Civil-Military Dynamics, Democracy, and International Conflict: A New Quest for International Peace (Advances in Foreign Policy Analysis)

CIVIL–MILITARY DYNAMICS, DEMOCRACY, AND INTERNATIONAL CONFLICT

Advances in Foreign Policy Analysis Series Editor: Alex Mintz Foreign policy analysis offers rich theoretical perspectives and diverse methodological approaches. Scholars specializing in foreign policy analysis produce a vast output of research. Yet, there were only very few specialized outlets for publishing work in the field. Addressing this need is the purpose of Advances in Foreign Policy Analysis. The series bridges the gap between academic and policy approaches to foreign policy analysis, integrates across levels of analysis, span theoretical approaches to the field, and advances research utilizing decision theory, utility theory, and game theory. Members of the Board of Advisors Allison Astorino-Courtois Steve Chan Betty Hanson Margaret Hermann Valerie Hudson Patrick James

Jack Levy Zeev Maoz Bruce M. Russett Donald Sylvan Steve Walker Dina A. Zinnes

Published by Palgrave Macmillan Integrating Cognitive and Rational Theories of Foreign Policy Decision Making Edited by Alex Mintz Studies in International Mediation Edited by Jacob Bercovitch Media, Bureaucracies, and Foreign Aid: A Comparative Analysis of the United States, the United Kingdom, Canada, France, and Japan By Douglas A. Van Belle, Jean-Sébastien Rioux, and David M. Potter Civil–Military Dynamics, Democracy, and International Conflict: A New Quest for International Peace By Seung-Whan Choi and Patrick James Economic Sanctions and Presidential Decisions: Models of Political Rationality By A. Cooper Drury

CIVIL–MILITARY DYNAMICS, DEMOCRACY, AND INTERNATIONAL CONFLICT A New Quest for International Peace

SEUNG-WHAN CHOI Assistant Professor of Political Science University of Illinois at Chicago and

PATRICK JAMES Frederick A. Middlebush Professor of Political Science University of Missouri, Columbia

CIVIL–MILITARY DYNAMICS, DEMOCRACY, AND INTERNATIONAL CONFLICT

© Seung-Whan Choi and Patrick James, 2005. All rights reserved. No part of this book may be used or reproduced in any manner whatsoever without written permission except in the case of brief quotations embodied in critical articles or reviews. First published in 2005 by PALGRAVE MACMILLAN™ 175 Fifth Avenue, New York, N.Y. 10010 and Houndmills, Basingstoke, Hampshire, England RG21 6XS Companies and representatives throughout the world. PALGRAVE MACMILLAN is the global academic imprint of the Palgrave Macmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd. Macmillan® is a registered trademark in the United States, United Kingdom and other countries. Palgrave is a registered trademark in the European Union and other countries. ISBN 1–4039–6485–8 Library of Congress Cataloging-in-Publication Data Choi, Seung-Whan. Civil–military dynamics, democracy, and international conflict : a new quest for international peace / Seung-Whan Choi and Patrick James. p. cm. Includes bibliographical references and index. ISBN 1–4039–6485–8 1. Peacekeeping forces. 2. Conflict management. 3. International relations. 4. Democracy. I. James, Patrick, 1957– II. Title. JZ6374.C46 2005 322
.5—dc22

2004057312

A catalogue record for this book is available from the British Library. Design by Newgen Imaging Systems (P) Ltd., Chennai, India. First edition: April 2005 10 9 8 7 6 5 4 3 2 1 Printed in the United States of America.

To my mother and the memory of my father Seung-Whan Choi To Carolyn, with much love Patrick James

[The foreign policy] decision-making process can be equated with one or a few leaders, with the roles of civil and military bureaucracies, legislatures, interest groups. . . . (Emphasis added) —Brecher 1999: 231

If it were left to me to decide whether we should have a government without a free press or a free press without a government, I would prefer the latter —Thomas Jefferson

Contents List of Tables List of Figures Acknowledgments

ix xi xiii

1 Introduction

1

2 A Review of Democracy, Peace, and Other Things The Democratic Peace and Its Critics The Recent Evolution and Prospects of the Democratic Peace

7 7 14

3 A New Look at International Conflict and Peace: Four Factors to Consider Intuition, or, How does It Work? A Flow Chart of the Foreign Policy Decision-Making Process Model Two Historical Exemplars Where To?

29 33 36

4 Research Design Hypotheses Measurement and Data Building Logit and Neural Network Models

39 39 43 47

5 Accounting for Militarized Interstate Disputes Logistic Regression Analysis Neural Network Analysis A Case Study: Bush Administration Diplomacy and the North Korean Nuclear Program

55 55 70

23 23

79

viii Contents

6 Accounting for Fatal Militarized Interstate Disputes Logistic Regression Analysis Neural Network Analysis 7 A Key Issue of Measurement: Military Expenditure and Civil–Military Relations Measuring Civil–Military Relations in Two Steps Empirical Implications: Do Civil–Military Relations Still Matter?

