitudes; beliefs, preferences, and judgments on a range of topics; abstract values or profiles reflecting individuals’ important goals), in the types of samples studied (IBM employees, representative national samples, teachers, students), and in their datagathering periods (from the late ’60s into the 21st century). The emergence of similar cultural regions across approaches affirms the reality of the systematic cultural value differences these approaches tap. The location of similar sets of countries in these regions in the Inglehart and Schwartz mappings also affirms the meaningfulness of countries as cultural units. The substantially similar ordering of countries on cultural orientations or dimensions, when different sub-samples (e.g., age or gender samples) are used to map countries in both the Schwartz and Inglehart analyses, further supports the idea of national cultures. Nonetheless, it is important to investigate other cultural units, such as ethnic groups, in future research. The research reported here, that used 52 ethnic subgroups from the ESS data as the unit of analysis, demonstrated that discrimination of the seven cultural orientations

216

shalom h. schwartz

holds at the ethnic group level and not only at the country level. This enables us to ask questions about cultural differences and similarities among ethnic groups, to compare the impact of nation with that of ethnicity, and to assess the effects of immigration to new countries on the culture of ethnic groups. The analyses revealed substantial empirical overlap between Inglehart’s tradition/secular-rational dimension and my autonomy/embeddedness dimension and between Inglehart’s survival/self-expression dimension and both my autonomy/embeddedness and egalitarianism/ hierarchy dimensions. Nonetheless, several countries exhibited substantially divergent rankings on each of these overlapping dimensions. Each dimension apparently captures some aspects of culture not captured by the others. For example, the level of conventional religious commitment apparently influences tradition/secular-rational scores strongly but has much less impact on autonomy/embeddedness scores. In contrast, autonomy-embeddedness scores apparently reflect the cultural emphasis on extended in-group bonds vs. pursuit of individual uniqueness more than tradition/secular rational-scores do. This difference between the cultural dimensions accounts for the divergent rankings of ex-communist countries on the two dimensions. The conceptual and empirical differences between the Schwartz and Inglehart dimensions can be exploited by using them together to derive deeper understandings of particular national cultures. Doing so may also contribute to our understanding of what each dimension and cultural orientation actually taps. We do not yet know what the optimal number of dimensions and orientations is for characterizing and broadly comparing cultures. Nor do we know which dimensions and orientations will be most fruitful. The findings from research using the approach presented here and the World Value Survey suggest, however, that studying basic values is a desirable path to follow. Values are particularly significant dimensions for comparing cultures because they affect so many different aspects of life. But other dimensions of cultural difference may also be important. The critical value dimensions are unlikely to be orthogonal. They evolve as preferences for resolving basic issues in managing life in society. It is not logical that preferences for resolving one issue are independent of those for other issues. Cultures that encourage autonomy in individual/group relations are unlikely to prefer hierarchy for managing human interdependence. Though they are not opposites, autonomy and hierarchy rarely appear together because they presume conflicting

a theory of cultural value orientations

217

views of human nature. Of course, we can derive orthogonal dimensions from data. But in doing so we miss the pull toward coherence in national cultures. To conclude, I quote an earlier summary of the points on which the approach to cultural dimensions presented here differs from others (Schwartz 2004:73): “(a) It derived the cultural orientations from a priori theorizing rather than post hoc examination of data. (b) It designated a priori the value items that serve as markers for each orientation. (c) It used as measures only items tested for cross-cultural equivalence of meaning. (d) It included a set of items demonstrated to cover the range of values recognized cross-culturally, a step toward ensuring relative comprehensiveness of cultural value dimensions. (e) It specified how the cultural orientations are organized into a coherent system of related dimensions and verified this organization, rather than assuming that orthogonal dimensions best capture cultural reality. (f ) It brought empirical evidence that the order of national cultures on each of the orientations is robust across different types of samples from each of a large number of nations around the world [and using different instruments]. These distinctive features increase the promise of this approach for future research.” References Apodaca, C. 1998. “Measuring Women’s Economic and Social Rights Achievement.” Human Rights Quarterly 20:139–172. Bellah, R.N., R. Madsen, W.M. Sullivan, A. Swidler, and S.M. Tipton. 1986. Habits of the Heart: Individualism and Commitment in American Life. New York: Harper & Row. Benedict, R. 1974/1946. The Chrysanthemum and the Sword: Patterns of Japanese Culture. New York: New American Library. Bond, M.H. 1996. “Chinese Values.” In Handbook of Chinese Psychology, edited by M.H. Bond. Hong Kong: Oxford University Press. Borg, I., and J.C. Lingoes. 1987. Multidimensional Similarity Structure Analysis. New York: Springer-Verlag. Clarke, H.D., A. Kornberg, C. McIntyre, P. Bauer-Kaase, and M. Kasse. 1999. “The Effect of Economic Priorities on the Measurement of Value Change: New Experimental Evidence.” American Political Science Review 93:637–647. Encyclopaedia Britannica Almanac 2005. 2004. Chicago: Encyclopedia Britannica. Ester, P., L. Halman, and B. Seuren. 1994. “Environmental Concern and Offering Willingness in Europe and North America.” In The Individualizing Society: Value Change in Europe and North America, edited by P. Ester, L. Halman, and R. de Moor. Tilburg, The Netherlands: Tilburg University Press. Etzioni, A. 1993. The Spirit of Community: The Reinvention of American Society. New York: Simon & Schuster. Fontaine, J.R.J., Y.H. Poortinga, and L. Delbeke. 2005. Structural Equivalence of the Values Domain across Cultures: Separating Sampling Fluctuations from Systematic,

218

shalom h. schwartz

Meaningful Variation. In Symposium on Measuring Basic Human Values, 1st European Association for Survey Research Conference, Barcelona, Spain, July. Freedom House (ed.). various years. Freedom in the World. Lanham, MD: University Press of America. Guttman, L. 1968. “A General Nonmetric Technique for Finding the Smallest Coordinate Space for a Configuration of Points.” Psychometrica 33:469–506. Gyekye, K. 1997. Tradition and Modernity: Philosophical Reflections on the African Experience. New York: Oxford University Press. Hofstede, G. 1980. Culture’s Consequences: International Differences in Work-Related Values. Beverly Hills, CA: Sage. ——. 2001. Culture’s Consequences: Comparing Values, Behaviors, Institutions, and Organizations across Nations. 2nd ed. Beverly Hills, CA: Sage. Huntington, S.P. 1993. “The Clash of Civilizations.” Foreign Affairs 72:22–49. Inglehart, R. 1997. Modernization and Postmodernization: Cultural, Economic and Political Change in 43 Societies. Princeton, NJ: Princeton University Press. ——, and W.E. Baker. 2000. “Modernization, Cultural Change, and the Persistence of Traditional Values.” American Sociological Review 65:19–51. ——, M. Basanez, J. Diez-Medrano, L. Halman, and R. Luijkx. 2004. Human Beliefs and Values: A Cross-Cultural Sourcebook Based on the 1999 –2002 Value Surveys. Buenos Aires: Siglo XXI Editores. Kohn, M.L., and C. Schooler. 1983. Work and Personality. Norwood, NJ: Ablex. Kurian, G.T. 1984. The New Book of World Rankings. New York: Facts on File Publishers. Lewis, B. 2003. What Went Wrong: The Clash between Islam and Modernity in the Middle East. New York: Harper Collins. Matsumoto, D.R. 2002. The New Japan: Debunking Seven Cultural Stereotypes. Boston: Intercultural Press. Naroll, R. 1973. “Galton’s Problem.” Pp. 974–989 in A Handbook of Method in Cultural Anthropology, edited by R. Naroll and R. Cohen. New York: Columbia University Press. Population Crisis Committee. 1988. Population Briefing Paper No. 20. Washington, DC: Population Crisis Committee, June. Putnam, R.D. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. ——. 2000. Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster. Raveh, A. 2000. “Co-plot: A Graphic Display Method for Geometrical Representations of MCDM.” European Journal of Operational Research 125:670–678. Schwartz, S.H. 1992. “Universals in the Content and Structure of Values: Theory and Empirical Tests in 20 Countries.” Pp. 1–65 in Advances in Experimental Social Psychology 25, edited by M. Zanna. New York: Academic Press. ——. 1994a. “Are There Universal Aspects in the Content and Structure of Values?” Journal of Social Issues 50:19–45. ——. 1994b. “Beyond Individualism/Collectivism: New Cultural Dimensions of Values.” Pp. 85–119 in Individualism and Collectivism: Theory, Method and Applications, edited by U. Kim, H.C. Triandis, C. Kagitcibasi, S.-C. Choi, and G. Yoon. Newbury Park, CA: Sage. ——. 1996. “Value Priorities and Behavior: Applying a Theory of Integrated Value Systems.” In The Psychology of Values: The Ontario Symposium, Vol. 8, edited by C. Seligman, J.M. Olson, and M.P. Zanna. Hillsdale, NJ: Erlbaum. ——. 1999. “Cultural Value Differences: Some Implications for Work.” Applied Psychology: An International Review 48:23–47. ——. 2003. “A Proposal for Measuring Value Orientations across Nations.” In Questionnaire Development Report of the European Social Survey (chap. 7). Retrieved from http://naticent02 .uuhost.uk.uu.net/questionnaire/chapter_07.doc.

a theory of cultural value orientations

219

——. 2004. “Mapping and Interpreting Cultural Differences around the World.” In Comparing Cultures, Dimensions of Culture in a Comparative Perspective, edited by H. Vinken, J. Soeters, and P. Ester. Leiden, The Netherlands: Brill. ——. 2005a. “Basic Human Values: Their Content and Structure across Countries.” Pp. 21–55 in Valores e comportamento nas organizações [Values and Behavior in Organizations], edited by A. Tamayo and J.B. Porto. Petrópolis. Brazil: Vozes. ——. 2005b. “Robustness and Fruitfulness of a Theory of Universals in Individual Human Values.” Pp. 56–95 in Valores e comportamento nas organizações [Values and Behavior in Organizations], edited by A. Tamayo and J.B. Porto. Petrópolis, Brazil: Vozes. ——. 2006. “Value Orientations: Measurement, Antecedents and Consequences across nations.” In Measuring Attitudes Cross-Nationally—Lessons from the European Social Survey, edited by R. Jowell, C. Roberts, R. Fitzgerald, and G. Eva. London: Sage. ——, and K. Boehnke. 2004. “Evaluating the Structure of Human Values with Confirmatory Factor Analysis.” Journal of Research in Personality 38:230–255. ——, A. Bardi, and G. Bianchi. 2000. “Value Adaptation to the Imposition and Collapse of Communist Regimes in Eastern Europe.” Pp. 217–237 in Political Psychology: Cultural and Cross Cultural Perspectives, edited by S.A. Renshon and J. Duckitt. London: Macmillan. ——, and A. Bardi. 1997. “Influences of Adaptation to Communist Rule on Value Priorities in Eastern Europe.” Political Psychology 18:385–410. ——, and M. Ros. 1995. “Values in the West: A Theoretical and Empirical Challenge to the Individualism-Collectivism Cultural Dimension.” World Psychology 1:99–122. ——, and T. Rubel. 2005. “Sex Differences in Value Priorities: Cross-Cultural and Multimethod Studies.” Journal of Personality and Social Psychology 89:1010–1028. ——, G. Melech, A. Lehmann, S. Burgess, and M. Harris. 2001. “Extending the CrossCultural Validity of the Theory of Basic Human Values with a Different Method of Measurement.” Journal of Cross Cultural Psychology 32:519–542. Tetlock, P.E. 1986. “A Value Pluralism Model of Ideological Reasoning.” Journal of Personality and Social Psychology 50:819–827. Triandis, H. 1995. Individualism and Collectivism. Boulder, CO: Westview. United Nations Population Division. 1998. World Population Monitoring. Earlier update of Website: http://unstats.un.org/unsd/mi. United Nations Women Watch. 1999. Fact Sheet on Women in Government. Update of Website: http://www.un.org/womenwatch/daw/public/percent.htm. Vanhanen, T. 1997. Prospects of Democracy. London: Routledge. Weber, M. 1958. The Protestant Ethic and the Spirit of Capitalism. New York: Scribners. Welzel, C., R. Inglehart, and H.-D. Klingemann. 2003. “The Theory of Human Development: A Cross-Cultural Analysis.” European Journal of Political Research 62:341–379. Williams Jr., R.M. 1968. “Values.” In International Encyclopedia of the Social Sciences, edited by E. Sills. New York: Macmillan. Women’s Social Equality. 1988. Population Briefing Paper No. 20, June 1988, Washington DC: Population Crisis Committee.

ISLAMIC CULTURE AND DEMOCRACY: TESTING THE ‘CLASH OF CIVILIZATIONS’ THESIS Pippa Norris and Ronald Inglehart In seeking to understand the causes of the events of 9/11 many popular commentators have turned to Samuel P. Huntington’s provocative and controversial thesis of a ‘clash of civilizations.’ This account emphasized that the end of the Cold War brought new dangers. “In the new world,” Huntington argued (1996:28), “. . . the most pervasive, important and dangerous conflicts will not be between social classes, rich and poor, or other economically defined groups, but between people belonging to different cultural entities. Tribal wars and ethnic conflicts will occur within civilizations . . . And the most dangerous cultural conflicts are those along the fault lines between civilizations . . . For fortyfive years the Iron Curtain was the central dividing line in Europe. That line has moved several hundred miles east. It is now the line separating peoples of Western Christianity, on the one hand, from Muslim and Orthodox peoples on the other.” For Huntington, Marxist class warfare, and even the disparities between rich and poor nations, have been overshadowed in the twenty-first century by Weberian culture. This influential account appeared to offer insights into the causes of violent ethno-religious conflicts exemplified by Bosnia, the Caucuses, the Middle East, and Kashmir. It seemed to explain the failure of political reform to take root in many Islamic states, despite the worldwide resurgence of electoral democracies around the globe. The framework seemed to provide a powerful lens that the American media used to interpret the underlying reasons for the terrorist attack on the World Trade Center. Commentators often saw 9/11 as a fullscale assault on the global hegemony of America, in particular, and a reaction by Islamic fundamentalists against Western culture, in general. Nevertheless, the Huntington thesis has been highly controversial. The claim of rising ethnic conflict in the post-Cold War era has come under repeated and sustained attack (Gurr 2000; Russett et al. 2000; Fox 2001; Chirot 2001; Henderson and Tucker 2001; Fox 2001). Many scholars have challenging the existence of a single Islamic culture stretching all the way from Jakarta to Lagos, let alone one that held values deeply incompatible with democracy (Kabuli 1994; Esposito and Voll 1996; Shadid 2001). What has been less widely examined, however,

222

pippa norris and ronald inglehart

is systematic empirical evidence of whether the publics in Western and Islamic societies share similar or deeply divergent values, and, in particular, whether any important differences between these cultures rest on democratic values (as Huntington claims) or on social values (as modernization theories suggest). This study seeks to throw new light on this issue by examining cultural values in seventy-five nations around the globe, including nine predominately Islamic societies, utilizing the World Values Survey/ European Values Survey (WVS/EVS) 1995–2001. Part I briefly outlines the Huntington thesis and the response by critics. Part II lays out the study’s research design including the core hypothesis, comparative framework, and survey data. Part III analyzes the evidence. The conclusion summarizes the results and reflects on their implications. The evidence confirms the first claim in Huntington’s thesis: culture does matter, and matter a lot: religious legacies leave a distinct and lasting imprint on contemporary values. But Huntington is mistaken in assuming that the core ‘clash’ between the West and Islamic societies concerns political values: instead the evidence indicates that surprisingly similar attitudes towards democracy are found in the West and the Islamic world. We do find significant cross-cultural differences concerning the role of religious leaders in politics and society, but these attitudes divide the West from many other countries around the globe, not just Islamic ones. The original thesis erroneously assumed that the primary cultural fault line between the West and Islam concerns government, overlooking a stronger cultural divide based on issues of gender equality and sexual liberalization. Cohort analysis suggests that as younger generations in the West have gradually become more liberal on these issues, this has generated a growing cultural gap, with Islamic nations remaining the most traditional societies in the world. The central values separating Islam and the West revolve far more centrally around Eros than Demos. Part I: The ‘Clash of Civilizations’ Debate The ‘clash of civilizations’ thesis advances three central claims. First, Huntington suggests that ‘culture matters’; in particular that contemporary values in different societies are path-dependent, reflecting long-standing legacies associated with core ‘civilizations.’ The concept of ‘civilization’ is understood by Huntington as a ‘culture writ large’: “It is defined both by common objective elements, such as language, history, religion,

islamic culture and democracy

223

customs, institutions, and by the subjective self-identification of people.” (Huntington 1996:41–43). Of these factors, Huntington sees religion as the central defining element (p. 47), although he also distinguishes regional sub-divisions within the major world religions, such as the distinct role of Catholicism in Western Europe and Latin America, due to their different historical traditions and political legacies. Second, the ‘clash’ thesis claims that there are sharp cultural differences between the core political values common in societies sharing a Western Christian heritage—particularly those concerning representative democracy—and the beliefs common in the rest of the world, especially Islamic societies. For Huntington, the defining features of Western civilization include the separation of religious and secular authority, the rule of law and social pluralism, the parliamentary institutions of representative government, and the protection of individual rights and civil liberties as the buffer between citizens and the power of the state: “Individually almost none of these factors was unique to the West. The combination of them was, however, and this is what gave the West its distinctive quality.” (1996:70–71) Other accounts have commonly stressed that the complex phenomenon of ‘modernization’ encompasses many additional social values that challenge traditional beliefs, notably faith in scientific and technological progress, belief in the role of economic competition in the marketplace, and the diffusion of modern social mores, exemplified by sexual liberalization and equality for women (Inglehart 1997; Inglehart and Baker 2000; Inglehart and Norris 2003). But Huntington’s claim is that the strongest distinguishing characteristic of Western culture, the aspect which demarcates Western Christianity most clearly from the Muslim and Orthodox worlds, concerns the values associated with representative democracy. This claim is given plausibility by the failure of electoral democracy to take root in most states in the Middle East and North Africa (see Midlarsky 1998). According to the annual assessment made by the Freedom House (2002), of the 192 countries around the world, two-thirds (121) are electoral democracies. Of the 47 countries with an Islamic majority, one quarter (11) are electoral democracies. Furthermore, none of the core Arabic-speaking societies in the Middle East and North Africa falls into this category. Given this pattern, in the absence of survey evidence concerning the actual beliefs of Islamic publics, it is commonly assumed that they have little faith in the principles or performance of democracy, preferring strong leadership and rule by traditional religious authorities to the democratic values of pluralistic competition, political participation, and political rights and civil liberties.

224

pippa norris and ronald inglehart

Lastly, Huntington argues that important and long-standing differences in political values based on predominant religious cultures will lead to conflict between and within nation-states, with the most central problems of global politics arising from an ethno-religious ‘clash.’1 It remains unclear whether Huntington is claiming that the core cleavage concerns Western democratic values versus the developing world, or whether the main contrast lies as a fault line between the West and Islam, but the latter has been the primary popular interpretation of the thesis, and the one which has aroused the most heated debate. Middle Eastern area studies specialists, scholars of the Koran, and students of Islamic law have contested a series of issues about the ‘clash’ thesis. Critics have challenged the notion of a single Islamic culture, pointing to substantial contrasts found among one billion people living in diverse Islamic nations, such as Pakistan, Jordan, Azerbaijan, Indonesia, Bangladesh, and Turkey, and the differences between Muslims who are radical or moderate, traditional or modern, conservative or liberal, hardline or revisionist (Hunter 1998; Esposito 1997; Fuller 2002). Observers stress the manifold differences within the Islamic world due to historical traditions and colonial legacies, ethnic cleavages, levels of economic development, and the role and power of religious fundamentalists in different states, claiming that it makes little sense to lump together people living in Jakarta, Riyadh, and Istanbul. Along similar lines, the idea that we can recognize a single culture of ‘Western Christianity’ is to over-simplify major cross-national differences, even among affluent postindustrial societies as superficially similar as the United States, Italy, and Sweden, for example the contrasts between Catholic Mediterranean Europe and Protestant Scandinavia, as well as among social sectors and religious denominations within each country. Moreover, setting this issue aside for the moment, even if we accept the existence of a shared ‘Islamic’ culture, scholars have also argued that the core values and teaching of the Koran are not incompatible with those of democracy (Kabuli 1994; Esposito and Voll 1996; Shadid 2001). Said (2001) decried Huntington’s thesis as an attempt to revive the ‘black-white,’ ‘us-them,’ or ‘good-evil’ world dichotomy that had been so prevalent during the height of the Cold War, substituting threats from ‘Islamic terrorists’ for those from ‘Communist spies.’ Western leaders, 1 International relations scholars have strongly challenged the evidence for Huntington’s claim that ethnic inter-state conflict has increased during the 1990s (Gurr 2000; Russett et al. 2000; Fox 2001; Chirot 2001; Henderson and Tucker 2001), although this body of work is not central to the argument presented here.

islamic culture and democracy

225

seeking to build a global coalition against the followers of Osama Bin Laden, took pains to distance themselves from the clash of civilizations thesis, stressing deep divisions within the Islamic world between the extreme fundamentalists and moderate Muslims. Leaders emphasized that the events of September 11th arose from the extreme ideological beliefs held by particular splinter groups of Al-Qaeda and Taliban fundamentalists, not from mainstream Muslim public opinion. Just as it would be a mistake to understand the 1995 bombing in Oklahoma City as a collective attack on the federal government by all Christian fundamentalists, rather than the work of a few individuals, it may inappropriate to view the attack by Al-Qaeda terrorists on symbols of American capitalism and financial power as a new ‘clash of civilizations’ between Islamic and Western cultures. As well as challenging the basic premises of the ‘clash of civilizations’ thesis, alternative explanations of radical Islamic fundamentalism suggest that the underlying root causes lie in deep disparities between rich and poor within societies, buttressed by the pervasive inequalities in political power in Middle Eastern regimes (Chirot 2001). Structural or neo-Marxist theories suggest that the best predictors of radical disaffection lie in uneven patterns of modernization around the world and the existence of pervasive inequalities within many Muslim societies. The most important cleavage may be between middle class, more affluent, educated and professional social sectors on the one hand,—the teachers, doctors, and lawyers in Cairo, Beirut and Islamabad—and the sub-strata of poorer, uneducated, and unemployed younger men living in Saudi Arabia, Libya, and Syria who, if disaffected, may become willing recruits to Islamic fundamentalist causes. Huntington distinguishes certain demographic characteristics of Islamic societies, notably the phenomena of the ‘youth bulge,’ but does not pursue the consequences of this generational pattern, in particular whether younger men from poorer sectors of society are particularly prone to political disaffection. Yet there are plausible alternative theories about the major cultural contrasts we could expect to find between Islam and the West. In work presented elsewhere (Inglehart and Norris 2003) we document how the modernization process has transformed values by generating a rising tide of support for equality between women and men in post-industrial societies, and greater approval in these societies of a more permissive and liberal sexuality, including tolerance of divorce, abortion and homosexuality. The version of modernization theory developed by Inglehart (1997) hypothesizes that human development generates changed cultural

226

pippa norris and ronald inglehart

attitudes in virtually any society, although values also reflect the imprint of each society’s religious legacies and historical experiences. Modernization brings systematic, predictable changes in gender roles. The impact of modernization operates in two key phases: i) Industrialization brings women into the paid work force and dramatically reduces fertility rates. Women attain literacy and educational opportunities. Women are enfranchised and begin to participate in representative government, but still have far less power than men. ii) The postindustrial phase brings a shift toward greater gender equality as women move into higher status economic roles in management and the professions, and gain political influence within elected and appointed bodies. Over half of the world has not yet entered this phase; only the more advanced industrial societies are currently moving on this trajectory. These two phases correspond to two major dimensions of cross-cultural variation: (i) A transition from traditional to secular-rational values; and (ii) a transition from survival to self-expression values. The decline of the traditional family is linked with the first dimension. The rise of gender equality is linked with the second. Cultural shifts in modern societies are not sufficient by themselves to guarantee women equality across all major dimensions of life; nevertheless through underpinning structural reforms and women’s rights they greatly facilitate this process (Inglehart and Norris 2003). If this theory is applied to cultural contrasts between modern and traditional societies, it suggests that we would expect one of the key differences between the Western and Islamic worlds to focus around the issues of gender equality and sexual liberalization, rather than the democratic values that are central to Huntington’s theory. Part II: Hypotheses, Comparative Framework, and Data To summarize, many issues arising from the ‘clash’ thesis could be considered, but here we focus upon testing two alternative propositions arising from the theoretical debate. Huntington emphasizes that the political values of democracy originated in the West with the separation of church and state, the growth of representative parliamentary institutions, and the expansion of the franchise. As such, he predicts that, despite the more recent emergence and consolidation of ‘Third Wave’ democracies in many parts of the world, democratic values will

islamic culture and democracy

227

be most deeply and widely entrenched in Western societies. If true, we would expect to find the strongest cultural clash in political values would be between the Western and Islamic worlds. In contrast, Inglehart’s modernization theory suggests that a rising tide of support for women’s equality and sexual liberalization has left a particularly marked imprint upon richer postindustrial nations, although traditional attitudes continue to prevail in poorer developing societies. Accordingly, given this interpretation, we also test the alternative proposition that any deep-seated cultural divisions between Islam and the West will revolve far more strongly around social rather than political values, especially concerning the issues of sexual liberalization and gender equality. The issues of cultural conflict and value change have generated considerable controversy but, as yet, almost no systematic survey data has been available to compare public opinion towards politics and society in many Middle Eastern and Western societies. Interpretations by area scholars and anthropologists have relied upon more qualitative sources, including personal interviews, observations and direct experience, and traditional textual exegesis of the literature, religious scriptures, and historical documents (see, for example, Lewis 2002). Recently commercial companies have started to conduct opinion polls that are representative of the public in a limited range of Muslim nations;2 Gallup’s survey examined attitudes towards other countries in nine Middle Eastern societies and the United States (Moore 2002), while Roper Reports Worldwide compared social values in the United States and Saudi Arabia (Miller and Feinberg 2002). The latest waves of the World Values Survey/European Values Survey (WVS/EVS), a global investigation of socio-cultural and political change, allow comparison of democratic values across a wide range of Western and Muslim nations, as well as in many other states.3 The

2 The main exceptions are the first-ever Gallup survey in nine predominately Islamic societies which was carried out to monitor reactions to the events of 9/11. Gallup surveyed 10,000 people in December 2001 and January 2002, with researchers conducting hour-long, in-person interviews in Saudi Arabia, Iran, Pakistan, Indonesia, Turkey, Lebanon, Kuwait, Jordan and Morocco. For details see http://www.gallup .com/poll/releases/pr020305.asp. In addition Roper Reports Worldwide conducted an annual worldwide survey from October 2001–January 2002 in 30 nations, including an urban sample of 1000 residents in the metropolitan areas in Saudi Arabia. For details of the Roper results see Miller and Feinberg (2002). 3 The following analysis draws upon a unique data base, the World Values Survey/ European Values Survey (WVS/EVS). We owe a large debt of gratitude to the WVS

228

pippa norris and ronald inglehart

study has carried out representative national surveys of the basic values and beliefs of publics in more than 70 nations on all six inhabited continents, containing over 80% of the world’s population. It builds on the European Values Surveys, first carried out in 22 countries in 1981. A second wave of surveys, in 43 nations, was completed in 1990–1991, a third wave was carried out in 50 nations in 1995–1996, and a fourth wave with more than 60 nations took place in 1999–2001.4 This total sample includes almost a quarter-million respondents, facilitating analysis of minority sub-groups, such as the Muslim populations living in Russia, India, Bulgaria, and Macedonia. This study focuses on analyzing attitudes and values in the last two waves of the survey, from 1995–2001. To test the evidence for the ‘clash of civilizations’ thesis, this study compares values at societal-level, based on the assumption that predominant cultures exert a broad and diffuse influence upon all people living under them.5 Classifying Cultural Regions In Huntington’s account nine major contemporary civilizations can be identified, based largely on the predominant religious legacy in each society: • Western Christianity (a European culture that subsequently spread to North America, Australia and New Zealand), • Islamic (including the Middle East, Northern Africa, and parts of South East Asia), • Orthodox (Russian and Greek),

and EVS participants for creating and sharing this invaluable dataset. Their names are listed in the WVS and EVS websites. For more information about the World Values Survey, see the WVS web sites http://wvs.isr.umich.edu/ and http://www .worldvaluessurvey.com. Most of the European surveys used here were gathered by the European Values Survey group (EVS). For detailed EVS findings, see Loek Halman, The European Values Study: A Sourcebook Based on the 1999/2000 European Values Study Surveys. Tilburg: EVS, Tilburg University Press, 2001. For more information, see the EVS website, http://evs.kub.nl. 4 Full methodological details about the World Values Survey/European Values Survey (WVS/EVS), including the questionnaires, sampling procedures, fieldwork procedures, principle investigators, and organization can be found at: http://wvs.isr.umich.edu/wvssamp.html. The four waves of this survey took place from 1981 to 2001, although it should be noted that all countries were not included in each wave. 5 In addition a distinct ‘Jewish’ culture could be identified, but Israel was not included within the current release of the WVS.

islamic culture and democracy

229

• Latin American (predominately Catholic yet with a distinct corporatist, authoritarian culture), • Sinic/Confucian (China, South Korean, Vietnam and Korea), • Japanese, • Hindu, • Buddhist (Sri Lanka, Burma, Thailand, Laos, and Cambodia), and (possibly) • Sub-Saharan Africa.6 Huntington treats states or societies as the core actors exemplifying these civilizations, although recognizing that populations with particular cultural and religious identities spread well beyond the border of the nation-state. Moreover some plural societies are deeply divided, so there is rarely a clean one-to-one mapping, apart from exceptional cases such as Japan and India. To analyze the survey evidence for these propositions, societies were classified into these categories, (see Table 1) based on the predominant (plurality) religious identities within each nation. The survey includes nine societies with a Muslim majority (ranging from 71 to 96 percent), including Jordan, Pakistan, Turkey, Azerbaijan, Bangladesh and Albania, Morocco, Iran and Egypt. This allows us to compare a range of states within the Islamic world, including semi-democracies with elections and some freedoms, exemplified by Albania, Turkey and Bangladesh, as well as constitutional monarchies ( Jordan), and suspended semi-democracies under military rule (Pakistan). Geographically these nations are located in Central Europe, the Middle East, and South Asia. In addition, the comparative framework includes 22 nations based on ‘Western Christianity’ (using Huntington’s definition to include both predominately Catholic and Protestant postindustrial societies, and countries like Australia and New Zealand which are not located regionally in the ‘West’ yet which inherited a democratic tradition from Protestant Britain). Other nations are classified into distinct civilizational traditions including Latin America (11), Russian or Greek Orthodox (12), Central European (10 nations sharing a common Western Christian

6 Although it should be noted that despite the centrality of the concept, the definition, labeling and classification of ‘civilizations’ remains inconsistent in Huntington’s work, for example it remains unclear whether Huntington believes that there is or is not a distinct African civilization, and the major discussion of types (pp. 45–47) excludes the Orthodox category altogether.

France Ireland Italy Malta Portugal Spain Portugal

Canada Denmark Finland Iceland Netherlands New Zealand Northern Ireland Norway Sweden Switzerland United States West Germany Moldova Montenegro Romania Russia Serbia Ukraine

Lithuania Poland Slovakia Slovenia

Bulgaria Georgia Greece Macedonia

Central Europe Croatia Czech Republic East Germany Estonia Hungary Latvia

Belarus Bosnia

Orthodox South Korea Taiwan

Sinic/Confucian Africa

Mexico Peru Uruguay Venezuela

Chile Vietnam Colombia China Dominican Rep El Salvador

Argentina Brazil

Latin America

Nigeria South Africa Tanzania Uganda Zimbabwe

SubSaharan

Note: This study compares 72 nation states and 75 societies, dividing states with distinctive historical traditions, cultural legacies and political institutions including the UK (Northern Ireland and Great Britain), Germany (East and West), and the Federal Republic of Yugoslavia (Serbia and Montenegro). The Catholic and Protestant societies are classified as ‘Western Christianity.’ In addition India and Japan are each treated as separate religious cultures. Source: The World Values Survey/European Values Survey (WVS/EVS), 1995–2001.

Bangladesh Egypt Iran Jordan Morocco Pakistan Turkey

Austria Albania Belgium Azerbaijan

Australia Britain

Islamic

Catholic

Protestant

Table 1. Classification of societies by the historically predominant religion

230 pippa norris and ronald inglehart

islamic culture and democracy

231

heritage with the West yet with the distinct experience of living under Communist rule), sub-Saharan Africa (5), South-East Asian (4 societies reflecting Sinic/Confucian values), plus Japan and India. In addition, ten societies contain a significant minority Islamic population (ranging from 4 to 27 percent), including Bosnia, Macedonia, Nigeria and India, although these nations have Orthodox, Protestant, or Hindu majority populations. In the multivariate regression models, each type of society was coded as a dummy variable and the ‘Western’ societies category was used as the (omitted) reference category. The models therefore measure the impact of living in each of these types of society, with controls, compared with living in the West. To rule out intervening variables, multivariate regression models compare the influence of predominant religious cultures in each type of society controlling for levels of human and political development. Modernization theories suggest that this process brings certain predictable shifts in cultural values, including declining belief in traditional sources of religious authority and rising demands for more participatory forms of civic engagement (Inglehart 1997; Inglehart and Baker 2000; Norris 2002). The WVS/EVS survey contains some of the most affluent market economies in the world, such as the US, Japan and Switzerland, with per capita annual incomes as high as $40,000; together with middlelevel industrializing countries such as Taiwan, Brazil, and Turkey, as well as poorer agrarian societies, such as Uganda, Nigeria, and Viet Nam, with per capita annual incomes of $300 or less. It also includes many different types of states, including established and newer democracies, semi-democracies, and non-democracies. Accordingly structural differences among societies are measured by the United Nations Development Program (UNDP) Human Development Index (HDI) 2000, (combining levels of per capita income, literacy and schooling, and longevity), and levels of democratization, which are classified based on the 1999–2000 Freedom House analysis of political rights and civil liberties.7 The structural differences among groups within societies are measured by the standard social indicators, including income (as the most reliable cross-cultural measure of socioeconomic status in different societies), education, gender, age, and religiosity.

7 These countries are ranked as equally ‘free’ according to the 2000–2001 Freedom House assessments of political rights and civil liberties Freedom House. 2000. Freedom in the World 2000 –2001. www.freedomhouse.org.

232

pippa norris and ronald inglehart

The latter was included to see whether the strength of religious beliefs influenced values more than the type of religious faith or identity (which, like being baptized as a Protestant or Catholic, can be purely nominal). To develop a religiosity scale, factor analysis was used with six indicators selected from the pooled World Values Survey/European Values Survey (WVS/EVS), namely the proportion of the population in different societies: (i) who say that religion is ‘very important’ in their lives, (ii) who find comfort in religion, (iii) who believe in God, (iv) who identify themselves as religious, (v) who believe in life after death, and (vi) who attend a religious service regularly. All these items tap values and beliefs common throughout the world’s religions and they were carried in all four waves of the WVS/EVS, to facilitate comparison over time. Factor analysis among the pooled sample (not reproduced here) showed that all the items fell into one dimension and formed a consistent and reliable ‘strength of religiosity’ scale (Cronbach’s Alpha = 0.48). After recoding, the scale was standardized to 100-points, for ease of interpretation, where the higher score represents the strongest religiosity. Measuring Political and Social Values Attitudes were compared towards three dimensions of political and social values: (i) support for democratic ideals and performance, (ii) attitudes towards political leadership, and (iii) approval of gender equality and sexual liberalization. As argued elsewhere (Norris 1999), an important distinction needs to be drawn between support for the ideals of democracy and evaluations of the actual performance of democracy. Evidence from previous waves of the World Value Study (Klingemann 1999; Dalton 1999) suggests that citizens in many countries adhere strongly to the general principles of democracy, such as believing that it is the best form of government and disapproving of authoritarian alternatives, and yet at the same time many remain deeply dissatisfied with the way that democratic governments work in practice. The phenomenon of more ‘critical citizens’ (Norris 1999) or ‘disenchanted democrats’ (Putnam and Pharr 2001) has been widely observed. To examine these dimensions, attitudes towards the principles and performance of democracy are measured in this study using the items listed in Table 2, where respondents are invited to express agreement or disagreement with the statements. It should be noted that the performance items do not ask people about their experience of democracy in

19.6

.862 .854

17.7

.780

.853

Democratic ideals

19.6

.881 .879

Religious leadership

15.7

.838 .721

Strong leadership

Note: Principal component factor analysis was used with varimax rotation and Kaiser normalization. The total model predicts 72.6% of cumulative variance. The democratic performance scale was reversed so that a positive response expressed greater satisfaction with democracy. Source: The World Values Survey/European Values Survey (WVS/EVS), Waves III and IV (1995–2001).

% Of total variance

V170 Democracies are indecisive and have too much squabbling V171 Democracies aren’t good at maintaining order V172 Democracy may have its problems but its better than any other form of government V167 Approve of having a democratic political system V200 Politicians who do not believe in God are unfit for public office V202 It would be better for [this country] if more people with strong religious beliefs held public office. V165 Approve having experts, not government, make decisions V164 Approve having a strong leaders who does not have to bother with parliament and elections

Democratic Performance

Table 2. Factor analysis of political values

islamic culture and democracy 233

234

pippa norris and ronald inglehart

their own country, such as how well their government works, but rather taps their expectations of how well democratic governments generally function in taking decisions and maintaining order. In addition, it is commonly assumed that one of the primary contrasts between Islamic and Western cultures relates to attitudes towards the role of religious leaders, who exercise power by virtue of their spiritual authority, or secular leaders who hold authority through elective office, reflecting deeper beliefs about the separation of church and state. We therefore also monitored support for the role of religious leaders in public life with the items listed in Table 2. Neither of these items cued respondents with any explicit reference to ‘democracy’ and indeed, in principle, there is no inconsistency in believing both in the value of spiritual authorities and in the principles of democracy, if the religious leaders exercise power through elected office, exemplified by Christian Democrat parties or politicians from the Christian far right. We also sought to compare attitudes towards preferences for strong leadership, measured by questions tapping support for non-democratic forms of government by experts or by leaders unaccountable to parliament or elections. Factor analysis confirmed that these political items did indeed fall into four distinct dimensions. Accordingly summary scales were constructed, each standardized to 100-points for ease of interpretation and consistent comparison across measures. Yet the alternative proposition is that the transformation of social values towards sexuality and women’s equality, which has profoundly affected the younger generation in postindustrial societies, may lie at the heart of any cultural clash between modern and traditional societies in general, and between the West and Islam in particular. In this regard, Huntington may have correctly identified the importance of civilizational values, but may have misdiagnosed the most crucial aspects of cultural differences. To explore this proposition we can compare support for gender equality, using a standardized scale developed elsewhere, also based on factor analysis, monitoring attitudes towards the roles of women and men in the workforce, education, politics, and the family.8 The Gender Equality items are similar to those commonly

8 The combined 100-pt gender equality scale is based on the following 5 items: MENPOL Q118: “On the whole, men make better political leaders than women do.” (Agree coded low); MENJOBS Q78: “When jobs are scarce, men should have more right to a job than women.” (Agree coded low); BOYEDUC Q.119: “A university education is more important for a boy than a girl.” (Agree coded low); NEEDKID Q110

islamic culture and democracy 0

10

20

30

40

50

60

235 70

Democratic Performance

80

90

68 68 86 87

Democratic ideals Religious leaders

62

39

61 61

Strong leaders Gender equality Homosexuality Abortion Divorce

100

82

55 53

12 48

25 35 Western

60

Islamic

Figure 1. Approval of political and social values in Western and Islamic societies. Note: Mean approval. See Table 3 for details. Source: (WVS/EVS), pooled sample 1995–2001.

contained in the more comprehensive psychological scales of sex roles. The gender equality scale was summed across the component items and standardized to 100-points for ease of interpretation. We also compared attitudes using 10-point scales monitoring approval or disapproval of three related dimensions of changing sexual mores, concerning homosexuality, abortion and divorce.

“Do you think that a woman has to have children in order to be fulfilled or is this not necessary?” (Agree coded low); SGLMUM Q112 “If a woman wants to have a child as a single parent but she doesn’t want to have a stable relationship with a man, do you approve or disapprove?” (Disapprove coded low). Three items used statements with Lickert-style 4-point agree-disagree responses, while two used dichotomies, and these items were all recoded so that higher values consistently represent greater support for gender equality. Principal component factor analysis revealed that all five items fell into a single consistent scale (not reproduced here), with a Cronbach’s Alpha of 0.54. For details of the construction, reliability, validity, and distribution of this scale see Ronald Inglehart and Pippa Norris. 2003. Rising Tide: Gender Equality and Cultural Change Around the World. New York: Cambridge University Press.

.03***

.02*** 117855

78 81 81 80 83 84 81 83 .20***

61 63 62 68 65 60 70 65 .18***

116629

86 87 80

68 68 63

49903

.30***

54 62 51 64 40 N/a 66 53 .49***

62 39 53

83223

.01

55 56 55 52 52 53 58 57 .19***

61 61 55

84932

.64***

64 67 75 62 64 61 63 67 .46***

82 55 67

135846

.51***

22 36 31 17 21 17 40 33 .45***

53 12 28

139841

.33***

46 48 23 33 22 25 46 38 .38***

48 25 36

139311

.37***

51 56 49 40 31 31 61 50 .33***

60 35 47

Democratic Political Values Liberal Social Values Approve of Approve of Disapproval of Disapproval of Approve of Approve of Approve of Approve of Democratic Democratic Religious strong leaders gender homosexuality abortion divorce Performance ideals leaders equality

Note: For the classification of societies see Table 1. All items have been scaled to 0–100. The significance of the difference between group means is measured by ANOVA (Eta) without any controls. *** Sig. P.000. Source: The (WVS/EVS), Waves III and IV (1995–2001).

Western Christianity Islamic All Other All Other Orthodox Central Europe Latin America Sinic/Confucian Sub-Saharan Africa Hindu Japanese ALL Difference between all group means Difference between Western and Islamic group N.

Type of society

Table 3. Mean scores on the political and social value scales

236 pippa norris and ronald inglehart

islamic culture and democracy

237

Part III: Analysis of the Results Table 3 compares the mean scores on these scales for each type of society, without any prior social or demographic controls, and the significance of the difference between societies. Figure 3 illustrates the contrasts between Western and Islamic societies. The results show that, contrary to the first hypothesis, Western and Islamic societies generally agreed on three of the four indicators of political values. Approval of how well democracy worked in practice was similar for those living in the West and in Islamic societies while, in contrast, far more critical evaluations were expressed in all other cultures around the globe, with the single exception of Japan. Similar patterns were evident when people were asked whether they supported democratic ideals, for example whether democracy was better than any other form of government. As others have reported (Klingemann 1999), in recent years high support for democratic ideals is almost universally found in most nations around the globe. Both Western and Islamic societies expressed similar levels of approval, while in contrast slightly less positive attitudes were evident elsewhere, with Sinic and Orthodox societies proving the least enthusiastic. Attitudes towards leadership by experts and by unaccountable government officials were also similar in Islamic and Western societies. Therefore the major political disagreement between Western and Islamic societies was found in attitudes towards the role of religious leaders, where Islamic nations proved far more favorable. Yet at the same time it would be an exaggeration to claim that this latter difference represents a simple dichotomous ‘clash of values’. Although it is true that many more Muslims than Westerners supported the idea of religious authorities, there was widespread agreement with this idea in many other parts of the world including Sub-Saharan Africa and Catholic Latin America. The West proved more secular in orientation, as did Central Europe, the Sinic/Confucian nations, and Japan. Yet comparing the simple means in each type of religious culture could be misleadingly if other endogenous factors are influencing the results, such as the level of democratization or economic affluence typically found in Western and Islamic societies. The multivariate OLS regression models presented in Table 4 therefore compare the impact of living within each type of religious culture after including controls for the societal-level of human and political development, and individual-level measures for age, gender, education, income, and strength of religiosity. In these models, each type of society was coded

2.6 –7.9 –5.3 –3.5 –3.1 –4.1 –2.5 –3.5 82.1 .01

95550

***

***

***

**

***

***

***

***

93965

.03 –.18 –.11 –.11 .01 –.05 –.06 .02

.06

.34 .25 .21 .24 .45 .43 .60 .49

.05

1.3 –8.9 –5.4 –6.1 1.4 –3.6 –8.9 3.3 68.8 .01

.27 .21 .17 .19 .37 .34 .47 .39

.06 –.18 –.12 –.08 –.03 –.07 –.02 –.03 ***

***

***

***

***

***

***

***

Approve of democratic Ideals 0–100 B St. Err. Beta Sig

45209

.33

9.7 5.2 0.1 3.8 –5.1 7.6 N/a –0.1 61.8 .32 .59

.41 .33 .27 .35 .79 .46 .00

.19 .09 .00 .05 –.03 .13

Favor religious leadership 0–100 B St. Err. Beta

N/s

***

***

***

***

N/s

***

***

Sig

64412

.06

–2.8 .35 5.5 .27 3.5 .24 3.3 .25 16.6 .47 4.3 .48 6.0 .62 1.7 .54 54.1 .01

B

–.06 .13 .08 .07 .16 .07 .05 .02

**

***

***

***

***

***

***

***

Favor strong leadership 0–100 St. Err. Beta Sig

Note: OLS regression models with blockwise entry with the political value scales as the dependent variables. The full model is illustrated in Table A1. Block 1 in all models control for the level of human development (Human Development Index 1998), level of political development (Freedom House 7-point index (reversed) of political rights and civil liberties 1999–2000), age (years), gender (male = 1), education (3 categories from low to high), income (10 categories), and religiosity. Block 2 then enters the type of society, based on the predominant religion, coded as dummy variables. Western societies represent the (omitted) reference category. The coefficients can be understood to represent the effect of living in each type of society compared with living in Western societies, net of all prior controls. Political value scales: For details see Table 2. Type of society: see Table 1. Sig. ***p.001; **p.01; *p.05. Source: All (WVS/EVS), pooled sample 1995–2001.

Type of society Islamic Orthodox Central European Latin American Sinic/Confucian Sub-Saharan African Hindu Japanese (Constant) Adjusted R2 Block 1 (Control variables only) Adjusted R2 Block 2 (Controls + type of society) N.

Scale

Approve of democratic Performance 0–100 B St. Err. Beta Sig

Table 4. Political values by type of society, with controls

238 pippa norris and ronald inglehart

islamic culture and democracy

239

as a dummy (0/1) variable. The Western category was excluded from the analysis, so that the dummy coefficients can be interpreted as the effect of living in these societies, after applying prior controls, compared with the effect of living in the West. The data was entered in blocks, including development and social controls in the first block, then the additional effects of the full model in the second block, including the type of society as well. The results show that after controlling for all these factors, contrary to Huntington’s thesis, compared with Western societies, support for democracy was marginally slightly stronger (not weaker) among those living in Islamic societies. This pattern was evident on three indicators: approval of the way democracy works in practice, support for democratic ideals, as well as disapproval for the idea of strong government leaders. It should be stressed that the difference on these items between Islam and the West were extremely modest in size, as shown by the strength of the standardized beta coefficient, and the statistical significance is largely the product of the large number of respondents in the pooled sample, but nevertheless the difference was in the contrary direction to that predicted by the Huntington thesis. Moreover, as observed earlier, even after introducing controls, lower support for democratic values was found in many other types of non-Western society, especially countries in Eastern and Central Europe, and Latin America, while the Sinin/Confucian states showed the greater approval of strong government. At the same time, after introducing all the controls, Islamic societies did display greater support for a strong societal role by religious authorities than do Western societies. This pattern persists despite controlling for the strength of religiosity and other social factors, which suggests that it is not simply reducible to the characteristics of people living in Islamic societies. Yet this preference for religious authorities is less a cultural division between the West and Islam than it is a gap between the West and many other types of less secular societies around the globe, especially in Sub Saharan Africa and, to a lesser extent, in Latin America. To examine these results in more detail, Figures 2 and 3 compares the location of each nation on these scales. Of all countries under comparison, Russia proved a striking outlier in Figure 2, displaying widespread disillusionment with the way that democratic processes worked, as well as little enthusiasm for democratic ideas. Other Orthodox societies also showed minimal faith in democracy, including the Ukraine, Moldova, Belarus, Georgia and Macedonia. A few other developing

240

pippa norris and ronald inglehart 80

Ice Ger Malta Aus Bng Egy Bos Iran Den NireLux Swe China JapGB Zim Est NZ ItaDomR Fin Yug Nor Chil Austi Mont Greece Ara SAfr Bul Arg Tai Geo Belg Mex Peru Swi Czech PhilBela Lat Lith India Mol Ukr Ven Slov Rom Aze

Approval of democratic perfomance

70

60

Fr Rus

PolBraz

50

SOCIETY

Tanz

Other Islamic

40

Western 60

70

80

90

100

Approval of democratic ideals

Figure 2. Democratic values. Source: (WVS/EVS), pooled sample 1995–2001. 80 Den Neth Swe Slov Ice EGermJapNireSp Hung Pol Lat Bul Lux Aus Can Rus Yug Bela Slovk Ukr Bng US Zim Chil Turk Rom SAfrVenUga Phil Czech

70

Disapproval of religious leaders

60 50 40

Jor Tanz

Greece

SOCIETY

Mor

Nigeria

Other

EgyIran

30

Islamic Western

20 40

50

60

70

80

Diapproval of strong leaders

Figure 3. Leadership attitudes. Source: (WVS/EVS), pooled sample 1995–2001.

islamic culture and democracy

241

countries from different cultures proved extremely critical of the way that democracy worked in practice, although showing greater support for democratic ideals, including Tanzania, Brazil and Poland. Many diverse cultures were located in the middle of the distribution, including Turkey and Jordan as Islamic societies, as well as the United States, Italy, and the Netherlands. Nations that gave the strongest endorsement for democratic ideals and practices included the Scandinavian societies of Denmark, Iceland, and Sweden, along with Germany and Austria, but high support was also registered in Muslim Bangladesh, Egypt and Azerbaijan. Therefore in general slightly lower levels of support for democracy were evident in some Eastern European states, notably in Russia, lending some confirmation for claims of a division between the Orthodox and Western worlds. But attitudes towards democratic principles and performance generally showed a broad distribution across many diverse cultural groups, providing minimal support for the stronger claim that the West is particularly distinctive to Islam in its faith in democracy. Indeed the difference between public opinion in Eastern and Western Europe could be explained equally satisfactorily as reflecting a residual hangover from the Cold War era, and the poor performance of electoral democracies and states in these nations, rather than being interpreted as the result of cultural legacies or the emergence of any ‘new’ ethno-religious cleavage. Figure 3 compared leadership attitudes by nation. Support for religious leaders was lowest in many secular societies in Scandinavia and Western Europe, as well as in certain nations in Eastern Europe like the Czech Republic. The United States proved distinctive, showing higher than average support for religious leaders, compared with other Western nations, while Greece was another outlier. At the other extreme, support for religious leaders was relatively strong in African societies including Nigeria, Tanzania, and South Africa, as well as the Philippines, all countries with strong religiosity. Compared with Western nations, many of the Islamic nations expressed greater support for the principle of religious authorities, but they were far from alone in this regard. There is also a fascinating split over the issue of strong leadership evident within the Islamic world; more democratic countries with greater political rights and civil liberties and parliamentary traditions, exemplified by Bangladesh, and Turkey, expressed greater reservations about strong leadership. To a lesser extent, Jordan also fell into this category. In contrast the public living in Islamic countries characterized by more limited political freedoms, less democratic states, and by strong

242

pippa norris and ronald inglehart

executives, expressed greater support for strong leadership, notably in Egypt, Iran, and Morocco. Yet so far we have not compared the alternative modernization thesis that the social values of gender equality and sexual liberalization could plausibly lie at the heart of any ‘clash’ between Islam and the West. The mean scores on these social attitudes in Table 3 reveal the extent of the gulf between Islam and the West, generating a far stronger cultural gap on these issues than across most of the political values. Regression models, including the same prior controls used earlier, show that many structural factors consistently help to predict attitudes, since egalitarian and liberal values are stronger among the young, women, the well-educated, and the less religious, as well as in modern societies with greater human and democratic development. After these controls are introduced, Table 5 shows that there remains a strong and significant difference across all the social indicators (including approval of gender equality, homosexuality, abortion and divorce) among those living in Western v. Islamic societies. Figure 4 shows the distribution of nations on the scales for gender equality and homosexuality in more detail. The results confirm the consistency of the sharp differences between Islam and the West on these issues. All the Western nations, led by Sweden, Germany and Norway, strongly favor equality for women and also prove tolerant of homosexuality. Many other societies show a mixed pattern, falling into the middle of the distribution. In contrast the Islamic nations, including Egypt, Bangladesh, Jordan, Iran and Azerbaijan, all display the most traditional social attitudes, with only Albania proving slightly more liberal. We lack time-series survey data that would allow us to trace trends in the post-war era, to see whether these cultural differences between societies have widened, as we suspect, due to the modernization process in post-industrial economies. Nevertheless, if we assume that people acquire their basic moral and social values as the result of the longterm socialization process, in the family, school and community, leading to generational rather than life-cycle effects, we can analyze these attitudes for different 10-year cohorts of birth. The results in Figure 5 confirm two striking and important patterns: first, there is a persistent gap in support for gender equality and sexual liberalization between the West (which proves most liberal), Islamic societies (which prove most traditional), and all other societies (which are in the middle). Moreover, even more importantly, the figures reveal that the gap between

–.18 *** –.17 *** –.09 *** .05 *** –.01 N/s *** 13 .03 *** –.09 ***

–0.67 0.24 0.24 –1.20 –2.10 –0.08 –0.05 –0.45 3.1 .23

103290

***

***

***

***

***

***

***

***

99980

–.18 –.26 –.18 –.11 –.13 –.05 –.05 –.06

.26

.05 .04 .03 .03 .07 .06 .08 .06

.05 .04 .03 .03 .06 .06 .08 .06

–.07 .03 .03 –.14 –.10 –.01 –.01 –.02 ***

N/s N/s

***

***

***

***

***

Approve of abortion 1–10 B St. Err. Beta Sig

.21

–1.9 –2.1 –1.6 –1.0 –2.9 –0.6 –1.2 –1.5 1.6 .20

Approve homosexuality 1–10 B St. Err. Beta Sig .05 .04 .03 .04 .07 .06 .08 .07

105432

.31 2

–0.25 –0.20 0.01 0.15 –2.30 0.29 –0.10 –0.05 2.16 .26

–.03 –.03 .01 .02 –.11 .03 –.01 –.01

N/s N/s

***

***

***

N/s

***

***

Approve of divorce 1–10 B St. Err. Beta Sig

Note: OLS regression models with blockwise entry with the social value scales as the dependent variables. The full model is illustrated in Table A1. Block 1 in all models control for the level of human development (Human Development Index 1998), level of political development (Freedom House 7-point index (reversed) of political rights and civil liberties 1999–2000), age (years), gender (male = 1), education (3 categories from low to high), income (10 categories), and religiosity. Block 2 then enters the type of society, based on the predominant religion, coded as dummy variables. Western societies represent the (omitted) reference category. The coefficients can be understood to represent the effect of living in each type of society compared with living in Western societies, net of all prior controls. Type of society: see Table 1. Gender equality scale: For details see fn.7. Sexual liberalization scales: “Please tell me for each of the following statements whether you think it [Homosexuality/ abortion/ divorce] can always be justified, never be justified, or something in-between, using this card from 1 (never justifiable) to 10 (Always justifiable).” Sig. ***p.001; **p.01; *p.05. N/s Not significant. Source: All (WVS/EVS), pooled sample 1995–2001.

.35 .30 .30 .25 .69 .42 .53 .52

Approve of gender equality 0–100 B St. Err. Beta Sig

Type of society Islamic –8.2 Orthodox –8.9 Central European –6.6 Latin American 2.6 Sinic/Confucian –0.3 Sub-Saharan African 7.3 Hindu 3.4 Japanese –14.4 (Constant) 32.7 Adjusted R2 Block 1 .26 (Control variables only) Adjusted R2 Block .33 (Controls + type of society) N. 63476

Scale

Table 5. Social values by type of society, with controls

islamic culture and democracy 243

244

pippa norris and ronald inglehart

90 Fin

Swe

Nor Ger Sp NZ EGer US Aust Peru Domin Ven Arg Chil Slov Cro Uru ElSai Mex Lith Czech Yug SAfr Braz Hung Alb Lat Jap Est Slovk Mont Pol Tanz Viet Uga Ukr China Bela India Tai Phil Aze Arm Iran Col

80

Gender Equality Scale

70

60

Society Other Islamic Western

Egy JOT

50

Rsq = 0.5581

40 0

1

2

3

4

5

6

7

8

Approval of homosexuality

Figure 4. Social values. Source: (WVS/EVS), pooled sample 1995–2001.

the West and Islam is usually narrowest among the oldest generation, but that this gap has steadily widened across all the indicators as the younger generations in Western societies have become progressively more liberal and egalitarian, while the younger generations in Islamic societies remain as traditional as their parents and grandparents. The trends suggest that Islamic societies have not experienced a backlash against liberal Western sexual mores among the younger generations, but rather that young Muslims remain unchanged despite the transformation of lifestyles and beliefs experienced among their peers living in postindustrial societies. Conclusion and Discussion The thesis of a ‘clash of civilizations’ has triggered something of a ‘clash of scholarship’ among those seeking to understand the causes and consequences of ethnic-religious conflict. This task has long been of interest to academe but it has received fresh impetus by the dramatic events and aftermath of 9/11. Alternative interpretations of these issues are important for themselves, but also because they carry important

50

60

70

80

Other

Islamic

Western Christian

Type of Society

Other

Islamic

Western Christian

Type of Society

Islamic

Western Christian

Type of Society

Islamic

Western Christian

Type of Society

COHORT

Other 1 1900–1916 1927–1936 1947–1956 1967–1976 1917–1926 1937–1946 1957–1966 1977–1984

2

3

4

5

6

7

Other 0 1900–1916 1927–1936 1947–1956 1967–1976 1917–1926 1937–1946 1957–1966 1977–1984

1

2

3

4

5

6

7

Figure 5. Support for sexual liberalization values by cohort and society. Source: (WVS/EVS), pooled sample 1995–2001.

1 1900–1916 1927–1936 1947–1956 1967–1976 1917–1926 1937–1946 1957–1966 1977–1984

2

3

4

5

6

7

40 1900–1916 1927–1936 1947–1956 1967–1976 1917–1926 1937–1946 1957–1966 1977–1984

Gender equality scale

Approval of divorce

Approval of homosexuality Approval of aabortion

90

islamic culture and democracy 245

246

pippa norris and ronald inglehart

policy implications, not least for how far differences between the United States and Middle Eastern states primarily reflect the views of political elites and governing regimes, or whether they tap into deeper currents of public opinion. To summarize the core components of the Huntington thesis, the claims are threefold: societal values in contemporary societies are rooted in religious cultures; the most important cultural division between the Western and Islamic world relates to differences over democratic values; and, in the post-Cold War era, this ‘culture clash’ is at the source of much international and domestic ethnic conflict. The comparative evidence from this study, drawing upon public opinion towards democracy in more than seventy societies around the globe in 1995–2001, suggests four main findings: (i) First, when political attitudes are compared (including evaluations of how well democracy works in practice, support for democratic ideals, and disapproval of strong leaders), far from a ‘clash of values,’ there is minimal difference between the Islamic world and the West. (ii) Instead the democratic ‘clash’ (if it can be called a clash) divides Post-Communist states in Eastern European (exemplified by Russia, Ukraine and Moldova), which display minimal support for democracy, from many other countries that display far more positive attitudes, including both Western and Islamic nations. This pattern could be explained equally well as reflecting the residual legacy of the Cold War and a realistic evaluation of the actual performance of democracy in these states, rather than by the reemergence of ethnic conflict based on the values of the Orthodox church which are, after all, part of Christendom. (iii) Support for a strong societal role by religious authorities is stronger in Islamic societies than in the West, but here it is not a simple dichotomy, as many other types of society also support an active role for religious leaders in public life, including the Sub Saharan African countries under comparison as well as many Catholic nations in Latin America. (iv) Yet there is a substantial cultural cleavage, although one underestimated by Huntington, in social beliefs about gender equality and sexual liberalization. In this regard, the West is far more egalitarian and liberal than all other societies, particularly Islamic nations. Moreover cohort analysis suggests that this gap has steadily wid-

islamic culture and democracy

247

ened as the younger generation in the West has gradually become more liberal in their sexual mores while the younger generation in Islamic societies remains deeply traditional. The results indicate that modern Western societies are indeed different, in particular concerning the transformation of attitudes and behavior associated with the ‘sexual revolution’ that has occurred since the 1960s, fundamental changes in the nature of modern families, and more expressive lifestyles. Equality for women has progressed much further, and transformed traditional cultural beliefs and values about the appropriate division of sex roles far more deeply, in affluent Western societies. But at the same time any claim of a ‘clash of civilizations,’ especially of fundamentally different political values held by Western and Islamic societies, represents an over-simplification of the evidence. Across many political dimensions examined here, both Islamic and Western societies are similar in their positive orientation toward democratic ideals. Where Islam societies do differ significantly from the West, in supporting religious authorities, they are far from exceptional around the world. Any black-and-white ‘Islam versus the West’ interpretation of a ‘culture clash’ as conveyed by the popular media is far too simple. We need to compare public opinion across more dimensions, and across a wider range of nations in the Middle East, Africa and Asia. Moreover it seems likely that different understandings of what “democracy” means, prevail in different societies: some people are attracted to democracy in so far as it seems linked with prosperity, while others emphasize individual liberty and freedom of expression. Nevertheless our findings indicate that one needs to be very cautious in generalizing from the type of regime to the state of public opinion in a given country. Support for democracy is surprisingly widespread among Islamic publics, even among those who live in authoritarian societies. The most basic cultural fault line between the West and Islam involves issues of gender equality and sexual liberalization. As another article in this journal indicates (see Inglehart, Norris and Welzel), these issues have important implications for democracy. But the democratic ideal is widely shared among Islamic publics. Western cultural norms may be rejected, but the goal of living under a democratic government is not—quite the contrary, it is endorsed by clear majorities of the public throughout Islamic society.