91 92 99

109 110 121

8 A Closer Look at Media Openness: From Institutional Democracy to Media Openness? Media Openness, Democracy, and Interstate Disputes Testing the Role of Media Openness Empirical Implications: The Power of Media Openness

140

9 The Quest for Peace Findings Policy Implications Some Final Thoughts

147 147 150 154

Appendix Notes References Index

155 157 175 193

129 130 137

List of Tables 2.1 2.2 4.1 4.2 5.1 5.2 5.3

5.4

5.5 5.6

6.1

6.2

6.3

Empirical Results of Oneal and Russett Empirical Results of Five Other Democratic Peace Studies Four New and Eight Replicating Hypotheses Rules of Operationalization for Variables Civil–Military Relations and Predicting MID Involvement, 1886–1992 Civil–Military Dynamics and Predicting MID Involvement, 1886–1992 Civil–Military Dynamics, Diplomatic Channels, and Predicting MID Involvement, 1886–1974 Civil–Military Dynamics, Political Communications, and Predicting MID Involvement, 1950–1974 Multicollinearity Diagnostics, 1950–1974 A New Quest for International Peace?: A Summary of Statistical Significance in MID Involvement Civil–Military Dynamics, Political Communications, and Predicting Fatal MIDs: Simplest, All Dyads Civil–Military Dynamics, Political Communications, and Predicting Fatal MIDs: Politically Relevant Dyads A New Quest for International Peace?: A Summary of Statistical Significance in Fatal MIDs

15 19 50 51 57 59

62

64 69

78

93

94

106

x List of Tables

7.1 7.2 8.1 8.2

Hypotheses and Operationalization Do Civil–Military Relations Still Matter? Media Openness or Institutional Democracy, 1950–1992? Substantive Effects of MID Involvement and Fatal MIDs, 1950–1992

119 123 141 143

List of Figures 3.1 4.1 5.1 5.2 6.1 6.2 7.1

A.1

Foreign Policy Decision-Making Process Model Linear, Logit, and Neural Network Model Fitting Three-Dimensional Plots of the Network Output against Two Inputs, 1950–1974 Marginal Effects of Each Independent Variable, 1950–1974 Three-Dimensional Plots of the Network Output against Two Inputs, 1950–1974 Marginal Effects of Each Independent Variable, 1950–1974 Two-Step Model Building Procedure for Civil–Military Relations and Neo-Kantian Peace Relationships for the Domestic Political–Social System: Societal Militarism, Civil–Military Relations, and Military Manpower System

30 52 72 74 101 103

111

155

This page intentionally left blank

Acknowledgments Each of the authors has much to be thankful for in the completion of this work. The Department of Political Science, University of Missouri, Columbia, gave Whan an understanding of how to think through significant problems in the study of political science, which has led him toward a research project on a new quest for world peace. The post-doctoral fellowship at the Centre for Security and Defence Studies, Norman Paterson School of International Affairs, Carleton University, Ottawa, provided Whan with a fine research environment in his temporary home in Canada. The Department of Political Science, University of Illinois at Chicago, supported Whan in the completion of this book with great care and professionalism. Pat has found the Political Science Department at the University of Missouri, Columbia, to be an ideal environment for serious research. The ability to collaborate with a promising doctoral student, Whan—now an accomplished scholar with a completed degree and tenure track at the University of Illinois at Chicago—is something great that has come with working at “Mizzou.” We are greatly indebted to Craig Cobane, A. Cooper Drury, Martin Edwards, Peter D. Feaver, Benjamin Fordham, Erik Gartzke, Mark Gasiorowski, Charles Gochman, Ryan C. Hendrickson, J. Joseph Hewitt, Kyung-Seong Jeon, Radford M. Neal, John R. Oneal, Brandon C. Prins, Robin Remington, Rafael Reuveny, Norrin M. Ripsman, Charles L. Sampson, Todd Sandler, Kenneth Schultz, Glen M. Segell, Mark Souva, Harvey Starr, Doug A. Van Belle, Paul Wallace, and Xinghe Wang for helpful commentaries on various parts of this book while at the manuscript stage. Of course, none of the fine scholars noted is to blame for any errors that one might find in this book. We accept responsibility for whatever shortcomings

xiv Acknowledgments

turn up within the pages that follow. However, as Billy Joel once said, “the mistakes you make are the only things you can truly call your own,” so perhaps we can claim some originality here. We are grateful to David Pervin, our editor, for valuable guidance at each stage of this project. Alex Mintz, Palgrave’s editor for the series in foreign policy analysis, is a superb scholar who offered good advice to us regarding the completion of this book. We are grateful to Yasemin Akbaba for research assistance. An earlier version of chapter 7 appeared in Armed Forces & Society as “Civil–Military Relations in a Neo-Kantian World, 1886–1992,” 2004, 30(2): 227–254. We thank Transaction Publishers for granting us copyright release to publish it here.