248

pippa norris and ronald inglehart References

Chirot, D. 2001. “A Clash of Civilizations or of Paradigms? Theorizing Progress and Social Change.” International Sociology 16(3):341–360. Dalton, Russell. 1999. “Political Support in Advanced Industrialized Democracies.” In Critical Citizens: Global Support for Democratic Governance, edited by Pippa Norris. Oxford: Oxford University Press. Esposito, John (ed.). 1997. Political Islam: Revolution, Radicalism or Reform? Boulder, CO: Lynne Reinner. Esposito, John L., and John O. Voll. 1996. Democracy and Islam. New York: Oxford University Press. Fox, J. 2001. “Two Civilizations and Ethnic Conflict: Islam and the West.” Journal of Peace Research 38(4):459–472. Freedom House. 2002. “Freedom in the World 2002: The Democracy Gap.” New York: Freedom House, www.freedomhouse.org. Fuller, Graham E. 2002. “The Future of Political Islam.” Foreign Affairs 81(2):48–60. Funkhouser, G.R. 2000. “A World Ethos and the Clash of Civilizations: A Cross-Cultural Comparison of Attitudes.” International Journal of Public Opinion Research 12(1):73–79. Gurr, Ted. 2000. Peoples versus States. Washington, DC: US Institute for Peace Press. Henderson, R.A., and R. Tucker. 2001. “Clear and Present Strangers: The Clash of Civilizations and International Politics.” International Studies Quarterly 45(2):317–338. Hunter, Shireen T. 1998. The Future of Islam and the West: Clash of Civilizations or Peaceful Coexistence? Westport, CT: Praeger. Huntington, Samuel O. 1996. “The West Unique, not Universal.” Foreign Affairs 75(6):28–34. Huntington, Samuel P. 1993a. “The Clash of Civilizations?” Foreign Affairs 72(3):22–49. ——. 1993b. “If not Civilizations, What? Paradigms of the Post-Cold War World.” Foreign Affairs 72(5):186–194. ——. 1996. The Clash of Civilizations and the Remaking of World Order. New York: Simon and Schuster. ——. 1997. “The Clash of Civilizations—Response.” Millenium—Journal of International Studies 26(1):141–142. Inglehart, Ronald. 1997. Modernization and Postmodernization: Cultural, Economic and Political Change in 43 Societies. Princeton, NJ: Princeton University Press. ——, and Pippa Norris. 2003. Rising Tide: Gender Equality and Cultural Change Around the World. New York: Cambridge University Press. ——, and Wayne E. Baker. 2000. “Modernization, Globalization and the Persistence of Tradition: Empirical Evidence from 65 Societies.” American Sociological Review 65:19–55. Kabuli, Niaz Faizi. 1994. Democracy according to Islam. Pittsburgh, PA: Dorrance Publications. Klingemann, Hans Dieter. 1999. “Mapping Political Support in the 1990s: A Global Analysis.” In Critical Citizens: Global Support for Democratic Governance, edited by Pippa Norris. Oxford: Oxford University Press. Lewis, Bernard. 2002. What Went Wrong? Western Impact and Middle Eastern Response. New York: Oxford University Press. Midlarsky, M.I. 1998. “Democracy and Islam: Implications for Civilizational Conflict and the Democratic Process.” International Studies Quarterly 42(3):485–511. Miller, Thomas A.W., and Geoffrey Feinberg. 2002. “Culture Clash.” Public Perspective 13(2):6–9. Norris, Pippa (ed.). 1999. Critical Citizens: Global Support for Democratic Governance. Oxford: Oxford University Press. ——. 2002. Democratic Phoenix: Political Activism Worldwide. New York/Cambridge: Cambridge University Press.

islamic culture and democracy

249

Putnam, Robert D., and Susan Pharr (eds.). 2001. Disaffected Democracies: What’s Troubling the Trilateral Countries? Princeton, NJ: Princeton University Press. Russett B.M., J.R. O’Neal, and M. Cox. 2000. “Clash of Civilizations, or Realism and Liberalism déjà vu? Some Evidence.” Journal of Peace Research 37(5):583–608. Said, Edwards. 2001. “A Clash of Ignorance.” The Nation. Shadid, Anthony. 2001. Legacy of the Prophet: Despots, Democrats, and the New Politics of Islam. Boulder, CO: Westview Press.

250

pippa norris and ronald inglehart Technical Appendix

Table A1. Illustration of the full regression model used in Tables 4 and 5 Approve of Democratic Performance

Developmental controls Level of human development (100-point scale) Level of political development Social controls Age (Years) Gender (Male = 1) Education (3 categories low to hi) Income (10 categories low to hi) Religiosity scale (100-pt low to hi) Type of society Islamic Orthodox Central European Latin American Sinic Sub-Saharan African Hindu Japanese (Constant) Adjusted R2 Block 1 (Control variables only) Adjusted R2 Block 2 (Controls + type of society)

B

St. Err.

Beta

Sig

–2.4

1.0

–.02

**

0.16

.06

.01

**

–0.05 0.41 1.56 0.01 –0.01

.01 .12 .07 .01 .01

–.05 .01 .08 .02 –.02

*** *** *** *** ***

1.3 –8.9 –5.4 –6.1 1.4 –3.6 –8.9 3.4 68.8 .01

.34 .25 .21 .24 .45 .43 .61 .50 .94

.03 –.18 –.11 –.11 .01 –.05 –.06 .02

*** *** *** *** *** *** *** ***

.05

Note: This illustrates the full OLS regression model, with blockwise entry, in this case with the approval of democratic performance 100-point scale as the dependent variable. Block 1 of the model controls for the level of development of the society and the social background of respondents. Block 2 then enters the type of society, based on the predominant religion, coded as dummy variables. Western societies represent the (omitted) reference category. The coefficients represent the effects of living in each type of society compared with living in Western societies, net of all prior controls. Democratic performance scale: For details see Table 2. Level of human development: Human Development Index (HDI) 2000, including longevity, literacy and education, and per capita GDP in $US PPP (UNDP Development Report 2000). Level of political development: (Freedom House 7-point index (reversed) of political rights and civil liberties 1999–2000) (www .freedomhouse.org). Type of society: see Table 1. Sig. ***p.001; **p.01; **p.05. N/s Not significant. Source: (WVS/EVS), pooled sample 1995–2001.

THE CULTURAL-ECONOMIC SYNDROME: IMPEDIMENTS TO DEMOCRACY IN THE MIDDLE EAST Brigitte Weiffen I. Introduction Among all the regions of the world, the Middle East is exceptional in its resistance to democratization. Since 1974, the year that Huntington (1991) identified as the starting point of the Third Wave of democratization, democratic transitions have occurred in nearly all the mayor world regions: Starting in Southern Europe, the Third Wave swept over a significant number of Latin American countries between 1979 and 1985, with the economically more successful military dictatorship in Chile following suit in 1989. During the second half of the 1980s, the wave of democratization spread to Asia, and it reached its peak after the fall of the Berlin Wall in 1989, tearing down the former Communist block. Freed from being torn between the two superpowers fighting for geopolitical dominance, African countries also began to democratize, and although few of them can be termed full democracies today, they have at least moved away from strict authoritarianism to some kind of semi-democratic or hybrid regime. This development is reflected in the most popular measures of freedom and democracy, like the Freedom House and the Polity Project ratings,1 where every region of the world has seen a rather significant improvement in its average level of democracy—except for one: The Middle East has remained more or less on the same level or has even seen a decline in level of democracy. This is even more striking when we consider the fact that economic well-being is regarded as the most important requisite for democracy (Lipset 1959; Diamond 1992). In a number of quantitative studies, per capita GDP (or GNP) has emerged as the most stable predictor of democracy (Lipset, Seong and Torres 1993; Burkhart and LewisBeck 1994; Barro 1999). That prosperous nations are more likely to be governed democratically than poor ones has therefore been established

1

For more information on the democracy measures, see section III.

252

brigitte weiffen

as “one of the most powerful and robust relationships in the study of comparative national development” (Diamond 1992:110). In terms of their per capita income, the majority of the states in the Middle East are rather wealthy, so that, according to the simple version of modernization theory sketched above, they should already have turned towards more democratic forms of government. But although there have recently been small steps of political opening in small countries like Bahrain, Qatar and Oman, the absolute monarchy remains the dominant form of government in the region. However, there is no direct causality between increasing wealth and democratic development. Some intervening effects that link the two developments have to be taken into account: More wealth usually leads to the growth of a middle class independent of the state, as well as to rising levels of education, a growing number of mass media and means of communication and therefore a larger spread of information. As a result of these changes, a pro-democratic political culture emerges, including a pluralistic and active civil society and the general quest for political participation (Inglehart and Welzel 2005; Welzel 2006). In the Middle East, some powerful obstacles stop the dynamics between increasing individual resources, the evolvement of emancipative values and the extension of freedom rights. The explanations for the democracy gap between the Middle East and the other world regions are either cultural or economic. The cultural explanation refers to historical legacies, especially to the dominant role of Islam in Middle Eastern societies, whereas the economic explanation for the democracy gap emphasizes oil wealth as the main barrier to democracy. This paper explores both the cultural and the economic explanation and proposes that their interaction in the Middle East is especially detrimental to democracy. II. Exploring Cultural and Economic Obstacles 1. The Cultural Explanation Since Samuel Huntington’s thesis of the “clash of civilizations” (Huntington 1993), the question whether countries under Islamic influence are more crisis-prone and less receptive to modernization has been widely disputed. With regard to the undeniable democracy gap in the Muslim world (Karatnycky 2002), the related thesis of incompatibility of Islam and democracy is contested.

impediments to democracy in the middle east

253

Among Western scholars, the main argument for incompatibility is the lack of separation between secular and religious issues or the fusion of religion and politics inherent in Islamic doctrine (Huntington 1984:208; Lewis 2001:28–29; Haynes 2001). In fact, Islam is often portrayed as an all-embracing system, not restricted to the spiritual sphere but with the claim to guide the whole life of the believers. But in Western perspective, secularization is regarded an indispensable element of the modernization process, and if it does not materialize, the other elements of modernization are blocked as well. Additionally, Western researchers point to the lacking respect for individual rights in the Muslim world, especially the preclusion and discrimination of women, as a major obstacle to democracy (Fish 2002; Donno and Russett 2004). On the other hand, other scholars assert that in principle Islamic religion and democracy are compatible. They claim that the blame for the democracy gap found in the Muslim world must not be put on religion, but has to be laid on the historical and political development of the respective countries. In large parts of what constitutes the Muslim world, despotic and patrimonial forms of rule have prevailed during centuries. What Max Weber called “sultanism” was characteristic of the Muslim civilization (Weede 1998): Arbitrary rulers governed with the support of slaves who completely depended on their mercy, without ever having to concede power to feudal lords or an emerging bourgeoisie.2 It is in this context where a fusion of patrimonial political tradition and Islamic religious doctrine took place: The despots claimed to rule “by grace of God”, leading to a religious legitimation of subordination to repressive rule. This, of course, is a perversion of Islamic doctrine: Originally, Islam solely demanded submission to Allah, and did not require obedience to despotic leaders (Weede 2000:179). It is thus only a certain interpretation of religious doctrine that sustains autocratic rule. This selfish abuse of doctrine is employed by conservative monarchic rulers to found their claim for dominance as well as by some movements of political Islam that arrogate for themselves to be the only ones following the true path (Zartman 1992:189; Zakaria 2003:147). Western advocates of compatibility therefore claim that it is not Islam itself, but a biased interpretation of Islam—pursued by Compared to other types of autocratic rule, the importance of slavery in Muslim empires is indeed striking: “Slaves, ideally torn from their kin context, can be more rationally deployed than free men, burdened with all their social bonds.” (Gellner 1981:57). 2

254

brigitte weiffen

Islamic conservatism or Islamic fundamentalism—that is incompatible with democracy. With regard to their position toward democracy, Islamic political thinkers can be grouped along two dimensions: On the one hand, there is a dichotomy between the ones who reject and the ones who endorse democracy, and on the other hand, there is a continuum between secular and purely religious orientation. A rough classification of existing positions along these two dimensions is displayed in Figure 1. According to Muslim scholars who reject democracy, this type of political regime is “antithetical to the Islamic way of life” (Moten 1996:106). In this view, basic prerequisites of democracy, namely individualism and the separation of politics from religion, are not compatible with core religious dogmas and rules set up by Islam. The idea of the nation state, fundamental for political development in the West, is irreconcilable to the Islamic concept of the umma, the community of all believers.3 Additionally, the concept of sovereignty of the people conflicts with the sovereignty of God. It is not the process of electing leaders, though, which is regarded as illegitimate, but its result, the idea of an assembly of human beings making their own laws. In Islam, the Sharia is regarded as unchangeable body of law given by God, and no one, not even the entire Muslim community, has the right to change explicit commands of God. (Mawdudi 1976:159–161) On the other hand, moderate Muslim thinkers endorse democratic ideas (Ibrahim 2007). Some support the idea of adapting Islamic doctrine to modern times, restricting its influence on politics. They accept pluralism, free elections and constitutionalism and in substance advocate the Western model of democracy. Others proclaim a specific Islamic path towards democracy. They argue that Islamic doctrine contains passages that can be interpreted as predecessors of democratic concepts or sources for a specific Islamic version of democracy. Esposito and Voll (1996) cite a broad range of attempts to make the Islamic heritage and democracy compatible. The dogma of God being the only legitimate sovereign, for example, might as well be interpreted as a prohibition of a hierarchical order among human beings, underlining the idea that even the ruler has to subordinate to God and his laws. A number of Islamic guidelines for social and political behavior, like

3 Tibi (1996:322; 329–336) cites various Muslim scholars maintaining these views. Some even equate democracy to “disbelief ”.

impediments to democracy in the middle east Endorsing democracy Secular orientation Religious orientation

democrats Islamic constitutionalists

255

Rejecting democracy nationalist or socialist authoritarians conservative/ traditional Islamic groups

Islamic democrats Fundamentalist religious orientation

Islamic populists

revolutionary Islamists

Figure 1. Democracy and Islam—positions of Muslim political thought.

the concept of consultation (shurah) and consensus (ijma) can lay the foundation for Islamic democracy. (Esposito and Voll 1996:23–28; Nafissi 2005:415) Combining the views on Islam and democracy, the scholars rejecting democracy on religious ground are either to be found among conservative Islamic groups, advocating classical Sunnite or Wahhabite Islam and supporting traditional rulers, or among revolutionary Islamists in favor of political Islam and opposing traditional rulers as well as all kinds of secular and “westernized” political systems. The representatives of the views reported by Esposito and Voll can be classified as Islamic democrats, although the amount of religious elements in their conceptions of political systems varies widely. Islamic constitutionalists are those who want to establish constitutional rule, but believe that Western European or American institutions are not suitable in a Muslim environment. They criticize secularism and argue that for Muslims, religion has to play a significant role in politics. That’s why they are searching for an Islamic version of democracy by reinterpreting the various Islamic principles as foundations for democracy. The priority of Islamic populists, on the contrary, is to bring down traditional or secular authoritarian rulers by requesting democratic elections. They appeal to the masses and pretend to fight for their interests. But often, the demand for democracy is mere lip service. They make use of democratic practice, while in reality pursuing undemocratic goals: the installation of their idea of an original Islamic state, relying on acclamation rather than on real participation. This places them rather close to their openly authoritarian counterparts, the revolutionary Islamists.

256

brigitte weiffen

Democrats have a secular orientation, looking to the experiences of the West as models in an effort to promote their countries’ development. Nationalist or socialist authoritarians of various persuasions represent the autocratic variant of secularism, considering the government a guiding force of the national economy and ideology (which may incorporate religious elements).4 While some of the positions reflect existing styles of government, others are only ideal-typical representations of theoretical positions in political thought. Empirically, the countries that belong to the Islamic civilization are a heterogeneous group: A lot of Muslims live in states where Islam does not play any role in politics. In fact, we only find a few examples of religious elites controlling political power, among them Iran and Afghanistan under the Taliban regime. But even under formally secular types of government, Islamic elements play an important role, when for example the legal system is based on the Sharia. Although religious doctrine itself is malleable and might after some readjustments and corrections of interpretations become fully compatible with democratic rule, so far the way the religious sources have been interpreted and the traditions prevalent in Islamic societies since the Middle Ages promote autocratic rule. This is illustrated by various quantitative studies showing that in general, Muslim countries are less likely to be governed democratically than others (Midlarski 1998; Barro 1999; Graeff 2000; Fish 2002). 2. The Economic Explanation The economic explanation for the democracy gap emphasizes oil wealth as the main obstacle to democracy. Resource-rich states like the oil countries in the Middle East are often termed “rentier states”, since they derive most of their revenues from external rents. Rents are paid by foreign actors, accrue directly to the state, and only a small fraction of the population is engaged in the generation of this rent, while the rest nevertheless might benefit from the distribution and utilization of it (Beblawi 1987). The characteristics of a rentier state make democracy less probable. Due to the income derived from the sale of oil, the govern-

4 Some, though not all, of the groups displayed in Figure 1 are described by Leca (1994:50–57), while I am responsible for completion and the grouping along the two dimensions.

impediments to democracy in the middle east

257

ments do not need to collect high taxes; in fact they often don’t collect taxes at all. As a consequence of that, the governments in those countries are confronted with fewer demands for accountability and representation by the public or can afford to ignore them (Gause 1995:291–293; Ross 2001:332). Whereas in history the rulers’ attempt to raise taxes has often led to demands for political participation, the inverted motto “No representation without taxation” seems to reflect political reality in the resource-rich states of the Middle East (Huntington 1991:65). Additionally, the state engages in spending on patronage, subventions and subsidies. Free education and health care, for example, are provided to the population. People indulged like this are satisfied with their lives and feel no need for political participation. There is no incentive to form associations or interest groups, and some of the governments even take deliberate action to depoliticize the population. If political groups aversive to the government still happen to form, the government is able to prevent them from becoming too strong. Oil wealth, after all, also enables them to spend more on internal security and sustain a large coercive apparatus (Ross 2001:333–336; Bellin 2004). Resource wealth also does not lead to the changes emphasized as essential for democratization by modernization theory. In order to promote democracy, material wealth has to unleash social changes, such as a higher level of education and rising occupational differentiation with shifts of the labor force into the industrial and, subsequently, the service sector. This economic diversification would also lead to a middle class independent of the state and equipped with bargaining power against the elites—whereas in oil-exporting countries the middle class is directly dependent upon the resources granted by the state (Ross 2001:336; Waterbury 1994:27–29). These processes of economic and social modernization usually do not occur in oil-wealthy states, as there seems to be no need to promote other economic sectors besides oil extraction (Ross 2001; Karl 1997). Oil wealth, in sum, makes possible a fairly high standard of living for the people, but enables the government to keep the public politically demobilized, either by fiscal generosity or by repression, and does not bring about the social changes that usually lead to political mobilization in favor of democracy. As quantitative studies demonstrate, oil wealth has indeed a strong negative impact on the prospects of democracy (Barro 1999; Ross 2001; 2004; Smith 2004).

258

brigitte weiffen 3. The Cultural-Economic Syndrome

The theories discussed above are not completely satisfying, as there are always examples that demonstrate the opposite: There are quite a number of countries with a large amount of Muslim population that have established or experienced democratic rule for some time, like Bangladesh, Pakistan and, recently, Indonesia. We also find—albeit fewer—oil-wealthy countries with democratic experiences, like Venezuela and the Republic of Congo. If we look at individual countries, the ones where a combination of Islamic cultural influence and economic structure characterized by resource wealth is at work seem to be most resistant to democratization. There obviously exists a cultural-economic syndrome exerting a particular deleterious impact on the chance to be governed democratically. In the countries affected by the cultural-economic syndrome, autocratic regimes benefit from a double legitimacy: Their religious orientation (ranging, of course, from mere public support of Muslim religious practice to an explicitly theocratic state) combined with their ability to provide their citizens with the amenities of a rentier state gives these polities a sound ideological and economic base. Subsequently we will analyze how the influences of economic and cultural obstacles complement or mutually reinforce each other. Even if one of the deleterious influences weakens, the other one still remains in place and blocks movements towards democracy. This dynamic takes place on the level of the government’s policymaking (macro level), involves intermediate—mainly social and religious—organizations (meso level), and finally, there are mechanisms involving the actions of the individual citizens (micro level). 3.1 The Macro Level: The Government’s Options As long as resources are abundant, autocratic regimes that derive their revenue from oil rents are in a comfortable position: They can postpone democratization indefinitely. But states whose resources are limited soon face the need to compensate for lower rents. There are basically two possibilities to act: Governments may choose to adapt to lower income by cutting expenditure, raising taxes or practice deficit spending. The alternative is to restructure and diversify the economy (Auty and Gelb 2001; Luciani 1994, 1995). Economic diversification is executed by building up new industries, trading companies or a banking sector. But economic modernization

impediments to democracy in the middle east

259

will bring about social changes, too. Education levels, occupational specialization and cooperation with foreign states will rise. As the regime is no longer able to buy consensus by distributing goods, services and incomes in exchange for little or nothing, it will need some kind of legitimation, and the citizens, in turn, will demand accountability and will want to influence political decisions which affect their lives and the business sector they work in. In order to increase its legitimacy, the regime might ponder complementing economic liberalization with political reforms in democratic direction. But either way, in countries where the cultural-economic syndrome is present, Islamic groups claiming a higher legitimacy are already waiting for their chance to gain ground. By promoting democracy, regimes may become more accountable, but at the same time are left more vulnerable. They may thus prefer to block the process of political opening, avoiding the danger to lose control over it (Leca 1994:74). Dismantling of the rentier state is costly for governments: Giving up widespread control over the economy and initiating privatization is an indicator of weakness. Additionally, rulers know that diversification leads to a multiplication of decision-making centers and claims for accountability. That is why countries may choose not to execute economic restructuring. In such a situation, the public will perceive more or less manifestly the financial crisis of the state, being affected by the reduction of subsidies or the rise of taxes. As the state is forced to abandon some of the ground it has occupied in society, competing forces like Islamic organizations have the opportunity to step into the breach and provide the population with key social services (Luciani 1994:146–147). This way they can win support, especially among the lower strata of the population, while at the same time they continue to taunt the regime with its illegitimacy and mismanagement of resources. Weakened states in such a situation are likely to act on the defensive and resort to repression. The harshening of autocratic rule and constant threat of political instability discourage investment and further prolong economic stagnation. On the other hand, a state that chooses to strengthen its fiscal basis not by restructuring the economy, but by cutting expenditure or increasing taxation may also consider fortifying its position by introducing or at least announcing democratic reforms, especially if it feels less threatened by revolutionary Islamic mobilization. To sum it up, governments forced to reduce or overcome rentierism may either choose to restructure the economy to set in motion a

260

brigitte weiffen

true modernization or try to carry on without restructuring. In both variants, moreover, they can choose to permit some degree of democratic opening. In case restructuring and political opening are initiated, the regime runs the highest risk of loosing power soon. In case that restructuring is carried out without permitting democracy, the challenge will be suspended until social changes have firmly taken root. By not restructuring but permitting democratic elements, actual movement towards democratic institutions will take some time, as democratization is controlled and modernization (with its consequences for social changes) is not actively promoted. And if the state neither restructures nor democratizes, but tries to sustain its rule by repression, it has apparently the best chance to survive. But differences between the various options are neutralized by Islamist challengers who, either by insurgent or democratic means, try to take power. So even if reasonable rulers want to lead their countries towards modernization and allow small steps of democratization, fundamentalist Islam still is a force that might block that process.5 3.2 The Meso Level: The Strength of Islamic Organizations During the last years, we have seen a wave of Islamic resurgence. Muslims rejecting the Western way of life resort to radical Islamic conceptions in order to define their culture and reaffirm their identity. Activists of that vein regard traditional monarchs as decadent and reproach them for their collaboration with the West. Their ideal is a truly Islamic regime. As already explained in the previous section, radical groups win support if they manage to replace the government in performing social tasks. As soon as they feel strong enough, and depending on the degree of political opening in their country, they either carry out insurrectional acts or initiate demands for democratization. Attacks on the government will demonstrate its vulnerability, but will also extinguish all tendencies of political liberalization. On the other hand, in case the government permits some political opening, political Islam tries to take part in the political game by hiding its real intentions and pretending to pursue democratic goals, in order to avoid being banned from the very beginning. But this camouflage

5 An illustrative example is Kuwait where the monarch has several times advanced more liberal positions than the Islamist majority in parliament.

impediments to democracy in the middle east

261

probably cannot last for long, since religious political groups are “nondemocrats of a peculiar kind” (Waterbury 1994:39): Whereas other types of non-democratic groups may change their preferences over time, for religious parties their views are dogmas, making them practically unable to negotiate or settle compromises. For a legal participation in political negotiations, religious parties would have to accept basic democratic—and mostly explicitly secular—rules (Waterbury 1994:41). The strength of Islamic organizations is further enhanced by the existence of transnational Muslim networks. Linked by the rejection of the legitimacy of the existing state entities, the Arab brothers from neighboring countries, guided by the idea of Pan-Arabism or endorsing the Islamic concept of the umma, interfere in the national political systems and foster Islamist movements (Leca 1994:62; Gause 1995:287–290). Ironically, Islamism is also nourished indirectly by Western powers interfering to protect authoritarian regimes. As the case of Algeria vividly demonstrates, the West often prefers persistent authoritarianism to democratic elections that hand power to Islamist forces, and this preference becomes even stronger if their access to resources is at stake. However, being barred from political competition, Islamist groups are likely to be strengthened further: Their grievances are now not only directed against an illegitimate national government, but they can additionally claim to be a nationalist movement against foreign domination and exploitation (Ayubi 1997:364; Fuller 2005:53–54). 3.3 The Micro Level: Citizens between Apathy and Alienation The behavior of individual citizens is the area where the mutual reinforcement of the properties of the rentier state and Islamic influence is most obvious. Both the political system and the predominant religion demand subordination, to authoritarian rulers or to God and religious leaders. Both political and religious systems use to portray themselves as unquestionable, and both at the same time take responsibility for the well-being of their members, hereby fostering passivity. Political passivity is a feature primarily caused by autocratic traditions and paternalistic, provident guidance by the rentier state. It is striking that even military defeats which in a number of other countries have led to the demise of authoritarian regimes did not stimulate people to call into question the regime. As long as sufficient rents accrue and material needs of the public are fulfilled by the benefits the state distributes, people remain indifferent towards politics. They do not

262

brigitte weiffen

respond with rebellion, but with apathy to despotic acts of repression (Waterbury 1994:26). When the state suffers from fiscal crisis and is not able to act as main source of well-being any more, people look for other sources of wellbeing and fulfillment. Religion has always played an important part in many Muslims’ lives, as its influence is not confined to a weekly visit to the mosque, but extends to daily life with its all-embracing proscriptions, obligations and recommendations.6 Resorting to religion in times of crisis is therefore a self-evident option. For people in material need, Islamic social organizations provide concrete help. For wealthier people who nevertheless might suffer from the loss of prestige of the formerly rich and successful state, Islam offers a way to redefine their identity and attain spiritual fulfillment and well-being. In short, once the impact of resource wealth on democratic prospects vanishes, the second obstacle, Islamic influence, is reinforced. The way how citizens’ behavior further develops depends on the magnitude of the fiscal crisis and the political measures that are taken to cope with it. If economic restructuring and/or political opening are initiated, modernization and related social change will surely increase political mobilization and arouse demands for political participation in the long run. At the very beginning of the reform process, people may first remain apathetic,7 as habits change slowly and they may still hope for the benefits or fear the repressive punishments of a revived rentier state. In case no economic restructuring is accomplished, either because the regime could overcome the budget deficit thanks to abundant resources and rising oil prices or because it does not want to be endangered by the consequences of modernization, the majority of the people is even more likely to remain apathetic and passive, awaiting what the government subsequently has to offer them. But a crisis of the rentier state not remedied by economic diversification may also cause the people to become alienated. While apathetic people usually remain indifferent towards politics and try to find happiness in private life or religious practice instead, alienation is characterized by intense negative

6 The Sharia is a comprehensive guide to conduct which is not confined to law properly so called, but also classifies acts on a moral scale, for example as recommended or reprehensible (Coulson 1964:83–84). 7 The onset of democratization from above in Qatar, for example, is regarded with scepticism by the conservative society.

impediments to democracy in the middle east

263

feelings towards the political regime and its policy. It is likely to lead the alienated fraction of the population to political action in order to denounce or attempt to change the situation.8 In the countries affected by the cultural-economic syndrome, this protest will in all probability be carried out in the name of Islam. III. Empirical Evidence To get an illustrative overview, we first look at mean democracy scores of different groups of countries. In order to compare predominantly Muslim countries with non-Muslim countries and oil countries with countries without oil wealth, dummy variables are created.9 Of course, this procedure generates an arbitrary cut-point. But it enables one to define the subset of countries where the cultural or the economic obstacles to democracy are in fact predominant. To measure the dependent variable, democracy scores are taken from two different sources. The Polity IV Project database of Marshall and Jaggers (Marshall and Jaggers 2002) offers annual country ratings of regime characteristics for 161 states, its most recent version running until 2001.10 The Polity database focuses mainly on the institutional attributes of regimes, like selection and accountability of the executive and regulation of participation, and depicts a country’s democratic and autocratic qualities on separate autocracy and democracy scales. By subtracting the 10-point autocracy scale from the 10-point democracy scale, a 21-point regime scale ranging from –10 (full autocracy) to 10

8 I use the concepts of apathy and alienation with reference to Almond and Verba (1989/1963) who employ them to define different degrees of incongruence between political culture and structure. When the political structure of rentierism breaks down, a gap between political culture and structure opens. If modernization is actively pursued and structure is redefined, adaptation of political culture will happen some day, although it is likely to lag behind. If, on the other hand, no modernization is enacted, structure is suddenly indefinite and political culture looses its foundation—opening the scene for new forces to advance their ideas. 9 A country is coded as predominantly Islamic when at least 70% of its population are Muslim (figures from the 1990s taken from Britannica Book of the Year 1998). A country is coded as oil country when fuel exports (mainly oil) account for more than 50 per cent of total exports of goods and services (percentages taken from World Development Indicators CD-Rom 2003, averaged over the years 1990–1999; due to missing data in the WDI, I add a code ‘1’ for United Arab Emirates and Iraq, referring to data on these countries in the CIA World Factbook 2003). 10 The dataset can be retrieved at http://www.cidcm.umd.edu/polity/.

264

brigitte weiffen

(full democracy) is created. An alternative measure of democracy is published by Freedom House annually. It focuses more on the political and civil rights of citizens. Countries are rated on a 7-point scale for both political rights and civil liberties, 1 representing the most free and 7 the least free.11 I add the political rights and civil liberties ratings and invert them, so that higher scores represent higher levels of freedom. This analyses use as dependent variable the mean regime score covering ten years, from 1992 to 2001, excluding the heyday of transitions to democracy around 1990 but nevertheless covering most of the 1990s. As a robustness check, the subsequent analyses were conducted with both measures of democracy. As we can see from Tables 1 and 2, there is a remarkable difference between mean democracy scores of Muslim and non-Muslim countries and oil-wealthy and non-oil-wealthy countries, respectively.12 Table 1. The cultural explanation: Comparing mean democracy scores 1992–2001 Muslim countries Non-Muslim countries

mean N mean N

Polity

Freedom House

–3.405 37 4.379 123

2.949 37 7.230 124

The Polity scale ranges from –10 to 10, the recoded Freedom House scale ranges from 0 to 12.

Table 2. The economic explanation: Comparing mean democracy scores 1992–2001

Oil-wealthy countries Non-oil-wealthy countries

mean N mean N

Polity

Freedom House

–4.570 20 3.600 140

2.795 20 6.736 140

The Polity scale ranges from –10 to 10, the recoded Freedom House scale ranges from 0 to 12.

11 For a description of the dataset, see, for example, Puddington 2007, Gastil 1991, or the organization’s web page: http://www.freedomhouse.org/ (where the full dataset is available). 12 T-Tests reveal that the differences between the means are highly significant.

impediments to democracy in the middle east

265

With the help of the binary variables, we form a group containing the countries that are affected by the cultural-economic syndrome. This group is the intersection of Muslim countries and oil-wealthy countries. Table 3 compares mean democracy scores again, showing that the cultural-economic syndrome group attains even lower average levels of democracy than the residual groups of countries that are predominantly Muslim without having oil wealth and countries that are oil-wealthy, but not predominantly Muslim. Subsequently, it is demonstrated that the joint effect of oil wealth and Islamic cultural influence is stronger than the two single variables and has an own predictive power beyond the sum of the two effects. Hypotheses like that can be tested by regression analysis. The advantage of regression analysis is that control variables can be included. In our case, it is useful to control for the determinant established as most stable predictor for level of democracy, which is economic wealth. In bivariate analyses regressing democracy on economic wealth, the oil countries with their relatively high level of per capita income but low degree of democracy tend to be outliers. First, we test each hypothesis—the cultural and the economic explanation as well as the cultural-economic syndrome—separately, entering the dummy variables constructed above in OLS regression models. Table 4 shows that the cultural-economic syndrome dummy is a very strong negative predictor of democracy. For both operationalizations of democracy, it is stronger than the Muslim country dummy and the oil country dummy alone. Using the Polity IV data, the negative impact Table 3. The cultural-economic syndrome: Comparing mean democracy scores 1992–2001

neither Muslim nor oil-wealthy countries predominantly Muslim countries without oil wealth oil-wealthy countries without Muslim predominance predominantly Muslim and oilwealthy countries

mean N mean N mean N mean N

Polity

Freedom House

4.534 117 –1.152 23 1.350 6 –7.107 14

7.315 118 3.765 23 5.567 6 1.607 14

The Polity scale ranges from –10 to 10, the recoded Freedom House scale ranges from 0 to 12.

266

brigitte weiffen Table 4. Impact of the cultural-economic syndrome on democracy: Dummy regressions Polity IV Data

Freedom House Data

Equation

(1)

(2)

(3)

(4)

(5)

(6)

N Constant

125 –10.558

125 –14.339

125 –16.224

126 –5.410

126 –7.278

126 –8.242

In GDP per capita PPP

1.906*** (5.062)

2.311*** (6.243)

2.521*** (7.235)

1.581*** (7.895)

1.783*** (9.122)

1.891*** (10.272)

–7.037*** (–6.784)

– –

– –

–3.511*** (–6.369)

– –

– –

Oil country dummy

– –

–8.453*** (–6.821)

– –

– –

–4.317*** (–6.601)

– –

Culturaleconomic syndrome dummy

– –

– –

–11.917*** (–8.296)

– –

– –

–6.137*** (–8.091)

0.405

0.406

0.476

0.494

0.503

0.561

Muslim country dummy

Adj. R2

First cell entries are unstandardized regression coefficients, t-ratios are displayed in parentheses. *** Statistical significance at the 0.01 level with a two-tailed test

of the cultural-economic syndrome is even stronger than the influence of economic wealth (see equation 3). Technically, the joint influence of Islam and oil wealth is a symmetric interaction effect: In our case, one variable is supposed to exert especially strong influence when the other one is strong as well. Therefore, the independent influence of the cultural-economic syndrome can also be demonstrated employing an interaction term between Muslim predominance and oil wealth.13 An inconvenience in the employment of interaction effects is the high correlation between at least one of the independent variables x and z and their interaction term x*z, leading to multicollinearity which makes it more difficult to identify the explanative power of each variable. A remedy to this problem is centering the predictor variables by putting them in deviation score form so that their means are zero (Aiken and West 1991).14

13 We now use the raw data that we employed for the construction of the country groups: percentage of Muslim population and fuel exports (mainly oil) as per cent of total exports of goods and services. To reduce skewness, the variables are transformed to their natural logarithms. 14 The only thing that changes when applying this rescaling procedure is the intercept (this is shown by the comparison of the first and the second equation in Table 5

impediments to democracy in the middle east

267

Regression analyses demonstrate the assumed positive influence of wealth on democracy: the wealthier a country, the higher its level of democracy. In Table 6, wealth measured in GDP per capita exerts the strongest influence on the Freedom House democracy scores. It is also shown that wealth based on oil exports constitutes an exception to the rule: High oil exports are a negative predictor of democracy. The same can be stated for the influence of Muslim population, which seems to be the strongest predictor of democracy in the equations working with the Polity data (Table 5). But additionally, the interaction between oil wealth and Muslim predominance is a significant explanatory variable for lower levels of democracy, although its t-ratio is weaker than the ones of the other predictor variables. Anyway, a typical strategy to evaluate the importance of the interaction term recommended by Jaccard and Dodge (2004:240) is testing the statistical significance of the change in R2 between the two equations. F statistics comparing Table 5. Impact of the cultural-economic syndrome on democracy (Polity Data): Regression with interaction effect Equation

(1)

N

125

N

(2)

(3)

125

125

Constant

–4.166

Constant

–9.231

–10.274

In GDP per capita PPP

1.547*** (4.270)

In GDP per capita PPP

1.547*** (4.270)

1.691*** (4.682)

In Muslim population

–1.407*** (–6.580)

In Muslim population centered

–1.407*** (–6.580)

–1.339*** (–6.313)

In fuel exports

–1.454*** (–5.386)

In fuel exports centered

–1.454*** (–5.386)

–1.306*** (–4.790)

– –

–0.303**0 (–2.335)

0.535

0.551

In Muslim pop. * In fuel exports Adj. R2

– – 0.535

In Muslim pop. centered * In fuel exports centered

First cell entries are unstandardized regression coefficients, t-ratios are displayed in parentheses. *** Statistical significance at the 0.01 level with a two-tailed test ** Statistical significance at the 0.05 level with a two-tailed test and 6); the relationships between the component variables of the interaction and the dependent variable are not affected. But the centering procedure reduces correlations between component variables and the interaction term, preventing the interaction from causing larger standard errors of the slope coefficients of its component variables—which usually reduces their significance levels (Aiken and West 1991:32–36).

268

brigitte weiffen

Table 6. Impact of the cultural-economic syndrome on democracy (Freedom House Data): Regression with interaction effect Equation

(1)

N

126

N

(2)

(3)

126

126

Constant

–2.339

Constant

–4.905

–5.499

In GDP per capita PPP

1.422*** (7.295)

In GDP per capita PPP

1.422*** (7.295)

1.504*** (7.789)

In Muslim population

–0.679*** (–5.950)

In Muslim population centered

–0.679*** (–5.950)

–0.643*** (–5.728)

In fuel exports

–0.772*** (–5.352)

In fuel exports centered

–0.772*** (–5.352)

–0.678*** (–4.659)

– –

–0.180*** (–2.604)

0.594

0.613

In Muslim pop. * In fuel exports Adj. R2

– – 0.594

In Muslim pop. centered * In fuel exports centered

First cell entries are unstandardized regression coefficients, t-ratios are displayed in parentheses. *** Statistical significance at the 0.01 level with a two-tailed test

the increase of R2 from equation 2 to 3 in both Tables 5 and 6 are significant at least on the 0.05 level, indicating that the inclusion of the interaction term adds a significant portion of explained variance. Beyond the influence of its component variables, the joint effect exerts an additional negative influence.15 Entering the interaction term makes the effect of oil wealth conditional on Muslim predominance, but it also makes Muslim population effects conditional on oil wealth. This nature of the interaction is further illustrated by Table 7 and 8 which display the effects of one component variable at different levels of the other component variable.16 In Table 7 15 In order to scrutinize the explanatory power of the interaction term, the difference of residuals created by the inclusion of the interaction can be examined. Comparing the unstandardized residuals of equations 2 and 3 from Table 6, a decrease of more than 0.7 points (on the Freedom House scale) is found for the following countries: Kuwait, Mali, Syria, Haiti, Ecuador, Turkmenistan, Burundi, Saudi Arabia, Iran, Venezuela, Algeria, Republic of Congo and Norway. The fact that nearly half of these countries belong to the cultural-economic syndrome group demonstrates that the interaction effect improves the model fit for the countries where the hypothesized joint effect of oil wealth and Muslim predominance exists. 16 Cf. Weede 1987 for the rationale of analyzing interaction effects this way; for technical details see Jaccard and Dodge 2004. Both tables draw on the Freedom House equations from Table 6. To make interpretation easier, though, the uncentered data were used for the calculation of the conditional effects.

impediments to democracy in the middle east

269

Table 7. Conditional effects of fuel exports on democracy score Where logged Muslim population is

There fuel exports effects are

0 (minimum) 1 2 3 4 4.61 (maximum) General effect in an additive model Corresponding standard error

–0.354 –0.534 –0.714 –0.894 –1.074 –1.184 –0.772 0.144

Table 8. Conditional effects of Muslim population on democracy score Where logged fuel exports are 0 (minimum) 1 2 3 4 4.58 (maximum) General effect in an additive model Corresponding standard error

There Muslim population effects are –0.330 –0.510 –0.690 –0.870 –1.050 –1.154 –0.679 0.114

which shows the effects of fuel exports on democracy moderated by different levels of Muslim predominance, we can see that the negative effect of fuel exports gets stronger with increasing fractions of Muslim population. Table 8 demonstrates the same for the conditional effect of Muslim population: A larger importance of fuel as export commodity leads to stronger negative impacts of Muslim influence on democracy. The last two rows of Table 7 and 8 enable us to evaluate the significance of the interaction by comparing it to the general effects and the corresponding standard errors of fuel exports and Muslim population, respectively (i.e. the effects in the simple additive model in Table 6, equation 2). As soon as the conditional effects deviate from an unconditional effect in an additive model by more than two standard deviations, we are dealing with a significant interaction (Weede 1987:365). In both Tables 7 and 8 the conditional effects exceed this range at the lowest as well as at the two highest levels displayed. This proves that the deleterious impacts of oil wealth and Muslim predominance in fact

270

brigitte weiffen

reinforce each other and that in countries where the cultural-economic syndrome is present, its negative impact on democratic performance attains strength beyond the obstruction to democracy found in countries where only one of the barriers exists. IV. Conclusion As theoretical reflections suggest and empirical analysis demonstrates, there actually exists a cultural-economic syndrome. In Muslim states without oil wealth, economic modernization usually takes place sooner or later, and its effects, complemented by globalization and the diffusion of liberal and democratic ideas associated with it place democratic reforms on the political agenda. In oil-wealthy states without Islamic influence, on the other hand, no mechanism able to replace governmental resource distribution with religious welfare organizations and material well-being with spiritual fulfillment and self-affirmation is at work, so that in times of crisis there is no alternative to finding a genuinely political solution. In countries where oil wealth and Islamic cultural tradition are at work, religious doctrine, political authoritarianism and wealth generated by external rents mutually reinforce each other in blocking the democratic option: People are neither free nor in need to pursue political representation. Nevertheless, in some of the smaller oil emirates, monarchs already bear in mind that oil reserves will end some day. First attempts to diversify the economy are made. Some even have carefully begun political opening (Herb 2003). Although in some of these cases elections have brought Islamist forces to the parliament, they have so far not seriously threatened the stability of the political regimes in their countries. Moreover, radical Islamist forces might be tamed by bringing forward their demands through regularized procedures of a legislative chamber instead of articulating them in heated religious circles or in the streets. If the strategies of economic restructuring and/or careful political opening are successful, they may serve as prototypes for development beyond rentierism in other countries of the region. The question is how Islam will react to the changes under way. If the present fundamentalist tendencies continue to be mass phenomena or even manage to topple one of the monarchies and install an Islamic

impediments to democracy in the middle east

271

state, chances of democracy in the region will diminish further.17 On the other hand, fundamentalism is just one current of political thought in Islam. In the age of globalization, the necessity to cooperate and trade with the rest of the world is no longer deniable. If economic prosperity is at stake, moderate positions might prevail against radical, isolationist interpretations of Islam. They might be able to reinforce democratic concepts contained in Islamic doctrine and thereby reconcile traditional elements with modernity or they might even lead their countries to a real secularization, making faith a private issue. Under such circumstances a democratic future of Muslim societies is possible. Although the mechanisms of oil wealth and rentierism usually retard modernization, at least until oil springs are exhausted, it does not prevent political leaders from thoroughly reflecting on the depletion of resources and implementing economic diversification. Similarly, Islam does not inherently make democracy impossible. It hinders democracy mainly as long as Islamic doctrine is interpreted by autocratically-minded leaders or would-be autocrats. Like in the Catholic Church, where political restraint finally won over the complicity of Catholic dignitaries with conservative autocrats in Southern Europe and Latin America, a victory of moderate over radical positions in Islam is at least a hope that promoters of democracy can cling to. References Aiken, Leona S., and Stephen G. West. 1991. Multiple Regression: Testing and Interpreting Interactions. Newbury Park: Sage. Albrecht, Holger, and Oliver Schlumberger. 2004. ‘ “Waiting for Godot”: Regime Change Without Democratization in the Middle East.’ International Political Science Review 25(4):371–392. Almond, Gabriel A., and Sidney Verba. 1989/1963. The Civic Culture. Political Attitudes and Democracy in Five Nations. Newbury Park/London: Sage. Anderson, Lisa. 2006. “Searching Where the Light Shines: Studying Democratization in the Middle East.” Annual Review of Political Science 9:189–214. Auty, Richard, and Alan H. Gelb. 2001. “Political Economy of Resource-Abundant States.” Pp. 126–144 in Resource Abundance and Economic Development, edited by Richard M. Auty. Oxford: Oxford University Press.

17 In fact, many scholars studying political regimes in the Middle East are currently leaning towards a more thorough analysis of the existing autocratic forms of rule instead of searching for faint hints of democratization (Albrecht and Schlumberger 2004; Anderson 2006).

272

brigitte weiffen

Ayubi, Nazih N. 1997. “Islam and Democracy.” Pp. 345–366 in Democratization, edited by David Potter, David Goldblatt, Margaret Kiloh, and Paul Lewis. Cambridge: Polity Press. Barro, Robert J. 1999. “Determinants of Democracy.” Journal of Political Economy 107(6):158–183. Beblawi, Hazem. 1987. “The Rentier State in the Arab World.” Pp. 49–62 in The Rentier State, edited by Hazem Beblawi and Giacomo Luciani. New York: Croom Helm. Bellin, Eva. 2004. “The Robustness of Authoritarianism the Middle East: Exceptionalism in Comparative Perspective.” Comparative Politics 36(2):139–157. Burkhart, Ross E., and Michael S. Lewis-Beck. 1994. “Comparative Democracy: The Economic Development Thesis.” American Political Science Review 88(4):903–910. Coulson, Noel J. 1964. A History of Islamic Law. Edinburgh: Edinburgh University Press. Diamond, Larry. 1992. “Economic Development and Democracy Reconsidered.” Pp. 93–139 in Reexamining Democracy, edited by Gary Marks and Larry Diamond. London/Newbury Park: Sage. Donno, Daniela, and Bruce Russett. 2004. “Islam, Authoritarianism, and Female Empowerment. What Are the Linkages?” World Politics 56(4):582–607. Esposito, John L., and John O. Voll. 1996. Islam and Democracy. New York/Oxford: Oxford University Press. Fish, Steven. 2002. “Islam and Authoritarianism.” World Politics 55(1):4–37. Fuller, Graham. 2005. “Islamists and Democracy.” Pp. 37–55 in Uncharted Journey. Promoting Democracy in the Middle East, edited by Thomas Carothers and Marina Ottaway. Washington, DC: Carnegie Endowment for International Peace. Gastil, Raymond D. 1991. “The Comparative Survey of Freedom: Experiences and Suggestions.” Pp. 21–46 in On Measuring Democracy. Its Consequences and Concomitants, edited by Alex Inkeles. New Brunswick/London: Transaction Publishers. Gause III, F. Gregory. 1995. “Regional Influences on Experiments in Political Liberalization in the Arab World.” Pp. 283–306 in Political Liberalization and Democratization in the Arab World. Vol. 1: Theoretical Perspectives, edited by Rex Brynen, Bahgat Korany, and Paul Noble. Boulder: Lynne Rienner. Gellner, Ernest. 1981. Muslim Society. Cambridge: Cambridge University Press. Graeff, Peter. 2000. “Ökonomische und nicht-ökonomische Gefahren für Demokratien.” Kölner Zeitschrift für Soziologie und Sozialpsychologie 52(2):226–245. Haynes, Jeff. 2001. Democracy in the Developing World. Africa, Asia, Latin America and the Middle East. Cambridge: Polity Press. Herb, Michael. 2003. “Emirs and Parliaments in the Gulf.” Pp. 84–90 in Islam and Democracy in the Middle East, edited by Larry Diamond, Mark F. Plattner, and Daniel Brumberg. Baltimore/London: Johns Hopkins University Press. Huntington, Samuel P. 1984. “Will More Countries Become Democratic?” Political Science Quarterly 99(2):193–218. ——. 1991. The Third Wave. Democratization in the Late Twentieth Century. Norman: University of Oklahoma Press. ——. 1993. “The Clash of Civilizations?” Foreign Affairs 72(3):22–49. Ibrahim, Saad Eddin. 2007. “Toward Muslim Democracies.” Journal of Democracy 18(2):5–13. Inglehart, Ronald, and Christian Welzel. 2005. Modernization, Cultural Change and Democracy: The Human Development Sequence. Cambridge: Cambridge University Press. Jaccard, James, and Tonya Dodge. 2004. “Analyzing Contingent Effects in Regression Models.” Pp. 237–257 in Handbook of Data Analysis, edited by Melissa Hardy and Alan Bryman. London: Sage. Karatnycky, Adrian. 2002. “Muslim Countries and the Democracy Gap.” Journal of Democracy 13(1):99–112. ——. “National Income and Liberty.” Journal of Democracy 15(1):82–93.

impediments to democracy in the middle east

273

Karl, Terry Lynn. 1997. The Paradox of Plenty. Oil Booms and Petro-States. Berkeley/Los Angeles: University of California Press. Leca, Jean. 1994. “Democratization in the Arab World: Uncertainty, Vulnerability and Legitimacy. A Tentative Conceptualization and Some Hypotheses.” Pp. 48–83 in Democracy Without Democrats? The Renewal of Politics in the Muslim World, edited by Ghassan Salamé. London: I.B.Tauris. Lewis, Bernard. 2001. The Multiple Identities of the Middle East. New York: Schocken Books. Lipset, Seymour M. 1959. “Some Social Requisites of Democracy: Economic Development and Political Legitimacy.” American Political Science Review 53:69–105. ——, Kyoung-Ryung Seong, and John C. Torres. 1993. “A Comparative Analysis of the Social Requisites of Democracy.” International Social Science Journal 136:155–175. Luciani, Giacomo. 1994. “The Oil Rent, the Fiscal Crisis of the State and Democratization.” Pp. 130–155 in Democracy Without Democrats? The Renewal of Politics in the Muslim World, edited by Ghassan Salamé. London: I.B.Tauris. ——. 1995. “Resources, Revenues, and Authoritarianism in the Arab World: Beyond the Rentier State?” Pp. 211–227 in Political Liberalization and Democratization in the Arab World. Vol. 1: Theoretical Perspectives, edited by Rex Brynen, Bahgat Korany, and Paul Noble. Boulder: Lynne Rienner. Marshall, Monty G., and Keith Jaggers. 2002. “Polity IV Project. Political Regime Characteristics and Transitions, 1800–2002. Dataset Users’ Manual.” Center for International Development and Conflict Management, University of Maryland. Mawdudi, Abu’l A’la. 1976. “Political Theory of Islam.” Pp. 158–171 in Islam: Its Meaning and Message, edited by Ahmad Khurshid. London: Islamic Council of Europe. Midlarsky, Manus I. 1998. “Democracy and Islam: Implications for Civilizational Conflict and the Democratic Peace.” International Studies Quarterly 42:485–511. Moten, Abdul Rashid. 1996. Political Science: An Islamic Perspective. New York: St. Martin’s Press. Nafissi Mohammad. 2005. “Reformation, Islam, and Democracy: Evolutionary and Antievolutionary Reform in Abrahamic Religions.” Comparative Studies of South Asia, Africa and the Middle East 25(2):407–437. Puddington, Arch. 2007. “The 2006 Freedom House Survey. The Pushback Against Democracy.” Journal of Democracy 18(2):125–137. Ross, Michael L. 2001. “Does Oil Hinder Democracy?” World Politics 53:325–361. ——. 2004. “Does Taxation Lead to Representation?” British Journal of Political Science 34(2):229–249. Smith, Benjamin. 2004. “Oil Wealth and Regime Survival in the Developing World, 1960–1999.” American Journal of Political Science 48(2):232–246. Tibi, Bassam. 1996. Der wahre Imam. Der Islam von Mohammed bis zur Gegenwart. München: Piper. Waterbury, John. 1994. “Democracy Without Democrats?: The Potential for Political Liberalization in the Middle East.” Pp. 23–47 in Democracy Without Democrats? The Renewal of Politics in the Muslim World, edited by Ghassan Salamé. London: I.B. Tauris. Weede, Erich. 1987. “Interaction Effects in Cross-National Studies. Human Capital Formation, Government Revenues and Economic Growth in Less Developed Countries.” Quality and Quantity 21(4):361–375. ——. 1998. “Islam and the West: How Likely is a Clash of these Civilizations?” International Review of Sociology 8(2):183–195. ——. 2000. Asien und der Westen. Politische und kulturelle Determinanten der wirtschaftlichen Entwicklung. Baden-Baden: Nomos. Welzel, Christian. 2006. “Democratization in the Human Development Perspective.” Comparative Sociology 5(2):259–290. Zakaria, Fareed. 2003. The Future of Freedom. Illiberal Democracy at Home and Abroad. New York: Norton. Zartman, I. William. 1992. “Democracy and Islam: The Cultural Dialectic.” Annals of the American Academy of Political and Social Science 524:181–191.

PART FOUR

INSTITUTIONS IN COMPARATIVE PERSPECTIVE

RUNNING UPHILL: POLITICAL OPPORTUNITY IN NON-DEMOCRACIES* Maryjane Osa and Cristina Corduneanu-Huci Introduction How is it possible for oppressed groups to mobilize to act on their grievances in a repressive, authoritarian state? The dominant political process theory of social mobilization and related studies of contentious politics primarily focus on protest in democratic societies (Diani 1996; Kitschelt 1986; Klandermans 1997; Meyer 2004; Rucht 1996). In democracies, civil rights are constitutionally protected, the mass media are uncensored, and dissent is tolerated. However, in non-democratic states, the political system is relatively closed, dissenters are persecuted, and the media is controlled. Can the political process framework equally illuminate mobilization in non-democracies? Empirical studies of non-democratic mobilization over-sample cases of popular contention at the moment of democratic transition. That is, social scientists select their cases from the tail of the distribution—instances where the authoritarian state already has been fatally weakened (Glenn 2001; Kuran 1991; Mueller 1999). Since data on repressive regimes are scarce, social action that takes place despite the risks imposed by a stable authoritarian state has been difficult to study. Recently, however, researchers have been making progress, examining a variety of non-democratic regimes in Latin American, Africa, and Asia (Bob 2002; Boudreau 2004; Brockett 2005; Schock 2005). Some have argued that the preconditions of a successful transition to democracy are found in courageous, often abortive, attempts by social actors to challenge non-democratic institutions and repressive practices (Noonan 1995; Pfaff 2006; Sandoval 1998). An investigation of conditions that facilitate mobilization in stable authoritarian regimes can thus correct for deficiencies in two separate (sometimes overlapping) scholarly literatures. The Western bias of social movement theory, particularly the political process framework, can be ameliorated through comparative analyses of political opportunity conditions in non-democracies. The democratization literature can be * Revised, with a new methodological addendum.

278

maryjane osa and cristina corduneanu-huci

extended by examining popular contention in repressive regimes where the ruling elites block, rather than succumb to, the democratic forces. Drawing on the comparative empirical studies, scholars are proposing theoretical formulations that are more sensitive to varying state and institutional formats (Hess and Martin 2006; Meyer 2004; Osa and Schock 2007). Even monolithic tyrannies may lie on cracked foundations. Fissures in the edifice of oppressive rule require inspection. The purpose of our research is to identify political opportunities that enable non-elites to resist, and to contend with, authoritarian governments at the peak of their power. We have studied twenty-four political situations in fifteen stable non-democracies, two thirds of which resulted in significant socio-political mobilizations, and one-third of which did not. Through qualitative comparative analysis, we seek the minimal set of conditions necessary for high risk/high cost activism and popular contention in non-democratic contexts. The Dimensions of Political Opportunity Social movement scholars have identified a number of different factors that create openings for political opportunity in an authoritarian setting. Most often mentioned are: elite divisions, changes in state repression, media access, and influential allies. There is no consensus regarding which factor or combination is necessary or sufficient to initiate mobilization in non-democratic settings. For example, a divided elite was judged to be an important factor creating political opportunity for challengers in 1989 China, the Philippines 1986, and South Korea 1987 (Calhoun 1994; Liang, Nathan, and Perry 2001; Schock 1999; Yun 1997; Zuo and Benford 1995). However, system-challenging mobilizations also occurred where the authoritarian elite was unified, e.g., the 1980 mobilization in Poland and the 1989 Leipzig demonstrations in GDR (Mueller 1999; Osa 2003). Complicating the analysis of political opportunity is the “repression/ protest paradox” which stems from the contradictory effects of political violence. Under some circumstances, regime violence successfully quells popular protest, but other times it provokes even larger mass collective action. Repression often has its intended effects, decreasing mobilization, as happened with the suppression of the student movement in China, or with the crackdown on Romanian miners 1977 (Calhoun 1994; Deletant 1998; Vasi 2001; Zuo and Benford 1995). On the other hand,

political opportunity in non-democracies

279

there are also cases like Burma, where the bloody suppression of student demonstrations led to an increase in protest mobilization as by-standers became outraged at the military regime (Shock 1999). Social scientists have attempted to resolve the repression/popular protest paradox (Brockett 1995; Mason and Krane 1989; Opp and Roehl 1990). From the rational choice perspective, Mason and Krane (1989) argue that the targeting strategy of a regime discriminates among potential victims of political violence, thus differentially affecting individual calculations regarding the decision to participate in popular protest. They posit that such regimes are strategically repressing either: 1) leaders of the opposition, 2) the rank-and-file members of opposition groups, or 3) the mass public, through indiscriminate state violence. Brockett (1995) disagrees with them, demonstrating that Mason and Krane’s thesis is empirically falsified by the experience of El Salvador and Guatemala (Brockett 1995:119–130). In particular, Brockett’s research shows that the third targeting strategy of indiscriminate state violence sustained over time does not increase mass support for oppositional collective action. He proposes instead that the popular response to state repression is influenced by the dynamics of the protest cycle. Brockett argues that intense repression will accelerate mobilization only during the ascendant phase of protest, i.e., during the opening of political opportunity. Later in the cycle, as exhaustion or cooptation takes a toll on movements, repression is likely to decrease protest participation. By considering repression as a joint factor of political opportunity during the upswing of the cycle, our study examines whether Brockett’s proposition holds true beyond the Central American cases he has examined. Finally, social networks are cited for their contribution to mobilization, but usually they are considered as an intervening variable between political opportunity and social movement formation. Most social movement research treats networks as “mobilizing structures,” or “meso-level groups, organizations and informal groups” that form coordinating loci “through which people mobilize and engage in collective action” (McAdam 1999). We have included it in our model of political opportunities because of an anomaly that we found during the collection and coding of the case study materials. Specifically, our study of the Romanian coal miners’ strike in 1977 turned up none of the hypothesized political opportunity variables. This dramatic and puzzling case is worth recounting here in some detail. In July-August 1977, thirty-five thousand Romanian miners organized a work stoppage, took a government minister hostage, and threatened

280

maryjane osa and cristina corduneanu-huci

First Secretary Nicolae Ceau escu with further actions if the government did not accede to their demands for improved working conditions and benefits. This protest took place at a time when there was no internal challenge to Ceau escu from within the party or military, there was increasing repression and deep infiltration and control of society by the Securitate secret police, external governments were supporting Ceau escu because of his independent foreign policy and determination to pay off foreign debts, and the dictator manipulated a far-reaching propaganda and media control operation. With so many obstacles to collective action arrayed against challengers, how could the Romanian miners organize such a broad-based and effective strike? To answer this question, we delved further into our source materials and investigated other miners’ strikes for clues (Ashwin 1999; Crowley 1997; Dennis 1969; Nash 1992). What we found pointed us to social networks. In the Valea Jiului mining region of Romania, Ceau escu’s economic policies created a unique and volatile environment. To increase Romania’s hard-currency exports of coal, Romanian authorities provided various incentives to attract thousands of workers to the mining region. Once resettled, they faced many difficulties if they attempted to return to their former homes. Indeed, Ceau escu’s policy included the abandonment—even the leveling—of many thousands of traditional villages. (This implemented Hrusciov’s 1959 rural systemization plan, eliminating the so-called “non-viable villages” (Deletant 1998:284–289).) When miners received work assignments, they were also allocated housing. As a result of the government’s placement of similarly assigned workers in the same housing blocs, the workers’ children also associated together in the schools. In addition, inter-marriage among these miners was a necessity, since further in- or out-migration from the region was seriously constrained. Consequently, the by-product of the regime’s economic policy was a social reality of multiplex networks: overlapping occupational, social, and residential affiliations. In addition, the specific organization of work for the miners concentrated them below ground in hazardous conditions that necessitated a high degree of solidarity and mutual concern for survival’s sake. The party operatives and supervisors exercised their authority over the workers from the mining enterprise offices. Thus the workplace itself offered a site for workers’ autonomous organization, deep below the earth’s surface (Vasi 2001). In the summer of 1977, the government cancelled previously generous retirement benefits for the miners and raised the retirement

political opportunity in non-democracies

281

age from fifty to fifty-five. On August 3, at a general meeting of the strikers at the large Lupeni mine, a list of demands was formulated to present to the authorities. They included: reduction of the work day from eight hours to six, restoring the retirement age to fifty, jobs for dependents (wives and daughters), improved medical care, and media coverage of the strike. The actions of the miners provoked a government crisis. First Secretary Ceau escu appointed a commission to negotiate with the miners that included top ranking government ministers, Politbureau members, and local mayors. When the commission arrived at the Lupeni mines, the officials were assaulted by workers when they tried to enter the mine manager’s office. The miners particularly distrusted the ranking member of the commission, Ilie Verde‚t, the Politbureau member in charge of the economy and Ceau escu’s close friend. They demanded that he call Ceau escu to appear in person to negotiate with the strikers. The miners then took Verde‚t hostage to put pressure on the dictator. Ceau escu, fearing that the unrest would spread beyond the mining sector, appeared at Lupeni to negotiate with striking miners. He promised to instate a six hour work day, build new factories to employ miners’ dependents, and take no retaliatory actions against strikers. These concessions brought the miners’ protest to an end. Shortly thereafter, the government began an extensive campaign of repression. The mining valley was sealed off by the army and secret police, strike leaders mysteriously disappeared, four thousand miners were relocated, and some were sent to forced labor camps (Deletant 1998:233). Thus, in the Romanian case, absent any other source of political opportunity, the unusual social circumstances of the miners fostered highly multiplex social networks and gave the laborers autonomous space below ground to plan their actions. The networks provided channels for uncensored communications and the social contacts needed to locate material resources. Analysis of the Romanian case leads us to propose the following hypothesis: under some conditions, networks can generate internally the resources to leverage political opportunity. Thus, we have included social networks as an independent variable to explore whether these contribute directly to political opportunity, rather than (as is usual in the literature) operate as an intervening variable within the mobilization process.