Chapter 1

Introduction Since the massive volume of democratic peace studies appeared in full force during the 1990s, several rounds of criticism have been directed toward their theoretical arguments and empirical findings. The present study seeks to contribute to that dialogue with a special emphasis on factors that go beyond the synthesis achieved in Russett and Oneal’s (2001) recent and prominent volume, Triangulating Peace: Democracy, Interdependence, and International Organizations.1 This work, which combines dyadic and systemic factors to create an overall picture of international conflict and peace, is a major step toward a theory that includes a full range of macro (i.e., systemic) and micro (i.e., unit) linkages, that is, macro–macro, macro–micro, micro– macro, and micro–micro linkages (Bunge 1996; James 2002). For example, consider these extensions: Bilateral rivalries might be dampened by increasing economic interdependence and organizational infrastructure at the regional level—a macro– micro connection. A micro–macro-oriented link would be from increasingly common pairs of democracies upward to more peaceful norms regarding negotiation and bargaining at the regional level (Mitchell 2002). Further examples would not be difficult to derive for Russett and Oneal’s state-of-the-art exegesis of neo-Kantianism. In the present study, it is argued further that civil–military dynamics, like the degree of civilian control over the military and whether or not conscription exists, along with factors related to political communications, such as the openness of media and availability of diplomatic channels for conflict management, are essential to a more comprehensive explanation for international conflict, crisis, and war. In this sense,

2 Civil–Military Dynamics

Triangulating Peace provides a foundation for the work at hand. Today’s world revolves around Washington, DC. This statement, in spite of expectations to the contrary based on imperial overextension and even exhaustion, is likely to remain true for decades to come. Thus, the present study places an emphasis on examples related to American foreign policy in making its key points throughout the remainder of this book. Implications from research findings as related to the United States, the unofficial system leader and manager for the foreseeable future, take on special interest. While the present volume seeks relevance to foreign policy in applied terms, as related to both the United States and the world as a whole, its fundamental motivation is to contribute to scientific knowledge through innovative theory and reproducible evidence. More specifically, Militarized Interstate Disputes (MIDs), which provide the focal point for so much of the systematic literature on conflict processes as related to the democratic peace, may be affected by factors not currently included in the most thorough dyadic or micro–micro specifications.2 The present study represents a commitment to identifying factors with high theoretical relevance to MIDs that remain outside of Russett and Oneal’s (2001) triangular model of neo-Kantian peace constructed so far. New factors, such as civil–military dynamics and political communications, are placed within a theoretical model of foreign policy decisionmaking to better explain MIDs and international conflict in general. Put differently, despite the seemingly impressive accumulation of findings about interstate dyads and conflict processes, an alternative to the emerging consensus can be summed up as follows: “even if there is a correlation between regime type and conflict or war proneness, the pattern might be produced by some third factor that has an impact on both war proneness and regime type” (Ray 1998: 36).3 More recently, in “The Flawed Logic of Democratic Peace Theory,” Rosato (2003)

Introduction

3

contends that existing democratic peace scholarship does not provide compelling causal mechanisms between regime type and conflict and calls for an investigation of other factors. Although some recent studies on the democratic peace take that possibility into account,4 this body of research still seems relatively “thin” in the range of factors explored—both domestic and international—that could explain the most persistent and salient results. The present study introduces additional, potentially important factors from the realm of comparative politics, most notably civil–military dynamics that include civil–military relations and military manpower system.5 It also focuses on political communications in terms of diplomatic channels and media openness.6 In other words, this study bridges across subfields of political science in pursuit of a more complete sense of the democratic peace. As will become apparent, the preceding set of regime-oriented and interstate factors, which are not as yet represented in any systematic way within the vast literature already noted, can help to account for the path from interstate dispute to peace. Taken together, these additional factors permit a complete set of linkages, from micro–micro to macro–macro, to be postulated and tested. Epigraphs from Brecher (1999: 231) and Jefferson that appear at the outset of this study provide the stimulus for further research. (Brecher’s observations, to be more exact, are part of his summing up of rational choice as an increasingly conventional approach to the study of international conflict.) Taken together, these commentaries convey the importance of the factors under scrutiny, namely, civil–military dynamics and political communications. Brecher and Jefferson also indirectly draw attention to how little is known about the abovenoted properties in combination with each other as related to international conflict. This study is the first attempt to link civil–military dynamics and political communications to the democratic peace, and a theoretical model of the foreign policy decision-making process, as it has developed so far, to enhance understanding of international conflict. In this sense, the

4 Civil–Military Dynamics

present study can be seen as a new quest for a more peaceful world in the millennium. As a new look at the democratic peace, a potential “Rosetta Stone” for International Relations, this study includes eight additional chapters. Chapter 2 reviews extant literature. Rather than going through all of the literature in detail, this chapter tries to briefly point out major arguments of the democratic peace studies—to summarize various criticisms in four points: (a) paradigmatic debates, (b) neglected issues related to covert intervention and democratization, (c) methodological predicaments, and (d) problematic assertions from rational choice theory. The chapter assesses ten major articles from seven prominent democratic peace scholars—including Oneal and Russett in response to their critics—and then concludes by calling attention to important “third factors” that have been missing from the democratic peace studies. Chapter 3 presents theoretical arguments about how interstate disputes occur in light of a model of the foreign policy decision-making process. The first section explains how foreign policy decision-making process model building came about in the first place. The second section presents a flow chart of the foreign policy decision-making process to convey ideas in a visual and potentially helpful way. Four new variables—(a) civil–military relations, (b) military manpower system, (c) diplomatic channels, and (d) media openness—are portrayed as key components of the model. The third section provides two historical exemplars: (a) Israeli Response to Egyptian Military Actions, May 1967 and (b) the KAL 007 Shootdown, September 1983. The fourth and final section is the conclusion. Chapter 4 presents hypotheses, data, and measurements along with logistic regression and neural network models. This research design chapter, in greater detail, focuses on the introduction of the four new elements outlined in the foreign policy decision-making process model. The four variables are intended

Introduction

5

to complement Russett and Oneal’s (2001) triangular peace model. Both standard logistic regression and potentially pathbreaking neural network models are constructed for the empirical testing discussed in chapter 5. The former technique is used to achieve consistency with the established literature, while the latter brings out some otherwise hidden, nonlinear aspects of the findings. Chapter 5 discusses empirical results in terms of various statistical model specifications, study periods, and methods. Using logit and neural network analysis, this chapter evaluates how MID involvement can be explained and understood in the context of the foreign policy decision-making model outlined in chapter 3. In addition, the role of diplomacy is examined through a contemporary case: U.S. interactions with North Korea over its nuclear program during the first term of President George W. Bush. Qualitative analysis is employed to look into the most persistent anomaly in the quantitative findings in the previous two sections, namely, the apparent connection of diplomatic channels with MID involvement. Chapter 6 discusses results from a data analysis of fatalities in MIDs.7 The purpose of this chapter is to learn more about the most intense and lethal MIDs, which is an intriguing issue for both the theory and practice of international relations. We analyze the determinants of fatal MIDs with a focus on the newly introduced variables from the foreign policy decision-making process model. The analysis provides a further test for both civil–military dynamics and political communications as signposts in the new quest for international peace. Chapter 7 explores a key measurement issue about civil– military relations. More specifically, we measure the degree of civilian versus military control in each state by utilizing the increase and decrease in military expenditure. Military expenditure, however, might be regarded as an indicator of the severity of threats to national security felt by national leaders—in other words, more serious threats could produce higher

6 Civil–Military Dynamics

military expenditures. We look into this possibility by endogenizing military expenditure and other variables such as military manpower. Chapter 8 elaborates the role of media openness in both theoretical reification and empirical testing. Empirical results in chapters 5 and 6 reveal that the world may become more peaceful and harmonious when interstate dyads interact with a high degree of media openness. This particular finding stands out within an exhaustive data analysis. The two chapters reflect on the foreign policy decision-making process model put forward in chapter 3, so they do not examine the underlying causal linkages between media and conflict. In addition, the study period for chapters 5 and 6 is relatively short, from 1950 to 1974. Chapter 8 attempts to deal with these limitations by assessing the power of media openness in a more comprehensive and nuanced manner. The ninth and final chapter summarizes research findings, derives policy recommendations, and suggests directions for future research. Among the four new factors, media appears not only to account most effectively for international conflict and peace, but also rivals or exceeds the standard components of neo-Kantianism. This result implies that, as compared with other, institutional aspects of democracy, media openness stands out as the most defining element of a democratic system. The importance of media in both theory and practice justifies the new quest for international peace that unfolds in this study.

Chapter 2

A Review of Democracy, Peace, and Other Things This chapter reviews extant literature on the democratic peace. After briefly summarizing the core propositions of the democratic peace, it presents a series of criticisms under four headings: (a) paradigmatic debates, (b) neglected issues related to covert interventions and democratization, (c) methodological predicaments, and (d) problematic assertions from rational choice theory. In response to its critics, the recent evolution and prospects of the democratic peace are covered as well. This chapter culminates in the suggestion that current scholarship on peace and conflict can benefit from a foreign policy decision-making process model that incorporates the four key elements noted at the outset of the book: (a) civil–military relations, (b) military manpower system, (c) diplomatic channels, and (d) media openness.