282

maryjane osa and cristina corduneanu-huci Data and Methods

The method of qualitative comparative analysis (QCA 3.0, Drass and Ragin 1986) developed by Charles Ragin (1987) is well-suited for an exploratory analysis of political opportunity conditions in non-democratic settings. First, it allows for detailed, contextual comparisons of more countries than is usual in comparative case studies. Second, QCA allows us to utilize a diverse set of data sources. For example, we use quantitative measures, such as event counts, to estimate governmental repression levels and we draw on the expertise of area scholars to interpret the qualitative evidence of domestic political divisions. Third, QCA assumes that social (and political) causality is non-linear, interdependent, and heterogeneous—a much more realistic set of assumptions for our cases than those underlying regression analysis. The use of standard statistical methods in this type of research is problematic for four reasons. First, since the boundaries of the population are unknown, sampling is impossible. Second, for quantitative analyses, the World Handbook data set is unreliable (Diani 1992; Diani and della Porta 1999). Finally, the protest event counts represent discrete values. For statistical analyses, we would have to transform these values into continuous variables as indices measuring magnitude or intensity. Since we are interested in the occurrence, or outcome, of mobilization, measurements of protest intensity or magnitude will not further the analysis. In short, we chose the method of qualitative comparative analysis in order to respect the contextual integrity of the cases, assume causal heterogeneity, and analyze moderately large and diverse sets of empirical instances (Amenta and Poulsen 1994; Ragin 1987, 1995). Case Selection Non-democracies obviously encompass a number of types: military dictatorships, monarchies, one-party states, and Leninist (or state socialist) regimes, to name a few. Yet, with the exception of traditional monarchies, they share an important characteristic: these societies lack a mechanism for regular, legitimate transfers of power sanctioned by those subject to the state. As a result, the basis of rule in non-democracies always involves a much higher degree of coercion than that in democratic states. And the institutions of coercion vary little among the types of authoritarian states.

political opportunity in non-democracies

283

Three features of authoritarianism act as barriers to collective action. First, political activity is usually confined to membership in a single mobilization party (e.g., the communist party in China or the USSR, the Ba’ath party in Syria). Second, the mass media are usually under some form of state control. Third, individual citizenship rights and due process are not guaranteed. By contrast, the democratic state facilitates social movement organization and mobilization: freedoms of speech, association, and the press are constitutionally guaranteed, and oppositional forms of political activism are considered legitimate (Osa and Schock 2007:139). For our study, we selected twenty-four cases of occurrence/non-occurrence of mobilization in fifteen stable non-democratic regimes. Case selection was not randomized. To obtain a random sample, we would need to enumerate, or define a population, of occurrence/non-occurrence of political mobilization in non-democratic contexts. This type of list does not exist, nor would it be feasible to compile. Like Goodwin (2000), who recognizes this problem with randomization, we select cases to maximize variation and we refrain from attempting statistical tests on our sample. All cases from our data set occur during the Cold War (1948–1989) but vary according to type of non-democratic regime (e.g., military regimes, state socialist regimes, bureaucratic authoritarian regimes) and region (including Eastern and Southern Europe, Latin America, the Middle East, sub-Saharan Africa, East Asia). What is a case? To set boundaries around the episodes of mobilization (or their absence) we began by graphing event counts from the World Handbook III (1985) to identify mobilization peaks and troughs. For fifteen non-democracies, we plotted two variables, political contention (a composite variable that includes strikes, demonstrations, riots, and armed attacks), and net repression (the difference between the number of political sanctions imposed and the relaxation of sanctions). Once we had located the periods of contention and quiescence, we turned to the descriptive area studies literature to learn more about catalytic events (or events of major significance) that would shed light on the mobilization contents. This secondary literature usually contained area experts’ periodizations that further explained the different phases of repression and contention. Finally, where area specialists contradicted the World Handbook data, we coded our cases based on the area analysts’ writings because of well known problems with data coding in the World Handbook (Diani 1992).

284

maryjane osa and cristina corduneanu-huci

Variables dependent variable. We defined “social mobilization” as sustained collective action opposing state policies by participants drawn from non-elite or repressed segments of society. This is a broader definition than Shock (1999) uses; he requires a high numerical threshold for participation (more than ten thousand participants). Since we wanted to include semi-clandestine attempts at oppositional mobilization (e.g., human rights organizing in Latin America) and neighborhood-level mobilizations (e.g., poor people’s movements in Honduras and Iran), we opted for a more inclusive definition. Although human rights organizing and the occupation of buildings may seem qualitatively different than the traditional indicators of mobilization (e.g., strikes, demonstrations), in authoritarian contexts, they are all manifestations of high risk/high cost political action. independent variables. The four dimensions of political opportunity that were first implied by Tarrow’s (1994) formulation are now regarded by many scholars as crucial factors of mobilization under authoritarianism. These are: state repression, elite divisions, influential allies, and media access/informational flow (Osa 2003; Shock 1999, 2005; Zuo and Benford 1995). These four variables were used for the initial model. After our analysis of the Romanian miners case, and cognizant of the literature’s emphasis on mobilizing structures in nondemocratic contexts, we decided to include social networks in the full model (Deletant 1998; Loveman 1998; Osa 2001; Oxhorn 1991). Since the QCA method proceeds by comparing and reducing conditions to their most parsimonious configurations, the correct selection of conditions is paramount for achieving reliable results. Thus, a short discussion on how the independent variables were identified is necessary.1 Four factors were coded as dummy variables: elite divisions, 1 For the first independent variable, dynamics of repression, we plotted the time series from the World Handbook, and we took the first difference (n– (n–1)). We took into account both long-term trends and annual value changes. The coding was then corroborated by the area literature’s findings on changes in repression. However, for the five cases after 1982 which were not covered by the World Handbook, we had to use area-specific secondary sources to identify changes in repression. The same coding protocol was followed for media access, since the quarterly version of the World Handbook detaches this variable from the aggregated repression index. For the qualitative variables (influential ally, elite divisions, social networks, mobilization), we coded 1 for presence and 0 for absence based on information from case studies by area experts. Each of these variables was coded independently by two coders and coding discrepancies were subject to further investigation. We did not impose thresholds in order to assess the degree of presence/absence for given variables.

political opportunity in non-democracies

285

influential allies, media access/information flow, and social networks. One condition was coded in terms of its dynamics: 0, if repression was decreasing; 1, if repression was increasing. This accords with extant research suggesting that it is not the repressive apparatus itself, but rather a change in its actions that determines the masses to mobilize (Shock 1999; Tarrow 1998). Elite divisions (divelit) were considered present if, after a review of the literature, evidence was found of competing factions within the ruling elite. Competitors within the ruling party or junta had to have a realistic chance of supplanting the faction in charge. The presence of a symbolic opposition within the polity or of powerless moderates was not enough to signal a “divided elite.” Repression is the only variable that was measured in terms of its dynamics (dynrepr). For each case, we calculated the net repression as a difference between the number of political sanctions and the count of relaxation of sanctions calculated from the quarterly data. We used the variable definitions specified in the World Handbook III codebook. Variable 18, “imposition of political sanctions,” includes actions taken by the government to neutralize or suppress a perceived domestic threat to the regime. The “relaxation of sanctions,” Variable 20, involves the modification or elimination of these restrictions. Influential allies (inflally) were identified also through a review of the area literature. These allies could include: religious organizations (e.g., Catholic Church in Brazil), foreign governments (US support of anti-Marcos demonstrations), international organizations (European Commission for Human Rights, UNHRC), transnational social movements (Comite para la Paz), or foreign political parties (Italian Communist Party support of Spanish opposition). Media access/information flows (medacc). All of the regimes included in the study tried to control the flow of public information, often by institutionalizing censorship and permitting only state-sponsored newspapers. We assessed the degree of media access, first, by examining the World Handbook data on the imposition/relaxation of censorship. In the cases where the World Handbook indicated that censorship controls had been relaxed or imposed, we went back to the original sources of the Handbook-coded data. The original newspaper reports provided more detailed information on imprisoned journalists or other press restrictions. Finally, the secondary literature provided fuller descriptions of the political context and information flows. We coded media access as present, if: there was a sustained relaxation of state censorship, or foreign journalists publicized opposition activities through factual

286

maryjane osa and cristina corduneanu-huci

reporting that was rebroadcast into the country by Radio Free Europe and Voice of America, or there was an active underground press that circulated uncensored materials. Most often, we found a combination of factors increasing information flows. For example, if a regime relaxed censorship of print media, authorities usually also put less effort into jamming Radio Free Europe broadcasts. Uncensored information on contention had to be available to the general public for media access to be considered present. Different types of social networks (socnet) were included in this rough measure. We coded networks present if case studies had shown that mobilization resulted from coordinated activities of specific, linked groups. For example, in the South African and the Poland cases, interorganizational networks were implicated in mobilization (Herbstein 1979; Hirschsohn 1998; Hirson 1979; Karis and Gerhart 1997; Osa 2003). In the Honduran and Chilean cases, area experts pointed to inter-connected, neighborhood groups relying on activists’ interpersonal and community ties to motivate collective action (Brockett 1994; King 1989; Schneider 1991). Data Set and QCA Procedures The Political Opportunity Structures in Non-Democracies (POSND) database (Osa and Corduneanu-Huci 2001) was constructed in MS Access and structured as five inter-linked tables. The first table contained information on state/political regime characteristics, such as regime structure, and notes concerning the political parties, unions, and economic organizations that were permitted under the regime. The second table coded case characteristics, including the temporal span, number of protest events, and graphs of the quantitative variables (e.g., repression). The third table was the event protocol, detailing individual events and recording bibliographic sources for the event information. The fourth table specified how variables were coded for each of the cases; binary values for the dependent and independent variables were set and coding justifications were noted. The final form was the table source, which contained author/date information for sources that documented each variable. Our database includes twenty-four cases, drawn from fifteen non-democratic countries. Eighteen of the cases resulted in social mobilization (socmob = 1) and six cases resulted in the non-occurrence of mobilization (socmob = 0). QCA takes the raw data matrix of nominal-scale variables coded in a binary form and reconstructs it as a truth table. (For the data

political opportunity in non-democracies

287

matrix, see Appendix 1.) The truth table comprises a number of rows (2n where n = the number of independent variables) representing all logical combinations of values on the independent variables. The number of empirical cases corresponding to each row can vary between 0 and n. The program first takes the raw data matrix and transforms it into a truth table in which the non-existent logical combinations are added to the empirical instances. The minimization algorithm starts with the raw data matrix and, by eliminating redundant factors via a step-wise comparison of case pairs, it produces the “prime implicants,” the minimum sets of conditions considered necessary for producing a certain outcome (0 or 1 on the dependent variable). The underlying logic of comparison for this algorithm is derived from John Stuart Mill’s methods of agreement and difference (Mill 1974). A potential problem with this method is the existence of contradictions, i.e., a pair of cases where the same set of causal conditions leads to opposite outcomes. Ragin (1995) suggests a number of options in dealing with this problem. The first option ignores the contradiction, given that the same set of conditions causes both a positive and negative outcome. In practice, the two contradictory cases are removed from the truth table. The second option, more generous analytically, is based on the assumption that the probability of occurrence of any given outcome in the contradiction is 0.5. Only one case from the contradictory pair is included in the truth table. So when the positive outcome is considered, the positive case is included; when the negative outcome is analyzed, the negative case is included. The third alternative is purely qualitative and consists in reexamining the contradictory cases for missed idiosyncrasies. It has the potential for adding new independent variables to the analysis. Since we had a single contradiction in our data set, we took option one and allowed the computer to ignore the contradiction for the analysis; consequently there were twenty-two cases used in configuring the truth table. In sum, by using the Boolean methods of qualitative comparative analysis, we can manage the complexity of our cases while considering the relevant factors in all their logical combinations. We can thus identify the political opportunity configurations in authoritarian countries that appear with the occurrence of the dependent variable, social mobilization. By analyzing these conjunctions, or “prime implicants,” we generate empirically grounded hypotheses that can be tested when good quality data become available.

288

maryjane osa and cristina corduneanu-huci Results

Since the evidence of social networks’ direct contribution to political opportunity was strong for only the Romanian case, we ran the analysis twice, once with the four political opportunity variables, and a second time with social networks added to the model.2 With the inclusion of social networks we achieve a model that is superior to the four political opportunity variables for the following reasons. First, when the social networks variable is excluded, the number of contradictory cases increases to eleven. According to De Meur and Rihoux (2002), a large number of contradictions—46% in the case of this model—indicates that the variable selection is problematic. Second, when we include the social networks variable and the logical cases (for which there are no empirical data) and run the strict minimization procedure, the QCA results are extremely parsimonious, resulting in only two prime implicants. Moreover, the empirical complement of the two prime implicants is identical to the logical complement. This means that the logic predicts what actually happens in our twenty-two cases. In short, when we exclude social networks, the number of contradictory cases jumps to eleven, almost half of the entire dataset, and there is a disjunction between the logical and empirical complements. For these reasons, we present below the results from the expanded model of the four political opportunity variables plus social networks. The results from the four variable model that we rejected are given in Appendix 2. Outcome One: Social Mobilization We first consider whether social mobilization (socmob) resulted from any combination of the five independent variables: dynamics of repression (dynrepr), divided elites (divelit), influential ally (inflally), media access (medacc), and social networks (socnet). In our dataset, we have eighteen cases with positive outcomes: East Germany 1953, Poland 2 The first model with the four political opportunity variables is specified in QCA notation, where the plus sign represents the logical operator “or,” as follows: SOCMOB = DYNREPR + DIVELIT + INFLALLY + MEDIAACC The results of QCA analysis using this model were uninterpretable, non-parsimonious, and laden with contradictions. According to De Meur and Rihoux (2002), this is an indication that the selection of variables for the model is poor. The results from this analysis are provided in Appendix 2. The final model adds social networks and is specified as: SOCMOB = DYNREPR + DIVELIT + INFLALLY + MEDIAACC + SOCNET This is the model that is used in the analysis presented in the body of the text.

political opportunity in non-democracies

289

1956, Portugal 1958, Portugal 1962, Spain 1962, Greece 1968, Honduras 1968, Poland 1968, Chile 1973, Argentina 1977, South Africa 1976, Brazil 1977, Romania 1977, Iran 1980, Chile 1983, Romania 1987, Burma 1988, and China 1989.3 The results of QCA truth table minimization identify four prime implicants, or combination of conditions, that are associated with the outcome, social mobilization in these eighteen cases. This first prime implicant shows that social mobilization occurred when challengers lacked media access if an influential ally (or allies) and social networks were both present. This scenario is covered by six cases in our database: Brazil 1977, Chile 1973, Iran 1980, Poland 1968, Portugal 1958 and Portugal 1962. State control of the media was total in most cases (Chile, Brazil and Portugal); already strict censorship was increasing in Poland where foreign journalists had been expelled from the country. Ironically, censorship was mildest in Iran: although the revolutionary government controlled what was published, reports of squatters’ protests did appear in the newspapers. Table 1. Prime Implicants socmobe = 1 N = 17 1st

medacc INFLALLY SOCNET1,2

or

2nd

dynrepr MEDACC DIVELIT SOCNET

or

3rd

DYNREPR MEDACC divelit inflally

or

4th

One of the following: DYNREPR medacc divelit SOCNET DYNREPR inflally divelit SOCNET

Lower case notation indicates absence of the condition; upper case notation indicates presence of the condition. 2 DYNREPR = Dynamics of Repression INFLALLY = Influential Ally MEDACC = Media Access DIVELIT = Division of Elites SOCNET = Social Networks 1

3 Since East Germany is one of the contradictory cases, it is dropped from the truth table. Consequently, only seventeen cases are included in the QCA analysis of the mobilization outcome.

290

maryjane osa and cristina corduneanu-huci

Challengers’ influential allies varied somewhat across the cases. In Brazil and Poland, the Catholic Church hierarchy offered support to grass-roots mobilization. International human rights organizations were active in Chile. Iranian “houses of labor,” were local institutions backed by labor unions and left-wing parties; these community-based organizations supported the poor people’s housing movement (Bayat 1997). In Portugal, suppressed political parties aligned with anti-Salazar groups (Fryer and Pinheiro 1961). In general, the external allies were long-standing institutions: local, such as the Iranian houses of labor; national, as in Portuguese historic political parties; or with international connections, such as the Church and human rights organizations in Latin America. Similarly, the networks involved in these mobilizations ran the gamut, from the “passive networks” of neighborhood residents in Iran, to loose coalitions of Brazilian students and workers linked through Catholic base communities, to inter-organizational networks of faith-based groups and human rights organizations in Chile. In sum, all the six cases share the same overall configuration of political opportunity but vary according to how the factors are expressed, especially in the types of allies and networks available to challengers. We expect that this variation would dictate differences in how allies and challengers interact, and it would also be related to the scope, size, duration and type of mobilization. Thus, while the absence of media access, availability of allies and social networks is associated with social mobilization under the first prime implicant, the form that this collective action takes varies. Specifically, social mobilization ranges from street demonstrations (Portugal, Iran), student protests (Poland, Portugal), strikes (Brazil), land and building seizures (Iran), hunger strikes and riots (Portugal), and underground human rights organizing to assist families of political prisoners (Chile). The second prime implicant shows that social mobilization occurred when repression decreased, the ruling elites were divided, challengers had access to media and social networks were activated. This seems to be a rather common scenario found in the social movements literature (Calhoun 1994; Osa 2001). Six cases in our database exhibited this configuration: Chile 1983, China 1989, Greece 1968, Honduras 1968, Spain 1962, and Poland 1956. This configuration starts with a lessening of repression. Decreasing repression often takes the form of a temporary reduction of regime violence due to international attention or a “thaw” in political relations stemming from such processes as de-Stalinization, détente, or

political opportunity in non-democracies

291

perestroika. Elite divisions may occur within the ruling party (China, Poland, Spain) or the military (Chile, Honduras, Greece). Information flows increase because of coverage by the foreign press (China, Poland), Radio Free Europe or BBC broadcasts that are not being jammed (Poland, Spain), loosening restrictions on local press as part of the overall reduction of repression (Chile, Greece, Honduras, Spain), and publication of uncensored materials by émigrés abroad (Greece, Poland). Again, social networks can take a number of forms: local neighborhood networks organized by social movement groups (Chile), student networks (China, Poland, Spain), underground networks of banned organizations (Greece), and coalitions of peasant organizations (Honduras). The collective mobilizations associated with this second prime implicant took the following forms: demonstrations and protest marches (Chile, China, Greece, Poland), riots and arson (Poland), strikes (Spain, Chile), hunger strikes (China), student protests (China, Poland, Spain), and property seizures (Chile, Honduras). The Chinese student mobilization and occupation of Tiananmen Square is a good illustration of this second prime implicant. First, in 1989, repression decreased as a result of the economic reforms of Hu Yaobang and Zhao Ziyang and a relaxation in the ideological field. Second, there was a strong division within the Communist Party elite between reformists and hard-line party elders. Third, large numbers of foreign journalists came to Beijing for the first Asian Development Bank meetings and Gorbachev’s upcoming state visit. There was some softening of state censorship; students also made use of “large character” posters and loudspeakers to broadcast information. In addition, Hong Kong news services, CNN and VOA were broadcasting reports of the students’ hunger strike and other protest events. Finally, the creation of cohesive student networks was facilitated by Beijing’s social geography that concentrated academic populations in small dormitories and neighborhoods (Calhoun 1994; Liang, Nathan and Link 2001; Zuo and Benford 1995). The third prime implicant shows that social mobilization occurred when increased repression was imposed by a unified (at least nominally) ruling elite, and challengers lacked influential allies but had access to media. Only three cases exhibit these more restrictive conditions for mobilization: Argentina 1977, South Africa 1976, and Burma/Myanmar 1988. The tightening of already restrictive conditions in these cases by a unified elite increased a sense of injustice and focused challengers’ outrage on the authorities. The strong impetus for mobilization in the

292

maryjane osa and cristina corduneanu-huci

configuration stemmed from the communications media informing the public of extensive injustices and the ability of oppositional groups to use the media for organizing. The expansion of information flows took the following forms: the foreign press exposed regime violence in detailed reports (Argentina and South Africa), shortwave broadcasts of opposition radio station publicized injustices and opposition demands (South Africa), the circulation of underground press or independent publications expanded rapidly (Burma, South Africa), and news of protest demonstrations appeared in the official press when editors were pressured by employees at state publishing houses (Burma). The outcomes of social mobilization in these cases occurred as: demonstrations (Argentina, Burma, South Africa), boycotts (South Africa), and strikes (Burma). The South African experience illustrates this set of conditions. Mobilization in Soweto in 1976 took place during a period of severe repression. Black leaders such as Nelson Mandela had been imprisoned since 1964 and the activities of the African National Congress were severely limited. The white ruling elite were united in support of apartheid, although leaders had begun to recognize that such a policy harmed the regime’s legitimacy. Protests began in Soweto when the government mandated the study of Afrikaans for public school students. News of the student demonstrations were broadcast on Radio Freedom, the short-wave station operated by the ANC. Besides uncensored radio broadcasts, information flows increased through the distribution of leaflets and underground literature. Foreign journalists also covered the resistance of secondary school students to the study of the Afrikaans language (Herbstein 1979; Karis and Gerhart 1997). The fourth prime implicant is similar to the third in that an unified elite is combined with increasing repression. However, instead of media access providing a mechanism for mobilization, social networks form the basis for oppositional activity. There are two variants of this prime implicant: version a) shows the absence of media access, version b) is absent the influential ally. Three cases are covered by the 4a variant: Chile 1973, Romania 1977, and Romania 1987. Four cases are covered by the 4b function term: Argentina 1977, South Africa 1976, Romania 1977, and Romania 1987. This configuration begins in the same fashion as the third prime implicant: increasing repressive actions taken by an undivided elite creates social tension and enhances collective feelings of injustice. Social networks are the means through which opposition to the regime coalesces. Networks take a variety of forms: informal women’s networks

political opportunity in non-democracies

293

(Argentina), inter-organizational networks of religious and human rights groups (Chile), multiplex labor/residential networks (Romania), and local networks of school-based, labor, and community groups (South Africa). Social mobilizations that occurred under this prime implicant included: organized assistance to victims of the regime by underground groups (Chile), strikes (Romania), protest marches and demonstrations (Argentina, Romania, South Africa), and boycotts (South Africa). An illustration of this configuration is Romania 1977, the anomalous case discussed at the beginning of this article. Here we found that despite severe and increasing repression by the Ceau escu regime, Romanian coal miners were able to use their multiplex (overlapping) social, occupational, and residential networks to mobilize mine workers for an extensive strike. This occurred in the absence of media access, so the distribution of information (and other resources) had to occur within social networks. Notice that a number of cases are covered by more than one function term. That is, different prime implicants can be associated with a single case. For example, the 1976 South African case is included as evidence for both the third prime implicant and the fourth prime implicant. This means that South African mobilization in 1976 resulted both from increasing repression, unified elite, media access, and no ally and from increasing repression, unified elite, social networks, and no ally. In short, the QCA analysis identifies two different configurations, or two sets of sufficient conditions, associated with mobilization in the South African case. Thus, if either the networks or media access had been absent, mobilization in Soweto still would have occurred. Outcome Two: Social Quiescence We now consider whether non-occurrence of social mobilization resulted from any combination of the five independent variables: dynamics of repression (dynrepr), divided elites (divelit), influential ally (inflally), media access (medacc), and social networks (socnet). In our dataset, we have six cases with negative outcomes: Spain 1952, Poland 1960, South Africa 1964, Uruguay 1973, Romania 1978, and Chile 1983lcp.4 The results of QCA truth table minimization identify two prime implicants,

4 Since Chile 1983lcp (low combative poblaciones) is the second contradictory case (paired with East Germany), it is dropped from the truth table. Consequently, only five cases are included in the QCA analysis of the no mobilization outcome.

294

maryjane osa and cristina corduneanu-huci Table 2. Prime Implicants socmobe = 0 N=5 1st

dynrepr medacc INFLALLY divelit socnet1,2

2nd

DYNREPR medacc inflally divelit socnet

or

Lower case notation indicates absence of the condition; upper case notation indicates presence of the condition. 2 DYNREPR = Dynamics of Repression INFLALLY = Influential Ally MEDACC = Media Access DIVELIT = Division of Elites SOCNET = Social Networks 1

or combination of conditions, that are associated with the negative outcome, non-social mobilization in these six cases. The first prime implicant shows that no mobilization occurred when repression was decreasing, elites were unified, there was a lack of media access and social networks, but an influential ally was present. This configuration describes a context where a non-democratic regime whose elite is unified reduces overall repression but maintains strong control of the media. Even if an influential ally is available for challengers, absent social networks, this configuration of political opportunity is insufficient to foster mobilization. This conjuncture is represented by one case in our database, Poland 1960. In this case, Poland’s Communist leader, Władysław Gomułka, had consolidated his regime following the upheavals of 1956. By 1960, Gomułka had eliminated his rivals within the Party, lessened the repression of 1958–1959, and although the Catholic Church was available as an ally, opposition networks that had formed in 1956 had been liquidated. Poland was quiet until 1966 (Osa 2001). The second prime implicant shows a combination of political opportunity variables that are common in repressive, non-democratic regimes. A unified elite increases repression and maintains strict control over media and information flows; challengers lack allies, oppositional networks, and media access. Sixty-seven percent of our cases of nonmobilization are covered by this function term: Romania 1978, South Africa 1964, Spain 1952, and Uruguay 1973. This finding shows that if the authoritarian state is able to maintain a unified elite and use the repressive apparatus to control the media, fragment society, and pre-

political opportunity in non-democracies

295

vent challengers from finding external support, the regime will not be subject to contentious events. Such a conclusion would have come as no surprise to Hitler, Stalin, Mao, Franco, Pinochet and the numerous dictators who have historically employed this strategy. Conclusion QCA minimization identified four prime implicants, or specific combinations of political opportunity variables resulting in mobilization, and two prime implicants associated with its non-occurrence in authoritarian regimes. Beyond the identification and discussion of the various POS combinations, some more general conclusions can be drawn. First, we are satisfied that the five variables—dynamics of repression, divided elite, influential ally, media access and social networks—capture the significant dimensions of political opportunity in the non-democratic cases. In all the instances of social mobilization some combination of our variables were found; in all the cases of non-occurrence of social mobilization the variables were missing, with the lone exception of the influential ally. That the QCA program found only one contradiction in our data set also suggests there were no specification problems. Second, what makes mobilization in a non-democratic context a high risk/high cost proposition is the propensity of such a regime to increase repression in order to quell contention. Paradoxically, mobilization results under increasing repression as well as under more relaxed conditions. We found that decreasing repression plays a role in creating political opportunity in one out of four possible conjunctures (prime implicant 2), although this condition might account for a larger percentage of mobilization episodes. On the other hand, increasing repression is a factor in two prime implicants (prime implicants 3 and 4 a,b). Our data provide additional cases to support Brockett’s contention that increasing repression at the beginning of a protest cycle will contribute to mobilization. However, since we do not collect data from the end of the cycle, his proposition is not falsifiable with our data set. In sum, although non-democratic regimes rely on repression to curtail political contention, our data show that this is only successful if the authorities are able simultaneously to prevent challengers’ media access and to inhibit the formation of politically-oriented social networks. Third, social networks clearly create political opportunities as well as respond to them. Although most of the scholarly attention to networks has been on their role as “mobilizing structures” for social movements,

296

maryjane osa and cristina corduneanu-huci

our research suggests that networks can produce opportunities as well as spread mobilization. (See also, Osa and Schock 2007.) Where communities contain cohesive, interlocking or multiplex social networks, feelings of group identity and solidarity are enhanced. If harsh, repressive measures are imposed upon such a community and a sense of injustice grows, these solidary groups become motivated to protest, despite the odds. The networks generate resources internally (information, material supplies, protection) that are deployed in highly politicized contexts to create opportunities for action. Finally, our analysis shows no necessary conditions of political opportunity in non-democracies. However, the four prime implicants discussed above identify different combinations, that is, four different sets of conditions considered sufficient to allow mobilization in authoritarian contexts. It is possible to reduce the prime implicants even further. We did a final QCA procedure to find the maximum reduction of political opportunity conditions. The program algorithm facilitates this by considering all the empirical cases (including the contradictions) and the subset of the logical cases that do not contradict empirical instances in the data set. This minimization algorithm identifies the key factors involved in producing the phenomena. For our dataset, these are: media access and social networks. In conclusion: our study examined the preconditions of democratization by identifying conditions under which political opportunities arise for challengers living under (seemingly unassailable) authoritarian governments. We found that the conventional strategies of non-democratic regimes for maintaining quiescence have both a logical and empirical foundation. Yet the greatest danger to stability in non-democracy occurs when society manages to overcome internal divisions and creates links between individuals and groups. Free and uncensored media also have an important role to play in confronting authoritarianism. The policy implications of this research are clear: if democratic governments want to encourage challenger movements within authoritarian countries, they should support cultural and educational exchanges that stimulate information flows and create networks, subsidize uncensored broadcasts such as Radio Free Europe and BBC, support émigré publications, and provide assistance to non-governmental organizations such as women’s rights groups and human rights associations that operate internationally. Conversely, our research suggests that efforts to engage non-democratic regimes economically may have little effect in stimulating democratic oppositions. Corporate leaders may find doing

political opportunity in non-democracies

297

business with autocrats easier than negotiating with short-term elected officials. And authoritarian governments easily channel profits through state-controlled organizations to the military or to regime cronies. Even if challengers are supported by external allies, political opportunity conditions are not easily produced when authoritarian governments are at the peak of their power. In fact, all the mobilizations we have studied ended with the suppression of contention. Nevertheless, the long-term outlook is hopeful: despite the numerous “failures” of collective action catalogued in our database, the majority of the countries we studied are now functioning democracies. Methodological Addendum (2007) Some social scientists remain skeptical of results obtained via QCA. Since the method relies on the logical sorting of conditions to identify causal configurations, it is highly sensitive to variable and case selection, as well as to data dichotomization (Amenta and Poulsen 1994; Becker 1998; Bollen et al. 1993; DeMeurs and Rhioux 2002; Goldthorpe 1997; Ragin 1987, 2000). Moreover, QCA has been criticized for not taking into account the temporal dimension of causation (Boswell and Brown 1999). Given these difficulties, how can we be certain that we have identified the relevant POS dimensions? Could there be some systematic bias operating in case selection of which we are unaware? Does the static nature of analysis miss important causal components? These concerns led us to further exploration of our cases with more conventional statistical methods in order to test the validity of the results presented in the original “Running Uphill.” Statistical Analysis on POSND Data Set In this Addendum, we report on the results of a statistical analysis using Maximum Likelihood Estimation (MLE) techniques on mobilization events in our fifteen countries for the period 1955 through 1982. Specifically, we test the marginal probability of the number of anti-regime protest events by using a negative binomial regression model. In order to utilize these techniques, we have had to reorganize and supplement our original data sets. The World Handbook III remains the source for event counts for Argentina, Brazil, Chile, China, East Germany, Greece, Honduras, Iran, Myanmar, Poland, Portugal, Romania, South Africa, Spain, and Uruguay. For the independent variables, we used data

298

maryjane osa and cristina corduneanu-huci

from the State Failure Task Force (Goldstone et al. 2000) and Polity IV (Marshall and Jaggers 20002), in addition to our original data set. For the QCA analysis, our cases were delimited temporally by identifying mobilization episodes; this required periodization of event sequences. Such a process is subject to error, even when rigorous protocols are followed. To check for error in case selection (or periodization), we have now completed a statistical analysis of all occurrences of anti-regime mobilization, and not just the events that make up our cases. In the MLE analysis, we used all the event data for our fifteen countries (1955–1982), disaggregating our initial dependent variable “social mobilization” into four separate dependent variables: social mobilization (total number of protest events), anti-regime protests, strikes, and riots. All of the reconstituted dependent variables were categorical variables in the form of event counts. Our panel contains 365 country-year observations (15 countries × 28 years—55 missing data points on two of our control variables, Log [GDP/capita] and economic growth). The five political opportunity variables—divided elites, influential ally, media access, repression dynamics and social networks—have been included in this analysis, although we have had to specify them differently in order to give them a quantitative basis. To operationalize the dimensions of repression and media access, we used the World Handbook data in a time series analysis of repressive measures enacted by the regime generally (dynamics of repression) and against journalists, in particular (media access). Net repression was computed from the WH variables by taking the number of repressive measures imposed by the regime and subtracting the number of relaxation of sanctions. We computed a similar measure for media access, by first reconstructing two yearly variables recording imposition of censorship and censorship relaxation respectively, based on the two WH quarterly series. Second, we performed the same computations for obtaining our media access variable as we had for repression, subtracting the relaxation of censorship incidents from the yearly number of censoring measures undertaken by the government. Thus, a lower value of this variable denotes greater access to a free media. Divided Elites and Influential Allies were equally challenging to quantify. The best approximation we found for divided elites uses a measure of elite homogeneity as a proxy for elite unity, which we code as the absence of elite divisions. The State Failure Task Force (Goldstone

political opportunity in non-democracies

299

et al. 2000) specifies ten categorical variables (ethnicity, clan membership, common ideology, bureaucratic membership, military status, entrepreneurial, mass revolutionary, intelligentsia, land ownership, no shared class characteristics) taking the values 1 (low) and 2 (high) if the ruling elite comprises a homogeneous group on a given dimension of homophily and 0 otherwise. We accept the argument that shared social characteristics increase the probability of elite unity (McPherson, Smith-Lovin, and Cook 2001). Thus, we compute our elite homogeneity/absence of divided elite measure as a dichotomous variable taking values of 1, if any of the nine dimensions of homophily records a value greater than 0, and 0 otherwise. If the tenth dimension—measuring no shared class characteristics of the ruling elites—is coded as 1 and the remaining nine homophily variables as 0, our elite homogeneity value has been coded as a 0. Since this is an indirect way to get at elite divisions, we are also considering the economic variable “trade openness” as a supplementary proxy measure for “divided elites.” Current research shows that trade openness, an economic measure of a country’s participation in international trade, is associated empirically with elite divisions in authoritarian regimes, although the mechanism of causation is not identified. Scholars suggest that the correlation between divided elites and protectionism (i.e., low values of trade openness) may stem from the inability of a heterogeneous ruling class to cooperate on upholding and enforcing the property rights and, more generally, the rule of law (Goldstone et al. 2000:10–11; Hegre 2000; Solingen 1998; Osa and Corduneanu-Huci 2005:192–193). To operationalize “influential ally,” we combine three SFTF variables: the number of a country’s memberships in universal, intercontinental and regionally defined organizations at a given moment in time. We total the number of all international organization memberships and dichotomize the new variable. If the level of participation in international organizations exceeds the sample mean (28.8), the variable takes value of 1; otherwise, it is coded as 0. As with “divided elites,” we can only measure the impact of influential allies on mobilization indirectly. The argument for considering an authoritarian regime’s participation in international organizations as a proxy for the “influential ally” is as follows: a non-democracy’s participation in IOs raises the salience of domestic human rights protections and the operation of the rule of law as the regime’s performance comes under greater international scrutiny. As violations and other abuses of autocrats become more

300

maryjane osa and cristina corduneanu-huci

widely known, the international organizations (or sub-divisions within them) can become potential allies for human rights and other activists.5 For example, Thomas has shown that non-democracies’ participation in an international normative framework articulating “respect for human rights and fundamental freedoms,” creates a “Helsinki effect” which strengthens democratic social movements and undermines the authoritarian state (2001:4). Non-democratic regimes participate in international institutions to enhance their status and legitimacy; however, the resulting international agreements “tend to entrap governments in commitments that exceed their original intentions or contradict their ideological claims” (Thomas 2001:266). The social networks variable is the most difficult one to measure across populations. Given the tendency of mobilization networks to be multiplex, the difficulty of identifying and coding all possible forms of mobilizational infrastructure, and the paucity of quantitative data, we can only indirectly measure the presence of social networks in antiregime mobilization. Since we assume that each episode of contention will provide opportunities to create or strengthen network connections, we operationalize this factor as the lagged value of the dependent variable (number of mobilization events). The variable values were determined using the event counts for our fifteen countries over the twenty-eight year period to create a pooled time series (unbalanced panel model). The five dimensions of political opportunity in non-democracies are important for anti-regime mobilization, we have argued. But we have not been able to test other independent variables within the structure of the QCA method. To challenge our POS thesis, we have included two other sets of variables—economic conditions and other political conditions—in the maximum likelihood estimation model. The economic factors that could affect social mobilization are: trade openness, GDP per capita, population size, GDP growth, inflation and inequality

On the other hand, a selection effect may be at work here: the number of claims against the autocrat and the probability of participating in international organizations might have a common latent cause. In this case, one would expect the two variables to be correlated: for example, the weaker the autocrat, the higher the level of international participation of a regime, and, in the same time, the higher the marginal probability of a mobilization event. If the selection bias were severe, a Heckman procedure would be appropriate to correct for this problem (Heckman 1979). However, when the selection bias is not extreme—which we believe to be the case here—OLS (or MLE) outperforms the Heckman technique (Stolzenberg and Relles 1997). 5

political opportunity in non-democracies

301

(Osa and Corduneanu-Huci 2005; Schock 1996). Since reliable and systematic data for inflation and inequality are unavailable for a majority of observations, we test only the first four variables. The trade openness variable is a special case. In addition to capturing elite homogeneity (as described above), trade openness has been found to impact internal political alignments and political stability (Goldstone et al. 2000; Katznelson and Shefter 2002; Milner 1997; Milner and Kubota 2005; Rogowski 1989). Consequently, the trade openness measure does “double duty” in this model. Finally, the two political variables control for the level of democracy, as well as for the periods of prolonged transition between succeeding governments (Marshall and Jaggers 2000). Social mobilization = exp (β0 + β1* Lag Social mobilization + β2* Repression + β3*Censorship + β4* Divided elites + β5*Influential ally + β6* Log GDP/capita + β7* Log population + β8* Growth + β9* Trade openness + β10* Democracy score + β11* Transition dummy + ε)

The negative binomial regression model tests the influences of our five POS/nondemocracy variables, and controls for other (potentially relevant) economic and political conditions on four dependent variables.6 These statistical controls are included to assess the internal validity of hypothesized causation so that we can determine whether—net of the other economic and political variables’ effect on our dependent variable—the POS variables are still significant. The dependent variables in this model disaggregate the types of contention represented by our initial dependent variable from the QCA analysis, the presence/absence of social mobilization. We have two strong findings regarding POS from the regression. First, the statistical analysis supports the relevance of our five POS dimensions. An increase in the level of repression and networks, as well as the presence of influential allies are significantly and robustly associated with an increase in the total number of mobilization events in the fifteen countries. (See Table 3.) Repression is statistically significant for all four dependent

6 We performed a series of alternative estimations for the same model: fixed-effects, negative binomial with clustered standard errors, GEE population-averaged model, etc. to control for fixed unit effect, heteroskedacity and AR(1) autocorrelation. We have also tested an alternative specification with first lags on the potentially endogenous variables: net repression and media access. The results presented here are generated by the random-effects estimation technique that was found to outperform fixed effects when submitted to a Haussman test.

302

maryjane osa and cristina corduneanu-huci Table 3. Negative Binomial Regression Model (MLE) Dependent variables

Independent variables Lag Social Mobilization (Proxy for Social Networks) Net incidence of Repression (Proxy for Dynamics of Repression) Net incidence of Censorship (Proxy for Media Access) Elite Homogeneity (Proxy for Divided Elites)

DV I Social mobilization (total event count)

DV II DV III DV IV Protests/Anti- Strikes Riots government (event 1count) (event count) demonstrations (event count)

.002*** (.0005) .02*** (.002)

.003*** (.0009) .007*** (.002)

.0005 (.002) .02*** (.003)

.003*** (.0009) .03*** (.002)

.04**

–.01

(.02) .29

(.03) –.50**

–.08***, .02 (ZINB) (.02) .08

(.19) .07 (.17)

(.22) .70*** (.27)

(.20) .26 (.19)

1.28***

1.23**

(.30) .53***

(.52) .26, .55**(ZINB) (.29) –3.34** (1.62)

.85**, .65 (ZINB) (.34) .36* (.20) –2.58** (1.23)

–.02***, .008(ZINB) (.008) .046*** (.02) –.17

–.01**, –.005(ZINB) (.006) .07*** (.01) –.20

(.35) –7.39** (2.96) 365 –460.90

(.28) –7.29*** (2.02) 365 –680.85

–.04***, .04 (ZINB) (.02) .15, –.27* (ZINB) (.14) Participation in .37*** international organizations (.13) (Proxy for Influential Ally) Log GDP/capita .28 Log Population

(.24) .44***

(.18) –2.32***, –.93(ZINB) (1.18) .003, .01** Trade openness –.007*, (also proxy for Divided Elites) –.001(ZINB) (ZINB) (.005) (.004) Polity democracy score .06*** .04*** (.009) (.01) .49*, .13(ZINB) Transitional period –.21, –.56*(ZINB) (.20) (.25) Constant –5.08*** –9.66*** (1.40) (1.80) Number of observations 365 365 Log-likelihood –1259.06 –790.58 Growth

(.14) –2.75*** (.85)

Note: random-effects negative binomial raw coefficients reported; standard errors shown in parentheses; *** p<0.01; ** p<0.05; * p<0.1. We are also reporting the coefficients and their significance generated by the Poisson process of the Zeroinflated negative binomial model (ZINB) in those cases in which these results diverged from the standard binomial model.

political opportunity in non-democracies

303

variables; the networks proxy variable and media access/censorship are significant for all dependent variables except for the number of strikes.7 The results obtained for the divided elites proxies are less clear cut. The elite homogeneity measure is statistically significant at the 0.05 level for strikes (DV III), and significant for total mobilization events (DV I) when estimated with a zero-inflated negative binomial model (ZINB). When we consider the trade openness measure as a proxy for elite divisions, divided elites becomes significant for total events, strikes and riots. ZINB estimation indicates that divided elites were significant for protests, but not for the remaining three dependent variables. Measurement error stemming from the inexact fit between the concept and the operationalization contributes to this lack of robustness in alternative estimation models. Nevertheless, the overall results for our five POS non-democracy variables provide a strong statistical confirmation for the selection of these independent variables for the QCA algorithm. A second finding from the MLE regression suggests the importance of disaggregating different forms of contention in order to identify various opportunity structures associated with them. For example, the effects of social networks and media access/censorship are significant for protests and riots, but not strikes (Table 3). It may be that protests and riots need the routine of previous events for network formation, whereas strikes can spontaneously generate the necessary infrastructure for dissent due to preexisting workplace relations. Equally likely, however, is the possibility that the operationalization of the networks variable is insufficiently sensitive to pre-existing forms of organization. Further, elite divisions seem to be more important than media access for strike opportunities, while the total social mobilization events (DV I) and strikes (DV III) are the only forms of contention associated with a regime’s increased visibility and participation in the international arena (the influential ally proxy). Since our conclusions can apply only to the country cases in our POSND data set (Osa and Corduneanu-Huci 2001), more data are needed for social movement scholars to test these results and generate new hypotheses.

7 Media access and elite homogeneity are unstable across different techniques of estimation, possibly because of the large percentage of 0s among its total values. For example, when estimated with a zero-inflated negative binomial model, media access changes sign and significance for social mobilization and riots, and elite homogeneity changes sign for social mobilization (see Table 3).

304

maryjane osa and cristina corduneanu-huci

Now let us consider the control variables. The four economic and two political variables were included in the full model in order to estimate the effects of all theoretically relevant factors affecting social mobilization in non-democracies. Political economists argue that per capita income, population size, and GDP growth will affect the mobilizing potential of citizens in a non-democratic regime (Alesina et al. 1996; Hibbs 1973; Huntington 1968; Londregan and Poole 1990; Przeworski et al. 2000; Venieris and Gupta 1983). In the MLE analysis, each of these factors is significant for the three types of mobilizations (DV II, DV III, DV IV). Although the procedure does not allow us to compare the POS variables and the economic variables in terms of relative significance, we can say with certainty that GDP per capita and economic growth have a greater substantive effect on mobilization than the POS variables, giving some support to modernization theories (Huntington 1968; Przeworski et al. 2000). When the GDP per capita rises, on average, the number of protests, strikes and riots also increase. Further, economic growth and trade openness are inversely correlated with mobilization event occurrence. The faster the economy grows and the more open its trade policies become, the fewer are the number of mobilization events (across all alternative specifications of the dependent variables). In short, while the regression shows that the economic factors affect the values on the dependent variable, the POS set is still significant. We conclude that, while economic conditions can increase the salience of anti-regime claims, mobilization in non-democracies also needs a set of structural openings in order to occur. Finally, we consider the two other political variables. Political scientists, especially in the democratic transitions literature, have argued that tipping points are important for social mobilization in a nondemocratic context (Kuran 1991; Lohmann 1994). Two variables are used to measure this possibility. First, Polity IV codes regimes in terms of their democratic characteristics. The democracy score is a categorical variable (range: 0 to 20) that comprehensively assesses the type of regime. Values greater than ten are democracies, and below or equal to ten are autocratic forms of government. Our selected countries are generally nondemocracies, so they are in the lower range but they can fluctuate over time. Results from the MLE analysis (Table 3) indicate that the closer a country is to the democratic transition (the threshold value of ten), the higher are the number of mobilization events. The second political variable codes whether the mobilization events occur during an interregnum period (between leaders or regimes) or not. The

political opportunity in non-democracies

305

effect of the prolonged government transition dummy is less clear-cut, showing an impact only on protest demonstrations, although the coefficient is not particularly robust. To sum up: Maximum likelihood estimation analysis has confirmed the significance of the five political opportunity variables necessary for mobilization in non-democracies: divided elites, influential ally, media access, repression dynamics and social networks. Testing the negative binomial regression model with four dependent variables shows the utility of disaggregating types of mobilization when considering political contention in authoritarian regimes. The full model shows that economic factors do contribute to mobilization but are not independent of political opportunities. The level of democracy does have an effect on mobilization occurrence but the interregnum variable does not. Combining Methodological Approaches Improved understanding of causality in social science has been accompanied by convincing calls for methodological pluralism (Mahoney 2008). Our interest in the multiple interactions of political conditions for social mobilization initially drew us to the QCA method. The big surprise of our foray into standard statistics was to see the strong complementarity between the QCA and MLE approaches. The regression analysis allows us a stronger basis for asserting the relevance of the five POS dimensions proposed in our 2003 article, “Running Uphill.” The MLE model also shows the importance of disaggregating the outcomes into types of mobilization for a fuller understanding of the causal configurations. But QCA more than holds its own, methodologically speaking. First, the QCA method does things that the statistical analysis cannot. Regression has well-known problems with degrees of freedom, multicollinearity and interpretability of results when trying to deal with multiple interaction terms—the area where QCA excels. For example, in order to include all the interaction terms among seven preconditions, a statistical regression model would include no fewer than one hundred variables, need millions of data points and lose degrees of freedom (Aiken and West 1991; De Marchi 2005; Jaccard and Turrisi 2003; Ragin 1987). Second, since MLE analysis is mainly interested in the marginal probability of event occurrence, it cannot account for the clustering of events under the same larger wave of mobilization. By contrast, QCA provides relevant information on the structure of opportunities at the onset of the event sequence. Finally,

306

maryjane osa and cristina corduneanu-huci

when different tests of significance are applied to the MLE model and generate contradictory results, QCA can help with the interpretation. For example, the Vuong test of model selection indicated that a zero inflated negative binomial model be considered as an alternative to the standard negative binomial model. (Beta coefficients are displayed in Table 3.) For media access and elite homogeneity, the coefficients change sign and significance across the two specifications (for DV I and DV IV in the first case, and DV I for the latter). The lack of robustness of the average effect on the dependent variables still preserves useful information about possible significance, but it complicates the interpretation of the results. On the other hand, the QCA Boolean logic allows for the specification of all the configurations (and cases) under which the presence or absence of the condition is necessary and/or sufficient for mobilization. In short, QCA complements the average effect analysis by further specifying the necessary and sufficient conditions among those statistically significant facilitators of mobilization in non-democracies. References Aiken, Leona S., and Stephen G. West. 1991. Multiple Regression: Testing and Interpreting Interactions. Newbury Park, CA: Sage Publications. Alesina, Alberto F., Sule Özler, Nouriel Roubini, and Phillip Swagel. 1996. “Political Instability and Economic Growth.” Journal of Economic Growth 1:189–211. Almeida, Paul D. 2003. “Opportunity Organizations and Threat-induced Contention: Protest Waves in Authoritarian Settings.” American Journal of Sociology 109:345–400. Amenta, Edwin, and Jane Poulsen. 1994. “Where to Begin: A Survey of Five Approaches to Selecting Independent Variables for Qualitative Comparative Analysis.” Sociological Methods and Research 23, 1 (August):22–55. Ashwin, Sarah. 1999. Russian Workers: The Anatomy of Patience. Manchester: Manchester University Press. Bayat, Asef. 1997. Poor People’s Movements in Iran. New York: Columbia University Press. Becker, Howard S. 1998. Tricks of the Trade: How to Think about Your Research While You’re Doing It. Chicago, IL: University of Chicago Press Chicago, Ill. Bob, Clifford. 2002. “Political Process Theory and Transnational Social Movements: Dialectics of Protest among Nigeria’s Ogoni Minority.” Social Problems 49:395–415. Bollen, Kenneth A., Barbara Entwisle, and Arthur S. Alderson. 1993. “Macrocomparative Research Methods.” Annual Review of Sociology 19:321–351. Boswell, Terry, and Cliff Brown. 1999. “The Scope of General Theory: Methods for Linking Deductive and Inductive Comparative History.” Sociological Methods & Research 28:154–185. Boudreau, Vincent. 2004. Resisting Dictatorship: Repression and Protest in Southeast Asia. New York: Cambridge University Press. Brockett, Charles. 1994. “The Structure of Political Opportunities and Peasant Mobilization in Central America.” In Social Movements in Latin America. The Experience of Peasants, Workers, Women, the Urban Poor, and the Middle Sectors, edited by J. Dominguez. New York and London: Garland Publishing, Inc.

political opportunity in non-democracies

307

——. 1995. “A Protest-Cycle Resolution of the Repression/Popular-Protest Paradox.” Pp. 117–144 in Repertoires and Cycles of Protest, edited by Mark Traugott. Durham: Duke University Press. ——. 2005. Political Movements and Violence in Central America. New York: Cambridge University Press. Calhoun, Craig. 1994. Neither Gods nor Emperors: Students and the Struggle for Democracy in China. Berkeley: University of California Press. Crowley, Steven. 1997. Hot Coal, Cold Steel: Russian and Ukrainian Workers from the End of the Soviet Union to the Post-Communist Transformations. Ann Arbor: University of Michigan Press. De Marchi, Scott. 2005. Computational and Mathematical Modeling in the Social Sciences. Cambridge, MA: Cambridge University Press. De Meur, Gisèle, and Benoît Rihoux. 2002. L’Analyse Quali-Quantitative Comparée: Approche, Techniques et Applications en Sciences Humaines. Louvain—La—Neuve, Belgium: Academia Bruylant. Deletant, Dennis. 1998. Ceauçescu si Securitatea. Constrangere si Disidenta in Romania Anilor 1965–1989. Bucharest: Ed. Humanitas. della Porta, Donatella, and Mario Diani. 1999. Social Movements: An Introduction. Oxford, UK: Blackwell Publishers. Dennis, Norman. 1969. Coal Is Our Life: An Analysis of a Yorkshire Mining Community. London and New York: Tavistock Publications. Diani, Mario. 1992. “Analyzing Social Movement Networks.” In Studying Collective Action, edited by M. Diani, and R. Eyerman. Newbury Park and London: Sage Publications. ——. 1996. “Linking Mobilization Frames and Political Opportunities: Insights from Regional Populism in Italy.” American Sociological Review 63:1053–1069. Drass, Kriss, and Charles C. Ragin. 1986. QCA: A Microcomputer Package for Qualitative Comparative Analysis of Social Data. Center for Urban Affairs and Policy Research, Northwestern University. Fryer, Peter, and Patricia McGowan Pinheiro. 1961. Oldest Ally. A Portrait of Salazar’s Portugal. London: Dennis Dobson. Glenn III, John K. 2001. Framing Democracy: Civil Society and Civic Movements in Eastern Europe. Stanford: Stanford University Press. Goldthorpe, John H. 1997. “Current Issues in Comparative Macrosociology: A Debate on Methodological Issues.” Comparative Social Research 16:1–26. Goodwin, Jeff. 2000. “Are Protesters Opportunists? Political Opportunities and the Emergence of Political Contention.” Paper presented at the Annual Meeting of the American Sociological Association, Washington, DC. Heckman, James. 1979. “Sample Bias as a Specification Error.” Econometrica 47:153– 161. Herbstein, Denis. 1979. White Man, We Want to Talk to You. New York: Africana Publishing Company. Hess, David, and Brian Martin. 2006. “Repression, Backfire, and the Theory of Transformative Events.” Mobilization 11:249–267. Hibbs, Douglas A. 1973. Mass Political Violence: A Cross-national Causal Analysis. New York: Wiley. Hirschsohn, Philip. 1998. “From Grassroots Democracy to National Mobilization: COSATU as a Model of Social Movement Unionism.” Economic and Industrial Democracy 19:633–666. Hirson, Baruch. 1979. Year of Fire, Year of Ash. The Soweto Revolt: Roots of a Revolution?. London: Zed Press. Huntington, Samuel P. 1968. Political Order in Changing Societies. New Haven, CN: Yale University Press. Jaccard, James, and Robert Turrisi. 2003. Interaction Effects in Multiple Regression. Newbury Park, CA: Sage Publications Inc.

308

maryjane osa and cristina corduneanu-huci

Karis, Thomas G., and Gail M. Gerhart. 1997. Nadir and Resurgence, 1964–1979. Vol. 5 of From Protest to Challenge. A Documentary History of African Politics in South Africa, 1882–1990. Bloomington and Indianapolis: Indiana University Press. Katznelson, Ira, and Martin Shefter (eds.). 2002. Shaped by War and Trade: International Influences on American Political Development. Princeton, NJ: Princeton University Press. King, Peter J. 1989. “Comparative Analysis of Human Rights Violations under Military Rule in Argentina, Brazil, Chile, and Uruguay.” Pp. 1043–65 in Statistical Abstract of Latin America, vol. 27, edited by J.W. Wilkie, and E. Ochoa. Los Angeles: University of California, UCLA Latin American Center Publications. Kitschelt, Herbert P. 1986. “Political Opportunity Structures and Political Protest: Anti-nuclear Movements in Four Democracies.” British Journal of Political Science 16:57–85. Klandermans, Bert. 1997. The Social Psychology of Protest. Oxford: Blackwell Publ. Ltd. Kuran. 1991. “Now Out of Never: The Element of Surprise in the East European Revolution of 1989.” World Politics 44 (October):7–48. Liang, Zhang, comp., Andrew J. Nathan, and Perry Link (eds.). 2001. The Tiananmen Papers: The Chinese Leadership’s Decision to Use Force Against Their Own People —in Their Own Words. New York: Public Affairs. Lohmann, Susanne. 1994. “The Dynamics of Informational Cascades: The Monday Demonstrations in Leipzig, East Germany, 1989–91.” World Politics 47:42–101. Londregan, John B., and Keith T. Poole. 1990. “Poverty, the Coup Trap, and the Seizure of Executive Power.” World Politics 42:151–183. Loveman, Mara. 1988. “High-Risk Collective Action: Defending Human Rights in Chile, Uruguay and Argentina.” American Journal of Sociology 104:477–525. Mahoney, James. 2008. “Toward a Unified Theory of Causality.” Comparative Political Studies 41:412–436. Mason, T. David, and Dale Krane. 1989. “The Political Economy of the Death Squads: Towards a Theory of the Impact of State-Sanctioned Terror.” International Studies Quarterly 33:175–98. McAdam, Doug. 1999. Political Process and the Development of Black Insurgency, 1930 –1970. 2nd ed. Chicago: University of Chicago Press. Meyer, David S. 2004. “Protest and Political Opportunities.” Annual Review of Sociology 30:125–145. Mill, John Stuart. 1974. Philosophy of Scientific Method. New York: Hafner Press. Milner, Helen V. 1997. Interests, Institutions, and Information: Domestic Politics and International Relations. Princeton, NJ: Princeton University Press. ——, and Keiko Kubota. 2005. “Why the Move to Free Trade? Democracy and Trade Policy in the Developing Countries.” International Organization 59:107–143. Mueller, Carol. 1999. “Claim ‘radicalization’? The 1989 protest cycle in the GDR.” Social Problems 46(4):528–547. Nash, June. 1992. I Spent My Life in the Mines: The Story of Juan Rojas, Bolivian Tin Miner. New York: Columbia University Press. Noonan, Rita. 1995. “Women Against the State: Political Opportunities and Collective Action Frames in Chile’s Transition to Democracy.” Sociological Forum 10:81–111. Opp, Karl Dieter, and Wolfgang Roehl. 1990. “Repression, Micromobilization, and Political Protest.” Social Forces 69:521–547. Osa, Maryjane. 2001. “Mobilizing Structures and Cycles of Protest: Post-Stalinist Contention in Poland.” Mobilization 6(October):40–161. ——. 2003. Solidarity and Contention: The Networks of Polish Opposition. Minneapolis: University of Minnesota Press. ——. 2005. “Linking Economic and Political Opportunities in Nondemocracies.” Pp. 171–201 in Economic and Political Contention in Comparative Perspective, edited by Maria Kousis and Charles Tilly. New York: Paradigm Publishers. ——, and Cristina Corduneanu-Huci. 2001. Political Opportunity Structures in NonDemocracies Database [computer files]. Columbia, SC.

political opportunity in non-democracies

309

——, and Kurt Schock. 2007. “A Long, Hard Slog: Political Opportunities, Social Networks and the Mobilization of Dissent in Non-Democracies.” Research in Social Movements, Conflicts and Change 27:123–153. Oxhorn, Philip. 1991. “The Popular Sector Response to an Authoritarian Regime: Shantytown Organizations since the Military Coup.” Latin American Perspectives 18:66–91. Pfaff, Steven. 2006. Exit-voice Dynamics and the Collapse of East Germany: The Crises of Leninism and the Revolution of 1989. Durham, NC: Duke University Press. Przeworski, Adam, Michael E. Alvarez, Jose Antonio Cheibub, and Fernando Limongi. 2000. Democracy and Development: Political Institutions and Well-Being in the World, 1950 – 1990. Cambridge, MA: Cambridge University Press. Ragin Charles C. 1987. The Comparative Method. Berkeley: University of California Press. ——. 1995. “Using Qualitative Comparative Analysis to Study Configurations.” Pp. 177–189 in Computer-Aided Qualitative Data Analysis, edited by U. Kelle. London: Sage Publications. ——. 2000. Fuzzy-Set Social Science. Chicago: University of Chicago Press. Rogowski, Ronald. 1989. Commerce and Coalitions: How Trade Affects Domestic Political Alignments. Princeton, NJ: Princeton University Press. Rucht, Dieter. 1996. “The Impact of National Contexts on Social Movement Structures: A Cross-Movement and Cross-National Comparison.” Pp. 185–204 in Comparative Perspectives on Social Movements: Political Opportunities, Mobilizing Structures, and Cultural Framings, edited by D. McAdam, J. McCarthy, and M. Zald. Cambridge: Cambridge University Press. Sandoval, A.M. Salvador. 1998. “Social Movements and Democratization: The Case of Brazil and the Latin Countries.” Pp. 169–201 in From Contention to Democracy, edited by M.G. Giugni, D. McAdam, and C. Tilly. Lanham, MD: Rowman and Littlefield, Publishers. Schneider, Cathy. 1991. “Mobilization at the Grassroots: Shantytowns and Resistance in Authoritarian Chile.” Latin American Perspectives 18:92–112. Schock, Kurt. 1996. “A Conjunctural Model of Political Conflict: The Impact of Political Opportunities on the Relationship between Economic Inequality and Violent Political Conflict.” Journal of Conflict Resolution 40:98–133. ——. 1999. “People Power and Political Opportunities: Social Movement Mobilization and Outcomes in the Philippines and Burma.” Social Problems 46:355–375. ——. 2005. Unarmed Insurrections: People Power Movements in Nondemocracies. Minneapolis: University of Minnesota Press. Stolzenberg, Ross M., and Daniel A. Relles. 1997a. “Assessing and Correcting Sample Selection Bias.” American Sociological Review 62:494–507. Tarrow, Sidney. 1998. Power in Movement. 2nd ed. Cambridge: Cambridge University Press. Thomas, Daniel C. 2001. The Helsinki Effect: International Norms, Human Rights, and the Demise of Communism. Princeton, NJ: Princeton University Press. Vasi, Ion B. 2001. “The Fist of the Working Class: The Social Movements of Jiu Valley Miners in Post-Socialist Romania.” Manuscript, Department of Sociology, Cornell University. Venieris, Yiannis P., and Dipak K. Gupta. 1983. “Sociopolitical and Economic Dimensions of Development: A Cross-Section Model.” Economic Development and Cultural Change 31:727–756. World Handbook of Political and Social Indicators III, 1948–1982. 1985. Charles L. Taylor, Principal Investigator. 2nd ed. Inter-University Consortium for Political and Social Research, Ann Arbor, MI. Yun, Seongyi. 1997. “Democratization in South Korea: Social Movements and their Political Opportunity Structures.” Asian Perspective 21:145–171. Zuo, Jiping, and Robert D. Benford. 1995. “Mobilization Processes and the 1989 Chinese Democracy Movement.” Sociological Quarterly 36:131–156.

SP52 EG53

PL56 PO58

PL60 PO62

SP62 SA64 GR68

HO68 PL68 CH73 UR73 AR77

1. 2.

3. 4.

5. 6.

7. 8. 9.

10. 11. 12. 13. 14.

Temporal span of the case

Spain 1952 to 1955 East Germany 1953–1 to 1954–3 Poland 1955 to 1958 Portugal 1958–1 to 1958–3 Poland 1960 to 1965 Portugal 1961–1 to 1962–4 Spain 1962–1969 South Africa 1964 Greece 1967–1 to 1969–4 Honduras 1968–1976 Poland 1966 to 1970 Chile 1973 to 1976 Uruguay 1973 to 1981 Argentina 1976 to 1979

ID code Country

Case #

0 1 1 1 1

0 1 0

0 1

0 1

1 0

Dynamics of Repression

1 0 0 0 1

1 0 1

0 0

1 0

0 1

1 1 1 0 0

1 0 1

1 1

0 1

0 0

1 1 0 0 0

1 0 1

0 1

1 1

0 1

Access Influential Divided to Ally Elites Media

Independent variables

Appendix 1: Political Opportunity Dataset

1 1 1 0 1

1 0 1

0 1

1 1

0 0

Social Networks

1 1 1 0 1

1 0 1

0 1

1 1

0 1

Social Mobilization

Dependent variable

310 maryjane osa and cristina corduneanu-huci

SA76

BR77 RO77

RO78 IR80

CH83

CH83– lcp

RO87 BU88

CN89

15.

16. 17.

18. 19.

20.

21.

22. 23.

24.

Chile (combative poblaciones) Chile (low combative poblaciones) Romania Burma/ Myanmar China

Romania Iran

Brazil Romania

South Africa

1986 to 1987 1988–1 to 1988–3 1989–2

1983–1987

1976–2 to 1977–2 1977 to 1978 1977–1 to 1977–4 1978 to 1980 1979–1 to 1981–1 1983 to 1987

0

1 1 1

0 1

1

1

0 0

0 0

0 0

1

1 0

0 1

1

1

0 0

0

1

0 1

1 0

0

1

0 0

1

1

0 1

0 0

0

1

1 0

0

1

0 1

1 1

1

1

1 1

0

1

0 1

1 1

1

political opportunity in non-democracies 311

312

maryjane osa and cristina corduneanu-huci Appendix 2: Model I—Four Political Opportunity Dimensions

Model: SOCMOB = DYNREPR + MEDACC + INFLALLY + DIVELIT Outputs Minimized: 1 Prime Implicants socmob = 1 N = 17 1st

dynrepr INFLALLY DIVELIT 1,2

or

2nd

DYNREPR medacc INFLALLY

or

3rd

DYNREPR MEDACC inflally divelit

Lower case notation indicates absence of the condition; upper case notation indicates presence of the condition. 2 DYNREPR = Dynamics of Repression INFLALLY = Influential Ally MEDACC = Media Access DIVELIT = Division of Elites 1

DOES A STRONG INSTITUTION OF RELIGION REQUIRE A STRONG FAMILY INSTITUTION? Kristen R. Heimdal and Sharon K. Houseknecht Introduction Although family and religion operate within the larger social context that includes other institutions, family and religion have a special affinity. The linkage between these two institutions tends to be strong and qualitatively different from other institutional relationships. The familyreligion connection is distinctive in large part because both institutions are characterized by cultural, rather than secondary interests (MacIver 1970:45). That is, while other institutions, such as polity or economy, are instruments for attaining other interests, family and religion are pursued for their own sakes, because they bring direct satisfaction. Further, both family and religion are devoted to organizing primary group relations, and they also are concerned with the person as a whole, whereas other institutions deal only with compartmentalized aspects of people’s lives. These various similarities between family and religion serve to strengthen the institutional ties between them (MacIver 1970 as cited in Pankhurst and Houseknecht 2000:6–7). The family and religion linkage has empirical as well as theoretical support. There is a large body of literature that documents the influence of religion on family (Christiano 2000; D’Antonio 1983; Jenkins 1991). And, although the bulk of the literature has focused on this causal direction (Thornton, Axinn and Hill 1992), there is some evidence for the impact of family on religion as well (Christiano 2000; Thornton 1985; Thornton, Axinn and Hill 1992). Indeed, family-religion linkages have been referred to as “complex bidirectional relationships” (Snarey and Dollahite 2001:650). While it is clear that family and religion influence each other, what has not been addressed is whether, as institutions, one has primacy over the other. The objective of this exploratory work is to shed some light on this matter. Because many of religion’s traditional functions have diminished over time or have been co-opted by other institutions (Berger and Berger 1983; Lechner 1991; Smith 1971; Swatos 1984),

314

kristen r. heimdal and sharon k. houseknecht

and because religion relies heavily on the family for the transmission of religious beliefs (Berger 1967; Christiano 2000; McGuire 1981; Westerhoff 1973; Wilson and Sandomirsky 1991), it is expected that it takes a strong family institution to produce a strong institution of religion and not the reverse. This is not to suggest that strength in family leads to strength in religion. In a context where families, as the first and most important agents of religious socialization, do not emphasize the salience of religion to family members, strong family could appear with moderate or weak religion. The goal here is not to see whether strength in family is a predictor of strength in religion, but rather to see whether strong family seems to be a requisite for strong religion. In other words, does it appear that strong family is a “necessary but not sufficient” requirement for strong religion? In looking at the primacy of the family institution vis-a-vis the institution of religion, it is important to consider other factors that may be related to the relative strength of these two institutions. This step is especially relevant when taking into account the diversity found among societies. Cultural, economic, and political variations, for example, are likely to be associated with strength in each institution. Thus, a second purpose of this research is to assess, in a preliminary way, whether there are contextual, macro-sociological variables that are associated with the relative strength of family and religion. There has been very little in the way of comparative analyses of the family and religion connection, and so the cross-societal approach that is used here will help to fill this gap. The forty-one countries that were part of the 1990 World Values Survey are studied using multiple sources of data, with country as the unit of analysis. Mahoney, Pargament, Tarakeshwar, and Swank (2001) found that the majority of recent work examining family-religion linkages used only single indicators for family and religion measures; thus, the inclusion of both attitudinal and behavioral measures of family and religion is a strength of this study. An additional contribution stems from the fact that much of the research examining the linkage between family and religion has been at the micro level. In their review of the literature, Thomas and Cornwall (1990) found a noticeable lack of a macro-sociological approach. The inter-institutional perspective underlying this work will extend the limited macro-level work that has been done.

does religion require family ?

315

Conceptual Framework Strength of Family Institution According to Berger and Berger (1983:3), the family institution “has been around since the dawn of history . . . was taken for granted for millennia, and . . . is surely the leading candidate for the status of basic institution in human society.” Yet these writers and some others take the stance on what has become a controversial issue that the strength of family has been declining. Popenoe (1988:8), for example, contended that the family is undergoing a process of “deinstitutionalization,” whereby family members grow increasingly autonomous, less bound to the group. He recognized, though, that the degree of family strength differs across societies. The prevalence of such things as divorce, singleparent households, nonmarital births, living alone, and never marrying indicates family disintegration and disorganization, according to Popenoe. Other proponents of the family decline thesis have identified inadequate child socialization and child poverty as components of family decline (see, for example, Whitehead 1993; Wilson 1993). Changes in traditional functions of the family, such as economic production, social control, and protective functions, also have been considered indicators of family breakdown. (Ogburn and Nimkoff 1955; Popenoe 1988). In this study, divorce rates and nonmarital birth rates are the two behavioral measures that indicate strength in the family institution. Scholars other than Popenoe (1988) have considered divorce as being a disruptive, disintegrative force within the family and as serving the needs of the individual over the family (Berger and Berger 1983; Williams 1970). The distinction between the effects of divorce on individual family members and its effects on families is an important one. Although divorce can be beneficial for certain family members, it indicates the breakup of some significant aspects of family organization when it occurs and also a weakness in the family institution as a whole, especially when the rate is high. The same holds in the case of legitimacy, which has been considered a cultural universal. Malinowski (1974 [1930]:58–59) noted, “. . . in all human societies the father is regarded as indispensable. The woman has to be married before she is allowed legitimately to conceive.” He referred to this phenomenon as the “Principle of Legitimacy.” Along these same lines, Berger and Berger (1983) identified nonmarital births as one source of family instability, and Williams (1970:93) deemed them a “major kind of disruption” for the family.

316

kristen r. heimdal and sharon k. houseknecht Strength of Institution of Religion

The institution of religion also is said by some to be undergoing change in a decline sense. Here the controversy often is framed in terms of the secularization thesis. Berger (1967:107) defined secularization as “. . . the process by which sectors of society and culture are removed from the domination of religious institutions and symbols.” Once considered an inevitable consequence of modernity, today the process of secularization is much debated. Berger (2001:445) himself recently noted that “. . . the world, with some notable exceptions . . . is as religious as it has ever been, and in some places is more religious than ever. This, however, does not mean that there is no such thing as secularization.” As with the family institution, there exists a continuum of strength, with varying levels of importance placed on religion by different societies. Strength in the institution of religion can be assessed through a number of factors, including the degree of influence religion has in relation to other institutions (cf. Williams 1970), the amount of participation in religion-oriented activities (Demerath and Hammond 1969; McGuire 1981; Swatos 1984), and the number of religious workers in a country (Nelson 1988). In this study, frequency of attendance at religious services and frequency of prayer are two of the measures that are used to indicate the degree of strength that characterizes the institution of religion in a given country. There has been a good deal of debate over the validity of religious attendance, especially when used in comparative studies. Swatos (1984), for example, pointed out that, in Scandinavian countries, the home is the primary place of worship, so relatively low attendance rates in these countries do not necessarily signify a weak institution of religion. Phillips (2004) found, too, that the meaning and significance of attendance varies across time and place. In other words, in addition to context, individual factors can play a role as well. Wilson (1992:196) noted that although “Church attendance gives all the appearance of being an objective indicator of spontaneous religious commitment,” people’s motivations for attending are different, and “may include everything from assertions of group solidarity to the local assumption that ‘going to church’ is required of those who would make claim to social standing.” Despite misgivings of this sort, attending worship services is a frequently used and generally accepted measure of people’s commitment to religion (McGuire 1981). While religious attendance is a “public” demonstration, prayer can be either a personal, private act of faith or one that can be easily shared

does religion require family ?

317

with others outside a formal setting. For this reason, as an indicator of strength of religion, prayer escapes problems of motivation. It also avoids other barriers to attendance such as the ability to get to a place of worship which may be hindered by transportation, health, and so on. Attitudes as Indicators of Strength The focus thus far has been on patterns of behavior as indicators of strength in the institutions of family and religion. It is important to note, though, that strength also is assessed attitudinally in this study. The importance that people ascribe to family and religion in their own lives serves as attitudinal indicators of strength in these institutions. Attitudes may be thought of as aspirations to certain values and norms. People are socialized as to what the societal values and norms are, and, when asked, may express public (and not private) goals and expectations. Thus, their behavior may not correspond with what they profess. As Williams (1970:413) noted, “A norm is a standard (not necessarily explicit) for the course that action should follow, not a description of the action that actually occurs” (emphasis in original). On the other hand, attitudes may be better indicators of strength than behavior, because behavior may be more subject to situational and sub-cultural influences. Attitudes also differ according to sub-cultures, but there are still larger societal goals and aspirations for which people aim. It should be noted that attitudes and behavior are two separate dimensions of family strength and also of religion strength. The potential lack of consistency between attitudes and behavior underscores the importance of taking both into account when measuring complex concepts. Just as it is important to consider the possible discrepancies between attitudes and behavior, it is important to understand the link between the individual and the larger social structure. Institutions can be conceptualized as having two components: a cultural aspect, which includes values, norms and ideologies, and a social structural aspect, which includes organizations and associations. As already stated, variations exist within societies and communities in the degree to which individual attitudes and behavior conform to the cultural values and norms. However, the process of internalization through socialization and social control render attitudes and behavior at the aggregate level reflective of a society’s prevailing values and norms. In the case of religion, scholars such as Kelley and De Graaf (1997), Ruiter and De Graaf (2006), and Scheepers, Te Grotenhuis, and Van Der Slik (2002) have used aggregate indicators to measure religious strength at the societal level. Aggregate

318

kristen r. heimdal and sharon k. houseknecht

indicators also have been used to measure family strength at the societal level (see, for example, Houseknecht and Sastry 1996). The Relative Strength of Family and Religion The same processes that may bring religion and family closer together also may put religion in a particularly vulnerable position relative to other institutions, as a result of a loss of functions. This is, in part, linked to Williams’ (1970) concept of institutional dominance, whereby one institution may be dominant over one or more other institutions in society. A traditional function of religion, legitimizing existing structures, has been diminishing in many societies (Smith 1971:197). And, speaking of Scandinavia specifically and Christendom generally, Swatos (1984) noted that in the past, the church was used by the state to handle a variety of tasks that are now managed otherwise. In addition to tangible aid, other institutions are competing with religion in providing moral guidance and authority (Berger and Berger 1983). In many societies, the family, too, may be relatively weak in relation to the economy and polity as a result of loss of functions over time. The question, though, is which institution has greater strength in relation to the other, family or religion? Religion provides symbolic legitimation for family patterns (cf. Berger 1967), and numerous familial events (such as weddings and funerals) are marked in religious contexts. It is also the case that religious factors can shape family dynamics, power, for example (Fontelas Rosado Nunes 2000; Fortuny Loret de Mola 2000), and that religious associations provide strength of community without which many families would struggle more than they do. Nevertheless, it seems plausible that the family institution is in a dominant position over the institution of religion. The institution of religion has always depended upon families to provide religious socialization—and this underscores a key way in which the institution of religion is dependent on the institution of family (cf. Westerhoff 1973). As McGuire (1981:43), pointed out, “a person is not born with a full-blown set of religious beliefs and practices; religion is developed and nurtured.” And Berger (1967:373) noted, “The family is the institutional area in which traditional symbols continue to have the most relevance in actual everyday living.” More than any other institution, the family is the primary source of inculcation for religious values and beliefs, as families “bring potential religious adherents into the world” and “give them their initial exposure to questions of faith”

does religion require family ?

319

(Christiano 2000:1–2). The family acts as a “critical support institution,” which facilitates the transmission of religious beliefs and identities from one generation to another (Wilson and Sandomirsky 1991:290). Although other institutions, such as education or polity, may play some role in instilling religious beliefs, in many societies these institutions play a very limited role at best and may even be prohibited from participating. Thus, it is the family that remains the earliest and most influential institution where matters of religion are concerned. According to D’Antonio (1985:395), the family can be seen as the “cross-generational lifeblood for most religious organizations.” And as McGuire (1981:46) stressed, “the family is not merely an agent of religious socialization . . . it is a fundamental unit of the institution of religion.” In sum, while the institutions of family and religion are closely interconnected, it can be argued, for the reasons just discussed, that the family is less dependent on religion than vice versa. It is expected that a strong family institution is necessary in order to have a strong institution of religion, but that a strong institution of religion is not necessary in order to have a strong family institution. This is not to say, however, that where there is strong family, there will be a strong institution of religion. On the contrary, a strong family institution can be associated with gradations in the strength of religion depending on the emphasis that families place on religion. It is expected that a strong institution of religion will appear only with a strong family institution, but that a strong family institution could be found across an array of strengths in religion, be they high, moderate or low. Situations in which both family and religion are strong may reflect a strong traditional culture. Finding strong family appearing with moderate or weak religion suggests that a secularization process characterizes religion but has not affected the family. A typology is used to assess the strength in the institution of family vis-à-vis the strength in the institution of religion. Intersecting strength of family and strength of religion on three dimensions (strong, moderate, and weak) produces nine categories that provide insight into the relative dominance between family and religion. Contextual Associations with Strength of Family and Religion To take account of diversity across societies, this study will look to see whether there are important contextual-level conditions that are associated with varying combinations of strength in family and religion

320

kristen r. heimdal and sharon k. houseknecht

(as suggested by the typologies). Catholic predominance; political economic structure, including communist/communist transition status and democracy score; level of development, as indicated by percent of population with high education and gross domestic product per capita; and degree of urbanization will be examined. Catholic Predominance The Catholic church’s bureaucratic tradition has shown itself to be highly resistant to modernizing processes (Berger 1967). Countries characterized by a Catholic majority tend to have a strong institution of religion (Iannaconne 1991). First, Catholicism’s emphasis on ritual translates into behavior such as church attendance and prayer (McGuire 1981). And we know that family plays a special integrating role in Catholic predominant countries (Gundelach 1994). Lenski (1961) argued that the Catholic church and kin group are mutually reinforcing organizations for devout Catholics, whereas the relationship between Protestant churches and the kin group is less complimentary. He also claimed that differing attitudes toward the immediate family held by Catholics and Protestants is associated with more durable immediate family relationships for Catholics than Protestants. Behaviorally, the Catholic Church’s proscription on divorce and strict teachings on nonmarital sexual relations should indicate both lower divorce and nonmarital birth rates. For these various reasons, it might be expected that Catholic predominance will be linked with strength of both family and religion. Contrary to this line of thinking, some evidence has been emerging which suggests that identification with Catholicism is becoming more weakly associated with attitudes, values, and behavior (D’Antonio 1985), including those that are family-related. In recent decades a supply-side model of religion has been promoted which posits that the degree of religious market regulation and the amount of religious competition determine the level of religiosity in a society (see, for example, Finke and Iannaccone 1993; Iannaccone 1991, 1994; Stark 1992; Stark and Iannaccone 1994; Stark and McCann 1993). Applying these principles, it has been found that, in religious markets where the Catholic church must compete with other religious organizations, the average Catholic is more devout than in a religious market monopolized by Catholicism (Stark 1992). According to Bruce (1992:170), religious pluralism may strengthen commitment to religious belief systems, as greater flexibility means “there is something for everyone.” However, it also should be

does religion require family ?

321

noted that Bruce acknowledged that pluralism may weaken religion because it exposes its human origins. Furthermore, the idea of religious markets is quite a debated one. There is growing evidence that questions its application (e.g., Bruce 2000; Chaves and Gorski 2001; Lechner 1996; Montgomery 2003; Pankhurst and Houseknecht 1999). Political Structure Communist/Communist Transition Status It is important to consider a country’s political economic system, specifically whether it is, or was until recently, communism, since communism is ideologically opposed to religion and religious activity. Although no country has managed to completely eliminate religion from the lives of its people, it was discouraged under the former Soviet regime. In some cases, religion was severely repressed (Gautier 1997), and, in other instances, incentives were used to limit religious activities (Kaariainen 1989). In countries dominated by the Communist party, churches were not allowed a social mission and volunteer efforts were illegal. Only since the early 1990s have religious organizations been able to establish programs assisting the needy in the former Soviet Union (Powell 1994). Furthermore, even though glasnost, perestroika, and democratization allowed for the opening and new construction of churches, mosques, and synagogues and for the availability of religious instruction for the young, this new religious freedom was not wholly embraced. As Powell (1994:303) reported, “There were frequent reports of recalcitrant town, city, and provincial officials who refused to acknowledge the new rights of the faithful: party activists or the chairmen of local soviet executive committees continued to obstruct groups attempting to register as a religious community, reopen or carry out repairs on a church, and so on.” In East Germany, for example, there is evidence that “new religious movements have encountered resistance by officials and have been unable to make effective inroads (Froese and Pfaff 2001:503). The evidence regarding the relationship between communism and the family institution is mixed. On the one hand, it has been argued that, as communist ideology dismantled or destroyed civil organizations, people retreated into their family life, since there was no other sphere through which to find fulfillment (Mason, Nelson, and Szklarski 1991). Families were considered the last bastion of civil life (Sandi 1992). On the other hand, communist regimes are suspicious of families and resent the family’s role in socialization (Cseh-Szombathy 1990). Furthermore, it has been argued that, in oppressive systems, frustration is often

322

kristen r. heimdal and sharon k. houseknecht

manifested at the individual level (Danziger 1991) with one result being increasing levels of family-related problems, such as divorce, child abuse, and illegitimacy. Using data on demographic trends related to marriage and divorce in Belarus around 1990, Vardomatskii and Pankhurst (2000) noted that the family was in a state of crisis, a situation that disrupts socialization. Disrupted socialization, in turn, means missed opportunities for the transmission of religious beliefs and values. While this research considered only Belarus, the findings might be applicable to other post-communist societies, given the similar circumstances that characterize many of them. It is expected, then, that strength in family and religion will be inversely associated with communist/communist transition status (which will be referred to from here on as communist/transition status). Level of Democracy Considering a country’s level of democracy offers a different kind of look into the association between political structure and the strength of family and religion—one of degree or intensity, rather than just type of political structure. Writing about the United States, Himmelfarb (1998) argued that democracy is associated with breakdown of the family, mentioning specific indicators of family strength such as nonmarital births and divorce rates. And Reichley (1986) noted that there will always be some degree of tension between religion and democracy, as religion claims to reveal universal truths, while democracy requires compromise. There is evidence, on the other hand, which suggests that democracy is positively associated with strength in family. The state power exercised by less democratic regimes has the ability to uproot neighborhoods and communities and to destabilize families (Fukuyama 1995). Writing about Eastern Europe, Rosenberg (1995) noted that the totalitarian regimes in these countries broke down the bonds of trust in families. Level of democracy also may be positively associated with strength in religion. It has been argued that a successful democracy may need religion. As Reichley (1986) stated, democracy depends on values that have no reliable source outside religion. Similarly, Frandkin (2000:90–91) argued that “successful political democracy will inevitably require moral instruction grounded in religious faith.” Given these two competing perspectives as regards the association between democracy on the one hand and strength of family and religion on the other, this work will explore both possibilities. One hypothesis

does religion require family ?

323

is that there is a positive relationship between level of democracy and strength in family and religion. The alternative hypothesis is that there is an inverse relationship between level of democracy and strength in family and religion. Level of Development There is a large body of research regarding development and religion. One of the outcomes often associated with development is secularization. While support for the secularization thesis has been mixed, there is a significant amount of research linking specific processes of development with declining levels of religion. This study will consider two them. First, an increased level of education has been found to be associated with a decline in religion, at both the individual and societal levels (see, for example, Campbell and Curtis 1994; Caplovitz and Sherrow 1977; Glock 1973; Mueller and Johnson 1975; Sigelman 1977; Wuthnow 1977). Second, an increase in real per capita income also has been linked with a decline in the importance of religion (Wuthnow 1977). These same forces may have a weakening effect on the family institution as well. It has been argued that development results in “extensive family disorganization” (Moore 1963:102). For one thing, development makes it very difficult for families to control the behavior of their members (Hargrove 1983). Specifically, prolonged education, a requisite for development, has been found to weaken parental authority (Chang 1996). In addition, education often necessitates departure from the family (Caplovitz and Sherrow 1977). Increases in real per capita income work in a similar manner, resulting in mobility as well as individualization, both of which may weaken family. It is expected, then, that countries characterized by weak family and weak religion will have higher levels of development than countries characterized by strong family and strong religion. Degree of Urbanization There is reason to believe that urbanization has a weakening effect on the family institution. According to Nimkoff (1965:354), “The evidence from around the world, with few exceptions, points to an increase in family disruption with increasing industrialization and urbanization.” Just as with development, urbanization weakens family control over member actions. The anonymity of urban areas lessens the consequences of behavior such as divorce and nonmarital births. Urbanization also weakens the institution of religion. Cox (1965:167) refers to

324

kristen r. heimdal and sharon k. houseknecht

urbanization and secularization as “twin tendencies,” and noted that “the cosmopolitan confrontations of city living exposed the relativity of the myths and traditions” once thought unquestionable, thus weakening religion (Cox 1965:1). Cox also pointed out that the increased mobility associated with urbanization leads to greater tolerance of new ideas and possibilities. It is expected that countries with weak family and weak religion will tend to be more highly urbanized. Data and Methods Data Data for five of the variables (family importance, religion importance, religious attendance, prayer, and education) come from the 1990 World Values Survey (Inglehart et al. 2000).1 In the 1990 wave of this longitudinal survey, face-to-face interviews with adult citizens eighteen years of age and older were conducted in forty-one countries, using, in most cases, stratified multi-stage probability sampling. Sampling exceptions include China, India, and Nigeria where the surveys are based primarily on the urban, literate population. The overall sample represents seventy percent of the world’s population. Data for the remaining variables were obtained from a variety of sources. Nonmarital birth information come from two sources, statistical information presented by de Guibert-Lantoine and Monnier (1997) in the journal, Population, and The Illustrated Book of World Rankings (Kurian 1997), which was originally obtained from the United Nations Population Division. Data for crude divorce rates were obtained from the United Nation’s Demographic Yearbook (1993a) and The Illustrated Book of World Rankings (Kurian 1997). The percentage of Catholics in each country comes from The World Quality of Life Indicators (O’Donnell 1991) and The Statistical Abstract of the World (Reddy 1994). Identification of communist/transition status was made using The World Factbook (1992), compiled by the Central Intelligence Agency. Democracy scores were provided by Paxton (2000). Figures for GDP per capita were obtained from the World Development Indicators (World Bank 1999) as well as the

1 While most of the surveys for this wave were carried out in 1990, the actual time frame ranges from 1988–1993.

does religion require family ?

325

United Nation’s Human Development Report (United Nations 1990; 1993b). Data for percent urbanized are from World Development Indicators (World Bank 1999). Whenever possible, figures for the variables are from 1990. When it was not possible to obtain 1990 figures, figures from the nearest available year were used. Sample Countries for this study were chosen on the basis of their inclusion in the 1990 World Values Survey.2,3 Although there is a total of forty-one countries, it should be noted that the number of countries varies for the different variables depending on the availability of data. There are forty-one countries for eight variables: family importance, religion importance, religious attendance, Catholic predominance, communist/transition status, level of democracy, GDP per capita, and degree urbanized. For the remaining four variables, there are fewer countries due to missing data: thirty-nine countries for high education; thirtyseven countries for crude divorce rate;4 thirty-six for nonmarital births, and thirty-three for prayer. Measurement Strength in the Family Institution The 1990 World Values Survey asked respondents to indicate how important family was in their lives: very important, quite important, not very important, or not at all important. This question serves as an attitudinal assessment of family strength in the present study. The figure used for each country is the percentage of individuals who responded that family was very important. Behaviorally, family strength for each

Because frequency of attendance at religious services was not asked in Estonia, Lithuanian, or South Africa in the 1990–1993 wave of the World Values Survey, figures for these countries were taken from the 1995–1997 wave. 3 These countries are: Argentina, Austria, Belarus, Belgium, Brazil, Britain, Bulgaria, Canada, Chile, China, Czech-Slovakia, Denmark, East Germany, Estonia, France, Finland, Hungary, Iceland, India, Ireland, Italy, Japan, Latvia, Lithuania, Mexico, Netherlands, Nigeria, Norway, Poland, Portugal, Romania, Russia, Slovenia, South Africa, South Korea, Spain, Sweden, Switzerland, Turkey, United States, West Germany. 4 Because divorce was not legal in Ireland until 1995, there is no crude divorce rate figure for this country. In Chile, no absolute divorce is permitted (Kreeft and Tacelli 1994). Seeking an annulment is the only way to “divorce” (IWRAW Publications 1995), and so this is the figure used for Chile. 2

326

kristen r. heimdal and sharon k. houseknecht

country is measured in two ways—the percentage of all births that are nonmarital and the crude divorce rate, measured as the number of divorces per 1,000 population. Strength in the Institution of Religion One indicator of strength in the institution of religion is attitudinal and is based on a question in the World Values Survey which asked respondents to indicate how important religion was in their lives. The percent of respondents within each country answering very important (rather than quite, not very, or not at all important) is used to indicate religious strength. Behaviorally, strength of religion is operationalized in two ways. The first is the percentage of people in each country who report attending religious services once a week or more, apart from weddings, funerals, and christenings, and the second is frequency of prayer. The percentage of people in each country who report that they pray to God often (rather than sometimes, hardly ever, only in times of crisis, and never) outside of religious services is the figure that is used. Contextual Variables A country is considered Catholic predominant if the population is at least eighty-five percent Catholic. Catholic predominant countries are coded as 1; all other countries are coded as 0. The variable, communist/communist transition status, refers to countries that were either communist in 1990, or making the transition out of communist regimes. Countries with this classification are coded as 1; all other countries are coded as 0. Level of democracy is assessed using Bollen’s democracy measure. As described by Bollen (1980), this measure includes three indicators of popular sovereignty through the electoral process: fairness of elections, whether the chief executive is elected, and effectiveness of means of selecting the legislative body; and three measures of political liberties: freedom of the press, freedom of group opposition, and government sanctions. Figures for this measure range from 0 (autocracy) to 100 (democracy). Using Bollen’s (1980) approach to calculate democracy scores, Paxton (2000) calculated scores for all the countries in the second wave of the World Values Survey. It is these figures that are used here. Gross domestic product per capita is gross domestic product converted to international dollars using purchasing power parity rates (PPP), which, according to the World Bank (1999), adjusts for such factors as the degree of openness of an economy, the relative sizes of tradable and nontradable sectors, and possible overvaluation of exchange rates.

does religion require family ?

327

Education is assessed using a World Values Survey question that asked, “At what age did you or will you complete your full-time education, either at school or at an institution of higher learning?” Because we are interested in relatively high levels of education, and for lack of a better measure, education is operationalized as the percentage of individuals who reported either completing or intending to complete their education at age 21 or older, the highest age category given. Asking respondents to report when they intend to finish their education is potentially problematic in that intention may prove to be inconsistent with behavior (i.e., the age at which respondents will actually complete their education), thus jeopardizing the validity of the measure. To help remedy this problem, only respondents who were 21 years of age or older at the time of the interview are included. Percent urbanized is measured by determining what percentage the midyear population of areas defined as urban is of the total population. Analytical Approach The first step in ascertaining the strength of the family-religion linkage is the calculation of correlations between the measures of strength in family and strength in religion. Correlations, however, are only able to capture whether there is a statistically significant linear association among the measures. A typology, in contrast, is able to take into account specific combinations of strength in family and strength in religion, and so we use typologies to explore whether one institution appears to have primacy over the other. Typologies will be constructed using two variables at a time—one indicator of family strength and one indicator of strength in religion. Because there may be discrepancies between attitudes and behavior, one typology includes only attitudinal measures, some only behavioral measures, and some a combination of both. To create these typologies, each variable is trichotomized (strong, moderate, and weak), thus producing nine categories into which the countries are classified.5 There is a total of three family variables and three religion variables (one attitudinal and two behavioral in each case), yielding nine separate typologies. Figure 1 provides an illustration of the structure of the typologies and

5 Since divorce was not legal in Ireland prior to 1995, Ireland will be classified as having “strong family” in the three typologies in which the crude divorce rate is the measure of family strength.

328

kristen r. heimdal and sharon k. houseknecht

highlights the key cells on which attention should primarily be focused in order to determine support for our major hypothesis, that strong family is a requisite for strong religion. Support would be indicated by a small number of countries in the moderate family/strong religion, weak family/moderate religion, and especially in weak family/strong religion categories. Further, countries classified as strong family should be distributed fairly evenly across the three degrees of strength of religion. However, if strong family is a prerequisite for strong religion, countries classified as having strong religion should be concentrated primarily under strong family. Two criteria guided the determination of cutting points for weak, moderate, and strong categories of the various measures of family and religion: 1) optimizing homogeneity within categories, and 2) having a sufficient number of cases within the categories for the analysis. Of the several possible approaches for defining cutting points, two were used in this study. First, we looked for natural breaks in the distributions and used them when they met the criteria. When the distributional curves were either smooth or presented too many breaks, we divided the range into three equal parts so as to avoid arbitrary selection. For this latter approach, we excluded extreme outliers before establishing the cutting points. The nine typologies are used to establish patterns of strength in the family institution vis-à-vis the institution of religion. The gamma statistic is calculated for each typology in order to assess the strength of association between each combination of strength in family and strength in religion variables. Gamma is the most appropriate measure of association here because the strength in family and strength in religion measures are at the ordinal level and because gamma is a symmetrical measure and so specification of which variable is independent and which is dependent is not required. We are cautious, however, about relying solely on the use of gamma to establish support for our major hypothesis, that a strong institution of religion requires a strong family institution. While a statistically significant gamma value does imply some degree of strength between the family and religion variables in the typologies, it is limited in its ability to capture support for our hypothesis. Our expectation is not that strength in family would predict strength in religion, but rather that strong family is a requisite for strong religion. In other words, since family is the major socialization agent for religion, in those countries where families do not consider religion to be all that important, strong family could appear with moderate or weak religion. Thus, in addition to gammas, we focus on the distribu-

does religion require family ? Strong

329

Strength of Family Moderate

Weak

*

*

Strong Strength of Religion Moderate

*

Weak

Key:

*

A small number of countries is expected in these cells Countries classified as having strong family should be distributed fairly evenly across the three degrees of strength in religion Countries classified as having strong religion should be concentrated primarily under strong family

Figure 1. Illustrative typology showing major analysis considerations.

tion of countries throughout the cells of the typologies for determining support for our hypothesis. The final step is to see whether there are contextual variables that are associated with different patterns of strength. In other words, what is the mean value (or percentage) of a contextual variable for the countries in each category of the typology? In each case, only those countries that fall within each category of the typology are considered. Results Descriptive Statistics Table 1 provides descriptive information for the sample in this crosssocietal study. One thing that stands out for most of the variables is the large range in figures. For example, the proportion of births that are nonmarital ranges from .5 to 58.3 percent; the crude divorce rate ranges from .5 to 4.7 per 1,000; the percent of respondents reporting

Continuous Measure—number of divorces per 1,000 population Ordinal Measure—strong (1.4 and below); moderate (1.5–3.0); weak (3.1 and above)

Divorce

Strength of Religion Religion Importance Continuous Measure—percentage of respondents within countries reporting religion is very important in their lives Ordinal Measure—strong (41.9% and above); moderate (21.4%–41.8%); weak (21.3% and below)

Continuous Measure—proportion of all births that are nonmarital for each country Ordinal Measure—strong (19.5% and below); moderate (19.6%–34.3%); weak (34.4% and above)

Continuous Measure—percentage of respondents within countries reporting family is very important in their lives Ordinal Measurea—strong (83.3% and above); moderate (72.6%–83.2%); weak (72.5% and below)

Definition

Nonmarital Births

Strength of Family Family Importance

Variable Name

41

37

36

41

Number of Countries

26.96

1.98

21.28

83.2

Mean

Table 1. Variables, definitions, and descriptive statistics

19.49

1.15

14.49

8.33

Std. Dev.

1.4

0.5

0.5

62.2

85.3

4.7

58.3

93.9

Minimum Maximum

330 kristen r. heimdal and sharon k. houseknecht

Percentage of respondents within countries reporting praying often Ordinal Measure—strong (27.0% and above); moderate (14.8%–26.9%); weak (14.7% and below)

Prayer

a

Bollen’s democracy score based on three measures of political participation and three measures of political liberties

Percenatage of respondents within countries completing education at age 21 or older

In current international dollars using purchasing power parity rates (PPP)

Percentage of population in each country residing in urban area

Democracy Score

High Education

GDP per capita

Degree Urbanized

All Ordinal Measures based on corresponding continuous measure

41

Countries communist in 1990 or undergoing a transition out of communism

Communist/ Transition

41

41

39

41

41

33

41

Countries 85% or more Catholic

Contextual Measures Catholic Predominant

Continuous Measure—percentage of respondents within countries attending religious services once or more per week Ordinal Measure—strong (29.3% and above); moderate (5.6%–29.2%); weak (5.5% and below)

Religious Attendance

68.30

10252.90

27.47

83.55

0.32

0.29

26.91

24.23

17.10

6084.69

15.57

22.41

0.47

0.46

18.11

21.12

25.50

1072.00

0.60

0

0

0

0.8

0.5

96.50

21449.00

61.40

100

1

1

81.0

83.7

does religion require family ? 331

332

kristen r. heimdal and sharon k. houseknecht

that religion is very important in their lives ranges from 1.4 to 85.3 percent; the percent of respondents with frequent religious service attendance ranges from .5 to 83.7 percent; and the percent of respondents reporting that they pray often ranges from .8 for 81.0 percent. For the contextual variables, democracy scores range from 0, the lowest possible, to 100, the highest possible; the percent of respondents completing education at age 22 or older (high education) ranges from .7 to 57.9 percent; GDP per capita ranges from $1072 to $21,449, and degree urbanized ranges from 25.5 to 96.5 percent. In contrast to the variables just mentioned, the range for percent of respondents reporting that family is very important in their lives is relatively narrow (62.0 to 93.8 percent). Family and Religion Linkage Correlation results (not shown here) reveal that attitudes regarding importance of family and religion in respondents’ lives are positively associated. Family importance also is positively associated with both behavioral measures of religion, religious attendance and prayer. Furthermore, a behavioral measure of family weakness, crude divorce rate, is inversely correlated with importance attributed to religion, as well as with religious attendance. Interestingly, family importance does not significantly correlate with the two behavioral measures of family strength, divorce and nonmarital births. Conversely, there is consistency between attitudes and behavior for religion, with religion importance having a high, positive correlation with both religious attendance and prayer. In sum, although the family variables are not strongly correlated, the attitudinal family variable (importance ascribed to the family) and one of the family behavioral variables (crude divorce rate) are rather highly correlated with the religion variables. The strongest correlations, though, are amongst the three religion variables. Using gamma to further assess the family and religion linkage, we find significant results beyond the .05 level in five out of nine typologies. Primacy of Family over Religion If the major hypothesis, that strong family is a requisite for strong religion is to be supported, we should find a small number of countries in certain key cells: moderate family/strong religion, weak family/moderate religion, and especially in weak family/strong religion. Based on the indicators available to us, the results in Table 2 show strong support for the expected pattern in all nine typologies. In four of them, there is

does religion require family ?

333

only one country in the weak family/strong religion category, and in the remaining five typologies, there are no countries in this category. Furthermore, the number of countries in the weak family/moderate religion category does not exceed two. These findings lend credence to the notion that where there is weak family, there is not likely to be strong, or even moderately strong, religion. Even in the moderate family/strong religion category, there are no more than three countries in any of the nine typologies, suggesting that even moderately strong family tends not to be enough to produce strong religion. Further, in four of the typologies, there are fewer countries in the moderate family/moderate religion category than the moderate family/weak religion category. Thus, moderately strong family may not be enough to produce even moderately strong religion. This is especially the case in the attitudinal typology, as there is just one country in the moderate family/moderate religion category but eight in the moderate family/weak religion category. Additional support for the hypothesis comes from the finding that in almost all of the typologies, countries classified as strong family are distributed fairly evenly across the three degrees of strength of religion categories. However, this is not the case for the distribution of countries in the strong religion category. They are concentrated primarily under strong family. Because the weak family/strong religion category is the crucial category for determining whether strong family is a likely requisite of strong religion, it merits further attention. In the three typologies with family importance as the measure of strength in family, the weak family/strong religion category is empty in two of the typologies; in the third, there is only one country in this category (Portugal). There are no countries in the weak family/strong religion category in the three typologies using the proportion of nonmarital births as the measure of family strength. In each of the three typologies with crude divorce rate as the family strength measure, there is one country in the weak family/strong religion category, and it is the same for all three typologies: the United States. This finding is consistent with the notions of religious exceptionalism (Warner 1993) and what we will call divorce exceptionalism in the United States. The United States is “unusually religious” (Kelley and De Graaf 1997) and also stands out as having one of the highest divorce rates in the world (United Nations 1993a). Not surprising, the consistent categories of strong family/strong religion and weak family/weak religion have fairly strong representation in all of the typologies. The countries that are found in the strong family/strong religion category of the various typologies, however,

334

kristen r. heimdal and sharon k. houseknecht

Table 2. Country-level results from nine typologies assessing relative strength of family and religion Typology Name

Family Importance and Religion Importance Strength of Family

Strength of Religion

Strong

Moderate

Weak

Total

Strong

Row = 88.9% Column = 32.0% (n = 8)

Row = 11.1% Column = 10.0% (n = 1)

Row = 0.0% Column = 0.0% (n = 0)

22.0% (n = 9)

Moderate

Row = 90.0% Column = 36.0% (n = 9)

Row = 10.0% Column = 10.0% (n = 1)

Row = 0.0% Column = 0.0% (n = 0)

24.4% (n = 10)

Weak

Row = 36.4% Column = 32.0% (n = 8)

Row = 36.4% Column = 80.0% (n = 8)

Row = 27.3% Column = 100.0% (n = 6)

53.7% (n = 22)

Total

61.0% (n = 25)

24.4% (n = 10)

14.5% (n = 6)

100.0% (N = 41)

Gamma = .823**

Typology Name

Family Importance and Religious Attendance Strength of Family

Strength of Religion

Strong

Moderate

Weak

Total

Strong

Row = 83.3% Column = 40.0% (n = 10)

Row = 8.3% Column = 10.0% (n = 1)

Row = 8.3% Column = 16.7% (n = 1)

29.3% (n = 12)

Moderate

Row = 58.8% Column = 40.0% (n = 10)

Row = 29.4% Column = 50.0% (n = 5)

Row = 11.8% Column = 33.3% (n = 2)

41.5% (n = 17)

Weak

Row = 41.7% Column = 20.0% (n = 5)

Row = 33.3% Column = 40.0% (n = 4)

Row = 25.0% Column = 50.0% (n = 3)

29.3% (n = 12)

Total

61.0% (n = 25)

24.4% (n = 10)

14.6% (n = 6)

100.0% (N = 41)

Gamma = .474*

Typology Name

Family Importance and Prayer Strength of Family

Strength of Religion

Strong

Moderate

Weak

Total

Strong

Row = 81.8% Column = 47.4% (n = 9)

Row = 18.2% Column = 20.0% (n = 2)

Row = 0.0% Column = 0.0% (n = 0)

33.3% (n = 11)

Moderate

Row = 63.6% Column = 36.8% (n = 7)

Row = 18.2% Column = 20.0% (n = 2)

Row = 18.2% Column = 50.0% (n = 2)

33.3% (n = 11)

does religion require family ?

335

Table 2 (cont.) Typology Name

Family Importance and Prayer Strength of Family Strong

Moderate

Weak

Total

Weak

Row = 57.6% Column = 15.8% (n = 3)

Row = 54.4% Column = 60.0% (n = 6)

Row = 18.2% Column = 50.0% (n = 2)

33.3% (n = 11)

Total

57.6% (n = 19)

30.3% (n = 10)

12.1% (n = 4)

100.0% (N = 33)

Gamma = .600**

Typology Name

Nonmarital Births and Religion Importance Strength of Family

Strength of Religion

Strong

Moderate

Weak

Total

Strong

Row = 40.0% Column = 9.1% (n = 2)

Row = 60.0% Column = 33.3% (n = 3)

Row = 0.0% Column = 0.0% (n = 0)

13.9% (n = 5)

Moderate

Row = 70.0% Column = 31.8% (n = 7)

Row = 20.0% Column = 22.2% (n = 2)

Row = 10.0% Column = 20.0% (n = 1)

27.8% (n = 10)

Weak

Row = 61.9% Column = 59.1% (n = 13)

Row = 19.0% Column = 44.4% (n = 4)

Row = 19.0% Column = 80.0% (n = 4)

58.3% (n = 21)

Total

61.1% (n = 22)

25.0% (n = 9)

13.9% (n = 5)

100.0% (N = 36)

Gamma = –.010

Typology Name

Nonmarital Births and Religious Attendance Strength of Family

Strength of Religion

Strong

Moderate

Weak

Total

Strong

Row = 62.5% Column = 22.7% (n = 5)

Row = 37.5% Column = 33.3% (n = 3)

Row = 0.0% Column = 0.0% (n = 0)

22.2% (n = 8)

Moderate

Row = 64.7% Column = 50.0% (n = 11)

Row = 29.4% Column = 55.6% (n = 5)

Row = 5.9% Column = 20.0% (n = 1)

47.2% (n = 17)

Weak

Row = 54.5% Column = 27.3% (n = 6)

Row = 9.1% Column = 11.1% (n = 1)

Row = 36.4% Column = 80.0% (n = 4)

30.6% (n = 11)

Total

61.1% (n = 22)

25.0% (n = 9)

13.9% (n = 5)

100.0% (N = 36)

Gamma = .245

336

kristen r. heimdal and sharon k. houseknecht

Table 2 (cont.) Typology Name

Nonmarital Births and Prayer Strength of Family

Strength of Religion

Strong

Moderate

Weak

Total

Strong

Row = 62.5% Column = 29.4% (n = 5)

Row = 37.5% Column = 42.9% (n = 3)

Row = 0.0% Column = 0.0% (n = 0)

27.6% (n = 8)

Moderate

Row = 63.6% Column = 41.2% (n = 7)

Row = 18.2% Column = 28.6% (n = 2)

Row = 18.2% Column = 40.0% (n = 2)

37.9% (n = 11)

Weak

Row = 50.0% Column = 29.4% (n = 5)

Row = 20.0% Column = 28.6% (n = 2)

Row = 30.0% Column = 60.0% (n = 3)

34.5% (n = 10)

Total

58.6% (n = 17)

24.1% (n = 7)

17.2% (n = 5)

100.0% (N = 29)

Gamma = .393

Typology Name

Divorce and Religion Importance Strength of Family

Strength of Religion

Strong

Moderate

Weak

Total

Strong

Row = 87.5% Column = 37.5% (n = 6)

Row = 0.0% Column = 0.0% (n = 0)

Row = 14.3% Column = 16.7% (n = 1)

18.4% (n = 7)

Moderate

Row = 44.4% Column = 25.0% (n = 4)

Row = 55.6% Column = 31.3% (n = 5)

Row = 0.0% Column = 0.0% (n = 0)

23.7% (n = 9)

Weak

Row = 27.3% Column = 37.5% (n = 6)

Row = 50.0% Column = 68.8% (n = 11)

Row = 22.7% Column = 83.3% (n = 5)

57.9% (n = 22)

Total

42.1% (n = 16)

42.1% (n = 16)

15.8% (n = 6)

100.0% (N = 38)

Gamma = .587*

Typology Name

Divorce and Religious Attendance Strength of Family

Strength of Religion

Strong

Moderate

Weak

Total

Strong

Row = 88.9% Column = 50.0% (n = 8)

Row = 0.0% Column = 0.0% (n = 0)

Row = 11.1% Column = 16.7% (n = 1)

23.7% (n = 9)

Moderate

Row = 29.4% Column = 31.3% (n = 5)

Row = 64.7% Column = 68.8% (n = 11)

Row = 5.9% Column = 16.7% (n = 1)

44.7% (n = 17)

does religion require family ?

337

Table 2 (cont.) Typology Name

Divorce and Religious Attendance Strength of Family Strong

Moderate

Weak

Total

Weak

Row = 25.0% Column = 18.8% (n = 3)

Row = 41.7% Column = 31.3% (n = 5)

Row = 33.3% Column = 66.7% (n = 4)

31.6% (n = 12)

Total

42.1% (n = 16)

42.1% (n = 16)

15.8% (n = 6)

100.0% (N = 38)

Gamma = .596**

Typology Name

Divorce and Prayer Strength of Family

Strength of Religion

Strong

Moderate

Weak

Total

Strong

Row = 75.0% Column = 50.0% (n = 6)

Row = 12.5% Column = 7.1% (n = 1)

Row = 12.5% Column = 25.0% (n = 1)

26.7% (n = 8)

Moderate

Row = 18.2% Column = 16.7% (n = 2)

Row = 81.8% Column = 64.3% (n = 9)

Row = 0.0% Column = 0.0% (n = 0)

36.7% (n = 11)

Weak

Row = 36.4% Column = 33.3% (n = 4)

Row = 36.4% Column = 28.6% (n = 4)

Row = 27.3% Column = 75.0% (n = 3)

36.7% (n = 11)

Total

40.0% (n = 12)

46.7% (n = 14)

13.3% (n = 4)

100.0% (N = 30)

Gamma = .393 * Significant at the .01 level (2-tailed) ** Significant at the .05 level (2-tailed)

tend to be greater in number and are more likely to be the same ones than is the case for those appearing in the weak family/weak religion category. Overall, countries tend to be fairly consistent with regard to where they fall in the nine typologies. It is interesting to consider some of the specific countries that fall into the strong family/strong religion and weak family/weak religion categories of the typologies. Ireland is the only country that appears in the strong family/strong religion category in all nine of the typologies. Ireland, then, reveals strength across all three family variables and across all three religion variables. Available data indicate that Brazil, Nigeria and Poland could be similar in this regard, but missing data for these countries preclude a final determination. Italy and Mexico also appear

338

kristen r. heimdal and sharon k. houseknecht

frequently in the strong family/strong religion category, and they are identical in their strength of family and religion patterns across the different typologies. The only exceptions are those typologies that include attitudes about the importance of religion. In these instances, Italy and Mexico are characterized as having only moderately strong religion. Latvia is the country most frequently seen in the weak family/weak religion category (in six of the nine typologies). Contextual Variables and Different Patterns of Relative Strength The means for democracy score, high education, GDP per capita, and percent urbanized are calculated, as well as the percentage of Catholic predominant and communist/transition countries for each cell (only the categories that contain two or more countries can be analyzed). The results are presented in Table 3. It was hypothesized that Catholic predominance would be linked with strong family and strong religion. The distribution of Catholic predominant countries in the typologies demonstrates this pattern quite clearly. In seven of the typologies, the strong family and strong, or in some cases moderate, religion categories have the highest percentage of Catholic predominant countries. In eight of the typologies, there are no Catholic predominant countries in the weak family/weak religion category. In the strong family/weak religion category, where few Catholic predominant countries were expected, six of the typologies have no Catholic predominant countries. Expectations also were that strength in family and religion would be inversely associated with communist/transition status, and this is the pattern that is found in the typologies. In six of the typologies, 100 percent of the countries in the weak family/weak religion category are characterized by communist/transition status. The association tends to be linear, especially when family and religion attitudes are considered. Interestingly, all three of the typologies for which these findings do not hold include the variable, nonmarital births. With the exception of one typology, the strong family/strong religion category does not exceed 20 percent communist/transition status in any of the typologies. Further, moderate family/moderate religion categories are no more than 40 percent communist/transition, and most are much lower. In five of the typologies, the mean democracy score is higher in the strong family/strong religion category than in the weak family/weak religion category, thus supporting the expectation of a positive association between democracy score and family/religion strength. In the

50.00 12.50 70.70 23.09 7103.00 63.36

60.00 10.00 74.81 20.99 7782.80 63.93

66.70 0.00 78.90 24.74 9822.78 70.82

Family Importance/Attendance Catholic Predominant (%) Communist/Transition (%) Democracy Score (mean) High Education (mean %) GDP per capita (mean $) Degree Urbanized (mean %)

Family Importance/Prayer Catholic Predominant (%) Communist/Transition (%) Democracy Score (mean) High Education (mean %) GDP per capita (mean $) Degree Urbanized (mean %) 14.30 14.30 97.95 26.54 14825.00 76.64

20.00 30.00 90.62 17.83 12275.20 75.85

44.40 11.10 91.83 20.67 12239.67 72.56

0.00 33.30 94.32 36.27 13931.67 81.97

0.00 0.00 99.76 45.94 16553.00 78.44

0.00 25.00 95.13 31.60 14545.50 78.76

0.00 50.00 79.17 35.05 2036.00 39.55

50.00 0.00 91.36 23.60 13709.00 82.15

60.00 40.00 87.26 23.42 11368.60 68.48

Mod Fam/ Mod Rel

Typology Categories

Strong Fam/ Strong Fam/ Strong Fam/ Mod Fam/ Strong Rel Mod Rel Weak Rel Strong Rel

Family Importance/Religion Importance Catholic Predominant (%) Communist/Transition (%) Democracy Score (mean) High Education (mean %) GDP per capita (mean $) Degree Urbanized (mean %)

Typologies/Contextual Variables

33.30 66.70 83.84 30.25 10372.67 68.20

0.00 75.00 79.04 34.45 8952.75 71.17

37.50 50.00 85.72 28.59 11206.75 71.69

50.00 0.00 91.24 11.20 11750.00 59.75

0.00 50.00 85.28 25.90 9821.50 77.40

0.00 100.00 35.41 42.15 4223.50 48.70

0.00 100.00 52.77 42.63 4961.67 56.40

16.70 66.70 71.38 31.75 7216.33 59.58

Mod Fam/ Weak Fam/ Weak Fam/ Weak Fam/ Weak Rel Strong Rel Mod Rel Weak Rel

Table 3. Distribution of contextual variables within categories of the typologies

does religion require family ? 339

100.00 50.00 93.49 11.50 7413.00 59.35

100.00 20.00 93.04 12.20 9080.80 58.28

60.00 20.00 88.90 18.86 10925.80 65.26

Nonmarital Births/Attendance Catholic Predominant (%) Communist/Transition (%) Democracy Score (mean) High Education (mean %) GDP per capita (mean $) Degree Urbanized (mean %)

Nonmarital Births/Prayer Catholic Predominant (%) Communist/Transition (%) Democracy Score (mean) High Education (mean %) GDP per capita (mean $) Degree Urbanized (mean %) 28.60 14.30 93.21 21.41 12837.29 71.97

9.10 36.40 87.70 22.59 11582.36 73.30

28.60 28.60 87.29 18.86 10837.46 66.43

0.00 80.00 75.73 38.78 8453.60 71.18

0.00 66.70 79.73 40.92 9785.67 69.55

15.40 46.20 85.40 30.19 10833.54 71.64

66.70 0.00 87.16 30.87 10281.00 85.00

33.30 0.00 77.44 27.73 10203.00 70.17

33.30 0.00 76.22 28.40 10471.00 72.43

50.00 0.00 99.62 9.20 16154.00 76.80

80.00 20.00 93.06 18.08 12766.40 74.28

100.00 0.00 91.16 17.15 10399.50 75.50

Mod Fam/ Mod Rel

Typology Categories

Strong Fam/ Strong Fam/ Strong Fam/ Mod Fam/ Strong Rel Mod Rel Weak Rel Strong Rel

Nonmarital Births/Religion Importance Catholic Predominant (%) Communist/Transition (%) Democracy Score (mean) High Education (mean %) GDP per capita (mean $) Degree Urbanized (mean %)

Typologies/Contextual Variables

Table 3 (cont.)

100.00 50.00 93.44 21.85 13212.50 62.25

50.00 50.00 93.53 24.43 12166.75 71.35

0.00 0.00 99.57 43.55 16262.00 81.45

0.00 33.30 94.32 36.27 13931.67 81.97

0.00 0.00 99.76 44.53 16579.75 82.70

0.00 25.00 95.59 38.40 14455.75 79.55

Mod Fam/ Weak Fam/ Weak Fam/ Weak Fam/ Weak Rel Strong Rel Mod Rel Weak Rel

340 kristen r. heimdal and sharon k. houseknecht

75.00 12.50 83.15 11.88 7454.88 59.51

83.30 16.70 82.49 15.12 7535.67 69.62

Divorce/Attendance Catholic Predominant (%) Communist/Transition (%) Democracy Score (mean) High Education (mean %) GDP per capita (mean $) Degree Urbanized (mean %)

Divorce/Prayer Catholic Predominant (%) Communist/Transition (%) Democracy Score (mean) High Education (mean %) GDP per capita (mean $) Degree Urbanized (mean %) 100.00 0.00 91.36 14.35 10246.50 54.45

60.00 40.00 77.49 24.00 6734.80 69.32

50.00 25.00 79.08 17.80 7410.00 66.65

25.00 75.00 70.70 25.53 8281.50 55.15

0.00 66.70 65.11 26.17 8035.33 56.70

50.00 50.00 77.59 21.80 8936.50 54.92

Note: Figures for categories with fewer than two countries were excluded.

66.70 16.70 77.69 14.22 5693.33 66.12

Divorce/Religion Importance Catholic Predominant (%) Communist/Transition (%) Democracy Score (mean) High Education (mean %) GDP per capita (mean $) Degree Urbanized (mean %)

11.10 11.10 96.46 25.19 14911.11 79.04

18.20 27.30 94.37 17.53 14604.64 76.39

20.00 20.00 98.80 22.80 15844.40 70.70

25.00 25.00 95.59 32.23 14800.00 79.98

0.00 0.00 99.76 45.94 16553.00 78.44

9.10 18.20 94.81 28.53 14926.73 79.91

0.00 100.00 63.89 47.87 6717.33 70.67

0.00 100.00 69.70 46.80 6647.50 70.95

0.00 100.00 73.33 45.48 6300.60 70.52

does religion require family ? 341

342

kristen r. heimdal and sharon k. houseknecht

remaining four typologies, the mean democracy score is lower in the strong family/strong religion category than in the weak family/weak religion category, showing support for the alternative hypothesis predicting that countries characterized by weak family and weak religion would tend to have higher democracy scores. The results, however, actually suggest a curvilinear association, with higher mean democracy scores in the moderate family/moderate religion category than in both the strong family/strong religion and weak family/weak religion categories in six of the eight typologies with available information. It was expected that countries with strong family and strong religion would have relatively lower proportions of highly educated persons, a pattern that is clearly demonstrated in the typologies. In all nine typologies, the mean percentage of highly educated persons is greater in the weak family/weak religion category than in the strong family/strong religion category. And there is a striking difference in the range of percentages, with the proportion of highly educated persons ranging from 11.50 to 24.74 percent in the strong family/strong religion category and from 31.75 to 47.87 in the weak family/weak religion category. Furthermore, in seven of the nine typologies, the highest proportion of highly educated persons is found in the weak family/weak religion category. The percentage of highly educated persons in the moderate family/moderate religion category falls between the percentage in the strong family and strong religion and weak family/weak religion categories in seven of the eight typologies with available information, which also is the expected finding. It was hypothesized that countries with weak family and weak religion would tend to have higher GDP per capita. The results show some support for this expectation. In five of the typologies, the mean GDP per capita is higher in the weak family/weak religion category than in the strong family/strong religion category. In general, though, the association between the relative strengths of family and religion and GDP per capita tends to be curvilinear. For six of the eight typologies that contain information for these categories, six reveal a curvilinear pattern, with the moderate family/moderate religion category having a higher mean GDP per capita than both the strong family/strong religion and weak family/weak religion categories. Turning to degree of urbanization, in five of the typologies, the mean percent of the population living in urban areas is higher in the weak family/weak religion category than in the strong family/strong religion category, which is the expected finding. In another typology, the figures for these two categories are nearly identical. Like in the

does religion require family ?