The Democratic Peace and Its Critics Based loosely on Kant’s Perpetual Peace (1795), the most prominent components of the democratic or neo-Kantian peace studies fall into three categories: democraticness, economic interdependence, and joint membership in international organizations. First, democracies are less likely to go to war against each other because of cultural and structural constraints. A common culture of negotiation and bargaining, as opposed to more immediate resort to violence, is one key component. Democratic

8 Civil–Military Dynamics

cultural norms favor discussion over coercion; even the latter, when it occurs, would tend to focus on court proceedings or perhaps peaceful protest in contrast to violent upheaval. The structural component focuses on checks and balances to conflict escalation, which exist not just in the United States but also effectively throughout all institutional democracies. Government structures in a democracy slow down the process of decision-making. A dictator, in other words, can act faster, all other things being equal, than a democratic leader (Maoz and Russett 1993; Russett 1993). Second, states are less inclined to wage war due to high trade or financial interdependence. In other words, economic incentives encourage peaceful relations (Pollins 1989a, 1989b; Oneal and Ray 1997; Oneal and Russett 1997, 1999a, 1999b, 2001; but see also Barbieri 2002). The foundation of this argument also lies in the concepts of “sensitivity” and “vulnerability” that combine to make up interdependence (Keohane and Nye 1977). States with economic linkages may be highly sensitive to each other’s actions; if connections are broken, they may be substantial in volume and create an interest in replacement. Even more intense are the points of vulnerability. If key commodities are involved, or if a situation arises in which a state’s credit is poor and loans are crucial to survival for the government (if not the country), conflict with those who have leverage will be inherently unappealing. Third, and finally, states are less likely to fight owing to greater joint membership in international organizations (Russett 1998; Russett et al. 1998; Oneal and Russett 1999c; Russett and Starr 2000; Russett and Oneal 2001). Perhaps, at one level, this factor anticipates the importance attributed to political communications in the present study. Common membership in international organizations creates additional nodes of communication as well as common interests. Moreover, marginal common interests that encourage creating or joining such entities can widen and deepen over time. A prominent historical example might be the series of institutions in Europe

Review of Democracy, Peace, and Other Things 9

after World War II, which started with the European Coal and Steel Community half a century ago and now embodied in the European Union (EU). So extensive are the common interests represented by the EU that the idea of war between or among such longtime rivals as Britain, France, and Germany seems virtually unthinkable. Support for the democratic or neo-Kantian peace as just described is substantial. The overwhelming majority of studies in each area focuses on interstate dyads in compiling evidence.1 Interstate dyadic analysis not only overcomes the ecological fallacy, but also helps accommodate relational variables such as power preponderance and alliance (Oneal et al. 2003). Despite the outpouring of successful studies of the neoKantian peace, a number of areas of criticism have emerged: (a) paradigmatic debates, (b) neglected issues related to covert interventions and democratization, (c) methodological predicaments, and (d) problematic assertions from rational choice theory. Each is addressed in turn. First, and perhaps most important, is criticism that emerges from the “paradigm war” between (neo)-realism and (neo)liberalism (Brecher 1999). Rather than normative and structural constraints, national interests at stake (e.g., a common external threat) are said to have induced peaceful coexistence among democratic countries (Farber and Gowa 1995, 1997; Mousseau and Shi 1999). The democratic peace, in other words, is the consequence of shared policy preferences or affinities on security (or alliances) among democracies during the Cold War era (Gartzke 1997, 1998, 2000). Peaceful relations among democracies are also associated with the absence of war in the bipolar world (Wolfson et al. 1998; James et al. 1999, 2000)—so the democratic peace may merely have been an artifact of Cold War tension. Although realist inputs into the democratic peace studies have enriched debate, they draw little attention to other potentially crucial elements that have appeared frequently in expositions within comparative politics. Of these studies, the most