343

case of GDP per capita, the overall association between the relative strengths of family and religion and degree of urbanization tends to be curvilinear. In five of the eight typologies with available information, the mean percentage for degree urbanized is higher in the moderate family/moderate religion category than in both the strong family/strong religion and weak family/weak religion categories. In sum, three of the six contextual variables clearly demonstrate the expected pattern in the typologies: Catholic predominance, communist/transition status, and high education. The results for democracy score were neither consistent with the hypothesis predicting a positive association, nor consistent with the hypothesis predicting an inverse association. Instead, a curvilinear association between democracy score and family/religion strength is observed. Results for GDP per capita and degree urbanized, too, reveal a curvilinear pattern, rather than the expected inverse relationship. The mean figures for these three variables in several of the typologies tend to be lower in the strong family/strong religion category and weak family/weak religion category than in the moderate family/moderate religion category. Given that democracy score, GDP per capita, and degree urbanized are significantly correlated with one another (not shown here), it is not surprising that they display a similar pattern in the typologies. A possible explanation for the curvilinear associations lies in the type of country most commonly found in the weak family/weak religion category, that is, communist/transition society. Further examination (not shown here) revealed that communist/transition status countries have the lowest mean figures of all the countries for these three contextual variables. This situation is especially the case for GDP per capita, in that the mean GDP per capita for communist/transition countries is half that of the other countries. Discussion and Conclusions The major goal of this study was to explore whether strength in family may be a requisite for strength in religion. We expected that the combination of a strong religious institution and a weak family institution would be a rare occurrence in the countries in our study. However, we did not think that a strong family institution would ensure a strong institution of religion. If families (as the major socializing agents of religion) do not emphasize the importance of religion to family members, then the religion institution is not likely to be strong, regardless

344

kristen r. heimdal and sharon k. houseknecht

of how strong the family institution is. The idea here is that strong family is a necessary but not sufficient requirement for strong religion. The typologies in this preliminary work show compelling support for this idea. If strong family is a requisite for strong religion, as our findings suggest, and if strength of family is declining in many societies, as Popenoe (1988) and some others have argued, the institution of religion may be threatened. Family decline could mean decline in the institution of religion in those places where strength of religion is moderate or strong. This may not necessarily be the case, however. A weakening of the institution of family might reflect structural changes in the family that are closely related to changing norms. The value dimension of culture, though, could very well remain potent in that there may still be a strong emphasis on companionship, children, intimacy, and so on. From this cultural perspective, we see that family strength may not be declining across the board and that there may be important pockets of strength, as in the value realm. Even in societies with weak or weakening family institutions, religious decline is not inevitable. Religious institutions may, for example, devise strategies that allow them to adapt to changes in the family institution. As Wilcox, Chaves, and Franz (2004) found, at least in the United States, most congregations are not reaching out to those in so-called unconventional family situations. However, those that offer activities for singles, single parents, and the divorced, for example, may be able to maintain members and attract new participants. Further, it is questionable whether the family institution is, in fact, undergoing significant decline. While the cross-sectional nature of this study does not allow examination of change over time, the findings reveal that, considering only the levels of family strength, in all nine of the typologies there are quite a few more countries classified as having strong family than there are as having weak family (in fact, in seven out of nine typologies, there are at least three times as many countries classified as having strong family than there are as having weak family). This pattern suggests that, even in a number of highly industrialized societies, family remains strong, thus making it possible for religion to maintain strength as well. Of course, there are a number of countries that have a weak family institution. Since our findings support the notion that countries with a weak family institution are likely to have a weak institution of religion, does this mean that both institutions will necessarily remain weak? We

does religion require family ?

345

do not believe this to be the case. Institutions are constantly changing, and change does not necessarily mean further weakness. Once thought to trigger decline in religious strength, there is evidence that differentiation of political and religious structures actually results in increased involvement in religion. Using case studies of Mormonism in 19th century Utah and Protestantism during the Second Great Awakening, Phillips (2004), for example, demonstrated that, when religious structures lose political and economic power, the meaning and importance of religious involvement increases. The post-communist countries of Central and Eastern Europe offer potential examples of change in the institutions of family and religion in the form of increased, rather than decreased, strength. Data (not shown here) from the 1995 wave of the World Values Survey suggest that, at least attitudinally, strength of family has increased in some Central and East European countries. In six of the nine post-communist Central and East European countries for which data are available, a higher percentage of respondents than in the 1990 wave report that family is very important in their lives. Growing strength of family may allow for increased strength of religion, as well. It was previously mentioned that Communist governments are suspicious of families and resent the family’s role in socialization (Cseh-Szombathy 1990). The postcommunist climate may allow families to be more effective agents of socialization. And, as the fear of openly practicing religion is removed, it is quite possible that parents will feel increased freedom to convey the importance of religion to their children. Data from the 1990 and 1995 waves of the World Values Survey (not shown here) show some support for this idea. In both waves, respondents were presented with a list of eleven qualities that children could be encouraged to learn at home and were asked to choose up to five qualities they considered to be especially important. In six of the eight post-communist countries for which figures are available for both 1990 and 1995, there is at least a modest increase in the percentage of respondents citing religious faith in 1995 compared with 1990. Finally, it is interesting to consider our finding that a strong family institution appears to be a requisite for a strong institution of religion in light of the fact, already mentioned, that the bulk of the research on the family/religion linkage has focused on the impact of religion on family. A widely touted generalization, for example, is that “The family that prays together stays together” (cf. Call and Heaton 1997; Larson and Goltz 1989). Of course, the unit of analysis in that research

346

kristen r. heimdal and sharon k. houseknecht

is either the individual, the couple, or the family. Our data suggest that, at least at the societal level of analysis, a strong family institution is possible without having strong religion. A question raised by our research, then, is, What is it other than religion that strengthens the family institution? Perhaps one reason scholars have linked religion so strongly with positive family outcomes is because the research to date has tended to focus on the United States which has a strong institution of religion and a weak family institution. It is well known, as previously mentioned, that the United States is an exception in almost every regard when it comes to religion. Should we expect similar findings at the micro level in societies characterized by a strong family institution and a weak institution of religion? In conclusion, based on the indicators available to us, our results suggest that a strong family institution is necessary in order to have a strong institution of religion. Given the various limitations of our data, though, our findings must be viewed as tentative. Data availability and comparability across many societies is a major challenge for comparative research. While we believe that our measures of strength of family and strength of religion are valid, we recognize that they are less than perfect. For example, we discuss the potential problems with one of our behavioral measures of strength of religion, frequency of attendance at religious services, elsewhere in the paper. And having only one attitudinal question each to assess attitudinal strength of family and attitudinal strength of religion means that these measures are not as rigorous as we would like. It would be good if future research could address some of these measurement limitations and also explore other indicators of strength in the family and religion institutions. Behaviorally, for example, strength in religion could be measured by the strength of religious networks, the number of religious workers in a country, or the degree of participation in religious-oriented activities other than attendance at religious services. Turning to study design, the exploratory nature of this work means that it is not very sophisticated in a statistical sense. The major advantage of exploratory research, though, is the generation of insight on topics that have not been highly researched, and that is the major contribution of our work. Scholarly attention needs to focus now on explanatory work that can more rigorously test the insights derived from this study. And, future attempts to assess whether family has primacy over religion also would be strengthened by a longitudinal approach that would allow

does religion require family ?

347

for an examination of change over time in each institution. In addition, scholars may want to consider units of analysis other than countries. Work of this sort is needed to confirm the results of our exploratory study. We hope our research will stimulate further scholarship on this important subject. References Berger, Peter L. 1967. The Sacred Canopy. Garden City, New York: Doubleday and Company, Inc. ——. 2001. “Reflections on the Sociology of Religion Today.” Sociology of Religion 62:443–454. ——, and Brigitte Berger. 1983. The War over the Family: Capturing the Middle Ground. Garden City, NY: Anchor Press/Doubleday. Bollen, Kenneth A. 1980. “Issues in Comparative Measures of Political Democracy.” American Sociological Review 45:370–390. Bruce, Steve. 1992. “Pluralism and Religious Vitality.” Pp. 195–210 in Religion and Modernization: Sociologists and Historians Debate the Secularization Thesis, edited by Steve Bruce. Oxford: Clarendon Press. ——. 2000. “The Supply-Side Model of Religion: The Nordic and Baltic States. Journal for the Scientific Study of Religion 39:32–46. Call, Vaughn R.A., and Tim B. Heaton. 1997. “Religious Influence on Marital Stability.” Journal for the Scientific Study of Religion 36:382–392. Campbell, Robert A., and James E. Curtis. 1994. “Religious Involvement Across Societies: Analyses for Alternative Measures in National Surveys.” Journal for the Scientific Study of Religion 33:215–229. Caplovitz, David, and Fred Sherrow. 1977. The Religious Drop-Outs: Apostasy Among College Graduates. Beverly Hills, CA: Sage. Central Intelligence Agency. 1992. The World Factbook. Washington, DC: Central Intelligence Agency. Chang, Jui-Shan. 1996. “What do Education and Work Mean? Education, Nonfamilial Work/Living Experiences and Premarital Sex for Women in Taiwan.” Journal of Comparative Family Studies 27:13–40. Chaves, Mark, and Philip S. Gorski. 2001. “Religious Pluralism and Religious Participation.” Annual Review of Sociology 27:261–81. Christiano, Kevin J. 2000. “Religion and the Family in Modern American Culture.” Pp. 43–78 in Family, Religion, and Social Change in Diverse Societies, edited by Sharon K. Houseknecht, and Jerry G. Pankhurst. New York: Oxford University Press. Cox, Harvey. 1965. The Secular City. New York: The Macmillan Company. Cseh-Szombathy, Laszlo. 1990. “Modeling the Interrelation between Macro-Society and the Family.” International Social Science Journal 42:441–9. D’Antonio, William V. 1983. “Family Life, Religion, and Societal Values and Structures.” Pp. 81–111 in Families and Religions, edited by William V. D’Antonio and Joan Aldous. Beverly Hills, California: Sage Publications. ——. 1985. “The American Catholic Family: Signs of Cohesion and Polarization.” Journal of Marriage and the Family 47:395–405. Danziger, James N. 1991. Understanding the Political World: An Introduction to Political Science. White Planes, New York: Longman Publishing Group. de Guibert-Lantoine, Catherine, and Alain Monnier. 1997. “La Conjoncture Demographique: L’Europe et Les Pays Developpes d’Outre-Mer.” Population 5:1187–1216.

348

kristen r. heimdal and sharon k. houseknecht

Demerath III, N.J., and Phillip E. Hammond. 1969. Religion in Social Context. New York: Random House. Finke, Roger, and Laurence R. Iannaccone. 1993. “Supply-Side Explanations for Religious Change.” Annals of the American Academy of Political and Social Science 527:27–39. Fontelas Rosado Nunes, and Maria José. 2000. “Women, Family, and Catholicism in Brazil: The Issue of Power.” Pp. 347–363 in Family, Religion, and Social Change in Diverse Societies, edited by Sharon K. Houseknecht, and Jerry G. Pankhurst. New York: Oxford University Press. Fortuny Loret de Mola, Patricia. 2000. “Converted Women Redefining Their Family Roles in Mexico.” Pp. 363–386 in Family, Religion, and Social Change in Diverse Societies, edited by Sharon K. Houseknecht and Jerry G. Pankhurst. New York: Oxford University Press. Frandkin, Hillel. 2000. “Does Democracy Need Religion?” Journal of Democracy 11:87–94. Froese, Paul, and Steven Pfaff. 2005. “Explaining a Religious Anomaly: A Historical Analysis of Secularization in Eastern Germany.” Journal for the Scientific Study of Religion 44:397–422. Fukuyama, Francis. 1995. “Social Capital and the Global Economy.” Foreign Affairs 74:89–103. Gautier, Mary L. 1997. “Church Attendance and Religious Belief in Postcommunist Societies.” Journal for the Scientific Study of Religion 36:289–296. Glock, Charles Y. 1973. Religion in Sociological Perspective: Essays in the Empirical Study of Religion. Belmont, CA: Wadsworth. Gundelach, Peter. 1994. “National Value Differences: Modernization or Institutionalization?” International Journal of Comparative Sociology 35:37–58. Hargrove, Barbara. 1983. “The Church, the Family, and the Modernization Process.” Pp. 21–48 in Families and Religions: Conflict and Change in Modern Society, edited by W. D’Antonio and J. Aldous. Beverly Hills, CA: Sage. Himmelfarb, Gertrude. 1998. “Democratic Remedies for Democratic Disorders.” Public Interest 131:3–24. Houseknecht, Sharon K., and Jaya Sastry. 1996. “Family ‘Decline’ and Child Well-Being: A Comparative Assessment.” Journal of Marriage and the Family 58:726–739. Iannaconne, Laurence R. 1991. “The Consequences of Religious Market Structure: Adam Smith and the Economics of Religion.” Rationality and Society 3:156–177. ——. 1994. “Why Strict Churches are Strong.” American Journal of Sociology 99:1180– 1211. Inglehart, Ronald, et al. 2000. World Values Surveys and European Values Surveys, 1981–1984, 1990 –1993, and 1995–1997 [Computer File]. ICPSR Version. Ann Arbor, Michigan: Inter-university Consortium for Political and Social Research (distributor). IWRAW Publications. 1995. IWRAW Country Reports: Chile. Retrieved September 21, 2000 from the World Wide Web: http://www.igc.org/iwraw/publications/countries/chile.html. Jenkins, Kip W. 1991. “Religion and Families.” Pp. 239–288 in Family Research: A Sixty-Year Review, 1930 –1990, Volume 1, edited by Stephen J. Bahr. Toronto, Canada: Lexington Books. Kaariainen, Kimmo. 1989. Discussion on Scientific Atheism as a Soviet Science. Helsinki: Suomalainen Tiedeakatemia. Kelly, Jonathan, and Nan Dirk De Graaf. 1997. “National Context, Parental Socialization, and Religious Belief: Results From 15 Nations.” American Sociological Review 62:639–659. Kreeft, Peter, and Ronald K. Tacelli. 1994. Handbook of Christian Apologetics. Downers Grove, IL: InterVarsity Press. Kurian, George. 1997. Illustrated Book of World Rankings. Armonk, NY: Sharpe Reference.

does religion require family ?

349

Larson, Lyle E., and J. Walter Goltz. 1989. “Religious Participation and Marital Commitment.” Review of Religious Research 30:387–400. Lechner, Frank J. 1991. “The Case against Secularization: A Rebuttal.” Social Forces 69:1103–1119. ——. 1996. “Secularization in the Netherlands?” Journal for the Scientific Study of Religion 38:252–264. Lenski, Gerhard. 1961. The Religious Factor. Garden City, New York: Doubleday and Company, Inc. MacIver, Robert M. 1970. On Community, Society, and Power. Chicago: University of Chicago Press. Mahoney, Annette, Kenneth I. Pargament, Nalini Tarakeshwar, and Aaron B. Swank. 2001. “Religion in the Home in the 1980s and 1990s: A Meta-Analytic Review and Conceptual Analysis of Links between Religion, Marriage, and Parenting.” Journal of Family Psychology 15:559–596. Malinowski, Bronislaw. [1930] 1974. “Parenthood, the Basis of Social Structure.” Pp. 51–63 in Family, Its Structures and Functions, edited by Ruth Laub Coser. New York: St. Martin’s Press. Mason, David S., Daniel N. Nelson, and Bohdan M. Szlarski. 1991. “Apathy and the Birth of Democracy: The Polish Struggle.” Eastern European Politics and Societies 5:205–233. McGuire, Meredith B. 1981. Religion: The Social Context. Belmont, California: Wadsworth Publishing Company. Montgomery, James D. 2003. “A Formulization of and Test of the Religious Economies Model.” American Sociological Review 68:782–809. Moore, Wilbert E. 1963. Social Change. Englewood Cliffs, New Jersey: Prentice Hall. Mueller, Charles W., and Weldon T. Johnson. 1975. “Socioeconomic Status and Religious Participation.” American Sociological Review 40:785–800. Nelson, Lynn D. 1988. “Religion and Foreign Aid Provision: A Comparative Analysis of Advanced Market Nations.” Sociological Analysis 49:49–63. Nimkoff, Meyer F. 1965. Comparative Family Systems. Boston, MA: Houghton Mifflin. O’Donnell, Timothy S., Amanda P. Lambert, Maureen P. Garelick, and Beth Browning. 1991. World Quality of Life Indicators, Second Edition. Santa Barbara, CA: ABL-CLIO. Ogburn, William F., and Meyer F. Nimkoff. 1955. Technology and the Changing Family. Boston, MA: Houghton Mifflin Company. Pankhurst, Jerry G., and Sharon K. Houseknecht. 1999. “Religion and Motherhood Orientation in Diverse Societies: The Influence of Religious Market, Religious Homogeneity and Religious Culture.” Paper Presented at the Annual Meeting of the Association for the Sociology of Religion, Chicago, Illinois. ——. 2000. “Introduction: The Religion-Family Linkage and Social Change—A Neglected Area of Study.” Pp. 1–40 in Family, Religion, and Social Change in Diverse Societies, edited by Sharon K. Houseknecht and Jerry G. Pankhurst. New York: Oxford University Press. Paxton, Pamela. 2000. Personal communication. April 21. Phillips, Rick. 2004. “Can Rising Rates of Church Participation be a Consequence of Secularization?” Sociology of Religion 65:139–153. Popenoe, David. 1988. Disturbing the Nest. New York: Aldine De Gruyter. Powell, David E. 1994. “The Religious Renaissance in the Soviet Union and its Successor States.” Chapter 12 in The Social Legacy of Communism, edited by James R. Millar and Sharon L. Wolchik. Washington D.C.: Woodrow Wilson Center Press and the Press Syndicate of the University of Cambridge. Reddy, Marlita A. (ed.). 1994. Statistical Abstract of the World. New York: Gale Research, Inc. Reichley, James A. 1986. “Democracy and Religion.” PS 19:801–806. Rosenberg, Tina. 1995. The Haunted Land: Facing Europe’s Ghosts After Communism. New York: Random House.

350

kristen r. heimdal and sharon k. houseknecht

Ruiter, Stijn, and Nan Dirk De Graaf. 2006. “National Context, Religiosity, and Volunteering: Results from 53 Countries.” American Sociological Review 71:191–210. Sandi, Ana Marie. 1992. “Restoring Civil Society in Central and Eastern Europe.” Futures 24:110–117. Scheepers, Peer, Manfred Te Grotenhuis, and Frans Van Der Slik. 2002. “Education, Religiosity and Moral Attitudes: Explaining Cross-National Effect Differences.” Sociology of Religion 63:157–176. Sigelman, Lee. 1977. “Multi-nation Surveys of Religious Beliefs.” Journal for the Scientific Study of Religion 16:289–294. Smith, Donald Eugene. 1971. Religion, Politics, and Social Change in the Third World. New York: The Free Press. Snarey, John R., and David C. Dollahite. 2001. “Varieties of Religion-Family Linkages.” Journal of Family Psychology 15:646–651. Stark, Rodney. 1992. “Do Catholic Societies Really Exist?” Rationality and Society 4:261–271. ——, and James C. McCann. 1993. “Market Forces and Catholic Commitment: Exploring a New Paradigm.” Journal for the Scientific Study of Religion 32:111–124. ——, and Laurence R. Iannaccone. 1994. “A Supply-Side Reinterpretation of the ‘Secularization’ of Europe.” Journal for the Scientific Study of Religion 33:230–252. Swatos Jr., William H. 1984. “The Relevance of Religion: Iceland and Secularization Theory.” Journal for the Scientific Study of Religion 23:32–43. Thomas, Darwin L., and Marie Cornwall. 1990. “Family and Religion in the 1980s: Discovery and Development.” Journal of Marriage and the Family 52:983–992. Thornton, Arland. 1985. “Reciprocal Influences of Family and Religion in a Changing World.” Journal of Marriage and the Family 47:381–394. ——, William G. Axinn, and Daniel H. Hill. 1992. “Reciprocal Effects of Marriage, Cohabitation, and Marriage.” American Journal of Sociology 98:628–651. United Nations. 1990. United Nations Human Development Report. New York: United Nations. ——. 1993a. Demographic Yearbook. New York: United Nations. ——. 1993b. United Nations Human Development Report. New York: United Nations. Vardomatskii, Andrei, and Jerry G. Pankhurst. 2000. “Belarus on the Cusp of Change: The Relationship between Religion and Family in a Newly Open Religious Market.” Pp. 149–174 in Family, Religion, and Social Change in Diverse Societies, edited by Sharon K. Houseknecht and Jerry G. Pankhurst. New York: Oxford University Press. Warner, R. Stephen. 1993. “Work in Progress toward a New Paradigm for the Sociological Study of Religion in the United States.” American Journal of Sociology 98:1044–1093. Westerhoff III, John H. 1973. “A Changing Focus: Toward an Understanding of Religious Socialization.” Andover Newton Quarterly 14:118–129. Whitehead, Barbara Dafoe. 1993. “Dan Quayle Was Right.” Atlantic 271:47–84. Wilcox, W. Bradford, Mark Chaves, and David Franz. 2004. “Focused on the Family? Religious Traditions, Family Discourse, and Pastoral Practice.” Journal for the Scientific Study of Religion 43:491–504. Williams, Robin M. 1970. American Society. Second edition. New York: Alfred A. Knopf. Wilson, Bryan R. 1992. “Reflections on a Many Sided Controversy.” Pp. 195–210 in Religion and Modernization: Sociologists and Historians Debate the Secularization Thesis, edited by Steve Bruce. Oxford, England: Clarendon Press. Wilson, James Q. 1993. “The Family-Values Debate.” Commentary 95:24–31. Wilson, John, and Sharon Sandomirsky. 1991. “Religious Affiliation and the Family.” Sociological Focus 6:289–309. World Bank. 1999. World Data 1995: World Bank Indicators on CD-ROM. Washington, D.C.: International Bank for Reconstruction and Development/the World Bank. Wuthnow, Robert. 1977. “A Longitudinal, Cross-National Indicator of Societal Religious Commitment.” Journal for the Scientific Study of Religion 16:87–99.

PART FIVE

SOCIAL PROCESSES

GLOBALIZATION AND INCOME INEQUALITY IN THE DEVELOPING WORLD Margit Bussmann, Indra de Soysa, and John R. Oneal* Recent research indicates that globalization promotes the growth of average incomes in developing countries (Borensztein et al. 1998; de Soysa and Oneal 1999; UNCTAD 1999; Ram and Zhang 2002; Dollar and Kraay 2002; Bhalla 2002; Ward and Gleditsch 2004), but the standard of living of the poor in these societies could decline if integration into the global economy adversely affects the distribution of income. Indeed, a number of academics have concluded that the fears of the anti-globalization movement are justified: multinational corporations increase income inequality in developing countries, further marginalizing the poorest of the poor (Korzeniewicz and Moran 1997; Alderson and Nielsen 1999; Milanovic 1999; Mazur 2000; Kentor 2001; Reuveny and Li 2003; Wade 2004). We reassess these claims by examining the effects of economic openness on income inequality within countries. While much of the current debate is largely focused on openness to trade (Dollar and Kraay 2002; Wade 2004; Wolf 2000), we pay special attention to the influence of foreign direct investment on the distribution of income in developing countries, using data recently published by UNCTAD (2000) and the World Bank (2000). Our results are easily summarized. We find no evidence to suggest that a large stock of foreign investment (or economically important trade) increases inequality in either developing or developed countries. Tests focused on the poorest 20% of people in each country confirm this result. Economic openness does not significantly affect the distribution of income in the periphery nor decrease the lowest quintile’s share of national income. If foreign investment and trade promote economic growth and do not shift income to richer segments of society in the process, it must be good for the poor in developing countries.

* Authors’ note: We gratefully acknowledge the helpful comments and sage advice of Erich Weede. We alone remain responsible for errors, however. John Oneal thanks the Center for Development Research (ZEF), University of Bonn, Germany, for a visiting fellowship that assisted in the completion of this project.

354

margit bussmann, indra de soysa, and john r. oneal

In the following section, we review previous research and various theoretical arguments on the effect of globalization on income inequality, and we justify our focus on the stock of foreign investment as the principal measure of a country’s integration into the global economy and our method of calculating this variable which distinguishes our study from others. Next we identify other influences that might affect the distribution of income within countries. This allows us to estimate the effect of foreign investment on inequality in a number of specifications in section three, ensuring that the omission of a key variable is not biasing our analyses. We present our conclusions in the final section. The Effect of Globalization on Income Inequality The process of globalization has long been of interest to social scientists in several disciplines. Early theories of neo-colonialism, advanced by sociologists especially, argued that multinational corporations retard economic growth in developing countries (Baran 1956; Chase-Dunn 1975; Frank 1969). Dependency on foreign investment was said to create an opportunity for exploitation that adversely affects economically- and politically-weak host countries. Later, dependency theorists recognized that some developing countries experience rapid development even when integrated into the global economy; but economic growth in these cases was said to be distorted and inegalitarian, adding to misery.1 Portes (1976; also Cardoso and Faletto 1979:75), for example, claimed that “sustained economic growth has been accompanied by rising social inequalities.” Multinational corporations were thought to increase income inequality in part by producing enclave economies, where relatively well-paid workers were surrounded by a larger number of marginalized poor. More importantly, foreign capitalists were said to prevail upon comprador elites to adopt policies that favored capital

As noted earlier, recent research indicates that globalization promotes growth in developing countries. Because UNCTAD made significant revisions in its data on foreign investment from 1995 to 2000, we re-estimated the analyses reported in de Soysa and Oneal (1999), which were based on the earlier data. The results confirm our earlier findings: larger foreign investment flows increase the growth rate of GDP per capita, foreign investment is more productive dollar-for-dollar than capital from domestic sources, and, contrary to dependency theory, a large stock of FDI does not adversely affect growth. 1

globalization & income inequality in the developing world 355 at the expense of labor.2 Multinational corporations were accused of using their economic power to create oligopolistic markets and subvert efforts at egalitarian development. Early empirical tests of dependency theory by Galtung (1971) and Rubinson (1976) found that the existence of a weak state and a concentration of exports in a few commodities or with a few developed countries were associated with increased inequality in the periphery, but Chan (1989) was unable to corroborate these findings. Later, the role of foreign direct investment (FDI) was emphasized by dependency theorists because “the most direct economic penetration by core nations of peripheral areas is through private investment by transnational corporations which directly own and control the process of production” (Chase-Dunn 1975:721). Statistical analyses provided substantial support for this view (Bornschier et al. 1978; Kohli et al. 1984; Bornschier and Chase-Dunn 1985). Indeed, the economic importance of foreign investment, indicated by the ratio of foreign investment stock to gross domestic product (GDP), was the best predictor of income inequality in developing countries in Chan’s (1989) comprehensive tests of prevailing theories. For dependency theorists, it was the accumulated effects of the multinationals’ penetration over time, not trade or volatile short-term flows of capital that mattered (Bornschier and Chase-Dunn 1985). Thus, the relationship between foreign investment and inequality seemed “one of the most robust quantitative, aggregate findings available” in sociology and political science (Evans and Timberlake 1980:532).3 Recent analyses reinforce the view that economic dependency on foreign direct investment increases income inequality (Alderson and Nielsen 1999; Kentor 2001; Reuveny and Li 2003).4

2 Bornschier et al. (1978:665) argued that “[t]he effect of dependence on income inequality is most likely due to its effects on the class structure of the country and the translation of this class structure into political power.” This essentially involved the suppression of organized labor and its political allies. 3 The determinants of foreign investment, as well as its consequences, have received the attention of those interested in the process of globalization (Oneal and Oneal 1988; Oneal 1994; Jensen 2003; Li and Resnick 2003). 4 Reuveny and Li (2003) find that FDI flows as a fraction of GDP averaged over a decade correlate positively with income inequality. The flow of foreign investment, however, may be volatile and is a weak measure of multinational corporations’ influence on a host economy. Average flows do not distinguish those countries continuously open to global forces; a country receiving FDI equal to 10% of GDP in each of 10 years would have the same score as a country receiving the full amount in the first

356

margit bussmann, indra de soysa, and john r. oneal

Neo-classical economists have largely ignored the effect of foreign investment on the distribution of income within countries. Solow (1956) suggested that foreign capital reduces the return on investment in a host country and raises the productivity of labor and hence the real wage. Since the ownership of capital is more concentrated than income from labor, this should equalize incomes and raise their average, at least in the absence of substantial restrictions on trade (Cooper 2001). Foreign investment (and trade) may also have beneficial effects by increasing competition in domestic markets. Profit margins are reduced and the prices of commodities lowered, redistributing wealth from producers to consumers (Mehlkop 2002). Moreover, FDI can benefit a wider class of entrepreneurial talent in developing societies by increasing the sources of capital and encouraging the development of financial markets. Restrictions on economic openness favor rent-seeking and create opportunities for corruption (Weede 2000). All these influences suggest that higher levels of foreign investment should reduce income inequality. The effects of foreign direct investment on income inequality may depend, however, on the policies adopted by host countries. Countries with high levels of inequality that do not change policies that advantage established elites could see a perpetuation or expansion of inequality because of what Birdsall (2002) calls “dynastic traps”. On the other hand, Dollar and Kraay (2002) report that in developing countries there is almost a one-to-one relationship between the growth of average incomes and increase in the incomes of the poorest quintile. If foreign investment, like capital from domestic sources, simply promotes growth, then the poor should benefit proportionately (Sylwester 2005). Thus, the effect of foreign investment on the distribution of income in developing countries seems uncertain. FDI may increase inequality by reinforcing the power of privileged groups and creating enclaves of well-paid employees of the multinational corporations surrounded by marginalized poor. Alternatively, increased competition in domestic

year and 0% for the remaining nine years. Alderson and Nielsen (1999) use the stock of FDI, the measure that we also advocate. However, they rely upon Ballmer-Cao and Scheidegger (1979) and UNCTAD (1995) for their data on foreign direct investment; although Ballmer-Cao and Scheidegger warned that their data should be “used with caution” (p. 122), and UNCTAD revised its data significantly after 1995. In addition, our use of FDI flows to calculate stocks in various years is preferable to Alderson and Nielsen’s interpolating between estimates of stocks, especially when these estimates are from different sources.

globalization & income inequality in the developing world 357 markets as a result of globalization could benefit consumers and less skilled workers in developing countries and reduce rent-seeking, thereby contributing to greater equality. Finally, these positive and negative effects of globalization might cancel out, or the net effect depends upon the policies of host countries. As Caves (1996:115) has suggested, “the distributional consequence of foreign investment in the long run remains a strictly unsettled issue.” Economists have more thoroughly considered the effect of trade on the distribution of incomes within countries. According to neo-classical theory, free trade should decrease inequality in developing countries because they have a comparative advantage in unskilled labor and trade increases the income of the factors of production used intensively by exporters. At the same time, inequality should increase in core countries where skilled workers are the primary beneficiaries. These straightforward conclusions may not hold, however, if more than two commodities and two factors of production are involved. Extensions of the Stolper-Samuelson theorem lead to such disparate results that the effect of trade on inequality is unpredictable, Cooper (2001) concludes. Recent empirical results are consistent with this view. Borsu and Glejser (1992) and Reuveny and Li (2003) report that the trade-to-GDP ratio is associated with more equitable income distributions in both core and peripheral countries; Edwards (1998), Higgins and Williamson (1999), and Mehlkop (2002) find no relation between trade and national income inequality; and Barro (2000) reports that economically important trade adversely affects inequality in developing nations. Other Sources of Income Inequality The most prominent theory of income inequality was advanced by Kuznets (1955), who suggested that there is an inverted U-shaped curve relating income inequality within countries to their average incomes. Kuznets noted that, at the beginning of the Industrial Revolution, most people farmed, an occupation characterized by uniformly low incomes. As people in Britain, Germany, and the United States moved to the cities, inequality initially increased because migrants earned higher wages in industry. Over time, wages in agriculture also rose as the supply of laborers in that sector declined. Thus, as industrialization progressed, incomes again became relatively equal but at a higher level. Collective bargaining and organized political action, facilitated by the character

358

margit bussmann, indra de soysa, and john r. oneal

of urban life, reinforced this demographic process. Kuznets suggested that economic development might similarly affect the distribution of income in peripheral countries in the contemporary period. A number of studies have found support for Kuznets’ theory (Weede and Tiefenbach 1981; Muller 1988; Higgins and Williamson 1999; Barro 2000; Bhalla 2002; Reuveny and Li 2003); but others have been contradictory or equivocal, especially in the analysis of time-series (Chan 1989; Anand and Kanbur 1993; Ravallion and Chen 1997; Deininger and Squire 1998). Of course, the effect of industrialization, like that of globalization, depends in part on the policies adopted by governments. Socialist states make equality a primary objective of policy. Democracies, too, are often thought to favor more equal incomes. As long as the median voter earns less than the national average, democratic leaders may be responsive to a demand for greater equality (Alesina and Rodrik 1994; Persson and Tabellini 1994; Weede 1997). The extent to which democracies might adopt policies to reduce inequality depends on the objective being pursued, however: equality before the law, equal opportunity, or equality of outcomes. Even if democratic governments do try to reduce inequality, the process may operate gradually and take time to secure its result, which may account for the lack of support for the simple hypothesis that democracies have more equal income distributions than do non-democratic countries (Kohli et al. 1984; Bollen and Jackman 1985; Barro 2000; Dollar and Kraay 2000; Reuveny and Li 2003). Muller (1988) reports that older democracies, but not newer ones, are more egalitarian than are non-democratic societies, though Chan (1989) and Weede (1989, 1990) cast doubt on this conclusion, too. Structural aspects of national economies—what Aldersen and Nielsen (1999) call an internal-developmental model—have been emphasized in recent research on income inequality. If Kuznets (1955) is right, the initial increase in inequality that results from industrialization should be greater if there is a large difference in the productivity of agricultural and industrial workers. Indeed, countries with a dualistic economy—modern industry and less technologically advanced farming—do have greater inequality (Nielsen 1994; Nielsen and Alderson 1995; Bourguignon and Morrison 1998; Alderson and Nielsen 1999). Incomes are also expected to be more unequal when agriculture’s share of national economic production is small. Inequality is relatively low within the agricultural sector because the skills of farmers are more uniform compared to the diversity in industry; hence, the larger the industrial sector, the greater is income inequality within a nation.

globalization & income inequality in the developing world 359 The distribution of a country’s population by age may also be important (Easterlin 1980; Higgins and Williamson 1999). A high proportion of children relative to adults should increase inequality because the birth rate is greatest among the poor and the poor consume a larger portion of their income than the rich (Bollen and Jackman 1985; Muller 1988; Nielsen and Alderson 1995; Alderson and Nielsen 1999). In addition, an abundance of youth increases competition for employment among unskilled workers and lowers their wages relative to older laborers. A young population also indicates a high population growth rate, which is associated with limited economic and political participation by women—an effect that should be most adverse for the poor. Finally, income inequality is often thought to be affected by the accessibility of education. Countries with broad-based public education are expected to have more equal distributions of incomes (Nielsen and Alderson 1995; Bourguignon and Morrisson 1998; Alderson and Nielsen 1999), though Higgins and Williamson (1999) find little support for this hypothesis. Earlier, Weede and Tiefenbach (1981) and Chan (1989) reported that widespread participation in the military is associated with greater equality in peripheral countries. Military training was thought to be an important source of education in less developed countries, but the benefit of military participation has not proven robust (Weede 1993). Results In assessing the consequences of globalization for the national distribution of income, we focus on the effect of foreign direct investment using regression analyses of pooled time-series for an unbalanced sample of 72 countries with about five observations per country on average over the period 1970–90. This allows us to make cross-sectional comparisons among the various countries and also to trace developments over time. The economic importance of foreign investment, indicated by the FDI stock-to-GDP ratio, best captures the influence of multinational corporations in a host economy because it indicates the strength of the historically accumulated weight of MNCs’ political and economic power. We also consider, however, the effects on inequality of trade-based measures of economic openness: the trade-to-GDP ratio and Sachs and Warner’s (1995) indicator of free-trade policies. Our tests directly measure the influence of multinational corporations—the principal agents of globalization—on inequality in host countries. Several recent studies

360

margit bussmann, indra de soysa, and john r. oneal

(Korzeniewicz and Moran 1997; Milanovic 1999; Sala-i-Martin 2002; Bhalla 2002; Bourguignon and Morrison 2002; Firebaugh and Goesling 2004) have inferred the consequences of globalization by comparing a period when the international economy was less integrated to a later time when globalization had advanced. Thus, we test the consequences of globalization for individual countries and not its aggregated effects in different periods of time. In our tests, we incorporate the Kuznets (1955) curve, the character of political institutions, and various aspects of the economy and society that have been emphasized in previous research. We use the most widely accepted data on income inequality (Deininger and Squire 1996), which are available for a large sample of countries but for particular years only. Our data on foreign direct investment is drawn from UNCTAD (2000) and the World Bank (2000). We calculated the stock of FDI in each year, 1970–1990, by adding the flow of foreign direct investments (in constant dollars) to or subtracting it from the stock of foreign capital in 1980—the only year for which estimates of FDI stocks are available—taking into account depreciation. We estimate the stock of foreign investment only in the ten years before and after the baseline year (1980) to limit the risk of introducing substantial measurement error into this key variable.5 By knowing the economic importance of foreign investment for each year, we can ensure that our key indicator of globalization is measured contemporaneously with the year in which income inequality was assessed by national survey.6 The definitions of our variables and the sources of our data are discussed in the appendix.7

5 Measurement error is introduced in using flows to estimate accumulated stocks because we do not have precise information regarding depreciation or inflation. We calculated depreciation using the accelerated method assuming that all capital had a half-life of ten years, and we assumed that the GDP deflator is valid for converting capital flows to constant dollars. 6 Thus, we follow the methods of Ravallion and Chen (1997) and Dollar and Kraay (2000). Other studies (Easterly 1999, Higgins and Williamson 1999, Reuveny and Li 2003) average data on inequality over a decade. Thus, observations for a country in, say, 1973, 1976, and 1977 would be averaged and regressed on independent variables drawn from the 1970s. This introduces uncertainty into the temporal sequence because a measure of inequality based on values from early in a decade is regressed on explanatory variables measured at later points in time. 7 Our data and the programs used to generate the results reported in the tables are posted at www.bama.ua.edu\~joneal\inequality.

globalization & income inequality in the developing world 361 We first estimate the effect of foreign direct investment on the Gini index, an aggregate measure of income inequality, for all 72 developed and less developed countries in various years, 1970–90. We take care to confirm that these analyses capture the experience of developing countries in particular. To ensure that the marginalization of the poor would be detected, we also perform several tests using the share of income received by the poorest 20% in each society. In most cases, we use ordinary least squares regression analyses of pooled time-series and report robust standard errors that take into account the clustering of our data by country. This produces consistent standard errors even in the presence of serial correlation and heteroskedasticity (Wiggins 1999). We re-estimate key analyses using a fixed-effects model with separate indicators for each country to determine whether our analyses of pooled data capture the actual experience of countries through time.8 Our simplest test is presented in the first column of Table 1. We regress the Gini index of national income inequality on the ratio of foreign direct investment to GDP, the natural logarithm of real per capita GDP and its square, and indicators that identify the geographical region in which each country is located. Contrary to the fears of those who oppose globalization, a large stock of foreign investment is not significantly (p < .58) related to the distribution of income. The inverted U-shape of the curve identified by the estimated coefficients of log (GDP per capita) and its square is consistent with Kuznets’ theory. Income inequality is low among the poorest countries, rises until the average income reaches $3804 (2000 constant dollars), and then falls with greater development. As we shall see below, however, the Kuznets curve accounts only for variation across countries, not within the time series.9 Our results confirm that inequality is particularly great in Latin

We do not use the panel-corrected standard error estimator (Beck and Katz 1995) because the number of countries in our pooled data is much greater than the number of observations in the individual time series. 9 Kohli et al. (1984) noted that Kuznets’ argument is not inconsistent with the theory of dependent development: Foreign investment, by promoting growth in the periphery, might increase income inequality in the early stage of industrialization. To insure that the effect of foreign investment was not being obscured, we regressed inequality on the FDI-to-GDP ratio without the two measures of income, log (GDP per capita) and its square. The estimated coefficient was far from statistical significance either with (p < .63) or without (p < .94) the regional indicators. 8

362

margit bussmann, indra de soysa, and john r. oneal

Table 1. Estimated coefficients from the regression of the Gini index of income inequality on foreign direct investment, 1970–1990 Variables

(1)

FDI/GDP

1.60 (2.89) 49.10** (18.09) –3.13** (1.09)

Real per capita income (ln) Real per capita income squared Socialist state Years of democracy Agricultural share of GDP Relative labor productivity Population under 15 Secondary school Enrollment rate Africa dummy Latin America dummy Asia dummy Oceania dummy Constant R2 N

– –

(2) 1.04 (2.99) 38.63 (20.36) –2.52* (1.25) –8.47*** (2.16) –0.014 (0.028)

–

–

–

–

–

–

– – 15.39*** 15.35*** (3.51) (3.69) 9.24*** 8.68** (2.65) (2.85) –1.45 –1.89 (2.47) (2.64) 0.11 –0.17 (2.36) (2.27) –148.69* –103.59 (82.12) (74.05) .60 .62 383 377

(3)

(4)

1.02 (2.61) 16.94 (17.36) –1.41 (1.03) –10.90*** (2.07)

0.83 (2.77) 15.97 (17.82) –1.41 (1.06) –11.04*** (1.46)

– –0.30* (0.13) 0.75* (0.29) 0.027 (0.172) –0.026 (0.049) 9.33*** (2.51) 6.16 (3.14) –3.02 (2.19) 0.94 (2.31) 3.06 (74.92) .72 322

– –0.31* (0.12) 0.68*** (0.099) – – 9.74*** (2.61) 6.98* (2.90) –2.92 (2.40) 1.19 (2.22) 10.57 (75.13) .74 325

Note: Numbers in parentheses are robust standard errors. * p < .05 ** p < .01 *** p < .001 (two-tailed tests)

America and Africa (Anand and Kanbur 1993; Tsai 1995; Alderson and Nielsen 1999; Higgins and Williamson 1999; Barro 2000; Dollar and Kraay 2000). Next, we add two political variables that might influence the national distribution of income. The first is a simple indicator of a socialist economy; the second is a count of the number of years a country had been democratic. Again, there is no evidence that foreign investment by multinational corporations increases inequality (p < .73). Not surprisingly, however, socialist states have more equal incomes than other

globalization & income inequality in the developing world 363 countries; the Gini index is 8.5 percentage points lower if a state is socialist. The magnitude of this effect indicates the importance in estimating the effects of FDI on income inequality of controlling for socialist policies.10 The longevity of democracy does not influence income inequality—a result consistent with most previous research.11 The log of real income and its square, controlling for the two political influences and the regional indicators, are still jointly significant (p < .02), though the log of real income individually just misses (p < .07) the conventional level of statistical significance. In the third column of Table 1, we estimate the effect of foreign investment while controlling for four economic and demographic characteristics that have been emphasized in recent research. The first is a gauge of the relative labor productivity (RLP) of the non-agricultural sector. When RLP is large, there is a clear indication of a dual economy. We also control for the share of GDP represented by the agricultural sector, the percentage of children enrolled in secondary schools, and the percentage of the population under age 15. These analyses, too, indicate that foreign investment does not affect income inequality (p < .70). The distribution of incomes is influenced by the economic importance of agriculture and its relative productivity, as expected, but not by the percentage of the population under 15 or the secondary-school enrollment rate. Thus, we find mixed evidence for Alderson and Nielsen’s (1999) internal-developmental model. In the last column of Table 1, we drop the insignificant variables and reestimate the coefficients. There is little change in the results. This is our best, most parsimonious account of national income inequality.12

10 Reuveny and Li (2003) report that FDI flows are positively correlated with inequality, but this finding may be spurious because they do not control for a state’s having a socialist economy. Because the FDI-to-GDP ratio is low in socialist countries, the effects of foreign investment may be conflated with the existence of a market economy. Hungary had the lowest inequality among Reuveny and Li’s cases. The fact that the Hungarians preferred to open their economy to FDI—not continue to repel it—is telling. Apparently they preferred higher standards of living with some inequality to greater equality at a lower average income. It is important to note, too, that the standard of living in socialist countries is influenced not only by income but also by privileged access to scarce resources, which is not taken into account in our analyses. 11 We substituted Jaggers and Gurr’s (1995) measure of institutional democracy for the count of the years that a country had been democratic, but the results were unchanged. 12 We tested whether the insignificance of the FDI variable is due to collinearity with any of the control variables, but the estimated coefficient of the FDI-to-GDP ratio remained insignificant when we excluded in turn each of the other independent variables. We also estimated a specification with yearly dummy variables to see whether

364

margit bussmann, indra de soysa, and john r. oneal

In both columns 3 and 4, there is still support for the Kuznets curve. The coefficients of real income and its square are jointly significant (p < .01 in column 4), though individually they are not.13 Tests of Robustness In Table 2, we check the robustness of our results. First, we re-estimate the parsimonious model in the last column of Table 1 using a fixedeffects model to capture unique national characteristics that might influence the distribution of incomes.14 Only variables that change over time can have an effect in fixed-effects analyses, so the socialist variable and the regional indicators drop out. The results in column 1 confirm that foreign investment does not affect inequality. The estimated coefficient of the FDI-to-GDP ratio is not significant at conventional levels (p < .10, two-tailed test), even though robust standard errors are not available for this estimator. In any event, the substantive impact on inequality is slight. An increase of one standard deviation in the FDI-to-GDP ratio (.12) is associated with a rise of less than one percentage point in inequality (4.68 * .12 = .56). This is small compared to the standard deviation of the Gini index (11.15 percentage points).15

our results are influenced by the gaps in the inequality data. The significance levels of the FDI-to-GDP ratio and the other variables were unaffected, and the yearly indicators were jointly insignificant. 13 Kentor (2001) suggests that multinational corporations exert their influence over the long term, so he assesses the effect of foreign investment stocks using a 10-year lag. It is possible that a large foreign presence would have a lasting effect on inequality because the distribution of income is relatively stable for most countries over time; but there is no reason to believe that foreign capital would not also have a proximate influence. To capture the effect of foreign capital over time on inequality, we repeated the analysis in column 4 using, first, five lagged values of the FDI-to-GDP ratio (t-1 through t-6) and, then, ten lagged values (t-1 through t-11) instead of just the contemporaneous measure. None of the individual terms in either analysis was significant, and the lagged terms were jointly insignificant (p < .84 for five lags, p < .34 for ten lags). 14 Hausman’s specification test indicates that a fixed-effects model is superior to an analysis with random effects (p < .01). It also is superior on theoretical grounds because we do not have a random sample of cases (Hsiao 1986, 43). There is, however, little difference in the results produced by the two methods. For doubts about the general usefulness of fixed-effects analyses, see King (2001). 15 This result is not much different from a test using random-effects reported by Alderson and Nielsen (1999). The estimated coefficient of the logarithm of FDI/GDP is 3.42 in their model 10. Thus, a one standard-deviation increase (.43) in their measure of foreign investment raises inequality by 1.47 percentage points, less than 17% of the standard deviation of the Gini index in their sample. This does not suggest “an important role for foreign capital penetration in the generation of inequality” (p. 627).

globalization & income inequality in the developing world 365 Table 2. Estimated coefficients from the regression of the Gini index of income inequality on foreign direct investment and trade, 1970–1990 Variable FDI/GDP FDI* real per capita income Trade/GDP Sachs & Warner openness Real per capita income (ln) Real per capita income squared Socialist state Agricultural share of GDP Relative labor productivity Africa dummy

(1) Fixed effects

(2) OLS

4.68 (2.77)

34.01 (46.97) –3.99 (5.84)

– – –6.13 (15.35) 0.25 (0.85) –0.17 (0.14) 1.02* (0.41)

Latin America dummy Asia dummy Oceania dummy Constant R2 Overall R2 Within R2 Between R2 N

74.34 (70.60)

325

.30 .07 .33

(3) OLS

(4) OLS

–

–

– –1.38 (2.15)

–

14.83 (17.43) –1.33 (1.04) –10.94*** (1.43) –0.31** (0.12) 0.67*** (0.10) 9.77*** (2.65) 6.89* (2.90) –2.84 (2.48) 1.30 (2.20) 14.50 (73.42) .74

10.96 (16.21) –1.11 (0.96) –14.39*** (2.01) –0.30* (0.13) 0.74*** (0.12) 10.18*** (2.44) 7.30** (2.59) –2.86 (2.41) 1.44 (2.00) 30.90 (69.46) .78

– 0.85 (1.49) 20.07 (17.39) –1.61 (1.03) –12.13*** (1.76) –0.24 (0.13) 0.71*** (0.10) 9.84*** (2.67) 7.90** (2.73) –2.41 (2.42) 1.19 (1.91) –11.58 (74.64) .74

325

343

317

Note: Numbers in parentheses for models 2–4 are robust standard errors. * p < .05 ** p < .01 *** p < .001 (two-tailed tests)

366

margit bussmann, indra de soysa, and john r. oneal

There is no evidence in our fixed-effects analysis that income inequality rises and falls with the average real income, as Kuznets suggested. The estimated coefficients for income and its square are jointly insignificant (p < .60).16 Bourguignon and Morrisson’s (1998) measure of relative labor productivity is the only statistically significant variable, and the 68 country indicators account for 90% of the variance in the Gini index—clear indications of the limits of our understanding of the causes of income inequality within countries. In the second column of Table 2, we show that the effect of foreign investment is the same in developing as in developed countries. In this OLS estimation, we include an interactive term (the logarithm of real income per capita * the FDI-to-GDP ratio) in the parsimonious specification shown in the last column of Table 1. This allows us to determine whether the effect of foreign investment varies with average real income, as Bornschier and Chase-Dunn (1985) argued. The estimated coefficient of the interactive term is, however, far from statistical significance. Thus, the influence of multinational corporations on national income inequality is not conditional on the host country’s level of development. Foreign investment does not adversely affect the distribution of income in either developing or developed countries. This test provides important assurance that the results reported in our other analyses are not biased against finding evidence of an adverse effect of globalization on inequality in developing countries due to over-sampling of wealthy countries. In the last two columns of Table 2, we substitute two alternative measures of globalization for the FDI-to-GDP ratio in our best speciThe Kuznets curve implies that economic growth will affect rich and poor countries differently. Growth in poor countries is expected to increase inequality; it should reduce inequality in rich countries. Indeed, the poorer a country is, the more growth should increase inequality. Similarly, the further a rich country is from the inflection point of the Kuznets curve, the more inequality should be reduced by economic growth. To test this, we created two interactive terms using countries’ economic growth rates over the previous five years and the deviation of a country’s income from the inflection point identified in column 1, Table 1. For countries whose income was less than the peak of the Kuznets curve, PoorGrowth = Growth * | AverageIncome—InflectionPoint | ; PoorGrowth = 0 for countries whose income was greater than the inflection point. The second measure (RichGrowth) was constructed in analogous fashion: growth was multiplied by the difference between average income and the inflection point of the Kuznets curve for rich countries; RichGrowth equalled zero otherwise. We then regressed the Gini index on these two interactive terms and the regional indicators. Neither measure of growth was near statistical significance, and both of the signs were contrary to expectations. There is no evidence, therefore, that economic growth is biased against low-income groups in poor countries. 16

globalization & income inequality in the developing world 367 fication. In column 3, we use the trade-to-GDP ratio and, in column 4, Sachs and Warner’s (1995) measure of economic openness; in both cases, we analyze the pooled time-series using OLS. The trade-to-GDP ratio indicates the economic importance of a nation’s exports and imports. It is the most common measure of economic openness; but because it is influenced by fundamental characteristics of a country (viz., its size and geographical location) as well as its economic policies, Sachs and Warner have categorized countries as open or closed based on the level of tariffs, the prevalence of quotas, and other indicators of protectionism. The results reported in the last two columns of Table 2 show that income inequality is unaffected by either of these alternative measures of globalization.17 As with the FDI-to-GDP ratio, we assessed whether the effect of trade on inequality was conditional on a country’s level of development. We added an interactive term with the logarithm of real income to the regressions with the trade-to-GDP ratio and Sachs and Warner’s (1995) index of economic openness. We do not report the results in the table, but in both cases the interactive term was insignificant. Thus, contrary to Barro (2000), we find no evidence that economically important trade or a governmental policy of free trade adversely affects income inequality in developing countries. Nor is there support for the prediction of the Stolper-Samuelson theorem that trade increases inequality in developed countries and decreases it in developing ones. Testing for Marginalization of the Poor Those interested in the effects of globalization are often specifically concerned about the consequences of international capitalism for the well-being of the poorest of the poor. There are theoretical grounds for this humanitarian interest. According to the theory of dependent development, foreign investment may foster growth but development will be distorted. In this view, multinational corporations create enclaves linked to the international economy, where workers earn relatively good wages, while the poor in remote regions are marginalized. Because the Gini index is a summary measure of inequality, higher incomes for a growing middle class could cause the index to decline even though

17 In a specification where we included the FDI-to-GDP ratio together with the tradeto-GDP ratio, both variables remained insignificant. They are correlated at r = 0.34.

368

margit bussmann, indra de soysa, and john r. oneal

peasants are earning lower wages.18 For this reason, we next analyze the effect of foreign investment on the share of income of the poorest 20% of society, as Deininger and Squire (1996) recommend. In column 1 of Table 3, we again use our best, parsimonious specification from Table 1, but substitute for the Gini index the lowest quintile’s share of income as the dependent variable. Foreign investment has no effect on the poor’s share of national income (p < .45). Nor is there evidence of a Kuznets curve. The only significant influences are the indicator of a socialist economy and the relative labor productivity of the non-agricultural sector. The poorest quintile in a socialist country receives 1.13 percentage points more income than in a capitalist country ceteris paribus—more than half the standard deviation of the quintile’s share of income (2.01 percentage points). An increase of one standard deviation in RLP has only a modest effect, lowering the poor’s share of income by 0.29 percentage points. In the second column of Table 3, we report the results of a fixedeffects test of the effect of foreign investment on the income share of the poorest quintile. The estimated coefficient of the FDI-to-GDP ratio is negative and significant at the .04 level. The effect of a one-standard deviation increase, however, is small: –0.16 percentage points, only 7% of the standard deviation of the lowest quintile’s share of income. This again indicates that statistical significances in our fixed-effects analyses are overstated because of the unavailability of robust standard errors for this estimator. The variance explained in this test confirms the limits of our understanding. The overall R-square is .07; within the time series, it is just .04. In the third column of Table 3, we again ensure that the effects of foreign investment do not differ for the developing and the developed countries. As before, the estimated coefficient of the interactive term (the logarithm of real income per capita * the FDI-to-GDP ratio) is insignificant. Foreign investment does not further marginalize the poor in developing countries.

18 Bornschier and Chase-Dunn (1985:123) suggest that only 20 percent of the population in a typical peripheral country are integrated into the world economy, with the rest being marginalized. If only a small proportion of countries’ populations benefit from globalization, analyses of the Gini index will capture any adverse effect of economic openness.

globalization & income inequality in the developing world 369 Table 3. Estimated coefficients from the regression of the poorest 20%’s share of income on foreign direct investment, 1970–1990 Variable FDI/GDP FDI* real per capita income Real per capita income (ln) Real per capita income squared Socialist state Agricultural share of GDP Relative labor productivity Africa dummy Latin America dummy Asia dummy Oceania dummy Constant R2 Overall R2 Within R2 Between R2 N

(1) OLS

(2) Fixed effects

–0.61 (0.79)

–1.30* (0.64)

– –2.49 (5.77) 0.13 (0.35) 1.13*** (0.33) –0.70 (3.36) –0.077* (0.038) –2.15** (0.78) –3.32*** (0.86) –0.80 (0.70) –1.20 (0.57) 19.26 (23.52) .54

– –3.10 (3.74) 0.16 (0.21)

293

–2.67 (3.32) –0.14 (0.096)

21.77 (17.02)

293

.07 .04 .03

(3) OLS –24.96 (17.08) 2.96 (2.13) –1.50 (5.89) 0.058 (0.37) 1.10*** (0.32) –0.57 (3.21) –0.075* (0.036) –2.10** (0.76) –3.23*** (0.81) –0.79 (0.69) –1.29 (0.55) 15.68 (23.67) .56

293

Note: Numbers in parentheses for models 1 and 3 are robust standard errors. * p < .05 ** p < .01 *** p < .001 (two-tailed tests)

Conclusion We have assessed the effect of globalization on the distribution of incomes in 72 countries, 1970–90, using the most widely accepted data on income inequality (Deininger and Squire 1996) and recently available information regarding foreign direct investment (UNCTAD 2000; World Bank 2000). In keeping with previous research, we focused on the

370

margit bussmann, indra de soysa, and john r. oneal

influence of foreign investment to gauge the consequences of globalization; but we also estimated the effects of two trade-based measures of economic openness: the trade-to-GDP ratio and Sachs and Warner’s (1995) indicator of free-trade policies. We used specifications incorporating the Kuznets (1955) curve, the character of political institutions, and various economic and demographic factors emphasized in recent research. In addition to our analyses of the Gini index, a summary measure of inequality, we assessed the effect of a large multinational presence on the income received by the poorest 20% in each society in order to ensure that we accurately determined the consequences of globalization for the poorest of the poor. We find no evidence that globalization has adversely affected national income inequality. The ratio of foreign direct investment to gross domestic product is unrelated to the Gini index in all eight of our tests; nor does foreign investment adversely affect inequality in developing countries in particular. The share of income received by the poorest 20% is uncorrelated with the economic importance of foreign investment in our two OLS analyses of the pooled time series. In a fixedeffects test of the lowest quintile’s share of income, the effect of FDI is significant statistically because of the lack of robust standard errors for this estimator; but substantively, the effect is small. Nor is either of the trade-based measures of openness associated with greater inequality. Neither the presence of multinational corporations nor international commerce worsens income inequality or increases the marginalization of the poor. If foreign direct investment increases average incomes (Borensztein et al. 1998; de Soysa and Oneal 1999; UNCTAD 1999; Ram and Zhang 2002; Dollar and Kraay 2002; Bhalla 2002) and does not adversely affect the distribution of income even in developing countries (Sylwester 2005), FDI must increase the incomes of the poor in these societies. Dollar and Kraay (2000, 2002) reach the same conclusion regarding the effects of trade. Further research is needed, however, on whether alternative economic strategies are particularly beneficial to the poor. Some may increase the absolute income of the poor by increasing their share of national income while others may benefit the poor by raising the economy’s growth rate. Our results indicate that the fears of the anti-globalization movement—that global economic integration comes at the expense of the poor—are not justified; and they reinforce the conclusion of recent research that globalization has reduced global income inequality (Bhalla 2002; Sala-i-Martin 2002; Firebaugh and Goesling 2004).

globalization & income inequality in the developing world 371 We investigated the effects of multinational corporations on income inequality in a variety of specifications. In these, we found only limited support for the Kuznets (1955) curve. Income inequality is curvilinearally related to the average real income only in cross-national estimations, not in the individual time-series. Socialist states do have more equal distributions of income, though this comes at the expense of growth (Barro 1991). The relative labor productivity of agriculture is associated with inequality, but it is hardly surprising that dual economies have high inequality. If there is inequality across economic sectors, there inevitably will be inequality across households. Taken as a group, our analyses make clear the limits of our understanding of the determinants of income inequality: We know more about what does not affect the distribution of incomes than what does. Appendix: Definitions of Variables, Sources of Data, and List of Countries Analyzed Dependent Variable: Income Inequality In Tables 1 and 2 we examine income inequality using the Gini coefficient, the most commonly used measure. The Gini index equals zero if everybody has the same income and 100 if one person possesses everything. In Table 3 our dependent variable is the poorest quintile’s share of national income. Both measures are taken from the Deininger and Squire (1996) data set, which contains a subset of observations that meet acceptable standards of data collection (e.g., they must be based on nationwide surveys and a comprehensive coverage of income sources). We restrict our sample to these high quality cases. We corrected Deininger and Squire’s Gini and quintile estimates to account for differences in the characteristics of the surveys: inequality is less when measurement is based on expenditures, not income; on income net of taxes, rather than gross income; and on household, not individual, surveys. Using the method of Dollar and Kraay (2000), we adjusted Deininger and Squire’s data based on regression analyses that identified the magnitudes of these differences. The summary statistics reported in the table at the end of this appendix are for the corrected data. Independent Variables: FDI-to-GDP Ratio We estimated the value of foreign direct investment in each year using UNCTAD’s (2000) revised data for the stock of FDI in 1980 and data on flows from UNCTAD (2000) and the World Bank (2000). To calculate annual values back to 1970 and forward to 1990, we converted foreign investment flows to constant dollars, accounted for depreciation using the accelerated method with a half-life of 10 years, and subtracted flows from or added them

372

margit bussmann, indra de soysa, and john r. oneal

to the stock of foreign direct investment in 1980. We calculated the average value for each year using the stock of FDI at the beginning and the end and divided this by the country’s real GDP (Heston et al. 1995). The FDI-to-GDP ratio indicates the importance of foreign investors, particularly multinational corporations, to the economy. Income We test for the inverted U-shaped Kuznets curve with the natural logarithm of real GDP per capita and its square based on purchasing power parity in international prices (Heston et al. 1995). The logarithmic transformation is performed to correct the skewed distribution of real GDP per capita. Democratic Experience and Socialist Countries Our democracy scores are taken from the Polity III data set ( Jaggers and Gurr 1995) which contains annual democracy and autocracy scores, based on 11-point scales. A summary score for each country-year was calculated by subtracting a state’s autocracy score from its democracy score, as Jaggers and Gurr recommend. Thus, this variable can range from +10 for countries that are purely democratic to –10 for authoritarian countries. The years of democratic experience were calculated by counting the number of years that the country had been a “coherent democracy” ( Jaggers and Gurr 1995), i.e., when the democracy-autocracy score was greater than +6. The dummy variable that identifies the socialist states is drawn from Barro (1991). In our sample these are Algeria, Rwanda, Tanzania, Zambia, and China. Economic Dualism and the Size of the Agricultural Sector We use two variables to account for the structure of a country’s economy. The first is Bourguignon and Morrisson’s (1998) measure of relative labor productivity (RLP) in agriculture with reference to the rest of the economy. This is a gauge of economic dualism and accounts for differences in productivity between the agricultural and the manufacturing and service sectors. RLP is defined as follows: (1 – agrGDP)* agremp RLP = ––––––––––––––––– agrGDP * (1 – agremp) where RLP is relative labor productivity, agrGDP is agriculture’s share of GDP and agremp is its share of employment.19 Data on the share of agriculture in employment is available from the UN Food and Agricultural Organization’s 19 RLP is similar to the measure of sector dualism used by Alderson and Nielsen (1999), which equals the absolute value of the percent of the labor force in agriculture minus agriculture’s share of GDP. Taking the absolute value seems inappropriate, however, because a state whose agricultural sector was more efficient than its industry could have the same score as a state with a more efficient industrial sector.

globalization & income inequality in the developing world 373 web site (apps.fao.org); data on agriculture’s share of GDP are available from the World Bank’s World Tables (1989–90, 1994). We also control for the size of the agricultural sector, where income is expected to be more equally distributed. Controlling for sectoral dualism, the size of the agricultural sector should be negatively related to income inequality. The size of the agricultural sector is operationalized as agriculture’s share of GDP. Age Structure of the Population and School Enrollment We include in some of our analyses a control for the share of total population that is under age 15. These data can be found on the World Bank’s web site (devdata.worldbank.org/hnpstats). The percentage of children enrolled in secondary schools is taken from UNESCO (various years), which provides data in five-year intervals. We created annual data by interpolating between the five-year estimates. Trade Trade is an alternative to foreign investment as an indicator of globalization. Our data regarding trade volume are taken from Heston et al. (1995). Our measure of economic openness equals exports and imports divided by GDP. We also employ the measure of free-trade policies created by Sachs and Warner (1995). They categorize countries as closed or open by taking into account the extent of non-tariff barriers, average tariff rates, black market exchange rates, the existence of a socialist economic system, and whether a state has a monopoly of major exports. A country is considered open if none of the above criteria apply. Regional Indicators The regional identifications of the Correlates of War project (Singer 1995) were used except that the United States and Canada were added to Europe so that the Latin American countries would be uniquely identified by the code for the Western Hemisphere. Table A1. Summary statistics of variables

Gini index of income inequality Poorest quintile’s share of income FDI/GDP Real per capita income (ln) Years of democracy Population under 15, % Secondary school enrollment rate Agricultural share of GDP, % Relative labor productivity Trade/GDP Sachs & Warner (1995) openness

Obs.