10 Civil–Military Dynamics

notable is the prominent structural realist exposition of Waltz (1979), which emphasized the distribution of capabilities among states as related to system propensity for war and bracketed internal characteristics of states. While extremely valuable in pointing out macro–macro patterns, structural realism had the effect of discouraging more micro-oriented theorizing about foreign policy. As a result, an infusion of “new blood” from fields that lie outside international relations, such as civil–military relations, is badly needed among students of international conflict, crisis, and war (Choi and James 2004). It is time to “color outside the lines” of the would-be triangular peace and not necessarily in a way inspired by the realist tradition. Second, the meaning of covert intervention for the democratic peace studies is still an unanswered question (Cohen 1994; James and Mitchell 1995). According to James and Mitchell (1995: 85), “while violence among democratic states is scarce, these states do fight one another—just in ways that are more concealed.” (Emphasis in the original.) Thus, democracies, in a clandestine way, become engaged in war against each other. This criticism, in a sense, also reflects realist perspectives in that covert interventions are often carried out in the name of national security. Mansfield and Snyder (1995a, 1995b, 1996) raise the possibility that, unlike democraticness, transition to democracy (i.e., democratization) is more likely to lead to war. Thus, promoting democratization as a policy prescription may be normatively or culturally counterproductive and even harmful to world peace, although the empirical results from Russett and Oneal (2001: 116–122) would argue strongly against having that concern. Third, methodological shortcomings pertaining to operationalization of variables and empirical models have hindered theoretical refinements of the democratic peace. To begin, measurement of the key concept of democracy creates a controversy because each operationalization produces somewhat different results (Oneal and Ray 1997). For example,

Review of Democracy, Peace, and Other Things 11

democraticness based on a joint regime score (i.e., democracy score minus autocracy score) for each dyad using the Polity data set differs from Dixon’s “weak link” assumption (i.e., choosing the lower score for democracy within each dyad) (Gurr et al. 1989, 1990, 1991; Dixon 1993, 1994; Jaggers and Gurr 1995; James et al. 1999, 2000; Oneal and Russett 2000). More importantly, the credibility of the Polity-based measurement of regime type remains open to question (Gleditsch and Ward 1997; Munck and Verkuilen 2002). The Polity data set is the most widely used source on the authority characteristics of modern polities. It is used to produce standard “democracy,” “autocracy,” and “democracy minus autocracy” scores based on five institutional attributes of democracy: (a) competitiveness of participation, (b) regulation of participation, (c) competitiveness of executive recruitment, (d) openness of executive recruitment, and (e) constraints on executive. Those scores, however, are determined mainly by the degree of constraint on the chief executive. As a result, “using the data summarily to classify modern polities as democracies directs attention away from the actual data that have been collected on authority patterns” (Gleditsch and Ward 1997: 381). In their recent and excellent review, Munck and Verkuilen (2002: 26) also observe that, “not only is virtually no theoretical justification [offered for generating the democracy, autocracy, and combined polity scores], but it also is open to criticism due to the index’s problems of conceptual logic.” In sum, the Polity index and related lineaments are not the final answers to the question of what democracy means, which further encourages a more extensive search for neglected factors that may account for international conflict and peace. Additional method-related concerns focus on statistical models. In addition to a simple logit model, several revised models, such as peace years correction (Beck and Katz 1995; Beck et al. 1998; Beck 2001; King and Zeng 2001), generalized estimating equation (GEE) (Diggle et al. 1994: 142–145,

12 Civil–Military Dynamics

151–152; Zorn 2001), simultaneous equations (James et al. 1999, 2000), fixed effects (Green et al. 2001), distributed-lags (based on Granger-causality testing) (Oneal et al. 2003), and neural network models (Zeng 1999, 2000; Beck et al. 2000; Lagazio and Russett 2004) have been used to deal with the challenges posed by various cross-sectional, time-series data sets. This type of data analysis inadvertently brings with it enormous statistical concerns such as time dependence (i.e., autocorrelation) and heteroskedasticity in the error terms, with higher levels of sophistication in the means of coping through each iteration of the research enterprise. One example pertains to highly skewed dependent variables, which are the norm in foreign policy outcomes as related to conflict processes. Whether wars, disputes, or other events are at issue, their rarity within interstate dyads is normatively desirable but statistically challenging. For such reasons, improvement of data analysis for these events continues to be a priority, with neural network analysis representing the state of the art. Although methodological developments continue to be very important, they can divert attention from real issues to somewhat narrow and inaccessible debates about measurement and statistical analysis. Rather than placing an emphasis on the search for further, plausible causes of international disputes (i.e., factors beyond the triangle identified so far), scholars have devoted a greater proportion of their time and efforts to complex methodological debates. For example, although Beck et al. (2000) introduced a highly sophisticated statistical model, neural network, into the studies of international conflict several years ago, it has not been widely utilized due to its complexity and relatively inaccessible nature. As an economist might describe it, the research enterprise on neo-Kantianism is imposing higher “barriers to entry” at each iteration. So, once again borrowing from the language of economics, there may be some “comparative advantage” in shifting efforts away from statistical methods and toward a more inclusive specification,