Mean

Std. Dev.

383 342 383 383 377 366 375 334 325 374 353

41.30 6.26 0.0779 8.531 48.20 31.20 64.55 13.16 3.36 0.482 0.654

11.15 2.16 0.1201 0.919 47.70 9.79 27.93 12.34 3.76 0.292 0.476

Min.

Max.

17.67 82.68 1.60 11.70 0 1.5845 6.178 9.803 0 181 16.51 49.40 2 119 0.33 58.84 0.93 39.88 0.035 1.513 0 1

374

margit bussmann, indra de soysa, and john r. oneal Table A2. List of countries analyzed

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

Algeria Australia Bahamas Bangladesh Barbados Belgium Bolivia Botswana Brazil Cameroon Canada Chile China Colombia Costa Rica Cote d’Ivoire Denmark Dominican Republic Egypt El Salvador Fiji Finland France Gabon Germany Ghana Greece Guatemala Honduras India Indonesia Iran Ireland Italy Jamaica Japan

37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72.

Jordan Korea, Republic of Lesotho Malaysia Mauritania Mauritius Mexico Morocco Nepal Netherlands New Zealand Nigeria Norway Pakistan Panama Peru Philippines Portugal Rwanda Seychelles Singapore Spain Sri Lanka Sweden Taiwan Tanzania Thailand Trinidad Tunisia Turkey United Kingdom USA Uganda Venezuela Zambia Zimbabwe

References Alderson, A.S., and F. Nielsen. 1999. “Income Inequality, Development, and Dependence: A Reconsideration.” American Sociological Review 64:606–31. Alesina, A., and D. Rodrik. 1994. “Distributive Politics and Economic Growth.” Quarterly Journal of Economics 109:465–91. Anand, S., and S. Kanbur. 1993. “Inequality and Development: A Critique.” Journal of Development Economics 1:19–43. Ballmer-Cao, T., and J. Scheidegger. 1979. Compendium of Data for World-System Analysis. Zürich: University of Zürich. Baran, P. 1956. The Political Economy of Growth. New York: Monthly Review Press.

globalization & income inequality in the developing world 375 Barro, R.J. 1991. “Economic Growth in a Cross-Section of Countries.” Quarterly Journal of Economics 106(2):407–433. ——. 2000. “Inequality and Growth in a Panel of Countries.” Journal of Economic Growth 5:5–32. Beck, N., and J.N. Katz. 1995. “What to Do (and Not to Do) with Time-Series CrossSection Data.” American Political Science Review 89:634–647. Bhalla, S. 2002. Imagine There’s No Country: Poverty, Inequality, and Growth in the Era of Globalization. Washington, DC: Institute for International Economics. Birdsall, N. 1998. “Life Is Unfair: Inequality in the World.” Foreign Policy 112:76–83. Bollen, K.A. and R.W. Jackman. 1985. “Political Democracy and the Size Distribution of Income.” American Sociological Review 50:438–57. Borensztein, E., J. de Gregorio, and J. Lee. 1998. “How Does Foreign Direct Investment Affect Economic Growth?” Journal of International Economics 45:115–36. Bornschier, V., and C. Chase-Dunn. 1985. Transnational Corporations and Underdevelopment. New York: Praeger Publishers. Bornschier, V., C. Chase-Dunn, and R. Rubinson. 1978. “Cross-National Evidence of the Effects of Foreign Investment and Aid on Economic Growth and Inequality: A Survey of Findings and a Reanalysis.” American Journal of Sociology 84:651–83. Borsu, A. and H. Glejser. 1992. “Do Protection, Schooling, Product per Head and Income Distribution Influence Growth?” European Economic Review 36:1235–39. Bourguignon, F., and C. Morrisson. 1998. “Inequality and Development: The Role of Dualism.” Journal of Development Economics 57:233–57. ——. 2002. “Inequality Among World Citizens, 1820–1992.” American Economics Review 92:727–744. Cardoso, F.H., and E. Faletto. 1979. Dependency and Development in Latin America. Berkeley, CA: University of California Press. Chan, S. 1989. “Income Inequality among LDCs: A Comparative Analysis of Alternative Perspectives.” International Studies Quarterly 33:45–66. Chase-Dunn, C. 1975. “The Effects of International Economic Dependence on Development and Inequality: A Cross-National Study.” American Sociological Review 40:720–38. Cooper, R.N. 2001. “Growth and Inequality: the Role of Foreign Trade and Investment.” Unpublished ms., Harvard University, April. de Soysa, I., and J.R. Oneal. 1999. “Boon or Bane? Reassessing the Productivity of Foreign Direct Investment.” American Sociological Review 64:766–782. Deininger, K., and L. Squire. 1996. “A New Data Set Measuring Income Inequality.” World Bank Economic Review 10:565–91. ——. 1998. “New Ways of Looking at Old Issues: Inequality and Growth.” Journal of Development Economics 57:259–87. Dollar, D., and A. Kraay. 2000. “Growth Is Good for the Poor.” Unpublished ms., World Bank, March. ——. 2002. “Spreading the Wealth.” Foreign Affairs 81:120–33. Easterlin, R.A. 1980. Birth and Fortune: The Impact of Numbers on Personal Welfare. New York: Basic Books. Easterly, W. 1999. “Life during Growth.” Journal of Economic Growth 4:239–75. Edwards, S. 1998. “Trade Policy, Growth, and Income Distribution.” AEA Papers and Proceedings 87:205–210. Evans, P.B., and M. Timberlake. 1980. “Dependence, Inequality, and the Growth of the Tertiary: A Comparative Analysis of Less-Developed Countries.” American Sociological Review 45:531–52. Firebaugh, G., and B. Goesling. 2004. “Accounting for the Recent Decline in Global Income Inequality.” American Journal of Sociology 110:283–312. Frank, A.G. 1969. Latin America: Underdevelopment or Revolution. New York: Monthly Review Press.

376

margit bussmann, indra de soysa, and john r. oneal

Galtung, J. 1971. “A Structural Theory of Imperialism.” Journal of Peace Research 8:81–117. Heston, A., R. Summers, D.A. Nuxoll, and B. Aten. 1995. Penn World Table (Mark 5.6a). Cambridge, MA: National Bureau of Economic Research. Higgins, M., and J.G. Williamson. 1999. “Explaining Inequality the World Round: Cohort Size, Kuznets Curves, and Openness.” NBER Working Paper 7224. National Bureau of Economic Research, July. Hsiao, C. 1986. Analysis of Panel Data. Cambridge: Cambridge University Press. Jaggers, K., and T.D. Gurr. 1995. “Tracking Democracy’s Third Wave with the Polity III Data.” Journal of Peace Research 32:469–82. Jensen, N.M. 2003. “Democratic Governance and Multinational Corporations: Political Regimes and Inflows of Foreign Direct Investment.” International Organization 57:587–616. Kentor, J. 2001. “The Long Term Effects of Globalization on Population Growth, Inequality, and Economic Development.” Social Problems 48:435–55. King, G. 2001. “Proper Names and Methodological propriety: Pooling Dyads in International Relations Data.” International Organization 55:497–507. Kohli, A., M.F. Altfeld, S. Lotfian, and R. Mardon. 1984. “Inequality in the Third World: An Assessment of Competing Explanations.” Comparative Political Studies 17:283–318. Korzeniewics, R.P., and T.P. Moran. 1997. “World-Economic Trends in the Distribution of Income, 1965–1992.” American Journal of Sociology 102:1000–39. Kuznets, S. 1955. “Economic Growth and Income Inequality.” American Economic Review 45:3–6, 17–26. Li, Q., and A. Resnick. 2003. “Reversal of Fortunes: Democratic Institutions and Foreign Direct Investment Inflows to Developing Countries.” International Organization 57:175–212. Mazur, J. 2000. “Labor’s New Internationalism.” Foreign Affairs 79:79–93. Mehlkop, G. 2002. Wirtschaftliche Freiheit, Einkommensgleichheit und physische Lebensqualität. Opladen: Leske + Budrich. Milanovic, B. 1999. True World Income Distribution, 1988 and 1993: First Calculations Based on Household Survey Alone. Development Research Group, World Bank, Washington. Muller, E.N. 1988. “Democracy, Economic Development, and Income Inequality.” American Sociological, Review 53:50–68. Nielsen, F. 1994. “Income Inequality and Industrial Development: Dualism Revisited.” American Sociological Review 59:654–77. ——, and A.S. Alderson. 1995. “Income Inequality, Development, and Dualism: Results from an Unbalanced Cross-National Panel.” American Sociological Review 60:674–701. Oneal, J.R. 1994. “The Affinity of Foreign Investors for Authoritarian Regimes.” Political Research Quarterly 47:565–88. ——, and F.H. Oneal. 1988. “Hegemony, Imperialism, and the Profitability of Foreign Investments.” International Organization 42:347–73. Persson, T., and G. Tabellini. 1994. “Is Inequality Harmful For Growth? Theory and Evidence.” American Economic Review 84:600–21. Portes, A. 1976. “On the Sociology of National Development: Theories and Issues.” American Journal of Sociology 82:55–85. Ram, R., and K.H. Zhang. 2002. “Foreign Direct Investment and Economic Growth: Evidence from Cross-Country Data for the 1990s.” Economic Development and Cultural Change 51:205–215. Ravallion, M., and S. Chen. 1997. “What Can New Survey Data Tell Us about Recent Changes in Distribution and Poverty?” World Bank Economic Review 11:357–82.

globalization & income inequality in the developing world 377 Reuveny, R., and Q. Li. 2003. “Economic Openness, Democracy, and Income Inequality: An Empirical Analysis.” Comparative Political Studies 36:575–601. Rubinson, R. 1976. “The World-Economy and the Distribution of Income within States.” American Sociological Review 41:638–59. Sachs, J.D., and A. Warner. 1995. “Economic Reform and the Process of Global Integration.” Brookings Paper on Economic Activity 1:1–118. Sala-i-Martin, X. 2002. “The Disturbing ‘Rise’ of Global Income Inequality.” NBER Working Paper no. 8904. Washington, DC: National Bureau of Economic Research. Singer, J.D. 1995. “System Membership List (System.dat).” Correlates of War Project, Department of Political Science, University of Michigan. Solow, R.M. 1956. “A Contribution to the Theory of Economic Growth.” Quarterly Journal of Economics 70:65–94. Sylwester, K. 2005. “Foreign Direct Investment, Growth and Income Inequality in Less Developed Countries.” International Review of Applied Economics 19:289–300. Tsai, P. 1995. “Foreign Direct Investment and Income Inequality: Further Evidence.” World Development 23:469–83. UNCTAD. 1995. World Investment Report: Transnational Corporations and Competitiveness. New York: United Nations Publications. ——. 1999. World Investment Report: Foreign Direct Investment and the Challenge of Development. New York: United Nations Publications. ——. 2000. World Investment Report: Cross-Border Mergers and Acquisitions and Development. New York: United Nations Publications. UNESCO. Various years. UNESCO Annual Statistical Yearbook. New York: United Nations Publications. Wade, R.H. 2004. “Is Globalization Reducing Poverty and Inequality?” International Journal of Health Services 34(3):381–414. Ward, M., and K. Gleditsch. 2004. “Globalization’s Impact on Poverty, Inequality, Conflict and Democracy.” In Globalization and Its Outcomes, edited by J. O’Loughlin, L. Staeheli, and E. Greenberg. New York: Guilford. Weede, E. 1989. “Democracy and Income Inequality Reconsidered.” American Sociological Review 54:865–68. ——. 1990. “Democracy, Party Government, and Rent-Seeking as Determinants of Distributional Inequality in Industrial Societies.” European Journal of Political Research 18:515–33. ——. 1993. “The Impact of Military Participation on Economic Growth and Income Inequality.” Journal of Political and Military Sociology 21:21–58. ——. 1997. “Income Inequality, Democracy and Growth Reconsidered.” European Journal of Political Economy 13:751–64. ——. 2000. Asien und der Westen. Politische und kulturelle Determinanten der wirtschaftlichen Entwicklung. Baden-Baden: Nomos. ——, and H. Tiefenbach. 1981. “Some Recent Explanations of Income Inequality.” International Studies Quarterly 23:255–82. Wiggins, V. 1999. “Comparing XTGLS with Regress Cluster ().” Stata Corporation, www.stata.com/support/faqs/stgls_rob.html. Wolf, M. 2002. “Doing More Harm than Good.” Financial Times (May 8th). World Bank. 2000. World Development Indicators CD-Rom. Washington, DC: World Bank. ——. Various. World Tables. Washington, DC: World Bank.

ENGLISH AS AN INTERNATIONAL LANGUAGE IN NON-NATIVE SETTINGS IN AN ERA OF GLOBALIZATION Masamichi Sasaki, Tatsuzo Suzuki and Masato Yoneda Introduction Language is an unquestionable prerequisite for human communication. Language is an “indispensable and universal component of the cultural system of all societies” (Barber 1982:3). As such language is an intrinsic element of sociology (cf. Hertzler 1965; Mesthrie et al. 2000), indeed an excellent exemplar of Durkheim’s “social representations” (Durkheim 1938). This paper will explore briefly the importance of language study to sociology. In the course of this discussion, we will also explore the claims of some who believe the dominance of English as an international language is only transitory. We will then turn to this paper’s principal focus, which is to examine English as an international language (or perhaps the international language). We will focus especially on the attitudes about English as an international language of non-English language speaking peoples and of peoples who speak English only as a second (or even third, and so on) language. Extensive empirical findings will be examined in an effort to predict the future direction of the spread of English as international language. There is extensive literature on language and this literature has myriad foci, ranging from social, to political, to economic; even to military, and beyond. And it is easy to see that a major consideration in much of the discussion of language concerns language dominance. What makes one language become dominant over others? What drives language dominance? Certainly political and economic circumstances play a major role as one language emerges over others throughout history (see, e.g., Bailey 1985; Barber 1982; Crystal 1997; Firth 1970; Hertzler 1965; Kachru 1986). At the same time, this paper is considering the question of international language dominance in the context of globalization, a phenomenon which has changed things very distinctly from the time of the Roman Empire for instance, or even the era of British colonialism. In terms of direct communication, for example, the telephone and the internet (email in particular) have made daily

380

masamichi sasaki, tatsuzo suzuki and masato yoneda

communication throughout the world a simple, straightforward, even incidental means of human communication. Such communications require language, and most often they require a common language, for convenience if for no other reason. English as International Language English is undeniably the dominant “international language,” at least at present. Indeed, Kachru (1986) pointed out that there are more nonnative than native users of English in the world. According to Power (2005), non-native English speakers now outnumber native ones 3 to 1 (also see Strevens 1982; Smith 1983; and Quirk and Widdowson 1985). We will discuss later the potential for English to remain dominant. How did English get where it is today? Brumfit (1982:1; also cf. Lieberson 1982; Noss 1983; Bryson 1990; Pennycook 1994; and Crystal 2004) summarizes well the source of English’s present dominance in the following: English is an international language in that it is the most widespread medium of international communication, both because of the number and geographical spread of its speakers, and because of the large number of non-native speakers who use it for part at least of their international contact. The predominance of English is mainly the result of two periods of world domination by English speaking nations: British imperialism in the nineteenth century, and the economic influence of the United States in the twentieth century. The combination of political influence and technological superiority acquired through these two successive movements has given English an advantage over other major imperial languages such as French or Spanish, while the relative geographical restrictions of Russian, Chinese in its many forms or Arabic have made these languages less influential internationally.

There is a difference between English as international language and previous languages that had dominant characteristics (such as Latin (see Wright 2004)), and of course that difference arises because we are now dealing with the context of globalization. The extraordinarily widespread use of daily international communication in this context means that the dominant language at present is used on a much more universal and widespread scale than ever before. The worldness of English, in both its global and local senses, implies relationships to the larger world and to the local context different from those of other languages. Given the dominant position of English in

english as an international language in non-native settings 381 the world and its connections both to inequitable economic systems and to the dominance of certain forms of culture and knowledge, there are inevitable questions to be asked here concerning language and inequality. (Pennycook 1994:34–35).

In this paper, our “questions” about language and inequality will be derived from the empirical work to be discussed shortly. This work concerns people’s attitudes toward English as international language and its dominance. These directly target the issues of the apparently inherent inequalities suggested by Pennycook. This dominance, though, was not inevitable. Historical circumstances were such that English coincidentally arose as international language, as dominant (cf. Melchers and Shaw 2003). Smith (1983:2) considers English an “international auxiliary language.” This begins to suggest a set of scholars who view English dominance in a different light. Another hint of this suggestion is given by Pennycook (1994:9), who states that English “is seen as neutral because it is assumed that once English has in some sense become detached from its original cultural contexts (particularly England and America), it is now a neutral and transparent medium of communication.” Bickley (1982:87) goes so far as to say that “English does not ‘belong’ to any one group of people. The use of English is always culture-bound, but the English language is not bound to any specific culture or political system.” So at least some scholars consider that the social and cultural accoutrements of English have lost much of their association with the dominant, international language. As Kachru (1986:vii) puts it, “Whether these . . . are real or imagined is not important; what is vital is the public attitude toward English, the love-hate relationship with the language” [emphasis added]. Other scholars have focused on English dominance as “linguistic capital.” Linguistic capital is not unlike social capital, cultural capital, or economic capital (cf. Bourdieu 1976, 1977; Bourdieu and Passeron 1990). As such, English “continues to provide unprecedented power for mobility and advancement to those native and non-native users who possess it as a linguistic tool” (Kachru 1986:14). Nonetheless, as the world globalizes, so too does communication, hence a single language becomes nearly imperative. That this language had imperialist origins becomes less and less an issue in its universal deployment. Not everyone agrees with this sentiment, however (cf. Quirk 1987). Deneire (1993:172), for example, states that “Resistance to Western domination in general and to American imperialism in

382

masamichi sasaki, tatsuzo suzuki and masato yoneda

particular has led to a revival of ethnonationalisms and to a rejection of English in favor of national languages.” Thus English as international language becomes embroiled in the debate over Westernization, a subject principally beyond our present scope. Some scholars, including Fishman (1998:27), argue that English dominance will “eventually wane in influence” (also cf. Al-Dabbagh 2005). In a long-term historical perspective, this is undoubtedly true, based upon past history. Many of these scholars focus on the notion of regionalization and see English dominance waning in the face of a potential rise of regional emphases. However, at present, no substitute languages appear on the horizon. Ironically, it could be globalization’s technology sector that ultimately contributes to the wane of English dominance. Machine or automatic language translation technology has had an amazingly rocky history. Many thought the world would have access to universal translators in the 1970s or 1980s. They have yet to come to fruition; however, as with many technologies, it is only a matter of time. If universal and accurate automatic translation were available, then reliance on a single dominant language would certainly wane; perhaps along with it English dominance. Attitudes toward English as International Language What do non-English speaking or non-native English speaking people feel about English as the international means of communication, the dominant language? Do these people want to learn English? Do these people feel “envious, resentful, or angry” (Crystal 1997:2). Crystal (1997:2) goes on to state: These feelings are natural, and would arise whichever language emerged as a global language. They are feelings which give rise to fears . . . and fears lead to conflict. . . . Political differences over language economics, education, laws and rights are a daily encounter for millions. Language is always in the news, and the nearer a language moves to becoming a global language, the more newsworthy it is.

Clearly the sources of feelings about language are extremely broad in scope, including, for instance, economics, politics, the law, policy making, and so on. Effectively this broad-based derivation is treated by the science of sociolinguistics (see, e.g., Kachru 1986, 1991; Baker 1992; Berns 1992; Ammon 1994; Pulcini 1997). Kachru (1991) called attitudes

english as an international language in non-native settings 383 (toward language) unplanned or invisible forces which can, in effect, become invisible policies. From a somewhat different perspective, Cooper and Fishman (1977:8) have pointed out that “Language attitude is sometimes studied in its own right and sometimes in connection with its relationship to language behavior.” These types of study of language attitudes are at least characteristically if not fundamentally different from the empirical studies being reported here. As Cooper and Fishman (1977:8) also point out, “Most research on the relationship between language attitude and second-language achievement has concentrated on the motivational aspects of language attitude or on the attitudes toward native speakers of the target language.” Language proficiency is often cited and studied in this regard; however, the present work does not speak to issues of motivation, achievement, proficiency or competency. Indeed, very little empirical work has ever before focused on the issue of Englishness, on the rise of English to the status of international language, as the principal conduit for international communications. What these and many other implied speculations have heretofore lacked is empirical evidence in general and evidence about attitudes toward English as international language in particular. (For further information on attitudes toward English among different nations, see, e.g., Berns 1988; Pride and Liu 1988; Pennington and Yue 1994; Shim 1994; Iwasaki 1994; Yong and Campbell 1995; van Essen 1997; Pulcini 1997; and Spolsky and Shhamy 1999.) Empirical Study and Analysis Between 1996 and 1998, Tokyo’s National Language Research Institute conducted nationwide cross-national personal interview surveys among 28 nations aimed at, among other things, determining the nations’ degrees of tolerance for English, both intra- and internationally. Sampling methods for most of these nations were probability sampling and quota sampling techniques (for details on the samplings for each surveyed nation, see Suzuki, Yanagihara and Yoneda 2003). The original questionnaire (see National Language Research Institute 1999) included 58 questions covering a number of themes and topics which included one’s language environment, one’s mother tongue, language in general, and one’s attitudes toward foreign languages, toward English, and toward Japanese.

384

masamichi sasaki, tatsuzo suzuki and masato yoneda

For purposes of the results we present here, 25 nations (except the Unites States, Australia, and the United Kingdom, where the mother tongue is English, the so-called “inner circle”; see Kachru 1985), were included in the analysis (see Figure 1). India, the Philippines, Nigeria, and Singapore are considered the so-called “outer circle” (see Kachru l985). This so-called “outer circle” refers to a group of nations with comparatively low level prevalence of English usage. Brazil, Argentina, Mongolia, Taiwan, China, Korea, Japan, Thailand, Indonesia, Vietnam, Egypt, Israel, Turkey, Italy, Spain, Portugal, France, Germany, Holland, Hungary, and Russia are considered to be in the so-called “expanding circle” (see Kachru 1985), where the prevalence of English usage is higher and increasing. Note that these nations “do not have a history of colonization by members of the inner circle” (Crystal 1997:54). Crosstabulations and correspondence analyses were conducted. The seven questions involved in this particular study are listed in Table 1.

Inner circle: Australia Canada New Zealand UK USA

Outer circle: Bangladesh Ghana India Kenya Malaysia Nigeria Pakistan Philippines Singapore Sri Lanka

Expanding circle: China Egypt Indonesia Israel Japan Korea Nepal Saudi Arabia Taiwan USSR Zimbabwe

Figure 1. Kachru’s (1985) “Concentric Circles of English”.

english as an international language in non-native settings 385 Table 1. Survey questions used for the analysis Question 7: Would you like/prefer to use your mother tongue when talking with foreigners in your country? 1 = yes 2 = no Question 8: Which language, your mother tongue or any foreign language, do you tend to use when talking with foreigners in your country? Please choose one from this card. The card showed the following options: 1. My mother tongue because I cannot speak any foreign languages. 2. My mother tongue although I can speak foreign languages. 3. Foreign languages. 4. No opportunity to talk with foreigners. Question 10: Including your mother tongue, what languages do you think will be essential or indispensable for international communication in the future? Respondents simply listed language(s) in answer to this question (multiple choices). Question 11: Including your mother tongue, what languages do you think will be essential or indispensable for communication within your country in the future? Respondents simply listed language(s) in answer to this question (multiple choices). Question 12: Including your mother tongue, what languages would you like your children to learn? If you do not have children, what languages would you choose if you did have children? Respondents simply listed language(s) in answer to this question (multiple choices). Question 18: English is said to be the world’s dominant or most influential language today. Do you agree or disagree? agree = 1 disagree = 2 Question 19: What do you think about English being the world’s dominant or most influential language today? The possible responses were: 1. I think it is good that English is dominant. 2. I do not think it is good that English is dominant, but I see no alternative. 3. I do not think it is good that English is dominant, and I think more should be done to use other languages.

Table 1 shows that the analyses described here involve Questions 7, 8, 10, 11, 12, 18, and 19. These numbers will be used throughout the following discussion to denote the questions as their results are discussed. The percentage distributions for each of the 7 questions for each of the 25 nations are shown in Tables 2 through 8.

386

masamichi sasaki, tatsuzo suzuki and masato yoneda

Table 2 shows the crosstabulation for Question 7 by nation. The bullets on the left indicate nations whose respondents said “No” to the question about preference for their mother tongue when talking with foreigners in their own nation much more often than “Yes.” In other words, in Singapore, the Philippines, Nigeria and Egypt, respondents were much less likely to prefer using their mother tongue to communicate with foreigners while in their own nations. Respondents in nearly all other nations answered “Yes” 60 or more percent of the time (with the exception of India at 53 percent). Indeed, many of the other respondents answered “Yes” 80 to 90 percent of the time. These findings are consistent with the fact that Singapore, the Philippines and Nigeria are outer circle nations. Egypt is an exception, as is India to some extent. Table 3 shows the crosstabulations for Question 8 by nation. Here the bullets are used to denote those nations that most frequently (35 percent or more) answered “Foreign Languages” to Question 8. These were Singapore, Holland, India, the Philippines, Israel, Turkey, and Egypt. Of course, the responses to this question, as can be seen in the table, are somewhat dictated by the percentages of respondents in these nations who cannot speak any other language and/or those who have no opportunities to speak with foreigners. In Nigeria, for instance, three-quarters of all respondents said they could not speak any foreign languages. Table 4 depicts the crosstabulations for Question 10 by nation. Responses were grouped into three categories: mother tongue, English, and “other major languages.” This table presents the results for those who selected English and those who did not. In terms of the indispensability of English for international communication in the future, virtually all respondents from all nations referenced English an average of 91 percent of the time. Only Brazil answered English less than 80 percent of the time (at 72 percent). Question 11 (see Table 5) asked respondents what languages they thought would be indispensable for communication in the future within their own nation. Here again, responses were grouped into three categories: mother tongue, English, and “other major languages.” This table presents the results for those who selected English and those who did not. Bullets in Table 5 denote those nations where respondents answered English less often than average (the average for English was 60 percent). These nations include Argentina, Spain, Portugal, Russia, Indonesia, Taiwan, China, and Japan, with responses for “English” ranging from

english as an international language in non-native settings 387 Table 2. Crosstabulations of Question 7 by nation Q7 Would you like/prefer to use your mother tongue when talking with foreigners in your country?

country Brazil Argentina Korea • Singapore Thailand France Germany Holland Hungary Italy Spain Portugal Russîa India Indonesia • Philippines Vietnam Mongolia Israel Turkey • Nigeria • Egypt

frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency %

Yes

No

944 87.7% 940 84.7% 879 87.9% 242 23.6% 749 74.9% 813 77.7% 746 74.7% 782 77.6% 961 90.2% 952 92.2% 878 87.8% 848 87.7% 903 86.7% 530 53.0% 698 69.4% 373 37.3% 919 91.9% 854 85.2% 821 82.1% 672 66.5% 299 29.9% 391 37.7%

132 12.3% 123 11.1% 120 12.0% 785 76.4% 251 25.1% 204 19.5% 180 18.0% 226 22.4% 103 9.7% 80 7.8% 120 12.0% 109 11.3% 139 13.3% 470 47.0% 308 30.6% 627 62.7% 81 8.1% 142 14.2% 179 17.9% 338 33.5% 702 70.1% 646 62.3%

N.A.

47 4.2% 1 .1%

30 2.9% 72 7.2% 2 .2% 2 .2% 10 1.0%

6 .6%

total 1076 100.0% 1110 100.0% 1000 100.0% 1027 100.0% 1000 100.0% 1047 100.0% 998 100.0% 1008 100.0% 1066 100.0% 1032 100.0% 1000 100.0% 967 100.0% 1042 100.0% 1000 100.0% 1006 100.0% 1000 100.0% 1000 100.0% 1002 100.0% 1000 100.0% 1010 100.0% 1001 100.0% 1037 100.0%

388

masamichi sasaki, tatsuzo suzuki and masato yoneda

Table 2 (cont.) Q7 Would you like/prefer to use your mother tongue when talking with foreigners in your country?

Taiwan China Japan Total

frequency % frequency % frequency % frequency %

Yes

No

683 60.0% 826 82.6% 790 78.1% 18493 72.3%

455 40.0% 169 16.9% 126 12.5% 6815 26.6%

N.A.

total

5 .5% 95 9.4% 270 1.1%

1138 100.0% 1000 100.0% 1011 100.0% 25578 100.0%

Table 3. Crosstabulations of Question 8 by nation Q8 Which language, your mother tongue or any foreign language, do you tend to use when talking with foreigners in your country? Mother Mother Foreign No tongue/ tongue/ languages opportunity Can’t speak Can speak to talk with foreigners any foreign foreign languages languages country Brazil

frequency % Argentina frequency % Korea frequency % • Singapore frequency % Thailand frequency % France frequency % Germany frequency % • Holland frequency % Hungary frequency % Italy frequency %

528 49.1% 631 56.8% 479 47.9% 211 20.5% 296 29.6% 532 50.8% 441 44.2% 151 15.0% 571 53.6% 495 48.0%

84 7.8% 147 13.2% 169 16.9% 97 9.4% 153 15.3% 211 20.2% 269 27.0% 192 19.0% 102 9.6% 146 14.1%

29 2.7% 111 10.0% 84 8.4% 598 58.2% 150 15.0% 222 21.2% 236 23.6% 610 60.5% 173 16.2% 285 27.6%

435 40.4% 201 18.1% 268 26.8% 119 11.6% 401 40.1% 73 7.0% 26 2.6% 55 5.5% 215 20.2% 106 10.3%

N.A. total

20 1.8% 2 .2% 9 .9% 26 2.6% 5 .5%

1076 100.0% 1110 100.0% 1000 100.0% 1027 100.0% 1000 100.0% 1047 100.0% 998 100.0% 1008 100.0% 1066 100.0% 1032 100.0%

english as an international language in non-native settings 389 Table 3 (cont.) Q8 Which language, your mother tongue or any foreign language, do you tend to use when talking with foreigners in your country? Mother Mother Foreign No tongue/ tongue/ languages opportunity Can’t speak Can speak to talk with foreigners any foreign foreign languages languages Spain

total

frequency % Portugal frequency % Russia frequency % • India frequency % Indonesia frequency % • Philippines frequency % Vietnam frequency % Mongolia frequency % • Israel frequency % • Turkey frequency % Nigeria frequency % • Egypt frequency % Taiwan frequency % China frequency % Japan frequency % frequency %

533 53.3% 517 53.5% 517 49.6% 274 27.4% 544 54.1% 84 8.4% 280 28.0% 447 44.6% 108 10.8% 307 30.4% 762 76.1% 223 21.5% 420 36.9% 442 44.2% 371 36.7% 10164 39.7%

144 14.4% 125 12.9% 63 6.0% 149 14.9% 132 13.1% 328 32.8% 328 32.8% 276 27.5% 139 13.9% 300 29.7% 138 13.8% 120 11.6% 85 7.5% 166 16.6% 76 7.5% 4139 16.2%

167 16.7% 156 16.1% 63 6.0% 423 42.3% 114 11.3% 554 55.4% 136 13.6% 120 12.0% 691 69.1% 361 35.7% 64 6.4% 365 35.2% 214 18.8% 118 11.8% 91 9.0% 6135 24.0%

156 15.6% 169 17.5% 399 38.3 154 15.4% 215 21.4% 34 3.4% 256 25.6% 153 15.3% 61 6.1% 42 4.2% 37 3.7% 329 31.7% 418 36.7% 268 26.8% 458 45.3% 5048 19.7%

N.A. total

1 .1%

6 .6% 1 .1%

1 .1% 6 .6% 15 1.5% 92 .4%

1000 100.0% 967 100.0% 1042 100.0% 1000 100.0% 1006 100.0% 1000 100.0% 1000 100.0% 1002 100.0% 1000 100.0% 1010 100.0% 1001 100.0% 1037 100.0% 1138 100.0% 1000 100.0% 1011 100.0% 25578 100.0%

390

masamichi sasaki, tatsuzo suzuki and masato yoneda Table 4. Crosstabulations of Question 10 by nation Q10 Including your mother tongue, what languages do you think will be essential or indispensable for international communication in the future? English

country

Brazil Argentina Korea Singapore Thailand France Germany Holland Hungary Italy Spain Portugal Russia India Indonesia Philippines Vietnam Mongolia Israel Turkey

frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency %

777 72.2% 951 85.7% 926 92.6% 976 95.0% 973 97.3% 1007 96.2% 960 96.2% 964 95.6% 957 89.8% 984 95.3% 944 94.4% 794 82.1% 912 87.5% 885 88.5% 879 87.4% 978 97.8% 977 97.7% 895 89.3% 958 95.8% 956 94.7%

Others 299 27.8% 159 14.3% 74 7.4% 51 5.0% 27 2.7% 40 3.8% 38 3.8% 44 4.4% 109 10.2% 48 4.7% 56 5.6% 173 17.9% 130 12.5% 115 11.5% 127 12.6% 22 2.2% 23 2.3% 107 10.7% 42 4.2% 54 5.3%

total 1076 100.0% 1110 100.0% 1000 100.0% 1027 100.0% 1000 100.0% 1047 100.0% 998 100.0% 1008 100.0% 1066 100.0% 1032 100.0% 1000 100.0% 967 100.0% 1042 100.0% 1000 100.0% 1006 100.0% 1000 100.0% 1000 100.0% 1002 100.0% 1000 100.0% 1010 100.0%

english as an international language in non-native settings 391 Table 4 (cont.) Q10 Including your mother tongue, what languages do you think will be essential or indispensable for international communication in the future? English Nigeria Egypt Taiwan China Japan total

frequency % frequency % frequency % frequency % frequency % frequency %

915 91.4% 881 85.0% 1032 90.7% 924 92.4% 902 89.2% 23307 91.1%

Others 86 8.6% 156 15.0% 106 9.3% 76 7.6% 109 10.8% 2271 8.9%

total 1001 100.0% 1037 100.0% 1138 100.0% 1000 100.0% 1010 100.0% 25578 100.0%

Table 5. Crosstabulations of Question 11 by nation Q11 Including your mother tongue, what languages do you think will be essential or indispensable for communication within your country in the future? English country

Brazil

frequency % • Argentina frequency % Korea frequency % Singapore frequency % Thailand frequency % France frequency % Germany frequency % Holland frequency %

651 60.5% 513 46.2% 743 74.3% 967 94.2% 626 62.6% 696 66.5% 726 72.7% 742 73.6%

Others 425 39.5% 597 53.8% 257 25.7% 60 5.8% 374 37.4% 351 33.5% 272 27.3% 266 26.4%

total 1076 100.0% 1110 100.0% 1000 100.0% 1027 100.0% 1000 100.0% 1047 100.0% 998 100.0% 1008 100.0%

392

masamichi sasaki, tatsuzo suzuki and masato yoneda

Table 5 (cont.) Q11 Including your mother tongue, what languages do you think will be essential or indispensable for communication within your country in the future? English Hungary

total

frequency % Italy frequency % • Spain frequency % • Portugal frequency % • Russia frequency % India frequency % • Indonesia frequency % Philippines frequency % Vietnam frequency % Mongolia frequency % Israel frequency % Turkey frequency % Nigeria frequency % Egypt frequency % • Taiwan frequency % • China frequency % • Japan frequency % Frequency %

633 59.4% 612 59.3% 490 49.0% 450 46.5% 202 19.4% 631 63.1% 393 39.1% 621 62.1% 850 85.0% 817 81.5% 717 71.7% 872 86.3% 801 80.0% 723 69.7% 230 20.2% 168 16.8% 460 45.5% 15334 59.9%

Others 433 40.6% 420 40.7% 510 51.0% 517 53.5% 840 80.6% 369 36.9% 613 60.9% 379 37.9% 150 15.0% 185 18.5% 283 28.3% 138 13.7% 200 20.0% 314 30.3% 902 79.8% 832 83.2% 551 54.5% 10244 40.1%

total 1066 100.0% 1032 100.0% 1000 100.0% 967 100.0% 1042 100.0% 1000 100.0% 1006 100.0% 1000 100.0% 1000 100.0% 1002 100.0% 1000 100.0% 1010 100.0% 1001 100.0% 1037 100.0% 1138 100.0% 1000 100.0% 1011 100.0% 25578 100.0%

english as an international language in non-native settings 393 16 to 49 percent. Clearly, this result is far from unanimous regarding respondents’ feelings about English usage in the future. There is certainly little question but that regional languages remain strong counterparts to English. Such languages include Spanish, Portuguese, Indonesian, Russian, Chinese, Japanese, Hindi, and Arabic (cf. Fishman 1982; AlDabbagh 2005). Table 6 shows the crosstabulations for Question 12 by nation. Question 12 simply asked respondents what language they would like their children to learn. Here too, responses were grouped into three categories: mother tongue, English, and “other major languages.” This table presents the results for those who selected English and those who did not. It is notable that nearly 88 percent of all respondents from all 25 nations selected English. Outliers are not marked on Table 6, but a close inspection reveals that Brazil, Portugal, Indonesia, and Taiwan all had respondents selecting a language other than English more than 20 percent of the time (with Taiwan at a striking 34 percent). Question 18 (Table 7) simply asked whether respondents agreed or disagreed with the statement that “English is said to be the world’s dominant or more influential language today.” Respondents from the 25 different nations agreed about 89 percent of the time. As in the previous table, outliers are not marked; however, if we select 19 percent as the cutoff point for those disagreeing, we identify the following nations: Hungary, Portugal, Russia, India, and China. These nations disagreed with the statement about English quite a bit more often than did the remaining 20 nations. The last table, Table 8, depicts the results of crosstabulations for Question 19 by nation. Respondents were asked how they felt about English being the world’s dominant or most influential language today. They could answer that (a) they think it is good that English is dominant; (b) they do not think it is good that English is dominant but they see no alternative; or (c) they do not think it is good that English is dominant and they think more should be done to use other languages. Interestingly, 59 percent chose the first and simplest answer: that it is good that English is dominant. Twenty-two percent chose the second option, while seventeen percent chose the third. Here again outliers are shown by bullets and include those nations whose respondents answered “Not good and other languages” more than 24 percent of the time. These nations include Brazil, Argentina, France, Spain, Russia, Turkey, and Egypt. Also notable in this crosstabulation are the responses of Korea, China, and Japan, with percentages for “Not good but no alternative”

394

masamichi sasaki, tatsuzo suzuki and masato yoneda Table 6. Crosstabulations of Question 12 by nation Q12 Including your mother tongue, what language, would you like your children to learn?

country Brazil Argentina Korea Singapore Thailand France Germany Holland Hungary Italy Spain Portugal Russia India Indonesia Philippines Vietnam Mongolia Israel Turkey

frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency %

English

Others

832 77.3% 938 84.5% 918 91.8% 928 90.2% 978 97.8% 971 92.7% 944 94.6% 938 93.1% 870 81.6% 925 89.6% 914 91.4% 704 72.8% 895 85.9% 912 91.2% 776 77.1% 937 93.7% 961 96.1% 899 89.7% 947 94.7% 964 95.4%

244 22.7% 172 15.5% 82 8.2% 101 9.8% 22 2.2% 76 7.3% 54 5.4% 70 6.9% 196 18.4% 107 10.4% 86 8.6% 263 27.2% 147 14.1% 88 8.8% 230 22.9% 63 6.3% 39 3.9% 103 10.3% 53 5.3% 46 4.6%

total 1076 100.0% 1110 100.0% 1000 100.0% 1027 100.0% 1000 100.0% 1047 100.0% 998 100.0% 1008 100.0% 1066 100.0% 1032 100.0% 1000 100.0% 967 100.0% 1042 100.0% 1000 100.0% 1006 100.0% 1000 100.0% 1000 100.0% 1002 100.0% 1000 100.0 1010 100.0%

english as an international language in non-native settings 395 Table 6 (cont.) Q12 Including your mother tongue, what language, would you like your children to learn?

Nigeria Egypt Taiwan China Japan total

frequency % frequency % frequency % frequency % frequency % frequency %

English

Others

959 95.8% 856 82.5% 749 65.8% 910 91.0% 825 81.6% 22448 87.8%

42 4.2% 181 17.5% 389 34.2% 90 9.0% 186 18.4% 3130 12.2%

total 1001 100.0% 1037 100.0% 1138 100.0% 1000 100.0% 1011 100.0% 25578 100.0%

Table 7. Crosstabulations of Question 18 by nation Q18 English is said to be the world’s dominant or most influential language today. Do you agree or disagree with this?

country Brazil Argentina Korea Singapore Thailand France Germany

frequency % frequency % frequency % frequency % frequency % frequency % frequency %

Agree

Disagree

N.A.

total

940 87.4% 1030 92.8% 948 94.8% 971 94.5% 953 95.3% 969 92.6% 911 91.3%

130 12.1% 80 7.2% 51 5.1% 56 5.5% 47 4.7% 67 6.4% 41 4.1%

6 .6%

1076 100.0% 1110 100.0% 1000 100.0% 1047 100.0% 1000 100.0% 1047 100.0% 998 100.0%

1 .1%

11 1.1% 46 4.6%

396

masamichi sasaki, tatsuzo suzuki and masato yoneda

Table 7 (cont.) Q18 English is said to be the world’s dominant or most influential language today. Do you agree or disagree with this?

Holland

Total

frequency % Hungary frequency % Italy frequency % Spain frequency % Portugal frequency % Russia frequency % India frequency % Indonesia frequency % Philippines frequency % Vietnam frequency % Mongolia frequency % Israel frequency % Turkey frequency % Nigeria frequency % Egypt frequency % Taiwan frequency % China frequency % Japan frequency % frequency %

Agree

Disagree

953 94.5% 858 80.5% 959 92.9% 861 86.1% 740 76.5% 838 80.4% 809 80.9% 885 88.0% 912 91.2% 946 94.6% 952 95.0% 844 84.4% 902 89.3% 956 95.5% 883 85.1% 1015 89.2% 762 76.2% 843 83.4% 22640 88.5%

55 5.5% 204 19.1% 73 7.1% 139 13.9% 224 23.2% 204 19.6% 191 19.1% 121 12.0% 88 8.8% 54 5.4% 40 4.0% 156 15.6% 108 10.7% 45 4.5% 152 14.7% 123 10.8% 225 22.5% 89 8.8% 2763 10.8%

N.A.

4 .4%

3 .3%

10 1.0%

2 .2% 13 1.3% 79 7.8% 175 .7%

total 1008 100.0% 1066 100.0% 1032 100.0% 1000 100.0% 967 100.0% 1042 100.0% 1000 100.0% 1006 100.0% 1000 100.0% 1000 100.0% 1002 100.0% 1000 100.0% 1010 100.0% 1001 100.0% 1037 100.0% 1138 100.0% 1000 100.0% 1011 100.0% 25578 100.0%

english as an international language in non-native settings 397 Table 8. Crosstabulations of Question 19 by nation Q19 What do you think about English being the world’s dominant or most influential language today?

country • Brazil • Argentina Korea Singapore Thailand • France Germany Holland Hungary Italy • Spain Portugal • Russia India Indonesia Philippines Vietnam Mongolia Israel • Turkey Nigeria

frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency % frequency %

Good

Not good but no alternative

Not good and other languages

393 41.8% 365 35.4% 412 43.5% 830 85.5% 830 87.1% 428 44.2% 511 56.1% 755 79.2% 529 61.7% 559 58.3% 256 29.7% 336 45.4% 418 49.9% 546 67.5% 664 75.0% 604 66.2% 705 74.5% 566 59.5% 645 76.4% 456 50.2% 651 68.1%

145 15.4% 280 27.2% 439 46.3% 103 10.6% 43 4.5% 217 22.4% 269 29.5% 86 9.0% 160 18.6% 209 21.8% 285 33.1% 250 33.8% 200 23.9% 177 21.9% 152 17.2% 211 23.1% 161 17.0% 221 23.2% 128 15.2% 222 24.6% 122 12.8%

399 42.4% 323 31.4% 95 10.0% 37 3.8% 80 8.4% 309 31.9% 89 9.8% 112 11.8% 167 19.5% 191 19.9% 308 35.8% 147 19.9% 220 26.3% 86 10.6% 69 7.8% 97 10.6% 80 8.5% 164 17.2% 51 6.0% 227 25.2% 183 19.1%

N.A. 3 .3% 62 6.0% 2 .2% 1 .1% 15 1.5% 42 4.6% 2 .2% 12 1.4% 7 .9%

1 .1% 20 2.4%

total 940 100.0% 1030 100.0% 948 100.0% 971 100.0% 953 100.0% 969 100.0% 911 100.0% 953 100.0% 858 100.0% 959 100.0% 861 100.0% 740 100.0% 838 100.0% 809 100.0% 885 100.0% 912 100.0% 946 100.0% 952 100.0% 844 100.0% 902 100.0% 956 100.0%

398

masamichi sasaki, tatsuzo suzuki and masato yoneda

Table 8 (cont.) Q19 What do you think about English being the world’s dominant or most influential language today?

• Egypt Taiwan China Japan total

frequency % frequency % frequency % frequency % frequency %

Good

Not good but no alternative

564 63.9% 627 61.8% 437 57.3% 283 33.6% 13367 59.0%

107 12.1% 241 23.7% 261 34.3% 483 57.3% 5172 22.8%

Not good and other languages 212 24.0% 147 14.5% 59 7.7% 40 4.7% 3892 17.2%

N.A.

5 .7% 37 4.4% 209 .9%

total 883 100.0% 1015 100.0% 762 100.0% 843 100.0% 22640 100.0%

ranging from 34 to 57 percent. Exceptionally low percentages for “Good” are seen for Argentina, Spain, and Japan. From these findings we can conclude that, in general, although the English language is regarded as the international language, how the importance of the English language is regarded in given nations has significant impact on the results for these 25 nations. Correspondence analysis is a statistical technique which is useful for those who collect categorical data; for example, data collected in social surveys. The method is particularly helpful in analyzing crosstabular data in the form of numerical frequencies and results in elegant but simple graphic displays in Euclidean space, thereby facilitating rapid understanding of the data. The correspondence analyses yielded the results shown in Figures 2a and 3. Figure 2a depicts the actual numeric positions of the 25 nations with reference to the results for Questions 7, 8, and 19. Using Questions 7, 8, and 19, we identify three clusters: A, B and C, as shown in Figure 2b. Cluster A is composed of nations whose speakers tend to use their native language when talking with foreigners, who do not think English’s dominance is good and who advocate greater use of other languages. Cluster A nations include Brazil, Argentina, Russia, France, Hungary, and Spain. Cluster B is composed of nations whose speakers tend to use their native language when talking with foreigners, and who do not think English’s dominance is good but feel that

english as an international language in non-native settings 399 .8 Brazil .6 .4 Russia Argentina Spain .2 Hungary 0.0

Nigeria

France

Egypt

Thailand Italy Mongolia Portugal Vietnam

Holland Turkey Indonesia Taiwan India Israel

Germany China

–.2 –.4

Singapore Philippines

Korea

–.6 Japan –.8 –1.0

–.5

0.0

.5

1.0

1.5

Figure 2a. .8 .6 .4

Cluster A

.2

Cluster C

0.0 –.2

Cluster B

–.4 –.6 –.8 –1.0

–.5

0.0

Figure 2b.

.5

1.0

1.5

400

masamichi sasaki, tatsuzo suzuki and masato yoneda

1.0 Singapore Israel Holland

.8 .6 .4

Egypt

Thailand

India

Philippines

.2

Indonesia Vietnam Turkey

–.0

Italy Nigeria

Brazil Portugal Russia

Germany Mongolia France

–.2

Taiwan

Hungary

China Argentina

–.4

Spain

Korea

Japan

–.6 –.6

–.4

–.2

–.0

.2

.4

.6

Figure 3.

there is no alternative. Cluster B nations include Japan, Korea, China, and Germany. Cluster C is composed of nations whose populace uses English when talking with foreigners and who think English dominance is good. Cluster C nations include Singapore, Egypt, the Philippines, India, Israel, Holland and Nigeria. Between Clusters A and B there are several nations with mixed elements, such as those that do not think that English dominance is good and that think more should be done to use other languages, and those that do not think that English dominance is good but that see no alternative. From Figure 2a, we can see that the X axis partitions the response categories of Questions 7 and 19. The positive side is composed of those nations that prefer to use English when talking with foreigners in their own nations and that think English dominance is good. The negative side is composed of those nations that use their mother tongue when talking with foreigners in their own nations and do not speak any foreign languages. The Y axis partitions those nations that think that English dominance is not good but that see no alternative (bottom), and those that also think that English dominance is not good and that think more should be done to use other languages (upper).

.8

english as an international language in non-native settings 401 Next we focus more on the environment of English language usage by introducing a different set of questions. For this analysis we used Questions 8, 10, 11, 12 and 19 (see Figure 3). Here, the X-axis can be interpreted as an indication of greater or lesser tolerance for English as a foreign language, with the negative side of the axis suggesting greater tolerance for English as a foreign language and the positive side suggesting lesser tolerance. (To estimate the degree of tolerance, Question 11, “Including your mother tongue, what languages do you think will be essential or indispensable for communication within your country in the future?” was used.) With regard to the Y-axis, we can think of this as an indication of attitudes toward English as an international language versus the use of English as it is perceived that there is no other alternative for international communication. Thus, the negative (bottom) half of the axis suggests nations that would rather see languages other than English used for international communications but that see no alternative to English, while the positive (top) half of the axis suggests nations more comfortable with English as the dominant international language and that use English when talking with foreigners in their own nations. In sum, the crosstabulation and correspondence analyses revealed three patterns: (a) nations whose speakers use their native language when speaking with foreigners, who do not think English dominance is good and who advocate greater use of other languages; (b) nations whose speakers tend to use their native language with foreigners, and who do not think English dominance is good but feel that there is no alternative; and (c) nations whose populace uses English when speaking with foreigners and who think English dominance is good. Conclusion These findings would appear to validate speculation about English being or becoming the dominant international language in our globalizing world and elucidating the actual attitudinal trends among nations regarding this phenomenon. These findings are certainly consistent with the general claim about English usage as international language. The findings further suggest that there are no immediate competitors to English presently on the horizon, indicating that English will continue to dominate international communication in the foreseeable future. As Pennycook (1994:8) has pointed out, those who use English as a foreign language are “the hardest to estimate [in numbers] but clearly

402

masamichi sasaki, tatsuzo suzuki and masato yoneda

[the] fastest growing section of world speakers of English.” As this study has shown, though, many of these English as foreign language users are not especially happy about it. Indeed, the data described here reinforce an extraordinary degree of ambivalence and/or discontent about using English. While this study has provided a wealth of new information about attitudes toward English as international language, the study too has perhaps raised more questions than it has answered. Nonetheless, this suggests that opportunities for further study should be strikingly apparent. And these opportunities include not only further analyses of the National Language Research Institute’s vast survey data but also new data gathering opportunities. References Al-Dabbagh, Abdulla. 2005. “Globalism and the Universal Language.” English Today 21(2):3–12. Ammon, Ulrich. 1994. “The Present Dominance of English in Europe.” Sociolinguistica 8:1–14. Bailey, Richard W. 1985. “The Idea of World English.” English Today 1:1–5. Baker, Colin. 1992. Attitudes and Language. Clevedon: Multilingual Matters. Barber, Bernard. 1982. “Introduction.” In Language Spread: Studies in Diffusion and Social Change, edited by Robert L. Cooper. Bloomington: Indiana University Press. Berns, Margie. 1988. “The Cultural and Linguistic Context of English in West Germany.” World Englishes 7:37–49. ——. 1992. “Sociolinguistics and the Teaching of English in Europe beyond the 1990s.” World Englishes 11:3–14. Bickley, Verner. 1982. “The International Uses of English: Research in Progress.” In English for International Communication, edited by Christopher Brumfit. Oxford: Pergamon Press. Bourdieu, P. 1976. “The School as a Conservative Force: Scholastic and Cultural Inequalities.” Pp. 192–200 in Schooling and Capitalism, edited by R. Dale, G. Esland and M. MacDonald. London: Routledge and Kegan Paul. ——. 1977. “The Forms of Capital.” In Education: Culture, Economy, Society, edited by A.H. Halsey, H. Lauder, P. Brown and A.S. Wells. Oxford: Oxford University Press. Bourdieu, P. and J.C. Passeron. 1990. Reproduction in Education, Society and Culture (2nd edition). London: Sage. Brumfit, Christopher. 1982. “English as an International Language I: What Do We Mean by ‘English’?” In English for International Communication, edited by Christopher Brumfit. Oxford: Pergamon Press. Bryson, Bill. 1990. Mother Tongue: The English Language. London: Penguin books. Cooper, Robert L. and Joshua A. Fishman. 1977. “Study of Language Attitudes.” The Bilingual Review iv:7–34. Crystal, David. 1997. English as a Global Language. Cambridge: Cambridge University Press ——. 2004. The Language Revolution. Cambridge: Polity Press. Deneire, Marc Gerard. 1993. “Democratizing English as an International Language.” Word Englishes 12:169–178.

english as an international language in non-native settings 403 Durkheim, Emile. 1938. Rules of Sociological Method. Glencoe, IL: Free Press. Essen, Arthur van. 1997. “English in Mainland Europe—a Dutch Perspective.” World Englishes 16:95–103. Firth, J.R. 1970. The Tongues of Men and Speech, 1937. Oxford: Oxford University Press. Fishman, Joshua A. 1998. “The New Linguistic Order.” Foreign Policy 113:26–34. Hertzler, Joyce O. 1965. A Sociology of Language. New York: Random House. Iwasaki, Yasufumi. 1994. “Englishization of Japanese and Acculturation of English to Japanese Culture.” World Englishes 13:261–272. Kachru, Braj B. 1985. “Standards, Codification and Sociolinguistic Realism: The English Language in the Outer Circle.” Pp. 11–30 in English in the World, edited by Randolph Quirk and H.G. Widdowson. Cambridge: Cambridge University Press. ——. 1986. The Alchemy of English: The Spread, Functions and Models of Non-native Englishes. Oxford: Pergamon Press. ——. 1991. “Liberation Linguistics and the Quirk Concern.” English Today 25:3–13. Lieberson, Stanley. 1982. “Forces Affecting Language Spread: Some Basic Propositions.” Pp. 37–62 in Language Spread, edited by Robert L. Cooper. Bloomington: Indiana University Press. Melchers, Gunnel and Philip Shaw. 2003. World Englishes. Oxford: Oxford University Press. Mesthrie, Rajend, Joan Swann, Andrea Deumert and William L. Leap. 2000. Introducing Sociolinguistics. Edinburgh: Edinburgh University Press. National Language Research Institute. 1999. Attitudes toward Japanese Language: A Tentative Codebook. Tokyo: National Language Research Institute. Noss, R.B. (ed.). 1983. Varieties of English in Southeast Asia. Singapore: Singapore University Press. Pennington, Martha C. and Francis Yue. 1994. “English and Chinese in Hong Kong: Pre-1997 Language Attitudes.” World Englishes 13:1–20. Pennycook, Alastair. 1994. The Cultural Politics of English as an International Language. London: Longman. Power, Carla. 2005. “Not the Queen’s English.” Newsweek, March 7:47–59. Pride, John B. and Liu Ru-Shan. 1988. “Some Aspects of the Spread of English in China since 1949.” International Journal of the Sociology of Language 74:41–70. Pulcini, Virginia. 1997. “Attitudes toward the Spread of English in Italy.” World Englishes 16:77–85. Quirk, Randolph. 1987. “The Question of Standards in the International Use of English.” Pp. 229–241 in Language Spread and Language Policy: Issues, Implications and Case Studies, edited by P.H. Lowenberg. Georgetown University Round Tables on Language and Linguistics. Washington, DC: Georgetown University Press. Quirk, Randolph and H.G. Widdowson (eds.). 1985. English in the World: Teaching and Learning the Language and Literatures. Cambridge: Cambridge University Press. Shim, Rosa Jinyoung. 1994. “Englishized Korean: Structure, Status, and Attitudes.” World Englishes 13:225–244. Smith, Larry E. (ed.). 1983. Readings in English as an International Language. Oxford: Pergamon Press. Spolsky, Bernard and Elana Shohamy. 1999. “Language in Israeli Society and Education.” International Journal of the Sociology of Language 137:93–114. Strevens, Peter. 1980. Teaching English as an International Language. Oxford: Pergamon Press. ——. 1982. “World English and the World’s Englishes; Or, Whose Language Is It, Anyway?” Journal of the Royal Society of Arts 5311:418–431. Suzuki, Tatsuzo, RyozoYanagihara and Masato Yoneda. 2003. “International Census on Japanese Language Usage—Sampling Design and Survey Conducted among Surveyed Nations.” Behaviometrika 58:7–29. Yong, Zhao and Keith Cambell. 1995. “English in China.” World Englishes 14:377–390. Wright, Roger. 2004. “Latin and English as World Languages.” English Today 20(4):3–13.

A NEW TEST OF CONVERGENCE THEORY Robert M. Marsh Introduction Jeffrey Alexander (1995:67) perceptively noted that modernization theory was first adumbrated, not by Parsons, Bellah, Eisenstadt, Kerr, Inkeles, Moore or Smelser. Rather, “[d]rawing from a centuries-long tradition of evolutionary and Enlightenment-inspired theories of social change, ‘modernization’ theory as such was born with the publication of Marion Levy’s [1949] book”, The Family Revolution in Modern China. Although Wallerstein (1976) declared modernization theory to be defunct, and this has been the taken-for-granted view of younger scholars ever since, I want to argue here (1) that Levy’s version of modernization theory may yet live on, unrecognized, in some versions of today’s more fashionable globalization theory, in the continuing interest in convergence among societies, economies and nations, and in the concept of the “convergence club” and (2) that one of Levy’s central propositions is empirically testable with data now available from the world’s societies. Levy’s convergence proposition is that over time, if and as the level of modernization increases, the level of structural uniformity among relatively modernized societies increases. Thus, at any given point in time, there is more structural variation among non-modernized than among modernized societies (Levy 1966:709). Let me put the difference as sharply as possible: some theories of globalization claim that to the extent societies are influenced by globalization processes there will be a growing homogenization of the world (Mattelhart 1983). The world polity/world society theory of John Meyer and his colleagues contends that in the post-World War II system of nation-states there are increasing pressures for legal, educational and other kinds of isomorphism (Boyle and Meyer 2002:77). Even peripheral societies “shift to . . . all the institutional apparatus of modern social organization” (Meyer 1980:115). In contrast, Levy would say that, despite the influence of globalization on both less and more modernized societies, the latter will converge more than the former.

406

robert m. marsh Levy’s Theory of Modernization

Levy (1966:11) defined modernization as the extent to which the members of a society use inanimate sources of power and tools to multiply the effects of their efforts. He then hypothesized that as a society’s level of modernization increases, the following changes will occur in the structure of the society. The units of society (e.g., roles, organizations) will become more specialized and therefore less self-sufficient; in other words, the degree of interdependence among units will increase. In the criteria governing role relationships and membership in organizations, there will be an increasingly universalistic ethic and predominantly universalistically defined relationships relative to particularistic ones. Levy drew upon Parsons’ pattern variables (Parsons 1951) and conceptualized a continuum from particularism to universalism. To the extent that social relationships and membership in organizations are based upon technical competence, what one can do, they are universalistic; to the extent they are based upon who one is, they are particularistic. Another structural change is an increasing degree of centralization relative to decentralization. Cognitive modernization—reliance on rationality and science in knowledge relative to non-rationality—will increase. Generalized media of exchange and markets will become more important. In less abstract terms, Levy specified other characteristics of modernized, relative to less modernized societies: longer life expectancy; increasing urbanization; education for an unknown future, since the expectation is one of continuous far-reaching change; fleeting casual contacts with strangers; less income inequality; and mass political participation. Before proceeding with Levy’s theory, I must clarify a widespread misunderstanding about the theory’s unit of analysis. The unit of analysis in world systems theory is the world system, e.g., the capitalist world economy, not individual societies. Modernization theory has often been charged with viewing societies as closed systems where the sources of change are intra-societal. To show that modernization theory is not a closed systems approach, I cite two representative statements of Levy’s. First, “the members of the relatively modernized societies with their increased facilities for communication, with their increased interest in sources of raw material, in markets, and the like [were interested] in the kind of exploration that would bring them into contact with the members of the relatively non-modernized societies” (Levy 1966:745).

a new test of convergence theory

407

Second, these contacts with the members of relatively modernized societies “made known to some of the members of relatively non-modernized societies structures previously unknown to them. Once these structures were known, . . . some would have been imported by the members of the relatively non-modernized societies” (Levy 1966:746). Levy was of course focused upon how changes worked themselves out within various modernizing societies, but he was very clear that the source of many of these changes was exogenous from the point of view of particular societies. The central proposition in Levy’s theory with which I am concerned is that if and as the level of modernization increases (a higher ratio of reliance upon inanimate energy and tools relative to animate energy), the level of structural uniformity among relatively modernized societies continually increases regardless of how diverse the original basis from which change took place in these societies may have been (Levy 1966:709). Simply stated, modernized societies will converge in each of the broad aspects of their structure noted earlier, and become more alike than is the case among non-modernized societies. An example of convergence in a demographic variable is that when total fertility is coded into 15 levels, each with a range of .50—from 8.0 to 8.49, 7.5 to 7.99, etc. down to 1.0 to 1.49—less developed nations are found in 10 of the 15 levels in 1950–1955, and in 13 of the 15 levels in 2000. In contrast, developed nations are found in only in five of the lowest fertility levels in 1950–55, and only in the three lowest levels in 2000 (Wilson 200: Table 2). In other words, over this half century, the less developed societies’ divergence in total fertility increased, while the developed societies became more alike, centered near a low mean fertility. To take another example, in the case of particularism-universalism, the hypothesis is not simply that modernized societies are relatively universalistic and non-modernized societies more particularistic. It is that the variation around universalism in modernized societies is less than the variation around particularism in non-modernized societies. More concretely, the variation around what one can do as a membership criterion in modernized societies will be less than the variation among the who one is criterion in non-modernized societies, where the substantive nature of the particularistic criteria vary greatly among such factors as race, ethnicity, gender, class, etc. Why does this convergence take place? Levy’s explanation is that modernization involves a shift from high levels of local self-sufficiency and low levels of interdependence to the opposite of these. As the units of society—roles, organizations, etc.—become more specialized, less self-sufficient, and therefore more interdependent, this interdependence

408

robert m. marsh

constrains the amount of structural variation among such units (Levy 1966:755–756). Structural isomorphism, another way to denote convergence, results from the requirement that all parts of a system maximize their intelligibility to one another. Otherwise, exchanges and communications are more problematic. The principle of isomorphism holds that social structures converge upon a common form to the extent that their interactions are frequent (Hawley 1984). Thus, modernized societies become more alike because their interactions are more frequent than those among non-modernized societies. The communications involved in these interactions call for a standardization of terms of reference, operating procedures and forms of organization. Sassen (1994), though not a modernization theorist, has pointed out that as “world cities” like New York and Tokyo become more economically interdependent, they become more culturally similar, and as the world’s major production centers, they contribute significantly to the increasing homogenization of goods and knowledge. Another mechanism of convergence can be found in demographic transition theory. Modernization typically involves significant population growth (death rates decline before birth rates). When the population size of a society increases, its division of labor also increases, and, via interdependence, reduces the number of possible forms that social arrangements may take, i.e., makes for more convergence (Manfredi 1964). Yet another set of mechanisms that produce convergence has been summarized as “the logic of industrialism” (Kerr et al. 1962; Kobrin 1977). Given modern science and its technological applications, and the efficiency and scale requirements inherent in modern methods of production and distribution, there are constraints that narrow the range of occupational and other social structural arrangements that will “work”. For example, a well-organized educational system is necessary to create a labor force appropriate to the industrial tasks to be performed. The basic idea is that the application of science and technology implies a level of standardization more universal than in non-modernized societies. This “technological imperative” argument is often dismissed by theories that see the emergence of a “post-industrial” or “postmodern” type of society that is radically discontinuous with industrial society. A deeper understanding of modernization theory can put the lie to this view. Modernization theorists were quite cognizant of long-run transitions, first from the primacy of agriculture to industrial manufacturing,

a new test of convergence theory

409

and subsequently from the later to a service and knowledge economy. When modernization researchers analyzed non-modernized and early modernizing societies, they were quite properly concerned with the first of these transitions. But the transition to a service and knowledge economy is discontinuous only if it can be shown to move in a fundamentally different direction than that specified by Levy. For example, Cardoso (1993, cited in Axford 1995:119) argued that the “third world” no longer exists, and has been replaced by a more complex division of labor based upon the ability of economies to produce goods and services related to information. I do not believe that this kind of argument renders modernization theory irrelevant. Modernization theorists emphasized educational and occupational upgrading, which can subsume new professional service and other “informational” occupations. Can the critics of modernization theory seriously claim that their postmodern and post-industrial societies repudiate Levy’s patterns of specialization and interdependence among occupations and organizations, universalism, rationality, centralization and generalized media of exchange? Credit cards, relative to money, are, after all, simply a more highly generalized medium of exchange. The extent to which “post-Fordism” has really been institutionalized is still controversial. However, there is some good evidence that “flexible specialization” has taken the place of mass production, and that there has been a shift from vertically-integrated, large-scale bureaucratic work organizations to smaller, flatter, “leaner” organizational structures. In detail, this stands in contradiction to the emphasis on standardization that was at the heart of the argument for the “logic of industrialism”. This is a sign that modernization theory needs to be revised in certain details. It is not a sign, in my view, that modernization theory is refuted. To some extent, the logic of industrialism is technology-driven (Inkeles 1998). Expressed in terms of Parsons’ (1971) concept of the cybernetic hierarchy, we can say that in the process of modernization, cybernetic elements high in information (social and cultural factors) cannot vary at random with respect to elements high in energy (technological factors). But neither Levy nor most of the other modernization theorists were technological determinists. When technology is diffused from one society to another, it passes through a cultural filter. Parsons insisted that the energic elements in the cybernetic hierarchy are necessary conditions, not controlling factors, in relation to informational elements. Despite its reliance on abstract scientific principles, technology has not a single type, but a range, of consequences for social structure, which partly

410

robert m. marsh

accounts for the fact that even the most highly modernized societies do not become identical. To some extent structural analysis has fallen into disfavor. One result of this is that it cannot be taken for granted that Levy’s strong statements about the increasing level of interdependence of parts in modernized societies, and its effects, will have meaning for today’s social scientists. To make this more concrete, let me cite two examples of interdependence noted by Levy. (1) If a society’s population has a high level of literacy and education, it will engage more in critical thinking and this will favor a democratic rather than authoritarian political system. (2) The probability of any given disruption spreading its effects to all parts of the system increases as a function of the degree of interdependence. Were he writing today, Levy could have illustrated the latter point by noting that cyberworms can wreak havoc on computer systems, spreading like a viral infection and causing local and world markets to crash. Some critics of modernization theory have argued that Levy exaggerated interdependence. They see modern societies as more loosely-coupled. But the real issue is one of the degree of interdependence. Levy does not claim that modernized societies have reached some limit of tightcoupling, but rather only that they are more tightly-coupled internally than are non-modernized societies. The critics have not provided evidence that demonstrates that modernized societies are as loosely-coupled as non-modernized societies. Earlier Research on Convergence William Form (1979) provided a thorough review of research from the 1950s to the mid-1970s on the relationship between the “technological imperative” and convergence. Although some studies Form reviewed did not support convergence, his overall summary of this body of research was that “the technological variable is cross-culturally quite robust in explaining social behavior on the job, in the factory, and in job-related social systems. At the societal level, technology strongly predicts the industrial and occupational division of labor. The literature bearing on worker resistance to factory technology clearly supports the convergence hypothesis” (Form 1979:23). Peacock, Hoover and Killian (1988) conducted a world systems analysis of convergence and divergence in economic development, measured by purchasing power parity (PPP) estimates of real GDP (RGDP) in

a new test of convergence theory

411

constant 1975 international dollars for each of 53 nations each year from 1950 to 1980. Following Wallerstein’s world systems theory (WST), societies were divided into three “zones”: the Core, the Semi-Periphery, and the Periphery. Wallerstein’s theory predicted convergence within each zone but divergence between zones. Wallerstein’s Core nations were basically the same as Levy’s modernized societies. Despite his contention that modernization theory was dead, Wallerstein’s WST made the same prediction as Levy of convergence among Core nations. They differed in that Wallerstein expected as much convergence within the Semi-Periphery and within the Periphery as within the Core, whereas Levy theorized more convergence among modernized than among nonmodernized (Semi-Periphery and Periphery) societies. Peacock, Hoover and Killian found that over the period from 1950 to 1980 there was increasing convergence in income shares relative to population among the Core nations, but this did not happen among the nations within the zones of the Semi-Periphery or the Periphery, where the trend was toward greater divergence. In other words, within-zone inequality of RGDP decreased only among the Core (most modernized) nations. Their findings thus support Levy more than Wallerstein. Because the neoclassical theory of economic growth predicted at least conditional convergence of national income per capita across the world’s economies (Solow 1956), economists and some sociologists have continued to study the question of convergence in between-nation per capita income and in income inequality. While income is only one of the aspects of social structure with which Levy’s theory is concerned, this body of research has some important implications. Some economists study what they call sigma-convergence: the cross-sectional dispersion of a variable, measured by the coefficient of variation, V, the ratio of a variable’s standard deviation to its mean. Convergence occurs if V, the dispersion, declines over time. Boyle and McCarthy (1999) found that V scores for GDP per capita for a wide group of World Bank countries from 1960 to 1992 show convergence, albeit at a slow rate, for “High Income” and “Upper Middle Income” countries, very slow to negligible convergence among “Lower Middle Income” countries, and divergence for “Low Income” countries. Similar findings were reported by Baumol, Blackman and Wolff (1989) and by Jones (1997:147–148). In the former study, while the V scores for variation in real GDP per capita in more developed societies were declining between 1950 and 1980, those for 23 less developed societies increased from 0.36 in 1950 to 0.53 in 1980. Using time series data from 1960 to 1990 for 113 market economies, Li

412

robert m. marsh

(1999) also found convergence in per capita income among the richest countries and divergence among the lowest income countries. Barro and Sala-i-Martin (1992) show that among the most modernized 20 OECD nations, the initially richer nations did grow more slowly, thus making for convergence. These studies are all consistent with Levy’s hypothesis of greater convergence among modernized than among non-modernized societies. Throughout his career Alex Inkeles studied convergence in societies both more and less modernized. His 1998 book, One World Emerging?, brought together a number of these researches. They range widely from the comparison of particular nations, e.g., the United States and the former Soviet Union, to more extensive cross-national comparisons of institutions like the family and schools, of occupational prestige hierarchies, and of due process guarantees written into national constitutions, and also to psychological similarities and differences in individual modernity, happiness, etc. Inkeles’s substantial contributions include the following. First, in every study he was at pains to distinguish (1) true convergence, where the societies being compared come to some single point with very little variation around their central tendency, from (2) instances of parallel rather than convergent change, when the societies remain as dispersed around a new norm as they were around the older norm, and (3) instances where the societies diverge over time and become even less alike. Second, Inkeles stressed how important it is to discover which forces induce convergent change and which resist convergence. He argues that cultural traditions and national history are major sources of divergence even among highly modernized societies. Third, Inkeles also differentiated his convergence theory from the world polity/world society theory of John Meyer and his colleagues (Meyer and Hannan 1979). The Meyer group argues that “[t]he intensification and expansion of world society and culture in the post-World War II framework, and the creation of many new nation-states highly dependent on this framework, have greatly increased world pressures for legal [and other kinds of ] isomorphism” (Boyle and Meyer 2002:77). The world polity is a system of rules legitimating the extension of authority of rationalized nationstates to control and act on behalf of their populations. Even peripheral societies “shift to modern forms of industrial and service economic activity; to modern state organizations; to modern educational systems; to modern welfare and military systems; in short to all the institutional apparatus of modern social organization” (Meyer 1980:115).

a new test of convergence theory

413

Inkeles disagreed and sided more with Levy’s view. Although the diffusion of modern patterns around the world is undoubtedly important, modern structural characteristics will be developed and adopted only if certain specific conditions are met, namely, that the societies are advanced industrial societies facing similar internal problems, e.g., an aging population. A pointed example of the contrast between theories is that Boli-Bennett (1980), a member of the Meyer group, stressed the convergence in the degree of state dominance in states both in non-modernized and modernized societies. Inkeles and Levy, in contrast, would call attention to the fact that among non-modernized societies, failed states and predatory states are found as well as stronger states, making for more diversity in state structures than among modernized societies. Thus, unlike the Meyer group, Inkeles, like Levy, expects convergence to occur more in modernized than in non-modernized societies. Comparisons limited to the advanced industrial societies that have shown convergence include Williamson and Fleming (1977). Over time, these societies had increases in: life expectancy for females at age 1, number of physicians per million population, number of social security programs available, number of years of compulsory education, and decreases in crude birth rates. They used V, the coefficient of variation, to measure convergence, and found that over time these industrial societies became more alike on each of these variables. The question of convergence and divergence is, thus, not the exclusive concern of modernization theory. As I have indicated, it has been central to world systems theory, world polity/world society theory, studies of world income distribution by economists, and it has had a new lease on life in various types of globalization theory. A further sign of its salience is the fact that the entire October–December 2005 issue of the International Journal of Comparative Sociology is devoted to “Convergence and Divergence in the Contemporary World System,” with articles both theoretical and empirical. Testing Levy’s Convergence Theory of Modernization To identify as “modern” societies those that are developing “modern institutions and lifestyles” and then to claim that these very societies are “converging” in various ways toward a common core of structural characteristics is tautological. I avoid tautology in the way Levy did, namely, by defining modernization by a deliberately narrow, objective,

414

robert m. marsh

technological variable. Levy’s (1966:11) conceptual definition of modernization was the extent to which the members of a society use inanimate sources of power and tools to multiply the effects of their efforts. Energy consumption per capita is one way to measure this, but I shall instead measure level of modernization by a variable that had a + .719 correlation with energy consumption in 1990, namely, the number of telephone mainlines per thousand population in a society. Caplow and Finsterbusch (1964), Barbera (1973) and others have also used telephones in this way. Telephones are a highly reliable indicator: all telephone systems have directories and these must be fairly accurate if the system is to function at all. Telephone statistics are based on direct enumeration and a self-verifying list. Indeed, for some countries, the numerator (telephones) is more accurately known than the denominator (population size). Another advantage is that in 1990 there were data on telephones for more societies (N = 201) than for energy consumption (N = 122). Having defined and measured a society’s level of modernization in this way, I can now treat as an empirical question the extent to which various economic, demographic, technological, political and social structural variables vary with modernization. Data and Measures The cross-societal variables used in this analysis come from the following data sets: Taylor and Jodice, World Handbook of Political and Social Indicators III: 1948–1977 (1983); United Nations Development Programme (hereafter UNDP), Human Development Report 2005 and various other years; World Bank, 2005 World Development Indicators and other years; Freedom House, Freedom in the World: The Annual Survey of Political Rights and Civil Liberties, various years; United Nations, Demographic Yearbook, various years; Bollen (1980); and Paxton (2004). Since convergence is a process that is theorized to occur in a variety of aspects of society, I selected variables that indicate a society’s level of economic development, its capitalist market economy, demographic patterns, technology, state and political democracy, cognitive modernization of its population, health, income inequality and poverty, gender patterns, and information and communications, as well as a composite measure of modernization. Some relevant variables could not be used because data were not available for many societies. A society’s number of telephone mainlines per thousand population at two points in time, 1965 and 1990, is my indicator of level

a new test of convergence theory

415

of modernization. It is a continuous variable. In 1965 it ranged from less than one telephone per thousand population (in Rwanda) to 481 (USA). In 1990 it ranged from less than one telephone per thousand population (Kampuchea) to 815 (Monaco). In the analysis to follow I shall divide all societies for which there are data on telephones into two categories: relatively non-modernized and relatively modernized. Although I am transforming a continuous variable—modernization, telephones—into a binary distinction, I follow Levy and emphasize the continuous variation between the polar extremes (e.g., Kampuchea and Monaco) by using the term relatively prior to each of the two types of societies. As Levy (1966:708) put it, “the distinction between ‘relatively non-modernized’ and ‘relatively modernized’ emphasizes at one and the same time the possibility of variations and the inevitability of common elements” as between the two types of societies. To avoid the selection bias problem of selecting societies on the basis of their ex post level of modernization, I categorize societies based upon their ex ante modernization level, at the beginning of the sample period, 1965. Thus, only the 21 societies with 100 or more telephone mainlines per thousand population in 1965 are categorized as relatively modernized. For the second ex ante time point, 1990, I take into account the rapid diffusion of telephones since 1965 and raise the minimum to 200 telephone mainlines per thousand population as the requirement for a society to be categorized as relatively modernized. Even with the higher cutting-point, of the 200 societies1 with data on telephones in 1990, the number that was relatively modernized had increased from 21 to 52. The other 148 societies with less than 200 telephones per thousand population are treated as relatively non-modernized. Information describing each of the variables to be used to test convergence theory will be presented below as each variable is introduced in the analysis based on Tables 1 and 2. In those tables two statistical measures are presented for each variable. The first is the mean, which allows us to compare the variable as between the relatively non-modernized and the relatively modernized societies, or between two points in time. For example, the first row in Table 1 shows that in 2003 the mean GDP per capita (in purchasing power parity or PPP $US) was 1 Of the 200 societies in the World Bank’s World Development Indicators data set for 1990, by 2005, 186 are independent states; the other 14 are overseas territories of France (e.g., French Polynesia), the Netherlands (e.g., Aruba), Denmark (e.g., Greenland), the United Kingdom (e.g., Bermuda), the United States (e.g., Puerto Rico), or a Special Administrative Region of the People’s Republic of China (e.g., Hong Kong).

416

robert m. marsh

$4,967 in the less modernized societies and $24,625 in the more modernized societies. The second statistic is the measure of convergence among societies. In comparing the relative homogeneity of less modernized and more modernized societies, for many development variables, as in the case of the variable, GDP per capita, just noted, the two categories of societies have very different means, and therefore it is misleading to compare the absolute magnitudes of the standard deviations (s.d.) from the mean. Although modernized societies have a larger s.d. for GDP per capita than less modernized societies, we would not infer from this that there is more variation among modernized than among less modernized societies. Instead, we are interested in the size of the s.d. relative to the size of the mean. The measure of relative variability I shall use is the coefficient of variation, V, which is the ratio of the standard deviation to the mean value of a variable. V is an absolute measure of dispersion in the sense that it is independent of the unit employed; it is a scale-invariant index of convergence. We can use it to compare the variabilities of distributions of variables from widely different domains. As Boyle and McCarthy (1997) point out, convergence can be measured by simply tracking the intertemporal change in the coefficient of variation for each cross-societal variable. Results Comparing Non-Modernized and Modernized Societies The first version of Levy’s convergence hypothesis that I shall test is: because processes of convergence in the past have been occurring more among relatively modernized than among non-modernized societies, it follows that at any given point in time, societies that are more modernized will have higher levels of structural uniformity than societies that are less modernized. This hypothesis is tested in Table 1. Societies are divided into the relatively non-modernized, which had fewer than 200 telephones per thousand population in 1990, and the relatively modernized, with 200 or more telephones. The question I then ask is: roughly ten years later—in or around the year 2000, depending on the year(s) for which data are available—is the variation among the less modernized societies in fact greater than that among the more modernized?

a new test of convergence theory

417

Table 1. Variation since 1990 of structural characteristics of societies that were relatively non-modernized or relatively modernized in 1990 Variable and Source

Level of Economic Development GDP/c (in PPP constant 2002 $ US) 2003a Energy efficiency (GDP per unit of energy use in PPP constant 2000 $ US per kg. Oil equivalent) 2002a Capitalist Market Economy Gross capital formation (% of GDP) 2000c Value of exports of goods and other market Services (% of GDP) 2000c Trade in goods (sum of merchandise exports and imports as % of GDP, in constant $ US) 2000c High-tech exports (% of manufacturing exports) 2000c Agriculture value added (% of GDP) 2000c Industry value added (% of GDP) 2000c Services value added (% of GDP) 2000c Labor Force and Occupational Structure % of the labor force unemployed 2000c Demographic Patterns Urban population (% of total population) 2003a Fertility rate (births per woman) 2005a Infant mortality rate (per 1,000 live births) 2003a Population under age 15 (% of total population) 2003a Population over age 65 (% of total population) 2003a

Relatively Non-Modernized Societies (<200 telephones/1,000 pop, 1990)

Relatively Modernized Societies (200+ telephones/1,000 pop, 1990)

N

Mean

V*

N

Mean

V*

122

$4,967

.831

38

$24,625

.410

$4.54

.531

33

$5.74

.329

133

21.33%

.354

37

23.68%

.253

137

38.02%

.618

37

54.14%

.596

143

65.10%

.513

39

82.69%

.694

106

7.59%

41

19.42%

.813

136

20.77%

.695

34

3.73%

.804

134

28.93%

.444

35

28.85%

.241

135

50.20%

.280

35

67.10%

.101

57

10.44%

.831

39

7.00%

.567

132

48.89%

.434

40

75.25%

.218

129

3.60

.461

38

1.65

.272

132

51.11

.768

39

6.03

.583

129

35.03%

.255

38

18.62%

.188

129

4.37%

.653

38

11.59%

.323

84

1.65

418

robert m. marsh

Table 1 (cont.) Variable and Source

Relatively Non-Modernized Societies (<200 telephones/1,000 pop, 1990) N

Technology Cellular telephone subscribers (per 1,000 population) 2003a The State and Political Democracy Government current tax revenue (% of GDP) 2002c General government consumption expenditure (% of GDP) 2000c Political democracy (0 = lowest, 100 = highest) 1995d Political rights index (0 = least free, 7 = most free) 2000e Civil liberties index (0 = least free, 7 = most free) 2000e Cognitive Modernization Public education expenditure, four levels of schooling (% of GDP) 2002a Public education expenditure (% of total government expenditure) 2002a Combined gross enrollment ratio for primary, secondary and tertiary education 2002–2003a1 Enrollment ratio, tertiary education 2000c Patents granted to residents (per million population) 2002a Health Public health expenditure (% of GDP) Public & private health expenditure /capita (US $)

117

Mean

170.49

Relatively Modernized Societies (200+ telephones/1,000 pop, 1990)

V*

N

Mean

V*

1.136

35

752.54

.222

48

22.18%

.432

27

33.18%

.293

130

15.50%

.503

37

18.75%

.231

130

58.78

.532

37

87.88

.283

132

3.99

.521

38

6.53

.207

132

3.97

.391

38

6.16

.198

95

4.69%

.545

35

5.58%

.256

71

16.24%

.356

24

14.19%

.258

128

64.84%

.262

40

92.23%

.128

87

18.96%

.810

39

51.37%

.329

64

7.69

2.275

37

120.03

1.503

2002c

145

3.13%

.542

42

2002c

145

$119

1.237

42

5.56% $1743

.319 .686

a new test of convergence theory

419

Table 1 (cont.) Variable and Source

Relatively Non-Modernized Societies (<200 telephones/1,000 pop, 1990) N

Number of physicians (per 100,000 population) 1990–2004a1 Income Inequality and Poverty Income share of top 10% of households 1990–2002c1 Ratio of income of richest 20% to poorest 20% 1992–2000a1 Gini coefficient of income inequality 1992–2000c1 % of pop living below national poverty line 1990–2002a1 % of pop living below 50% of median income 1999–2000a1 Gender: Particularism-Universalism Primary school net enrollment ratio (female as % of male) 2002–2003a1 Secondary school net enrollment ratio (female as % of male) 2002–2003a1 Tertiary school net enrollment ratio (female as % of male) 2002–2003a1 Female % of combined gross enrollment ratio for three levels of education 2003a Male % of combined gross enrollment ratio for three levels of education 2003a Females as % of total labor force 2003a Female labor force participation (% of male rate, age 15+) 2003a Female profession al and technical workers (% of male) 1992–2001a1

131

Mean

110.51

Relatively Modernized Societies (200+ telephones/1,000 pop, 1990)

V*

N

Mean

1.13

39

288.03

V*

.406

22

34.04%

.224

9

24.90%

.115

92

12.00

.846

30

5.83

.284

92

42.75

.246

30

32.76

.158

72

39.13%

.445

–

–

–

–

–

–

22

10.00%

.312

116

95.2%

.092

37

99.89%

.011

98

96.7%

.216

38

102.26%

.028

111

98.44%

.516

36

138.11%

.319

124

63.35%

.313

37

95.97%

.135

124

65.59%

.237

37

90.49%

.111

122

53.62%

.279

38

51.15%

.191

122

65.94%

.262

38

71.26%

.184

49

46.57%

.322

35

50.06%

.186

420

robert m. marsh

Table 1 (cont.) Variable and Source

Relatively Non-Modernized Societies (<200 telephones/1,000 pop, 1990) N

Ratio of estimated female to male earned income 2002a Gender-related development index 2003a Female % of seats in parliament 2004a Gender empowerment 1992–2004a1 Information and Communications Newspaper circulation (per 1,000 population) 2000c Radios (per 1,000 population) 2000c TV sets (per 1,000 population) 2000c Information and communications technology expenditure (% of GDP) 2003c Information and communications technology expenditure ($ US per capita) 2003c Composite Measures of Development Human development index 2003a

Mean

Relatively Modernized Societies (200+ telephones/1,000 pop, 1990)

V*

N

Mean

V*

115

.507

.278

36

.581

.210

104

.630

.263

34

.915

.041

.654

38

.253

33

126 46

13.35% .470

21.70% .705

.489 .171

29

61.84

.784

29

245.57

.540

33

268.50

.648

8

1,025.65

.549

148

168.13

.887

52

527.83

.392

.418

26

.831

26

.244

40

42 43 132

5.50% $165.12 .649

6.29% $1,736.27 .908

.290 .442 .051

* V, the coefficient of variation, is the standard deviation divided by the mean. 1 When a range of years, rather than a single year is listed, the year of each country’s latest survey or data point within the time range is used. Sources: a) United Nations Development Programme (UNDP), Human Development Report, various years (www.hdr.undp .org). c) World Bank, World Development Indicators, various years (www.worldbank.org). d) Bollen-Paxton: Kenneth Bollen (1980) on pp. 387–388 published 1965 democracy scores for 123 societies. Pamela Paxton, using Bollen’s measure of democracy, assigned 1995 democracy scores to 170 societies, and kindly provided the author with them in a personal communication on April 20, 2004. e) Freedom House, Freedom in the World: The Annual Survey of Political Rights and Civil Liberties, various years.

a new test of convergence theory

421

The first variable in Table 1 supports Levy, since it shows that in the year 2003, the variation, V, in GDP per capita is less among the modernized (.410) than among the non-modernized societies (.831). The much lower mean level of economic development of the less modernized societies is associated with greater variation in their GDP per capita. Among the modernized societies with higher GDP per capita, the uniformity in their GDP per capita is also higher. The second economic development variable provides a measure of a society’s energy efficiency by “showing comparable and consistent estimates of real GDP across countries relative to physical inputs (units of [commercial] energy use)” UNDP 2005: 355). Flavin and Durning (1988) noted that “[m]ost market-economy countries did improve their energy efficiency by between 20 and 30 percent” between 1973 and the mid-1980s (cited in Welsh and Butorin 1990:370). The GDP produced per unit of energy used was $5.74 in the modernized societies, somewhat higher than the $4.54 in the non-modernized societies: the modernized societies were more energy efficient. The variation is greater in non-modernized (.531) than in modernized societies (.329), so Levy’s hypothesis is again confirmed. The next domain concerns aspects of a society’s capitalist market economy. Both modernization and globalization theories recognize that given the demise of socialist command economies, the principal type of economy in an increasing number of societies is some form of a capitalist market economy. The first variable is gross capital formation as a percent of GDP, a standard economic variable thought to affect subsequent economic growth. Levy (1966:767) saw the need for capital formation as a basis for further increases in productivity. The mean for gross capital formation is somewhat higher in the more modernized societies (23.68% of GDP) than in the less modernized (21.33%), and V is higher for the less modernized societies (.354) than for the more modernized societies (.253). The next variables concern international trade flows, and measure a society’s integration into the world economy. The mean value of exports of goods and other market services as a percent of GDP is much higher in modernized (54.14%) than in non-modernized societies (38.02%), and, as Levy predicted, the variation in this variable is greater in the non-modernized (.618) than in the modernized societies (.596). The next variable is trade in goods—the sum of merchandise exports and imports as a percent of GDP. While the mean for trade in goods is higher in modernized societies (82.69%) than in non-modernized societies (65.10%),

422

robert m. marsh

contrary to Levy, the variation is greater among modernized (.694) than among non-modernized societies (.513). Consider next high-technology exports as a percent of all manufacturing exports. High-technology exports are products with high R & D intensity in such industries as aerospace, computers, pharmaceuticals, scientific instruments, and electrical machinery. As expected, modernized societies have more high-tech exports (19.42% vs. 7.59%), and modernized societies have less variation (.813) than non-modernized societies (1.65). The next three variables give the value added (as percent of GDP) respectively for the agricultural, the industrial, and the service sectors of a society’s economy. (The sum of the three sectors is 100% for value added.) The means for these variables show a familiar pattern: agriculture value added is much higher in non-modernized societies; industry value added is almost the same in non-modernized as in modernized societies; and services value added—increasingly the hallmark of advanced capitalist economies—is higher in the modernized societies. The expected pattern of more variation among non-modernized societies holds for both industry and services value added, but agricultural value added varies more among the modernized (.804) than among the non-modernized societies (.695). In summary, Levy’s hypothesis fares quite well with regard to the capitalist market economy variables, with five out of seven confirmations. As a labor force and occupational structure variable, we next consider the percent of the labor force unemployed as of the year 2000. The unemployment rate is lower in the more modernized societies, and Levy is supported by the fact that the variation in unemployment is less (.567) in modernized than in non-modernized societies (.831). We turn next to demographic variables. The urban population is a much higher percent of the population in modernized than in non-modernized societies, and the variation in this variable supports Levy: the .218 in the modernized is lower than the .434 in the non-modernized societies. The number of births per woman is much higher (3.60) in non-modernized than in modernized societies (1.65). Levy is confirmed because the variation in fertility rate is higher (.461) in non-modernized than in modernized societies (.272). The infant mortality rate is much higher in the non-modernized societies, and, once again, it is the non-modernized societies that have more variation (.768 vs. .583). Two other demographic variables have to do with the age structure of a population. Less modernized societies, at an earlier stage of the demographic transition, tend to have higher rates of population growth,

a new test of convergence theory

423

which makes for a higher proportion of the population under 15 years old and a lower proportion over age 65 than in modernized societies. The means for these two variables in Table 1 confirm this. Moreover, the more modernized societies have less variation than the less modernized in the percent under age 15 (V = .188 vs. .255) and also less variation in the percent of the population over age 65 (.323 vs. 653). All five of the demographic variables thus support Levy. Since the late 1990s cellular telephones have been even more of a technology growth industry than regular telephones. But the growth in cell telephone subscribers per thousand population was so disproportionate that by 2003, the more modernized societies had a mean of 753 and the non-modernized only 170. Among the non-modernized societies there was much more variation (1.136) around the low mean for cell phones, while the modernized societies tended toward more uniformity (.222) around the mean for their much more numerous cell phones. The next section of Table 1 tests Levy’s hypothesis with regard to the state and political development. As for the state, Levy (1966:17–18) theorized that “the further the process of modernization goes, the greater is the degree of centralization that must be achieved if stability is to exist”. It is the increasing level of interdependency of the parts and of the members of modernized societies that accounts for the increase in centralization. The state increases in significance relative to other types of organizations in society. The first variable used to test this is the government’s current tax revenues, excluding grants, as a percent of GDP in 2002. The assumption is that stronger, more centralized states have larger tax revenues. Table 1 shows that, as Levy expected, tax revenues are larger (33% of GDP) in modernized than in non-modernized societies (22%), and the variation in tax revenues is greater in the non-modernized societies (.432) than in modernized societies (.293). The second variable is general government consumption expenditure as a percent of GDP in 2000, another common indicator of state strength. Levy is confirmed: the mean is higher and the variation is lower (.231) in modernized societies than in non-modernized societies (.503). As for political development, Levy argued that although modernized societies can and do achieve higher levels of centralization than in nonmodernized societies, non-authoritarian forms of centralization displace some of the authoritarian forms. Mass political participation increases, as does democracy in the following sense: although even democratic societies, like most societies, remain hierarchically organized, especially if they are large-scale, “certain rights and privileges of individuals of

424

robert m. marsh

lower ranks in the hierarchy are protected from or guaranteed by those of higher ranks in the hierarchy” (Levy 1966:299–300). To measure political democracy, I first present Paxton’s 1995 democracy scores, derived from her use of Bollen’s (1980) well-known index. This index includes three indicators of popular sovereignty through the electoral process (fairness of elections, whether the chief executive is elected, and the effectiveness of the means of selecting the legislature); and three indicators of political liberties (freedom of the press, freedom of group opposition, and government sanctions). Bollen democracy scores range from a low of 0 to a high of 100. As much of the literature has already demonstrated, the mean democracy scores are much higher in modernized than in non-modernized societies (88 vs. 59). The variation in democracy scores is less (.283) in modernized than in nonmodernized societies (.532). The Freedom House indexes of political rights and civil liberties both range from a low of 0, least free, to a high of 7, most free. Both show the clear trend toward more freedom in the modernized than in the non-modernized societies. The difference in variation for political rights supports Levy: the .521 among nonmodernized is greater than the .207 among modernized societies. The same is true for civil liberties, where the variation is greater among the non-modernized (.391) than among the more modernized societies (.198). All five of the state and democracy variables support Levy. Levy’s abstract Parsonsian proposition held that modernization brings about an increasing reliance on rationality and science as cognitive standards relative to traditionalism in knowledge and thinking processes. This does not mean that non-rational and unscientific thinking disappears, even in highly modernized societies. It only means that the ratio of rational and scientific cognition to its opposite type will increase. Data that would precisely test this idea of cognitive modernization are hard to come by. The closest approximation I can provide in Table 1 is with variables concerning formal education expenditures and enrollments, and a measure of patent innovations. The first of these variables is public educational expenditure for pre-primary, primary, secondary and tertiary schooling, as a percent of GDP in 2002. It includes both capital expenditures on construction, renovation, etc. and current expenditures on goods and services, e.g., salaries, books, etc. Modernized societies have only a slight edge over non-modernized societies in this regard (6% vs. 5%), but Levy is confirmed by the fact that the modernized societies are more uniform in their educational spending (V = .256 vs. .545). The same public spending on four levels

a new test of convergence theory

425

of schooling variable is next measured as a percent of total government expenditure, rather than as a percent of GDP. Although the mean for public education spending as a percent of all government spending is actually lower in the modernized societies, the variation in this variable supports Levy: it is higher among non-modernized (.356) than among modernized societies (.258). The next two variables concern educational enrollments. I present a composite variable, the combined gross enrollment ratio for primary, secondary and tertiary education. The modernized societies have a much higher mean gross enrollment ratio, and also less variation in enrollment ratios (.128) than the non-modernized societies (.262). Given the need for highly educated professional and technical personnel, and for scientific activity, in modernized societies, it is especially important to see if tertiary education enrollment ratios support Levy. Enrollment is much higher (51% vs. 19%) in the modernized societies, and the variation in tertiary education enrollments in modernized societies (.329) is much less than in the non-modernized societies (.810). To measure scientific and technological innovation, I present the number of patents granted to residents (as opposed to non-residents) per million population in 2002. Levy is supported: far more patents are granted to residents in modernized than in non-modernized societies (120 vs. 8 per million population), and while there is much variation around the mean in both modernized and non-modernized societies, the former are more uniform in this respect (V = 1.503) than the non-modernized societies (2.275). Levy does well in the area of cognitive modernization, with all five tests confirming his hypothesis. In the area of health, my first variable is public health expenditures as a percent of GDP in 2002. Public health spending rises from a mean of 3.13% of GDP in non-modernized to 5.56% in modernized societies, and Levy is confirmed in that the variation is greater (.542) in nonmodernized than in modernized societies (.319). Next, we consider public and private health expenditures per capita (in current US dollars in 2002). Modernized societies spend much more on health than non-modernized societies: $1,743 vs. $119, and the variation in this spending is greater in the non-modernized (1.237) than in the modernized societies (.686). The third health variable is the number of physicians per 100,000 population. Physicians include graduates of a faculty or school of medicine who are working in any medical field (UNDP 2005:359). The results for physicians again support Levy: Although there are 288 of them per 100 thousand population in modernized societies, in contrast to 111 in

426

robert m. marsh

non-modernized societies, the uniformity is greater in the modernized (.406) than in the non-modernized societies (1.13). All three health variables confirm the convergence hypothesis. The next set of variables tests Levy’s hypothesis as it bears on income inequality and poverty. Levy (1966:265–279) hypothesized that there is a shift from a relatively unequal distribution of income in non-modernized societies toward a more equal distribution in modernized societies. An implication of this is that poverty rates are higher in less modernized societies. The first measure of income inequality is the income share of the top ten per cent of households. In the period from 1990 to 2002, as hypothesized, the richest decile received 34% of the income in non-modernized societies, in contrast to only 25% in the modernized societies, and the variation was less (.115) in the latter societies than in the former (.224). Levy is also confirmed in regard to the ratio of the income of the richest 20% to that of the poorest 20%. This ratio declines from 12 in non-modernized societies to 6 in modernized societies, and the variation is less in the modernized (.284) than in the non-modernized societies (846). The Gini coefficient of income inequality also supports Levy. The modernized societies have less income inequality (mean Gini = 32.8) than the non-modernized (42.8), and the variation in Gini coefficients is greater in the non-modernized (.246) than in the modernized societies (.158). The data on poverty rates in the UNDP 2005, Tables 3 and 4 are measured in a somewhat different way for the non-modernized than for the modernized societies. For the former, it is the “national poverty line”, the percent of the population living below the poverty line as defined by the authorities in each country. For the modernized societies, the measure is “the percent of the population living below 50% of the median adjusted disposable household income”. Keeping this difference in mind, we can tentatively conclude that the percent living in poverty is much higher in the non-modernized (39%) than in the modernized societies (10%). And, as expected by Levy, the variation in poverty rates is greater in non-modernized (.445) than in non-modernized societies (.312). The four tests for convergence in inequality and poverty thus all support Levy’s hypothesis. I have conceptualized the next domain of variables in Table 1 as Gender: Particularism-Universalism. In these tests I am putting words in Levy’s mouth, but I believe I am making a correct inference from his theory as to what he would have said in the area of feminist theory and modernization. As noted above, in the criteria governing role relation-

a new test of convergence theory

427

ships and membership in organizations, to the extent that membership is based upon what one can do, it is universalistic; to the extent that it is based upon who one is, it is at the particularistic pole. Membership particularism can be based upon gender, among other things. Levy’s 1966 book was written at the very beginning of the movement for feminism and feminist theory. It has an entry for “sex”, but no entries for “gender” or “feminism” in its index. Women are discussed mainly in connection with “sex role differentiation”. For example, in “every society the inculcation of role differentiation on the basis of sex is always an actual barrier . . . to puristically universalistic treatment of individuals” (Levy 1966:659). This can be re-stated as: sex role differentiation is a barrier to universalistic criteria for membership in organizations because it favors one sex more than the other. Thus, I believe I am in no way distorting Levy’s theory when I claim that if he were writing today, starting from his stated proposition that as modernization increases, societies will develop an increasingly universalistic ethic and predominantly universalistically defined relationships relative to particularistic ones, he would have gone on to logically derive from this the further proposition that insofar as organizational membership criteria of a universalistic type take the place of those of a particularistic type, women’s participation in various types of organizations and occupations, and their rewards for such membership, will approximate more closely those of men. It is regrettable that most feminist theorists are either oblivious of, or hostile toward, modernization theory. In the matter of the relationship of gender to modernization, it is logically inconsistent to favor feminism while rejecting modernization theory. The first five tests of Levy’s hypothesis concern gender differences in participation in schooling at various levels. Primary school net enrollment for females as a percent of males has a mean of 95% in non-modernized societies, and virtually 100% in modernized societies. Levy is confirmed in that the variation is less (.011) in modernized than in non-modernized societies (.092). Secondary school net enrollment shows females almost as likely as males (97%) to be enrolled, in non-modernized societies, and to be more likely than males in modernized societies (102%). The smaller variation in modernized societies (.028) than in non-modernized (.216) again supports Levy. At the tertiary school level the mean net enrollment shows this pattern even more markedly: in non-modernized societies, women participate in higher education 98% as much as men, and do so much more than men in modernized societies (138%). There is

428

robert m. marsh

more variation of tertiary enrollment in non-modernized (.516) than in modernized societies (.319). The next pair of variables offers a summary measure: a comparison of the percent of females to the percent of males in the combined gross enrollment ratio for primary, secondary, and tertiary levels of education. There are two axes of comparison: gender and modernization level. Gender does not produce large differences in enrollment: almost the same percent of women (63%) as of men (66%) in non-modernized societies are enrolled in these three levels of education, and a somewhat larger percent of women (96%) than of men (90%) in modernized societies are enrolled. The large differences are due to modernization, not gender. The enrollment rates for women rise dramatically from 63% in nonmodernized to 96% in modernized societies, and the male enrollment rates similarly undergo a sharp rise from 66% in non-modernized to 90% in modernized societies. The variation in modernized societies is less than in non-modernized societies both for females (.135 vs. .313), and for males (.111 vs. .237). Looking at the educational enrollment variables by level of education as a whole, the remarkable fact is perhaps how little gender particularism there is in either non-modernized or modernized societies. To see this, understand that if universalistic standards of membership totally displaced particularistic standards, the mean for each cell would be 100%. Women would be 100% of men, i.e., as frequent as men, in each level of education and in all levels taken together. The departures from the 100%, or gender equality, standard for each level of education favor males by not more than five percentage points in non-modernized societies, and in modernized societies the departures from 100% are even smaller at the primary and secondary levels, and in the only large departure from 100%—that at the tertiary level—it is women whose participation greatly exceeds men’s. Women increasingly seek gainful occupations and participate in the labor force. The next variable, females as a percent of the labor force, actually shows a higher participation rate for women in non-modernized (54%) than in modernized societies (51%). More to the point, Levy is confirmed in that there is more variation in women’s labor force participation in non-modernized (.279) than in modernized societies (.191). A different way to get at the relationship between gender and work is by the female labor force participation rate as a percent of the male rate. The mean rates show a different pattern than for the previous variable, females as a percent of the total labor force. Women’s rate in non-modernized

a new test of convergence theory

429

societies, 66% of that for men, rises to 71% of men’s rate in modernized societies. The larger variation for this variable in non-modernized societies (.262) than in modernized societies (.184) supports Levy. In analyzing gender particularism vs. universalism in the occupational realm, it is not enough to consider only women’s labor force participation. Women could have a high rate of participation, but compared to men’s, this could be in lower status, less desirable jobs. I therefore compare next where on the particularism-universalism scale women’s participation in higher status occupations fits. The variable is female professional and technical workers as a percent of males. Women are in professional or technical occupations 47% as often as men in non-modernized societies, and this rises to 50% in modernized societies. Since there is less variation in modernized (.186) than in non-modernized societies (.322), Levy is supported. What is the ratio of estimated female earned income to estimated male earned income? Universalism is the extent to which women’s income approximates men’s. Although the ratio is not high in either non-modernized or modernized societies, it is higher in the latter (.581) than in the former societies (.507). The variation is greater in the non-modernized (.278) than in the modernized societies (.210). As a composite index, the United Nations Development Programme constructed a gender-related development index (GDI), which measures “average achievement in the three basic dimensions captured in the human development index—a long and healthy life, knowledge and a decent standard of living—adjusted to account for inequalities between men and women” (UNDP 2005:356). The index can range from 0 to 1. Women’s development on these three dimensions comes much closer to men’s in modernized (.915) than in non-modernized societies (.630), and the variation supports Levy because it is greater (.263) in non-modernized than in modernized societies (.041). Having examined women’s education, occupations, and income relative to men’s, we next consider a measure of women’s power: the percent of seats in parliament (a lower or single house or an upper house or senate, where appropriate) held by women. The percent of seats held by women rises from 13% in non-modernized to 22% in modernized societies, and Levy is confirmed in that there is more variation in this variable in the non-modernized societies (.654 vs. .489). There is also a UN composite index of gender empowerment, which gets at gender inequality in three basic dimensions of empowerment: economic participation and decision-making, political participation and

430

robert m. marsh

decision-making, and power over economic resources (UNDP 2005:356). The index can range from 0 to 1. It shows that women have a much higher mean level of empowerment in modernized (.705) than in non-modernized societies (.470), and the variation among modernized societies (.171) is less than that in non-modernized societies (.253). On every one of the 12 tests of gender particularism-universalism, Levy’s hypothesis is confirmed. The next conceptual domain in Table 1 concerns information and communications. Implied in what we have already seen in Levy’s theory is that when larger-scale non-modernized societies attempt to be highly centralized, especially in generalized ways rather than for only a limited purpose, they tend to become unstable. This is due mainly to the fact that they have not yet attained a high level of (mass) communications. Communications through the mass media (newspapers, radios, television, etc.) are more extensive in modernized societies, as is seen in the next three variables. Newspaper circulation per thousand population increases from 62 in non-modernized societies to 246 in modernized societies, while Levy’s predicted greater variation is seen more in non-modernized (.784) than in modernized societies (.540). Modernized societies have 1,026 radios per thousand population, in contrast to 269 in non-modernized societies, and the variation is less in the former than in the latter (.549 vs. .648), which confirms Levy. There are 528 TV sets per thousand people in modernized and only 168 in non-modernized societies, and the latter also have more variation in TV sets (.887 vs. .392). The World Bank’s 2005 World Development Indicators includes another communications variable relevant to modernization: information and communications technology (ICT) expenditures. This includes external spending on information technology products purchased by businesses, households, governments, and educational institutions from vendors or organizations outside the purchasing entity; internal spending on internally customized ICT software, capital depreciation, etc.; and spending on telecommunications and other office equipment. These ICT expenditure data cover the world’s 55 largest buyers of such technology among countries and regions, accounting for 98 percent of global ICT spending. The first of these ICT variables in Table 1 is ICT expenditures as a percent of GDP. As expected, it reveals that these expenditures are a higher percent of GDP in modernized (6.29%) than in non-modernized societies (5.50%). The variation among non-modernized societies (.418) is greater than that among modernized societies (.290). When these ICT expenditures are expressed in US$ per capita rather than as a percent of GDP, the

a new test of convergence theory

431

mean for the modernized societies is much higher ($1,736) than for the non-modernized ($165); the variation supports Levy: .831 in nonmodernized in contrast to .442 in modernized societies. Levy’s hypothesis is consistently supported in all five of the information and communications variables. The last variable in Table 1 is the UNDP’s composite human development index (HDI), which measures the same three basic dimensions of human development introduced earlier in the gender-related development index—a long and healthy life, knowledge, and a decent standard of living. Whereas the gender-related index gets at the inequalities between men and women on these variables, the HDI does this for each society’s population as a whole. Like the gender-related index, HDI can range from 0 to 1. As expected, the HDI is much higher in modernized (.908) than in non-modernized societies (.649), and the greater uniformity in modernized (.051) than in non-modernized societies (.244) also supports Levy. Strictly speaking, Table 1 is a series of tests of similarities and differences among societies at a single point in time, not a test of convergence. Convergence tests require data over time. It is now time to supplement the static picture with the longitudinal data that alone can demonstrate actual convergence. We shall test the second version of Levy’s convergence hypothesis: among societies already relatively modernized, if and as their level of modernization increases further, the level of structural uniformity among them will also increase. The minimum requirement for testing convergence is that we have data for at least two points in time, which can be called time t1 and time t2, respectively. Since convergence, if it occurs at all, may require not just a few years, but decades, t1 and t2 should be as far apart as data availability permits. Another difference from Table 1 is that the hypothesis now refers only to societies that were already relatively modernized at time t1. Since convergence is expected to occur among modernized societies (as they continue to modernize further), we are no longer concerned with the relatively non-modernized societies. In Table 1 my cutting-point for modernization level was 200 telephone mainlines per thousand population as of 1990. In Table 2, to meet the extended time line requirement, I define as modernized only those societies that as early as 1965 already had 100 or more telephone mainlines per thousand population, putting aside the much large number of societies that had fewer than 100 telephones in 1965. Only 21 of the 126 societies for which telephone data are available (Taylor and Jodice 1983) were modernized as of 1965: Australia, Austria, Belgium,

432

robert m. marsh

Canada, Czechoslovakia, Denmark, Finland, France, the German Federal Republic (West Germany), Iceland, Israel, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Sweden, Switzerland, the United Kingdom and the United States. Time t1 for each variable in Table 2 is 1965 or as soon after 1965 as data are available (e.g., 1970, 1975, or 1980). Time t2 is the most recent time for which data are available, e.g., the 1990s, 2000 or later. Table 2 is a true test of convergence for at least two reasons. First, the level of modernization among the 21 societies did in fact continually increase: the mean number of telephones rose from 235 in 1965 to 468 in 1990 and 574 in 2003. Second, by comparing the measure of variation (V) at time t1 with that for time t2 for each variable in Table 2 we can discover whether these 21 societies already modernized at t1 become more convergent, or alike, between t1 and t2. Table 2. Did the 21 societies already modernized in 1965 become more alike as they reached higher levels of modernization during the period since 1965? Variable and Source

N

Mean

V*

20 21

$17,486 $28,656

.157 .190

20 21

$4.26 $5.60

.317 .290

1975b 2000c

20 21

24.70% 21.96%

.198 .115

1980c 2000c

20 21

33.70% 46.01%

.551 .651

1990c 2003c

19 20

48.53% 61.02%

.487 .579

1990c 2000c

19 21

15.11% 19.85%

.618 .413

Level of Economic Development GDP/c (in PPP constant 2002 $ US) 1975a 2002a Energy Efficiency (GDP per unit of energy use in PPP constant 2000 $ US per kg. oil equivalent) 1990a 2002a Capitalist Market Economy Gross Capital Formation (as % of GDP) Value of exports of goods and other market services (as % of GDP) Trade in goods (sum of merchandise exports and imports as % of GDP, in current $ US) High-tech exports (as % of manufacturing exports)

a new test of convergence theory

433

Table 2 (cont.) Variable and Source

N

Mean

V*

1980c 2000c

18 19

5.17% 3.06%

.554 .765

1980c 2000c

18 19

36.74% 29.67%

.107 .190

1980c 2000c

18 19

58.09% 67.27%

.068 .089

Labor Force and Occupational Structure % of the labor force unemployed 1990–92c1 2000–02c1

20 21

5.43% 5.92%

.610 .393

21 21

73.88% 80.00%

.157 .138

21 21

2.186 1.695

.222 .218

21 21

17.95 4.29

.270 .246

21 21

25.75% 18.52%

.141 .172

21 21

12.62% 12.90%

.193 .165

21 19

17.71 803.21

.959 .193

20 18

35.26% 36.01%

.175 .230

20 21

17.50% 19.79%

.409 .207

Agriculture value added (as % of GDP) Industry value added (as % of GDP) Services value added (as % of GDP)

Demographic Patterns Urban population (as % of total population)

1975a 2003a Fertility rate (births per woman) 1970–75a1 2005a Infant mortality rate (per 1,000 live births) 1970a 2003a Population under age 15 (as % of total population) 1970c 2003a Population 65+ years old (as % of total pop) 1980f 2003a Technology Cellular telephone subscribers (per 1,000 population) 1990a 2003a The State and Political Democracy Government tax revenue (as % of GDP) 1970b 2002c General government consumption expenditure (as % of GDP) 1970b 2000c

434

robert m. marsh

Table 2 (cont.) Variable and Source Political democracy (0 = lowest, 100 = highest) 1965d 1995d Political rights index (0 = least free, 7 = most free) 1973e 2000e Civil liberties index (0 = least free, 7 = most free) 1973e 2000e Cognitive Modernization Public education expenditure, four levels of schooling, (as % of GDP) 1978b 2002a Public education expenditure (as % of total government expenditure) 1990a 2002a School enrollment ratio, primary and secondary levels (as % of population 5–19 years old) 1970b 1977b % of 20–24 year-olds enrolled in tertiary education 1965c 1992c Patents granted to residents (per million population) 1998a 2002a Health Public health expenditure (as % of GNP) 1978b (as % of GDP) 2002a Number of physicians (per million population) 1970b 1990–2004a1 Income Inequality Income share of top 10% of households 1962–73b1 2000c Income share of bottom 40% of the population 1950–75b1 1987–97c1

N

Mean

V*

21 19

94.06 97.43

.183 .055

21 21

6.52 7.00

.203 .000

21 21

6.48 6.57

.211 .091

21 20

6.36% 5.87%

.228 .213

11 17

13.13% 12.65%

.221 .188

21 21

87.14% 88.48%

.083 .088

19 17

15.37% 41.12%

.478 .286

21 21

169.90 161.48

1.219 1.177

21 21

3.81% 6.68%

.545 .137

21 21

1,491 3,329

.232 .317

16 7

26.07% 24.38%

.153 .119

16 19

17.84% 21.17%

.195 .168

a new test of convergence theory

435

Table 2 (cont.) Variable and Source

N

Mean

V*

15 19

37.90 30.82

.196 .163

19 21

40.07% 61.32%

.514 .324

19 21

38.25% 49.26%

.379 .267

21 21

38.76% 43.66%

.117 .073

15 19

4.26% 2.92%

.753 .579

15 19

6.66% 5.34%

.575 .487

15 19

17.53% 13.99%

.376 .341

15 19

38.77% 36.52%

.137 .141

15 19

77.41% 82.52%

.116 .066

15 19

52.98% 57.51%

.145 .095

1970b 2000c

20 19

342.95 293.74

.335 .456

1980c 2000c Information and communications technology expenditure (as % of GDP) 2000c 2003c

20 21

395.97 617.95

.274 .214

19 19

7.22% 6.55%

.210 .210

Gini coefficient of income inequality

1965–70b1 1992–2000c1

Gender: Particularism-Universalism Gross enrolment in tertiary education—% females 1990c 2000c Gross enrolment in tertiary education—% males 1990c 2000c Females as % of total labor force 1980c 2000c % females in agricultural occupations 1990–92c1 2000–02c1 % males in agricultural occupations 1990–92c1 2000–02c1 % females in industrial occupations 1990–92c1 2000–02c1 % males in industrial occupations 1990–92c1 2000–02c1 % females in service occupations 1990–92c1 2000–02c1 % males in service occupations 1990–92c1 2000–02c1 Information and Communications Newspaper circulation (per 1,000 population) TV sets (per 1,000 population)

436

robert m. marsh

Table 2 (cont.) Variable and Source

N

Mean

V*

Information and communications technology expenditure ($ US per capita) 2000c 2003c

19 19

$1,748 $1,951

.403 .341

Composite Measures of Development Human Development Index 1975a 2000a

20 20

.851 .937

.024 .014

* V, the coefficient of variation, is the standard deviation divided by the mean. 1 When a range of years, rather than a single year is listed, the year of each country’s latest survey or data point within the time range is used. Sources: a) United Nations Development Programme (UNDP), Human Development Report, various years (www.hdr.undp.org). b) WHPSI: Taylor and Jodice (1983). c) World Bank, World Development Indicators, various years (www.worldbank.org). d) Bollen-Paxton: Kenneth Bollen (1980) on pp. 387–388 published 1965 democracy scores for 123 societies. Pamela Paxton, using Bollen’s measure of democracy, assigned 1995 democracy scores to 170 societies, and kindly provided the author with them in a personal communication on April 20, 2004. e) Freedom House, Freedom in the World: the Annual Survey of Political Rights and Civil Liberties, various years. f ) UNDY: United Nations Demographic Yearbook. UN Statistics Division, Department of Social and Economic Affairs, New York (http://unstats.un.org/unsd/demographic/default.htm).