Review of Democracy, Peace, and Other Things 13

since ideas for the latter can arise from a much wider range of sources.2 Fourth, and finally, the introduction of rational choice theory into the democratic peace studies has focused directly on the triangle of factors within neo-Kantianism. Bennett and Stam (2000a) employ a game-theoretic model to deal with the democratic peace. Strategic behavior by rational states that pursue self-interest is assumed for interstate dyads, so expected utility maximization is anticipated. International conflict, however, cannot be explained purely by national actors’ selfinterested behavior based on estimated costs and benefits. Misperception, misunderstanding, and miscalculation play ongoing roles in international conflict (Stoessinger 2001). Bueno de Mesquita et al. (2003; for an earlier iteration of the model, see Bueno de Mesquita et al. 1999) present a gametheoretic model of strategic interaction that represents the state of the art in the field. This model offers the most compelling explanation to date for the democratic peace. The model assumes that national leaders are rational, selective, and unified on war-related decision-making. Although the model is effective in explaining efforts by democratic leaders to stay in office and select their adversaries in an optimal way (which ends up being consistent with the democratic peace), it does not address the concerns at the center of the present study, that is, interactions between civilian and military leaders, military manpower system, diplomatic channels, and media openness. If incorporated into the future models, any one of the preceding factors would have the potential to alter equilibria and possibly underlying visions of how foreign policy is derived and implemented. For contemporary students of foreign policy, the triumph of rational choice theory can be sensed in Fearon’s (1995: 382) bold statement that “the main theoretical task facing students of war is not to add to the already long list of arguments and conjectures but instead to take apart and reassemble these diverse arguments into a coherent theory fit for guiding empirical

14 Civil–Military Dynamics

research.” Despite the fact that rational choice theory has made the democratic peace studies more elegant as well as parsimonious, it still does not address what may turn out to be key factors in explaining international disputes. Reassembling the “old” may be Fearon’s priority, but the emphasis here will be on the “new.” Perhaps a metaphor can help explain why: If one is certain that all the pieces of a puzzle are available, then putting them together is the obvious priority. The argument here, however, is that one or more pieces may still be missing; locating these on the basis of sound principles then should take precedence.

The Recent Evolution and Prospects of the Democratic Peace Among scholars of peace and security, Oneal and Russett consistently have attempted to incorporate the by-products of criticism into their research. Oneal and Russett can be regarded as “core” democratic peace advocates because of the cumulative nature of their efforts. Thus, it is worth reviewing their recent works to see how they have responded to critics. In doing so, we can assess how the democratic peace studies as a whole have evolved, because Oneal and Russett have been setting the agenda for some time. We choose to compare their latest five articles, published during the period from 1999 to 2003. Since the five articles are an evolutionary “manifesto” of their cumulative learning about the democratic peace during the last decade, this substitutes effectively for reviewing the full extent of their previous work. The main focus of this review is to assess the degree of convergence in empirical findings. Table 2.1, which incorporates the year of publication, study period, table number, statistical method, and results obtained for democraticness, economic interdependence, and joint membership in international organizations, should suffice for the preceding purpose.

Table 2.1. Empirical Results of Oneal and Russett Published year Study period Table Method 2003

1885–1992

1

Distributed-lagsa a

Democraticness Economic interdependence International organizations ***

***

***

2001

1886–1992

1

GEE Fixed effectsa Fixed effects w/distributed-lagsa

*** *** ***

*** _ *

n/a n/a n/a

1999c

1886–1992 1886–1992 1886–1939 1886–1992

1

GEEa Peace years correctiona GEEa GEEb

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

*** ** ** **

_ ^^^ _ *

1999b

1870–1989 1870–1989 1870–1989 1870–1939 1950–1985 1950–1985

1

Peace years correctiona GEEa GEEa GEEa GEEa GEEa

***/^^^ ***/_ ***/*** ***/*** **/_ */_

* * ** ** ** *

n/a n/a n/a n/a n/a n/a

Peace years correctiona Peace years correctionb GEEb

*** *** ***

_ * **

n/a n/a n/a

1999a

1950–1992

2 1

a

All dyads. b Politically relevant dyads. GEE: generalized estimating equation. * p  .05; ** p  .01; *** p  .001; ^^^ p  .001, wrong sign; _: not statistically significant; n/a: not applicable.

15

16 Civil–Military Dynamics

Table 2.1 reveals some variation in the empirical findings of Oneal and Russett. With more extensive data collection, the study period has been extended; in addition, statistical methods have become more sophisticated—from simple logit to distributed-lags models. It seems obvious from the table that availability of more extended data series has contributed to evolution of the democratic peace research. Analysis for the years 1886–1992 reveals significant effects for democracy and interdependence, although a controversial and even questionable fixed effects model for the shorter years 1951–1992 may not support the liberal peace (Oneal and Russett 2001: 475; Russett and Oneal 2001). Of course, the question then becomes which era to heed in developing theory and policy. The post–Cold War era, in particular, would seem inherently more relevant than the days of the Congress of Vienna or even the World Wars or the Cold War itself. Thus, data analysis for the 1990s, as it becomes feasible through data collection, promises to be the most interesting of all. Democraticness (as measured in the Polity data) seems to be statistically significant on a consistent basis.3 However, it should be noted that, as Oneal and Russett’s (1999b) findings indicate, the results in table 2.1 for low democracy (on the left side of the slash) and high democracy (on the right side) are not consistent. Oneal and Russett (1999b: 223) report that, “holding DEML [lower democraticness score in a dyad] constant, the greater the democracy of the more democratic state (DEMH, [higher democraticness score in a dyad]), the more likely the occurrence of a dispute.” (Emphasis added.) This result is consistent, of course, with the overarching characterization of results by Oneal and Russett themselves: two democracies are peaceful, two autocracies less so, and one of each will fight much more. Given recent foreign policy actions by the United States, the preceding statistical result invites speculation about the impact of very high discrepancy in dyads for democraticness where power discrepancy also is significant. Could it be that states