Let us now consider the results of Table 2. In the interests of space, since the variables in Table 2 are mostly the same as those in Table 1, I shall simply enumerate (1) the variables that support Levy, by showing convergence between time t1 and time t2, followed by (2) the variables that do not show convergence, thereby disconfirming Levy. Evidence for Convergence among Modernized Societies Among the 21 societies already modernized in 1965, the following changes show convergence over the subsequent time period. (1) Between 1990 and 2002, as energy efficiency (the GDP produced per unit of energy consumed) increased, the variation in energy efficiency among the 21 highly modernized societies decreased from V = .317 to .290. In other words, these societies became more alike in their energy efficiency. Among the seven capitalist market economy variables, only two support Levy. (2) Gross capital formation as a percent of GDP actually declined

a new test of convergence theory

437

in these highly modernized societies, but more to the point, the variation in this variable declined from .198 in 1975 to .115 in 2000. (3) High-tech exports as a percent of all manufacturing exports increased between 1990 and 2000, while the variation on this variable decreased from .618 to .413. (4) The variation among these 21 highly modernized societies in the percent of the labor force that was unemployed declined from .610 in 1990–1992 to .393 a decade later. Four of the five demographic changes support Levy. (5) As the percent of the total population that was urban increased over the long time span from 1975 to 2003, the variation in this variable declined. (6) Over this same time period, as the mean fertility rate declined, convergence in fertility rates increased slightly. (7) Infant mortality rates declined precipitously, while the variation decreased from .270 in 1970 to .246 in 2003. (8) The 21 societies became more homogeneous between 1980 and 2003 in the percent of their population over 65 years old. Thus, as these modernized societies continued to increase their level of modernization, they also became more alike in their level of urbanization, fertility and infant mortality rates, and percent of the population more than 65 years old. (9) The technology variable, cellular telephone subscribers per 1,000 population, is a text book example of Levy’s convergence hypothesis. Between 1990 and 2003, cell phones increased from 18 to 803 per thousand people. Over the same 13-year period, the uniformity in cell phone frequency also increased markedly, as is shown by the sharp decline in variation from .959 to .193. Convergence is the common pattern for the state and political democracy variables. (10) As general government consumption expenditures (as a percent of GDP) rose in the 21 modernized societies, variation in this measure of state strength declined from .409 in 1970 to .207 in 2000. (11) The Bollen democracy scores, already high in 1965, increased somewhat over the subsequent three decades, and convergence in democracy scores also increased. The level of political rights (12) and of civil liberties (13), also already high in 1973, rose higher by 2000, and over this period variation declined significantly from .203 to .000 for political rights, and from .211 to .091 for civil liberties. Among the cognitive modernization variables, Levy’s hypothesis is confirmed in four of the five tests. (14) Public education spending for four levels of education ass a percent of GDP converged slightly from .228 in 1978 to .213 in 2002. (15) The modernized societies also became

438

robert m. marsh

more alike in public education expenditures as a percent of total government spending. (16) The percent of 20–24-year-olds enrolled in tertiary educational institutions sharply increased over the period from 1965 to 1992, and concurrent with this the variation declined from .478 to .286. (17) These 21 societies varied greatly in both 1998 and 2002 in the number of patents granted to residents per million population: the standard deviation was larger than the mean. But Levy is supported in that the degree of variation declined somewhat, from 1.219 to 1.177. (18) Public health expenditures as a percent of GNP or GDP rose between 1978 and 2002, while the variation in these expenditures declined from .545 to .137. Convergence is seen for all three of the income inequality variables. (19) The mean share of national income going to the top ten percent of households declined somewhat over the period from 1962–1973 to 2000, and the variation in this respect also declined, from .153 to .119. (20) The earliest year for which data are available on the income share received by the bottom 40 percent of the population was between 1950 and 1973; the latest year in the World Bank data ranges from 1987 to 1997. When we compare these two time periods, the income share going to the bottom 40 percent of the population increased while the variation in his variable decreased from .195 to .168. (21) The mean Gini coefficient of income inequality fell between 1965–1970 and 1992–2000, while the variation decreased from .196 to .163. Thus, as the modernized societies of 1965 became even more modernized during the subsequent decades, they became (a) less unequal in all three of these measures of income distribution, and (b) more homogeneous in these aspects of income distribution. Levy is confirmed in all but one of the nine tests for gender particularism-universalism. (22) Between 1990 and 2000, the modernized societies dramatically increased the mean in the percent females in the gross enrollment in tertiary education. The female percent rose from 40% to 61%, while the variation decreased from .514 to .324. (23) The percent of males enrolled in tertiary education also increased, but less markedly: from 38% to 49%. Variation in the percent males declined from .379 to .267. (24) From 1980 to 2000, females as a percent of the total labor force increased, while the variation decreased from .117 to .073. Six variables chart the participation of women and of men in agricultural, industrial and service occupations. In these modernized

a new test of convergence theory

439

societies, it is not surprising that for both sexes the rate of employment declines in the agricultural and industrial sectors and increases in the service sector. What is noteworthy is that convergence increased in five of the six variables. Specifically, (25) the variation in the percent of women employed in agriculture declined from .753 in 1990–1992 to .579 in 2000–2002, (26) the variation in men’s employment in agriculture dropped from .575 to .487; (27) the variation in women’s employment in industrial occupations fell from .376 to .341; (28) the variation in women’s employment in service occupations declined from .116 to .066; and (29) the variation in men’s employment in service occupations decreased from .145 to .095. Thus, shifts in employment between the agricultural, industrial and service sectors over the decade of the 1990s, when examined by gender, show almost uniformly that the modernized societies were becoming more homogeneous as they continued the upward course of their modernization. The four information and communications variables show two confirmations of Levy. (30) As the number of TV sets per thousand population increased dramatically between 1970 and 2000, the variation decreased from .274 to .214. (31) Information and communications technology expenditures in US dollars per capita increased during the brief period from 2000 to 2003, and variation decreased from .403 to .341. Finally, (32) these modernized societies’ mean Human Development index scores rose from .851 in 1975 to .937 in 2000, and over that quarter of a century, the variation decreased from .024 to .014. What remains to report are the less numerous instances of disconfirmation of Levy in Table 2. Once again, in the interests of space, I list them briefly. The general picture for these variables is that although the 21 modernized societies were becoming more modernized during the time period from t1 to t2, more importantly they do not give evidence of convergence. Instead, the variation increases, which indicates these modernized societies were diverging, becoming less alike. Levy is thus disconfirmed by the fact that convergence does not increase between t1 and t2 for the following variables: GDP per capita, the value of exports of goods and services as a percent of GDP, the sum of merchandise exports and imports as a percent of GDP, value added by agriculture, by industry and by services as a percent of GDP, and the percent of the population under age 15. Nor did convergence increase for the following variables: government tax revenue as a percent of GDP, school enrollment ratios at the primary

440

robert m. marsh

and secondary levels, and number of physicians per million population. One gender particularism-universalism variable failed to show increases in convergence: the percent of males in industrial occupations. Finally, Levy’s convergence hypothesis was not confirmed for newspaper circulation and for information and communications technology expenditures as a percent of GDP. To conclude the analysis of the data, one question is: how many of the tests of Levy’s convergence hypothesis in both Tables 1 and 2 were confirmed? Table 1’s 51 variables provide 51 tests, Table 2’s variables an additional 45, for a total of 96 tests. Of these 96, convergence was demonstrated in 81, i.e., 84.4% were confirmations. Since only Table 2 compared data over time and was therefore a stricter test of convergence, we may give it more weight. Of its 45 tests, 32, or 71.1%, were confirmed. These confirmation rates provide rather strong support for the convergence theory of modernization. Discussion To put my main findings in the appropriate context, I distinguish Levy’s general modernization theory from his specific proposition about convergence. What Levy presented as general modernization theory has many similarities to what the United Nations Human Development Programme in 2005 called Millennium Development Goal Indicators. These “end-states” included such development goals as enlarging people’s choices, a long and healthy life, acquiring knowledge, having access to resources needed for a decent standard of living, and achieving gender equality. These UN “human development indicators” are being monitored on a regular basis to assess how well each nation is achieving these goals. Contrary to what in some scholarly quarters has been called the demise of modernization theory, modernization as “the development project” is still taken quite seriously in other quarters. In such parts of the world as the Asian tigers, and more recently China and India, “the idea of development remains a powerful inspiration” (Edelman and Haugerud 2005:50). Inglehart and Welzel (2005) have formulated a revised theory of modernization and tested it with data over time from the World Values Surveys. The data presented in this paper support Levy’s general modernization theory in the sense that they show in Table 1 that the mean level of development around the year 2000 is higher among the relatively

a new test of convergence theory

441

modernized than among the relatively non-modernized societies. This is true for level of economic development, measures of capitalist market economy, demographic variables, the state and political democracy, most measures of education and cognitive modernization, health, income inequality and poverty, most measures of gender universalism, mass media of communications, information technology, and the composite human development index. The same support for Levy’s general modernization theory was shown in Table 2, where we saw that among the 21 societies already modernized as of 1965, as their level of modernization (measured by telephones per thousand population) continued to rise in subsequent years, their mean level of development on most variables moved in the predicted direction. The main concern of this paper, of course, has been not Levy’s general modernization theory but his convergence proposition, that if and as the level of modernization increases, the level of structural uniformity among relatively modernized societies will also increase, and therefore, at any given point in time, there is more structural variation among nonmodernized than among modernized societies. This proposition has been largely confirmed by my data. Some scholars rejected modernization theory because it saw social change as linear (Segal 1992). I have not disaggregated my data into a yearly time series. Were I able to do this, I should probably observe for given variables annual fluctuations up and down around the trend line. Summarizing their year-by-year data from 1950 to 1981 on V scores for real GDP per capita among modernized societies, Baumol, Blackman and Wolff (1989:310) note that “the secular movement toward convergence may reverse itself at least in the short or intermediate term.” But my data are nevertheless linear in the sense that between the two points in time (which are for the most part separated by from ten to twenty or more years), the trend does move in the predicted linear direction. Does the ratio of confirmations to disconfirmations vary among the domains of social structure considered in Tables 1 and 2? There are theoretical reasons for expecting some domains to show more convergence than others. Fukuyama (1992) contended that convergence would take the form of global economic liberalism (the collapse of state socialism) and of global liberal democracy. Some of his critics countered with the argument that “the urge to be more whole-hearted consumers of the products of Western capitalism weighed more heavily than any love of democratic values” (Axford 1995:46; Offe 1991) and

442

robert m. marsh

therefore we should expect less convergence in political democracy than in liberal capitalism. I have variables from the domains of both political democracy and the capitalist market economy in Tables 1 and 2. Although Offe and Axford contend there is more convergence in the capitalist market than in democracy, my results show the opposite: more convergence in the democracy variables (6/6 or 100%) than in the capitalist market variables (7/14 or 50%). Inkeles (1998:19–24) proposed a classification of five major “elements” that can be ordered, on the basis of our research findings, from greater to lesser degrees of convergence. Most convergent are “modes of production and patterns of resource allocation”, by which Inkeles refers to Levy’s criteria of increasing reliance on inanimate sources of energy, science and technology. Next most convergent are “institutional arrays, forms and processes” like the increasing levels of rational bureaucratic modes of organization. Less convergent are (1) patterns of social relationships in the areas of kinship, class and leisure, and (2) systems of popular attitudes, values and behavior. Most divergent are systems of political and economic control. For example, France and the United States show many convergent patterns, but continue to diverge in such respects as the centralization of the French and the decentralization of the American state. To ask whether my variables in Tables 1 and 2 support Inkeles presupposes that my variables and domain classifications can be mapped onto Inkeles’ five domains. In some ways they cannot. I have a domain for demographic variables, which Inkeles omits altogether. The most congruent my variables can be made with Inkeles’ domain of “institutions” is by selecting variables from three institutional sectors in Tables 1 and 2: education (in my cognitive modernization domain), health, and the mass media (in my information and communications domain). My domains of gender particularism-universalism and of income inequality and poverty correspond somewhat to Inkeles’ “social relationships of kinship and class” domain. I have no variables for Inkeles’ domain of “popular attitudes and values”. For Inkeles’ domain of “modes of production and resource allocation”, I can use my capitalist market economy variables and the energy efficiency variable. Finally, Inkeles’ “political and economic control” domain can be at least partially indicated by my state and political democracy domain. Ranking Inkeles’ domains according to his view of their descending order of convergence, we have:

a new test of convergence theory

443

Inkeles

My Domains

Confirmations

Modes of production, resource allocation

Capitalist market economy & energy efficiency

Institutions

Education, health, mass media

15/18 = 83%

Social relationships: class, kinship

Inequality, gender

27/28 = 96%

Popular attitudes, values

None

Political and economic control

State and political democracy

9/16 = 56%

9/10 = 90%

The domain Inkeles thinks is least convergent, political and economic control, is the domain where I find one of the highest levels of convergence (90%). My least convergent domain is the capitalist market economy (56%), where Inkeles expected the greatest convergence. The two domains Inkeles believed show intermediate convergence (institutions and social relationships) have high levels of convergence in my data. Thus, my data are consistent with the general idea that the extent of convergence is higher in some areas of social structure than in others. But there is little agreement between the differentials in convergence in my data and those in various theories that assert which domains have greater or lesser convergence. Further research is clearly needed on this crucial point. Table 1 compared non-modernized with modernized societies at a given point in time, and showed a very high rate of confirmation of the convergence hypothesis (49/51 or 96% of the tests). Relative to the less modernized societies, the modernized societies do seem to constitute a “convergence club” (Bsaumol, Blackman and Wolff 1989; Li 1999). Table 2, which compared the 21 societies already modernized in 1965 as they moved toward even higher levels of modernization in the subsequent decades, is a stricter test of convergence, and it showed convergence less often (32/45 or 71% of the tests). Thus, even among the societies called members of the “convergence club,” there is evidence on some variables of divergent rather than convergent change. Because the time interval between t1 and t2 is not the same for all the variables in Table 2, I computed the rate of convergence or divergence by taking the mean change per year for each variable. Industry value added has

444

robert m. marsh

the highest rare of divergence. Its V increased 78%, from .107 in 1980 to .190 in 2000, or 3.9% per year on average over the twenty years. The variables with the next highest rates of divergence were agriculture value added (1.9% per year), services value added (1.5%), trade in goods (1.5%), and number of physicians per million population (1.4%). On these and some other variables, the “convergence club” was less homogeneous in 2000 than it had been in 1970 or 1980. At the other extreme, the variables that show the most rapid convergence are: cell telephones (6.1% mean decline in variation per year), information and communications technology expenditure in US dollars per capita (5.1%), per cent of women in service occupations (4.3%) political rights and per cent of women enrolled in tertiary education (3.7% each), per cent of males in service occupations (3.5%), high tech exports (3.3%), and public health expenditures (3.1%). It was in these areas that the 21 most modernized societies most rapidly became more homogeneous. Note that while the variables with the most rapid rates of divergence are mostly clustered in only one of my conceptual domains—the capitalist market economy—the variables that converged most rapidly show little clustering. Instead, they are spread through several domains: technology, gender particularism-universalism, political democracy, the capitalist market economy, and health. An important task for future research is to explain why some of these aspects of social structure rapidly converge, while other rapidly diverge. I conclude by suggesting the implications of my findings supporting Levy’s convergence theory for current theories of globalization. During the Cold War, many students of development worked within the paradigm that grouped societies as First, Second or Third World. The collapse of the Communist Second World has been accompanied by claims that the divide between the First World (modernized, capitalist societies) and the Third World (poor and developing societies) is also obsolete. “The conception of an interdependent, interacting, global manufacturing system cuts across the old view of a world consisting of nation-states. . . . But the new world that has superseded it . . . does not lend itself to the simple identification of . . . rich and poor, industrialized and non-industrialized” [and modernized and non-modernized] (Harris 1986:200). Many globalization theorists now hold this view, and talk of worldwide convergence. My findings that variation in social structure, very broadly understood, is greater among less modernized than among more modern-

a new test of convergence theory

445

ized societies go against this view. Convergence, I suggest, is not some uniform process operating throughout the world. Rather, the fact of great variability among the less developed societies means that while some are converging toward the patterns of the developed world, others are diverging. Moreover, the fact that convergence is greater among modernized societies means that few of them are lagging so markedly below the mean for modernized societies that they are “easy targets” for the more successfully developing societies to overtake. Finally, my findings also agree with the literature suggesting that the OECD and some other countries constitute a “convergence club”, i.e., a set of societies that experience convergence in relation to one another, but not in relation to the broader constellation of poorer, less modernized societies. References Alexander, Jeffrey C. 1995. “Modern, Anti, Post and Neo.” New Left Review 210:63–101. Axford, Barrie. 1995. The Global System. New York: St. Martin’s. Barbera, Henry. 1973. Rich Nations and Poor in Peace and War. Lexington, MA: D.C. Heath. Barro, Robert J., and Xavier Sala-i-Martin. 1992. “Convergence.” Journal of Political Economy 100(2):223–251. Baumol, W.J., S. Blackman, and E.N. Wolff. 1989. Productivity and American Leadership: The Long View. Cambridge, MA: MIT Press. Boli-Bennett, John. 1980. “Global Integration and the Universal Increase of State Dominance, 1910–1970.” Pp. 77–107 in Studies of the Modern World-System, edited by Albert Bergesen. New York: Academic Press. Bollen, Kenneth. 1980. “Issues in the Comparative Measurement of Political Democracy.” American Sociological Review 45(3):370–390. Boyle, Elizabeth H., and John W. Meyer. 2002. “Modern Law as a Secularized and Global Model.” Pp. 65–95 in Global Prescriptions, edited by Yves Dezalay and Bryant G. Garth. Ann Arbor, MI: University of Michigan Press. Boyle, G.E., and T.G. McCarthy. 1997. “A Simple Measure of B-Convergence.” Oxford Bulletin of Economics and Statistics 59(2):257–264. ——. 1999. “Simple Measures of Convergence in Per Capita GDP.” Applied Economic Letters 6(6):343–347. Caplow, Theodore, and Kurt Finsterbusch. 1964. A Matrix of Modernization. New York: Bureau of Applied Social Research, Columbia University. Cardoso, Fernando H. 1993. “North-South Relations in the Present Context: A New Dependency?” In The New Global Economy in the Information Age, edited by Martin Carnoy, Manuel Castells, Stephen Cohen and Fernando H. Cardoso. Basingstoke, PA: Penn State Press. Edelman, Marc, and A. Haugerud. 2005. The Anthropology of Development and Globalization. Malden, MA: Blackwell. Flavin, C., and A. Durning. 1988. “Raising Energy Efficiency.” Washington, DC: Worldwatch Institute. Form, William. 1979. “Comparative Industrial Sociology and the Convergence Hypothesis.” Annual Review of Sociology 5:1–25.

446

robert m. marsh

Freedom House. 2003 and other years. Freedom in the World: The Annual Survey of Political Rights and Civil Liberties. New York: Freedom House; New Brunswick, NJ: Transaction Publishers. Fukuyama, Francis. 1992. The End of History and the Last Man. London UK: Hamish Hamiton. Harris, Nigel. 1986. The End of the Third World. London: Tauris. Hawley, Amos. 1984. Human Ecology: A Theoretical Essay. Chicago, IL: University of Chicago Press. Inglehart, Ronald, and Christian Welzel. 2005. Modernization, Cultural Change, and Democracy. Cambridge, UK: Cambridge University Press. Inkeles, Alex. 1998. One World Emerging? Convergence and Divergence in Industrial Societies. Boulder, CO: Westview Press. Jones, Charles I. 1997. “Convergence Revisited.” Journal of Economic Growth 2:131–153. Kerr, Clark, J.T. Dunlop, F.H. Harbison, and C.A. Myers. 1962. Industrialism and Industrial Man. Cambridge, MA: Harvard University Press. Kobrin, Stephen J. 1977. “Industrialization and Variation in Social Structure: An Empirical Test of the Convergence Hypothesis.” Proceedings of the Industrial Relations Research Association, 29th Meeting (1976):177–185. Levy Jr., Marion J. 1949. The Family Revolution in Modern China. Cambridge, MA: Harvard University Press. ——. 1966. Modernization and the Structure of Society. Princeton, NJ: Princeton University Press. Li, Qing. 1999. “Convergence Clubs: Some Further Evidence.” Review of International Economics 7(1):59–67. Manfredi, John F. 1964. “Societal Complexity and Limited Alternatives.” Pp. 159–164 in Fact and Theory in Social Science, edited by E.W. Count and G.T. Bowles. Syracuse, NY: Syracuse University Press. Mattelart, Armand. 1983. Transnationals and the Third World: The Struggle for Culture. South Hadley, MA: Bergin and Garvey. Meyer, John W. 1980. “The World Polity and the Authority of the Nation-State.” Pp. 109–135 in Studies of the Modern World System, edited by Albert Bergesen. New York: Academic Press. ——, and Michael Hannan. 1979. National Development and the World System. Chicago IL: University of Chicago Press. Offe, Claus. 1991. “Capitalism by Democratic Design?” Social Research 58(4):42–54. Parsons, Talcott. 1951. The Social System. Glencoe, IL: The Free Press. ——. 1971. The System of Modern Societies. Englewood Cliffs, NJ: Prentice-Hall. Peacock, W.G., G.A. Hoover, and C.D. Killian. 1988. “Divergence and Convergence in International Development.” American Sociological Review 53:838–852. Sassen, Saskia. 1994. Cities in a World Economy. Thousand Oaks, CA: Pine Forge Press. Segal, David R. 1992. “From Convergence to Chaos.” Research in Social Movements, Conflicts and Change 14:1–11. Solow, Robert M. 1956. “A Contribution to the Theory of Economic Growth.” Quarterly Journal of Economics 70:65–94. Taylor, Charles L., and David A. Jodice. 1983. The World Handbook of Political and Social Indicators III: 1948–1977. Cologne, Germany: Zentralarchiv für Empirische Sozialforschung. United Nations Department of Social and Economic Affairs. Statistical Office. 1981–1985. Demographic Yearbook. New York: United Nations. United Nations Development Programme. 2005 and various years. Human Development Report. New York: Oxford University Press. Wallerstein, Immanual. 1976. “Modernization: Requiescat in Pace.” Pp. 131–135 in The Uses of Controversy in Sociology, edited by Lewis A. Coser and Otto N. Larsen. New York: Free Press.

a new test of convergence theory

447

Welsh, Brian, and Pavel Butorin (eds.). 1990. Dictionary of Development. New York: Garland. Williamson, J.B., and J.J. Fleming. 1977. “Convergence Theory and the Social Welfare Sector: A Cross-National Analysis.” International Journal of Comparative Sociology 18:242–253. Wilson, Chris. 2001. “On the Scale of Global Demographic Convergence 1950–2000.” Population and Development Review 27(1):155–171. World Bank. 2005. 2005 World Development Indicators. Washington, DC: World Bank.

NOTES ON CONTRIBUTORS Margit Bussmann is Assistant Professor at the Department of Politics & Management; University of Konstanz, Germany. Her research interests include causes of international and domestic conflict, economic interdependence, and development. She has recently published articles in European Journal of International Relations, International Studies Quarterly, and Journal of Conflict Resolution. Cristina Corduneanu-Huci is a PhD candidate in Political Science at Duke University. She is interested in political contention, democratization, and the political economy of development. Her research has appeared in Comparative Sociology and in the volume, Economic and Political Contention in Comparative Perspective, edited by Maria Kousis and Charles Tilly. Mattei Dogan, a sociologist and political scientist, is a Senior Fellow at the National Center for Scientific Research in Paris and Professor Emeritus at the University of California, Los Angeles. He has been a visiting professor at the University of Trento, Indiana University, Yale University, Institute of Statistical Mathematics in Tokyo, and the University of Florence. He is co-founder of the Society for Comparative Research. For many years he was a member of the editorial board of the Revue Francaise de Sociologie and other academic journals. Author, co-author, editor or co-editor of 35 books, he has also published more than 100 articles in academic journals. He is listed in Who’s Who in France? and in Contemporary Authors. Formerly chair of the IPSA research committee on political elites, until 2002, and of the ISA research committee on comparative sociology. Janet Harkness holds the Donald and Shirley Clifton Chair in Survey Science at the University of Nebraska and is Director of the Survey Research and Methodology Program at Nebraska. She is also affiliated with the Center for Survey Methods and Analysis in Mannheim, Germany. Kristen R. Heimdal is a Senior Research Associate at Strategic Research Group in Columbus, Ohio and a Ph.D. candidate in the Department of Sociology at The Ohio State University. Her dissertation investigates

450

notes on contributors

how cohabiting and married couples negotiate financial matters and handle kin work and relations. She is co-author (with Sharon K. Houseknecht) of “Cohabiting and Married Couples’ Income Organization Approaches: Sweden and the United States”, Journal of Marriage and Family (2003). She has presented papers at the annual meetings of The American Sociological Association and the Society for the Scientific Study of Religion. Sharon K. Houseknecht is Associate Professor Emerita of Sociology at the Ohio State University. Recent publications include “The Impact of Marital Disruption and Conflict on Children’s Health”, Youth and Society (2006 with Darcy W. Hango); “Explaining Teen Childbearing and Cohabitation: Community Embeddedness and Primary Ties”, Family Relations (2005 with Susan K. Lewis); “Marital Disruption and Accidents/Injuries among Children”, Journal of Family Issues (2005 with Darcy W. Hango). She co-edited (with Jerry G. Pankhurst), Family, Religion and Social Change in Diverse Societies, Oxford University Press (2000). Current research interests focus on depictions of children and war in the media; exploration of the dynamics of support and opposition to family legislation; and a cross-societal study of the influence of religious market diversity, religious homogeneity, and religious culture on religiosity and social values. Ho-fung Hung is an assistant professor of Sociology at Indiana University, Bloomington. He researches and publishes on contentious politics, globalization, nationalism, and social theory. He is currently working on a book-length project that delineates China’s particular form of modernity by examining how the neo-Confucianist ideology shapes its trajectory of state formation and popular contention from the eighteenth century to the present. He is also editing a book on the impact of the contemporary rise of China on the global capitalist system. Ronald Inglehart is a professor of political science and program director at the Institute for Social Research at the University of Michigan. He helped found the Euro-Barometer surveys and directs the World Values Surveys. His research deals with changing belief systems and their impact on social and political change. His most recent books are (with Pippa Norris) Rising Tide: Gender Equality in Global Perspective (Cambridge University Press, 2003); (with Pippa Norris) Sacred and Secular: the Secularization Thesis Revisited (Cambridge University Press, 2004); and (with

notes on contributors

451

Christian Welzel) Modernization, Cultural Change and Democracy: The Human Development Sequence (Cambridge University Press, 2005). He also edited Mass Values and Social Change: Findings from the Values Surveys (Leiden: Brill Publishers, 2003) and Human Beliefs and Values: A Cross-Cultural Sourcebook based on the 1999 –2001 Values Surveys (Mexico City: Siglo XXI, 2004). Author of more than 220 publications, he has been a visiting professor or visiting scholar in France, Germany, The Netherlands, Switzerland, Japan, South Korea, Taiwan, Brazil and Nigeria and has served as a consultant to the U.S. State Department and the European Union. For more information, see Inglehart’s website: http://wvs.isr.umich .edu/ringlehart/index.html and the World Values Survey website: http://www.worldvaluessurvey.org Valeriy Khmelko (Doctor of Science in Sociology; Institute of Philosophy of the Ukraine Academy of Science, sociology department, 1988) is Professor of Sociology at the University of Kiev-Mohyla Academy, Chair of the sociology department (2008) and President of the Kiev International Institute of Sociology (since 1992). He was previously Chair of the department (1992–2000) and Dean of the Faculty of Social Sciences and Technologies (2000–02). He was a Visiting Professor of Sociology at the Johns Hopkins University in 1993. He is currently involved in the Editorial Boards of three journals: Sociology: Theory, Methods, Marketing (Ukrainian Academy of Science); Polish Sociological Review (Polish Sociological Association); and International Journal of Sociology. His major publications are: Social Orientation of Personality [ In Russian; Kiev, 1988]. Production as a Societal Process (with the collaboration of V. Tolstykh, V. Mezhuev, V. Mazur, E. Agababian) [ In Russian; Moscow, 1986]. “The Russian Factor and Territorial Polarization in Ukraine” (with the collaboration of D. Arel), The Harriman Review, v. 9, No. 1–2, Spring 1996. “Transformations of Social Structure in Ukraine during the Decade of Independence”. Studia Socjologiczne (Warszawa), 2004, v. 4 (v. 175). He is editor of the book: Social Structure and Personality: Studies by Melvin L. Kohn and his collaborators (In Ukrainian; Kiev, 2007). Melvin L. Kohn (Ph.D., Sociology, Cornell University, 1952) is Professor of Sociology at the Johns Hopkins University. He was formerly Chief of the Laboratory of Socio-environmental Studies of the (U.S.) National Institute of Mental Health (1960–85). He is author of Class and Conformity: A Study in Values (Dorsey Press, 1969; 2nd ed., University

452

notes on contributors

of Chicago Press, 1977), Work and Personality: An Inquiry into the Impact of Social Stratification (co-authored with Carmi Schooler, Ablex, 1983), Social Structure and Self-Direction: A Comparative Analysis of the United States and Poland (co-authored with Kazimierz M. Slomczynski, Blackwell, 1990; 2nd ed., IFiS Publishers [Institute of Philosophy and Sociology of the Polish Academy of Sciences], 2006), and Change and Stability: A CrossNational Analysis of Social Structure and Personality (Paradigm Publishers, 2006; Chinese edition published by the Social Sciences Academic Press of the Chinese Academy of Social Science, 2007). He is also editor of Cross-National Research in Sociology (Sage, 1989) and co-editor, with Aleksandra Jasinska-Kania and Kazimierz M. SÜomczynski, of Power and Social Structure: Essays in Honor of Włodzimierz Wesołowski (University of Warsaw Press, 1999). Translations of his books and collections of his papers have been published in Italian, German, Polish, Chinese, and Ukrainian. Several of his articles have been translated and published in these languages as well as in Danish, French, Japanese, Russian, and Swedish. He is currently doing a collaborative study of social structure and personality of urban China in transition. The first article to come from this research was recently published by Kohn and his colleagues, Lulu Li, Weidong Wang, and Yin Yue in Comparative Sociology. Robert M. Marsh is Professor Emeritus of sociology at Brown University. His recent publications include The Great Transformation: Social Change in Taipei, Taiwan Since the 1960s (M.E. Sharpe, 1996); “Weber’s Misunderstanding of Traditional Chinese Law”, in the American Journal of Sociology, 106.2 (2000); and articles on social change, national identity, ethnicity, and social class identification, published in the International Journal of Comparative Sociology (1999, 2000), Comparative Sociology (2002) and The Sociological Quarterly (2003). His current research involves testing social capital theory in Taiwan, using recent survey data on organizational participation and trust as the expressions of social capital. He has taught at National Tsing Hua University in Taiwan and has been a Guggenheim Foundation Fellow and a recipient of grants from the Chiang Ching-Kuo Foundation and the National Science Council, Republic of China. Peter Ph. Mohler is Director of the Center for Survey Methods and Analysis in Mannheim, Germany and a former Secretary General of the International Social Survey Program.

notes on contributors

453

Pippa Norris is the McGuire Lecturer in Comparative Politics at the John F. Kennedy School of Government, Harvard University. Last year, she also served as Director of the Democratic Governance Group at UNDP in New York. Her work compares democracy, elections and public opinion, political communications, and gender politics in many countries worldwide. A well-known public speaker and prolific author, she has published almost three dozen books. This includes a series of volumes for Cambridge University Press: A Virtuous Circle (2000, winner of the 2006 Doris A. Graber award), Digital Divide (2001), Democratic Phoenix (2002) and Rising Tide (with Ronald Inglehart, 2003), Electoral Engineering (2004), Sacred and Secular (with Ronald Inglehart, 2004, winner of the Virginia Hodgkinson prize), and Radical Right (2005). Her most recent books are Driving Democracy: Do power-sharing institutions work? (for CUP) and an edited volume for UNDP, Making Democracy Deliver: Governance for Human Development. John R. Oneal, b. 1946, PhD in Political Science (Stanford University, 1979) is Professor Emeritus of Political Science at The University of Alabama, Tuscaloosa. He has published research on decision making, imperialism, burden sharing in alliances, U.S. public opinion, and the causes of war as well as economic development. He was a Fulbright Scholar and Fellow at the Norwegian Nobel Institute in 2000. Noriko Onodera is a Senior researcher at the Public Opinion Research Division, Broadcasting Cultural Research Institute at NHK, Tokyo. Since joining NHK in 1980, she has been involved in research at the Public Opinion Research Division mainly on sampling and survey methods. She conducted Japanese surveys for the International Social Survey Programme (ISSP) from 1995 to 2002. Maryjane Osa studies the impact of social networks on political mobilization in non-democracies. She is the author of Solidarity and Contention: Networks of Polish Opposition as well as numerous scholarly articles. Professor Osa teaches in the Sociology Department at Northwestern University. Vladimir I. Paniotto, Doctor of Science (1987) in Sociology (Institute of Philosophy of the Ukraine Academy of Science), has been Professor of Sociology at the University of Kiev Mohyla Academy since 1992.

454

notes on contributors

He has also been General Director of the Kiev International Institute of Sociology since 1992. He was Chief of the Section of Computer Simulation of Social Processes, Institute of Sociology, Ukrainian Academy of Sciences (1987–1991), and Visiting Professor of Sociology at the Johns Hopkins University (1993 and 1995). He was the National representative in Ukraine of the European Society of Market and Opinion Research (ESOMAR, 1996–2005), and is currently involved in the activities of the Ukrainian Association of Marketing (vice-president), Sociological Association of Ukraine (member of Government Body), American Sociologocal Association (ASA), AAPOR and WAPOR. He is currently involved in the Editorial Boards of several journals. He has published 10 books and about 160 articles and chapters. Among the books (all in Ukrainian or Russian) are: Sociological Data Analysis (with V. Maksimenko and N. Kharchenko). Kiev: Publishing house “KM Academy”, 2004. The Experience of Social Process Modeling (an edited volume). Kiev: Naukova Dumka, 1989. Masamichi Sasaki is Professor of Sociology at Chuo University. He was President of the International Institute of Sociology from 1997 to 2001 and was (Founding) Editor of Comparative Sociology. He co-edited, with Jing Tiankui and Li Peilin, Social Change in the Age of Globalization (Brill, 2006); co-authored with Tatsuzo Suzuki, Social Attitudes in Japan: Trends and Cross-National Perspectives (Brill, 2nd edition, 2003); edited Values and Attitudes across Nations and Time (Brill, 1998); and co-edited, with Alex Inkeles, Comparing Nations and Cultures (Prentice Hall, 1996). His research interests are comparative sociology, the sociology of religion, and social trust. Shalom H. Schwartz is Emeritus Professor of Psychology at the Hebrew University in Jerusalem. He is immediate past president of the International Association for Cross-Cultural Psychology and a fellow of the American Psychological Association. He has published books and articles in social psychology, cross-cultural psychology, political psychology, sociology, education, law, and economics. The State of Israel recently awarded him its most prestigious civilian prize for his career research in psychology. Schwartz’s theory of a near-universal structure of relations of conflict and congruence among basic human values has been validated in research in over 75 countries. It has been used by him and others to examine how individuals’ values relate to attitudes, behavior, subjective well-being, identification, personality, interpersonal prob-

notes on contributors

455

lems, and socialization. His work on cultural value orientations maps the distinctive cultures of 74 countries and identifies 8 world cultural regions. This research examines reciprocal influences of the cultural value orientations of nations and social structural variables, historical sources of cultural differences, relations of culture to social policy, and cultural influences on individual behavior within societies. Tom W. Smith is Director of the General Social Survey and of the Center for the Study of Politics and Society at the National Opinion Research Center, University of Chicago. He is co-founder and former Secretary General of the International Social Survey Program. Indra de Soysa is Professor of Political Science at Norwegian University of Science and Technology (NTNU) in Trondheim, Norway, and Associate Scholar at the Center for the Study of Civil War at the International Peace Research Institute, Oslo (PRIO). He has published articles in International Studies Quarterly, Journal of Peace Research, International Organization, World Development, and Conflict Management and Peace Science. His research interests include the implications of globalization, the causes of conflict, the resource curse, and political and economic development. Tatsuzo Suzuki is Professor Emeritus at the Institute of Statistical Mathematics, Tokyo, Japan. He wrote, with K. Takahashi, Sampling Method (in Japanese) (Tokyo: Asakura Shoten, 1998) and also wrote, with C. Hayashi and M. Sasaki, Data Analysis for Comparative Social Research (NorthHolland, 1992). Erich Weede, Professor Dr., born 1942, taught sociology at the University of Bonn until his retirement in fall 2004. He had acquired academic degrees in psychology and political science. In 1982/83 he was president of the Peace Science Society (International), and in 1985/86 vice-president of the International Studies Association. He has produced more than 200 publications in German or English, among them 11 books. His topics concern war, the rise and decline of nations, Asian civilizations, economic growth and income inequality. His books include Economic Development, Social Order and World Politics (Lynne Rienner, Boulder, CO 1996) and The Balance of Power, Globalization, and the Capitalist Peace (for the Friedrich-Naumann-Foundation, Liberal Verlag, Berlin 2005). He still serves as a member of the editorial boards of six scientific journals: Electronic Journal of Sustainable Development, International Interactions, Journal

456

notes on contributors

of Conflict Resolution, Korea and World Affairs, New Asia, and Pacific Focus. He is a member of the Mont Pelerin Society. Brigitte Weiffen, M.A. (2001) in Political Science, University of Bonn, Germany, is a research associate at the Department of Politics and Management, University of Konstanz. From 2005 to 2007, she collaborated on the research project “The International Organization of the Democratic Peace”, carried out by the Peace Research Institute Frankfurt and the Institute of Political Science at the University of Tübingen, Germany. From 2001 to 2005, she worked as research assistant at the Department of Sociology, University of Bonn. Her research interests include comparative democratization, international organizations, political science methodology, Latin American politics and Muslim societies. She has published in Comparative Sociology, Review of International Studies, European Political Science, World Development and various edited volumes. Masato Yoneda is Senior Researcher at the Department of Language Information and Resources at the National Institute for the Japanese Language. His works include “Survey of Standardization in Tsuruoka, Japan: Comparison of Results from Three Surveys Conducted at 20-Year Intervals”, Japanese Linguistics 2:24–39, 1997, and “International Census on Attitudes toward Japanese Language”, in The Study of Methodology in Cross-National Survey Research: From a Perspective of Cultural Diversity (Kwansei Gakuin University, 2004:151–186).

INDEX Adelman, I. 25 Africa 6, 23, 26, 30, 36, 37, 40, 192, 196, 197, 215, 277 South 286, 289, 291, 292, 293, 297, 362 Agricultural sector 3, 98, 100, 104, 105, 108, 109, 121, 357, 358, 363, 371–373, 404, 422, 438, 439, 444 Albania 229, 242 Alderson, A.S. 358, 405 Alexander, J. 405 Algeria 261 Al-Qaeda 225 Osama bin Laden 225 Almond, G. 25 Apter, D. 16 Argentina 39, 289, 291, 292, 297, 384, 386, 393, 398 Aristotle 18, 20 Asia 16, 19, 22, 26, 37, 40, 197, 251, 277 East 111 South 192, 196, 197, 215 South East 23, 228, 231 Asian Tigers 99, 440 Asynchronic comparison 2, 29, 30 Australia 37, 39, 228, 229, 431 Austria 39, 241, 431 Authoritarian conservatism 124–126, 129, 130, 410, 423 Authoritarian regime 18, 19, 23, 24, 26, 232, 247, 251, 255, 256, 261, 270, 277, 278, 282, 284, 287, 294–297, 299, 300, 305, 372 Bureaucratic 23 Authority (religious) 223, 241, 246, 247, 313, 323, 100–102, 108 Autocracy 253, 256, 258, 259, 261, 271, 372, 112 Automatic translation 382 Azerbaijan 224, 229, 241, 242 Bahrain 252 Bangladesh 224, 229, 241, 242, 258 Banks, J. 35 Barro, R.J. 357, 367, 372 Baumol, W.J. 441 Belarus 239

Belgium 28, 39, 431 Bellah, R.N. 405 Bendix, R. 21, 30 Berger, P.L. 315, 316, 318 Binary comparison 20–23, 42, 265, 286, 415 Explicit 20 Implicit 20 Birdsall, N. 356 Bolivia 194, 197 Bollen, K.A. 326, 413 Borsu, A. 357 Bosnia 221, 231 Bourguignon, F. 372 Bornschier, V. 366 Boyle, G.E. 411, 416 Brazil 22, 26, 30, 38, 40, 285, 290, 297, 384, 386, 393, 398 Britain 21, 22, 39, 432 Bruce, S. 320, 321 Bulgaria 186, 188, 228 Burma (see also Myanmar) 229 Bussmann, M. 8 Cambodia 229 Canada 16, 39, 190, 373, 432 Caplow, T. 414 Ceau‘escu, N. 281, 293 Central Europe (see also Europe) Chad 22 Chan, S. 355, 358, 359 Chase-Dunn, C. 366 Chile 26, 35, 40, 286, 289, 290, 292, 297 China 21, 22, 30, 38, 39, 40, 186, 228, 264, 278, 283, 289, 290, 297, 384, 386, 393, 400, 440, 97–99, 100, 112, 118 Cities 101, 102, 113 Clash of Civilizations 5, 6, 225, 228, 244, 247, 252 Coefficient of dispersion 28 Cold War 221, 224, 241, 283 Post- 5, 221, 246 Coleman, J.S. 16, 20 Comparative (analysis) Global pproach a 15 Segmentation 15 Single case study 15–17, 19

458

index

Communication 379–381, 386, 401 Confirmatory factor analysis 129, 132, 142, 162 Convergence theory 405–449 Cooper, R.N. 357 Copernicus, N. 14 Co-plot multidimensional scaling technique 191 Corduneanu-Huci, C. 7 Correspondence analysis 384, 398, 401 Costa Rica 22, 38 Cox, H. 323, 324 Cross-cultural survey 45, 76, 410 Cross-national survey 17, 28, 45, 47, 66, 74, 98, 119, 139, 371, 383, 412 Cross-sectional survey 119–126, 130–132, 136, 137, 142, 152, 153, 359 Cross-societal variable 414, 416 Crosstabulations analysis 384, 386, 393, 398, 401 Cultural-economic syndrome 258–270 Culture 53, 71, 106, 171–221, 222, 224, 246, 252, 260, 313–316, 317, 319, 344, 381, 412 African 5 Confucian 5 East Central Europe 5 Greek or thodox 228 English peaking s 5 Islamic 5, 6, 224, 225, 234, 256, 265, 270 South sAian 5 Western European 5, 196 Western 23, 223, 224, 225, 234 Cutright, P. 34, 35 Cybernetic hierarchy 409 Czechoslovakia 33, 35, 432 Czech Republic 241 Darwin, C. 13 Democracy 5, 6, 8, 10, 16, 25, 29, 35, 194, 198, 200, 208, 209, 221–250, 251–276, 277, 304, 320, 324, 326, 332, 338, 342, 343, 363, 372, 414, 423, 424, 437, 441, 442, 444 In the Middle East 6 Non- 296, 299, 301, 303, 358 Democracy gap 252, 253, 256 Democracy-totalitarianism dichotomy 19 Democratic values 222–227, 246, 441 Democratization 6, 198, 199, 208, 231, 237 Demography 22, 174, 198, 202, 407, 408, 414, 422, 423, 437, 441

Deneire, M.G. 381 Deng Xiaoping 99, 112 Denmark 28, 38, 39, 432 Descriptive facts 17 Descriptive statistics 329–332 Developed countries 10, 98, 359, 361, 366–368, 353, 355, 407, 411, 413 Developing countries 10, 20, 27, 353–356, 361, 366–368, 370, 407, 413, 444, 445 Development 32, 33, 99, 416, 427, 429, 431, 440 (socio) economic 32, 36, 109, 183, 199, 200, 202–204, 207, 215, 224, 410, 414, 421, 441 Political 34, 35, 114, 231, 237, 423 Diachronic comparison 2, 29, 30 Dictatorial government 33 Dimension of inequality 148 Dimension of political opportunity 278–281, 295, 300 Dimension of orientation 124, 126, 128, 129, 130 Dimension of personality 122, 124, 132, 141, 149, 152, 154, 156 Dimension of social stratification 142, 150 Dogan, M. 1, 2 Dollar, D. 370 Dumont, R. 16 Durkheim, E. 18, 379 Dynamics of repression 288, 293, 295, 298 Eastern Europe (see also Europe) 119, 197, 239, 241, 246 Eckstein, H. 15, 16 Ecological environment 2, 36 Ecological problem 37 Education 8, 31, 35, 40, 53, 60, 73, 100, 101, 125, 135, 141, 142, 149, 150, 167, 257, 319–327, 332, 338, 343, 359, 405– 444 Edwards, S. 357 Egypt 41, 229, 241, 242, 384, 386, 393, 400 Eisenstadt 405 El Salvador 279 Elite 246, 256–258, 284, 285, 290, 291, 292, 299, 354, 356 Divisions 7, 278, 284, 285, 288, 291, 293, 295, 298, 299, 303, 305 Homogeneity 298, 299, 301, 303, 306 Unified 7, 291–294

index Energy consumption 33 England (see also Britain) 22 English 63, 66, 68, 75, 379, 383–386, 393, 398, 400, 401 As an int. language 379, 401 Dominance 379, 380, 381, 382, 393, 398, 400, 401 Non-speaking 379 Esposito, J. 254, 255 Estonia 186 Europe 13, 28–40, 182, 194, 212, 224, 373 European Social Survey 180, 211 European Value Survey 185, 198, 222, 227, 228, 231, 232 Fascism 22 Family institution 313–351 Family-religion linkages 313, 314, 327 Finland 22, 38, 39, 432 Fishman, J.A. 382, 383 Flora, P. 29 Forrest, J. 30 France 30, 31, 38, 384, 393, 398, 432, 442 Frandkin, H. 322 Freedom House 251, 264, 267, 414, 424 Fukuyama, F. 441 Fundamentalism 224 Christian 225 Islamic 221, 225, 254 Taliban 225, 231 Galtung, J. 355 Gauss curve 28 Gender 53, 60, 113, 407, 414, 427, 428, 431, 439, 441 Gender (In)equality 5, 6, 125, 126, 129, 137, 139, 140, 146, 148, 154, 164, 222, 226, 227, 232, 234, 235, 242, 246, 247, 428, 429, 440 Gender particularism-universalism 426–430, 438, 440, 442, 444 Gender related development index 429, 431 Georgia 239 Germany 21, 22, 39, 46, 52, 53, 63, 66, 71, 99, 107, 241, 242 East 33 Nazi 22 Ghana 16, 38 Gini index 361–371, 426, 438 Globalization 8–10,107, 271, 353–361, 366, 367, 379–382, 405

459

Globalization theory 405, 413, 421, 444 Gorbachev, M. 291 Greek Cypriot culture 228 Grosser, A. 15 Guatemala 35, 279 Haney, M. 123 Harkness, J. 2 Hayek, F.A. 100 Health 35, 207, 257, 317 Heimdal, K.R. 8 Higgins, M. 357, 359 Himmelfarb, G. 322 Hofstede, G. 4, 5, 215, 173, 178, 192 Holland (see also Netherlands) 384, 386, 400 Honduras 289, 290, 297 Hong Kong 291 Houseknecht, S.K. 8 Hudson, M. 16 Hung, H.-F. 3 Hungary 33, 39, 384, 393, 398 Huntington, E. 36 Huntington, S. (see also clash of civilizations) 5, 6, 40, 97, 109–111, 221, 222, 223–229, 234, 239, 246, 251, 252 Hu Yaobang 291 Iceland 241, 432 Income 142, 149, 150, 162, 231, 237, 304, 323, 357–368, 411, 412, 426, 429, 438 Distribution 8, 110, 357–359, 361, 362, 364, 366, 413, 426, 438 (In)equality 29, 110, 200, 357–367, 406, 411, 414, 426, 438, 441, 442 India 16, 22, 30, 38, 39, 41, 228, 229, 231 Indonesia 35, 224, 384, 386, 393 Industrialism 408–412, 438, 439, 440 Industrialization 21, 33, 35, 226, 323, 344, 357, 358 Industrial Revolution 22, 357 Britain 21 France 21 Industrial society 408, 413 Information and communications technology 414, 430, 431, 439, 440–442, 444 Inglehart, R. 4–6, 187–189, 440 Inkeles, A. 405, 412, 413, 442, 443 Institution of religion 313–351 Internal development model 358, 363

460

index

Internal diversity 28 Degree of linguistic homogeneity 28 International Language (see also language) 9 International communication 380, 382, 383, 386, 401 International Social Survey Program 46, 72, 73 International trade 299 Openness 299–304 Iran 229, 242, 256, 289, 290, 297 Ireland 39 Islamic democracy 254, 255 Islamic organizations 260–262 Islamic society 6, 42, 222–225, 237–247, 256 Islamic world 5, 222–229, 241, 246 Israel 384, 386, 400, 432 Italy 22, 28, 37, 39, 224, 241, 384, 432 Jaccard, J. 267 Jaggers, K. 263, 372 Japan 20–22, 30, 41, 119, 120, 124, 130, 152, 165, 188, 194, 196, 229, 231, 237, 384, 386, 393, 398, 400, 432 Jones, L.V. 411 Jordan 224, 229, 241, 242 Kachru, B.B. 380–384 Kamarck, A.M. 36 Kashmir 221 Kerr, C. 405 Khmelko, V. 3 Klingemann, H.D. 198 Kohn, M.L. 3, 4, 119, 133–139, 153, 158, 162, 167 Korea 384, 393, 400 Kuznets, S. 357, 358, 360, 361, 370 Curve 368, 371, 372, 361, 364 Language 9, 379–405 Arabic 393 Chinese 393 English see English German 63, 66, 68, 72 Hindi 393 Indonesian 393 Japanese 47, 52, 53, 68, 71, 72, 75, 393 Portuguese 393 Russian 393 Spanish 393 Laos 229

Latin America 18, 23, 26, 36, 37, 40, 192, 194, 197, 215, 223, 229, 237, 239, 246, 251, 271, 277, 284, 290, 373 Lebanon 16 Lenski, G. 320 Level of democracy (see also democracy) 198, 199, 200, 237, 251, 265, 267, 301, 305, 322–326 Level of development (see also development) 27, 37, 99, 192, 320, 323, 366, 367, 440, 444 Level of education (see also education) 252, 257, 259, 323, 428, 437 Level of modernization (see also modernization) 405, 406, 414, 415, 432, 437, 441 Level of poverty 28, 29 Levy, M.J. 21, 405–449 Libya 225 Lijphart, A. 17 Linguistic capital 381 Lipset, S.M. 41 Longitudinal comparison 121–126, 132, 133, 142, 145, 150, 152, 158, 162, 164, 324, 346 Longitudinal measurement model 124, 126, 132, 133, 142, 153, 154, 162 Longitudinal survey 2, 8, 29, 121–124, 132, 150, 152, 158, 164, 324, 431 Lorenz curve 28 Luxembourg 432 Macedonia 192, 228, 231, 239 Mach, B. 125, 129 Mahoney, J. 314 Malinowski, B. 315 Malthus, D.R. 30 Mao Zedong 99, 100, 112 Marsh, R.M. 10 Marshall, M.G. 263 Martz, J. 23 Mason, D.S. 279 Maximum Likelihood Estimation 297, 300, 305 Mehlkop, G. 357 Mexico 39, 40 Meyer, J. 405, 412, 413 Middle East 23, 192, 194, 196, 197, 221–229, 246, 247, 251–275, 283 Mill, J.S. 13, 24 Mises, L. von 99 Modernization 10, 20–22, 173,

index 223–226, 242, 252, 253, 257, 258, 260, 262, 407–440 Cognitive 10, 406, 414, 424, 425, 437, 441, 442 Modernization theory 10, 225, 227, 231, 252, 257, 304, 408–411, 413, 427, 440, 441 Mohler, P. 2 Moldova 239 Mongolia 384 Montaigne, M. de 13 Montesquieu, C. 14 Moore, B. 405 Morocco 22, 39, 229, 242 Morrison, C. 372 Multiple language (see also language) 2, 3 Multiple language survey 76 Multiplex (social) networks 280, 281, 293, 296, 300 Multi-stage random sampling 122 Myanmar 291, 297 Netherlands 39, 241, 432 Newton, I. 14 New Zealand 228, 229, 432 Nicaragua 35 Nielsen, F. 358 Nigeria 16, 38, 231, 241 Nimkoff, M.F. 323 Norris, P. 5 Norway 22, 38, 242, 432 Offe, C. 442 Oman 252 Onodera, N. 2 Oneal, J.R. 8, 353 Osa, M. 7 Ottoman Empire 194 Pakistan 22, 224, 229, 258 Pan-Arabism 261 Paniotto, V.I. 3 Pankhurst, J.G. 322 Parsons, T. 18 Paxton, P. 324, 326 Path analysis 199, 200, 203 Peacock, W.G. 410, 411 Pearson, K. 60 Peters, B.G. 16 Pejovich, S. 106 Personality 119–170 Peru 39, 194, 197 Philippines 35, 241

461

Poland 33, 35, 39, 119–141, 152, 156, 165, 241, 278, 286, 288–290, 293, 294, 297 Political development 231, 423 Political opportunity 277–315 Political participation 223, 252, 257, 359, 362, 406, 423, 429 Political parties 285, 286, 290 Political process 277 Political structure 321–322 Political values 227, 232–236 Pooled time-series 359, 361, 367, 370 Popenoe, D. 315, 344 Population growth 30 Portes, A. 354 Portugal 35, 289, 290, 297, 384, 386, 393 Powell, D.E. 321 Presidential regime 26, 38 Probability sampling 383 Proximate condition of life 152, 156, 162, 166, 167 Przeworski, A. 24 Putin, V. 101–104, 109, 111, 113 Pye, L. 16, 23 Qatar 252 Qualitative comparative analysis 278, 282, 287 Quantitative comparative analysis 282, 286, 298, 300 Quota sampling techniques 383 Radical social change 119–137, 141, 144, 151, 152, 156, 158, 160, 165–167 Ragin, C.C. 23, 282, 287 Rational choice perspective 279 Regional disparities 29 Regional indicators 363, 364, 373 Regression analysis 359, 361, 367, 371 Religion 24, 222, 223, 232, 253, 255, 261, 262, 313–351 Buddhism 196 Catholicism 8, 196, 223, 224, 229, 232, 237, 246, 271, 290, 294, 320, 321, 324–326, 338, 343 Hinduism 196 Islam 6, 196, 222, 224, 225, 227, 234, 239, 241, 242, 244, 247, 253 Protestantism 196, 345 Religious culture 224, 231, 237, 246 Religious socialization 314, 318, 319

462

index

Rentier state 256, 258, 259, 261, 262, 270, 271 Replication of relationships 17 Response category 45–53, 71 Response scale 45–52, 58, 63 Reuveny, R. 357 Riggs, F. 16, 26 Roman Empire 379 Romania 191, 278, 279, 280, 281, 284, 288, 289, 292–294, 297 Rubinson, R. 355 Ruiter, S. 317 Russia 38, 39, 41, 97–106, 109, 110, 228, 239, 241, 246, 384 Sachs, J.D. 359, 367, 370, 373 Said, E. 224 Sala-i-Martin, X. 412 Sartori, G. 15 Sasaki, M. 9 Sassen, S. 408 Saudi Arabia 225, 227 Scandinavia 18, 224, 241 Scheepers, P. 317 Schwartz, S.H. 4, 5, 192, 211, 215, 216 Schwartz Value Survey 180 Science of sociolinguistics 382 Secularism 42, 255, 256 Shevtsova, L. 109 Shils, E. 20 Sigma-convergence 411 Singapore 176 Smith, T.W. 381 Social class 119, 120, 134, 135, 140, 141, 148, 149, 152, 154, 156, 161, 162, 164–167 Social mobilization 22, 256, 262 Social networks 7 Social structure 119–122, 131, 132, 134, 140, 141, 144–148, 150–158, 161, 165–167 Socialist economy 362, 368, 373 Socialist state 358, 362, 363, 368, 371, 372 South Africa 241, 286, 289, 291–294, 297 South Korea (see also Korea) 26, 278 Soviet Union (see also USSR) 97–99, 104, 106, 119, 121, 122, 151, 321 Soysa, I. de 8, 353 Spain 28, 35, 39, 41, 289, 290, 293, 294, 297, 384, 386, 393, 398 Sri Lanka 38, 229 Standard deviation 364, 368

Stolper-Samuelson theorem 357, 367 Strauss, L. 19 Suzuki, T. 9 Swafford, M. 123 Sweden 22, 188, 192, 224, 241, 242, 432 Switzerland 22, 39, 41, 231, 432 Synchronic comparison 30, 31 Syria 39, 225 Taiwan 384, 386, 393 Tanzania 241 Teune, H. 16 Thailand 16, 39, 229, 384 Theory of materialism-postmaterialism 173 Theory of cultural value orientations 173–223 Cultural dimension 173, 198, 202–208, 214, 217 transnational cultural groupings 192 value priorities 173, 178, 180 Theory of work values 173 Third Wave 226, 228, 251 Third World 19, 32, 40 Thomas, D.C. 300, 314 Tocqueville, A. de 23 Totalitarianism 42 Transition 1, 104, 107, 110, 120, 134, 140, 152, 153, 165, 226, 251, 264, 277, 301, 304, 320–326, 338, 343 Theory 23 Tunisia 22 Turkey 20, 39, 224, 229, 231, 241, 384, 386, 393 Typology(-ies) 18, 20 Deductive 17 Ideal type 17, 23, 175 Empirical 1 Inductive 17 of political systems 19, 38 of egimes r 19 of social actors 19 Tyrannical regime 18, 19 Ukraine 119–170 UNCTAD 353, 360, 371 United Nations Development Program 414, 429, 440 Human Development Index 429, 431, 439, 441 United States (see also dimension of orientation) 20, 22, 26, 30, 37–41, 119, 120, 123, 124, 126, 129, 130,

index 133, 134, 152, 165, 195, 224, 227, 241, 246, 373, 432 Urbanization 8, 32, 35, 39, 40, 406, 437 Uruguay 293, 297 USSR (see also Russia) 22 Stalinism 25 Vanhanen, T. 29, 199, 200, 207 Vardomatskii, A. 322 Verba, S. 15 Vernet, D. 31 Veyne, P. 26 Vietnam 39, 384 Da Vinci, L. 14 Wallerstein, I. 405, 411 Warner, R.S. 359, 367 Weber, M. 14, 17, 20 Weberian typology of legitimacy (see also typologies) 17, 18, 253 Weede, E. 3, 358, 359 Weiffen, B. 6 Western democracy see democracy

463

Western Europe (see also Europe) 23, 29, 195, 223, 228, 241, 255 Westernization 35 Williams Jr, R.M. 315, 317, 318 Williamson, J.B. 357–359 World Bank 29, 36, 97, 353, 360, 371, 373 World Handbook data 282, 283, 285, 297, 298 World polity/world society theory 405, 412, 413 World system 406, 410 World system theory 411, 413 World Value Survey (see also European Value Survey) 185, 198, 208, 216, 222, 227, 231, 232, 314, 324–327, 345 Wright, E. 134 Yeltsin, B. 101, 102 Yemen 192 Yoneda, M. 9 Yugoslavia 28, 39 Zimbabwe

186, 188, 192

INTERNATIONAL STUDIES IN SOCIOLOGY AND SOCIAL ANTHROPOLOGY 79. HARRIS, R. & M. SEID (eds.). Critical Perspectives on Globalization and Neoliberalism in the Developing Countries. 2000. ISBN 90 04 11850 0 80. HOWARD, G.J. & G. NEWMAN (eds.), Varieties of Comparative Criminology. 2001. ISBN 90 04 12245 1 81. NDEGWA, S.N. (ed.), A Decade of Democracy in Africa. 2001. ISBN 90 04 12244 3 82. JREISAT, J.E. (ed.), Governance and Developing Countries. 2002. ISBN 90 04 12247 8 83. KEITA, M. (ed.), Conceptualizing/Re-Conceptualizing Africa. The construction of African Historical Identity. 2002. ISBN 90 04 12420 9 84. BERG, R. VAN DEN, Nyoongar People of Australia. Perspectives on Racism and Multiculturalism. 2002. ISBN 90 04 12478 0 85. DOGAN, M. (ed.), Elite Configurations at the Apex of Power. 2003. ISBN 90 04 12808 5 86. SENGERS, G., Women and Demons. Cult Healing in Islamic Egypt. 2003. ISBN 90 04 12771 2 87. WILSON, H.T. The Vocation of Reason. Studies in Critical Theory and Social Science in the Age of Max Weber. Edited and with a Foreword by T.M. Kemple. 2003. ISBN 90 04 13631 2 88. ZEGEYE, A. and R.L. HARRIS, Media, Identity and the Public Sphere in Post-Apartheid South Africa. 2003. ISBN 90 04 12633 3 89. INGLEHART, R. (ed.), Human Values and Social Change. Findings from the Values Surveys. 2003. ISBN 90 04 12810 7 90. BEN-RAFAEL, E. (ed.), Sociology and Ideology. 2003. ISBN 90 04 13104 3 91. AL-HAJ, M., Immigration and Ethnic Formation in a Deeply Divided Society. The Case of the 1990s Immigrants from the Former Soviet Union in Israel. 2004. ISBN 90 04 13625 8 92. AMINEH, M., Central Eurasia in World Politics. Conflict, Security, and Development. 2004. ISBN 90 04 12809 3 93. VINKEN, H., J. SOETERS & P. ESTER (eds.), Comparing Cultures. Dimensions of Culture in a Comparative Perspective. 2004. ISBN 90 04 13115 9 94. ASSIÉ-LUMUMBA, N.T., Cyberspace, Distance Learning, and Higher Education in Developing Countries. Old and Emergent Issues of Access, Pedagogy, and Knowledge Production. 2004. ISBN 90 04 13121 3 95. HARRIS, R.L. and M. Seid, Globalization and Health. 2004. ISBN 90 04 14145 6 96. DOGAN, M. (ed.), Political Mistrust and the Discrediting of Politicians. 2005. ISBN 90 04 14530 3 97. MÉRIOT, S.-A., Nostalgic Cooks. Another French Paradox. 2006. ISBN 90 04 14346 7 99. EISENSTADT, S.N., The Great Revolutions and the Civilizations of Modernity. 2006. ISBN 90 04 14812 4 100. NIEUWENHUYS, E. (ed.), Neo-Liberal Globalism and Social Sustainable Globalisation. 2006. ISBN 90 04 15159 1

101. SCHRYER, F., Farming in a Global Economy. A Case Study of Dutch Immigrant Farmers in Canada. 2006. ISBN 90 04 15170 2 102. SHLAPENTOKH, D. (ed.), Russia between East and West. Scholarly Debates on Eurasianism. 2006. ISBN 90 04 15415 9 103. DAY, R.B. & J. MASCIULLI (eds.), Globalization and Political Ethics. 2006. ISBN 90 04 15581 3 104. ESMER, Y. & T. PETTERSSON (eds.), Measuring and Mapping Cultures. 25 Years of Comparative Value Surveys. 2007. ISBN 978 90 04 15820 7 105. PATTERSON, R. (ed.), African Brain Circulation. Beyond the Drain-Gain Debate. 2007. ISBN 978 90 04 15885 6 106. AMINEH, M.P. (ed.), The Greater Middle East in Global Politics. Social Science Perspectives on the Changing Geography of the World Politics. 2007. ISBN-13: 978 90 04 15859 7, ISBN-10: 90 04 15859 6 107. SASAKI, M. (ed.), Elites: New Comparative Perspectives. 2007. ISBN 978 90 04 16402 4 108. RIETJENS, S.J.H. Civil-Military Cooperation in Response to a Complex Emergency. Just Another Drill? 2008. ISBN 978 90 04 16327 0 109. SASAKI, M. (ed.), New Frontiers in Comparative Sociology. 2009. ISBN 978 90 04 17034 6