Review of Democracy, Peace, and Other Things 17

scoring especially high on democraticness might be more proactive, allowing for the capability to do so, when it comes to regime change? The obvious reference here is the United States in relation to Afghanistan and especially Iraq. The war with Afghanistan might be explained as a simple “action– reaction” process, with the United States seeking to destroy the home base of the Taliban in response to the 9/11 disaster. War with Iraq, however, is not explained in such a facile manner and controversy continues within the United States itself as to whether appropriate justification existed for the use of force to overthrow Saddam Hussein. One intriguing possibility, then, is an interaction effect between quasi-hegemonic power and highly discrepant scores on democracy. The statistical findings for DEMH reported above might encourage further thinking in that direction. Statistical significance for economic interdependence in table 2.1 seems to vary with the statistical methods employed. Fixed effects and peace years correction models seem to weaken the impact of economic interdependence and even remove its significance. For example, in their article implementing the first peace years correction model, based on Barbieri’s (1998) data, Oneal and Russett (1999a: 430) observe that “we find no evidence with this specification and set of cases for interdependence reducing the likelihood of a militarized dispute,” although it should be pointed out that Russett and Oneal (2001) have conducted a compelling series of data analyses that do produce a pacific effect for trade. One salient explanation for a potentially conflict-prone effect (albeit in some potentially narrow context) may be that interdependence is a double-edged sword. As noted above, sensitivity and vulnerability create a context in which interdependent actors have an incentive to preserve an amicable relationship. The sheer volume of contact embodied in interdependence, however, also creates recurring opportunities for conflict. If other factors become sufficiently salient, such as domestic political imperatives, disputes and wars are by no means out of the question for closely connected states. All other

18 Civil–Military Dynamics

things being equal, interdependence should discourage interstate conflict, but in some contexts it might create friction as well (see Barbieri 2002). Joint membership in international organizations, the third component of neo-Kantianism, needs further rigorous testing in light of mixed results. Evidence for the pacific effects of IGOs becomes more impressive, however, when the increasing trend in the number of international organizations is factored out. Like interdependence, perhaps international organization membership can have bidirectional effects with some potential for conflict in opposition to a generally positive and cooperationinducing impact. Familiarity can breed contempt and states in many common international organizations might find themselves, for whatever reasons, on opposite sides of any number of issues. Persistent conflict then might even tend to enhance the potential for interstate violence. In this respect, Gallarotti (2001: 386) aptly points out that, since international organizations can provide justification to members’ foreign policy positions, these entities “are often used to promote and magnify conflict.” Among the five articles, the empirical results of the most recent one (Oneal et al. 2003) appear very impressive: All three neo-Kantian peace hypotheses are supported strongly. One reason for that superior performance under testing might be an improved methodological technique, the distributed-lags model, which takes into account endogeneity among variables, that is, possibly important reciprocal relations among the three neo-Kantian peace factors, realist factors, and the probability of interstate disputes. On balance, it seems that the distributedlags model has been underutilized in the current scholarship of international conflict and peace. Perhaps it provides the “bottom line” on the neo-Kantian factors; if so, preceding speculation about confounding effects would seem to be off base, at least from an empirical point of view. Table 2.2 presents the empirical results of five other democratic peace studies, which when compared with Oneal and Russett’s works reveal other developments. Based on a logit

Table 2.2. Empirical Results of Five Other Democratic Peace Studies Author (year)

Study period

Table

Method

Democraticness

Description

Hensel et al. (2000)

1816–1992

1

Descriptive statistics

Low probability of rivalry onset

Confirm the robust and pacifying effects of democraticness in terms of rivalry onset

3

Descriptive statistics

Low frequency of democratized rivalries

Confirm the robust and pacifying effects of democratized rivalries

4

Logit regression

Statistically significant

High democracies are about 3 times more likely than high autocracies to resolve their MIDs through compromise

Bivariate statistics

Statistically significant

The democratic peace propositions are re-evaluated and their results show statistical significance

Mousseau (1998)

1816–1992

Maoz (1998)

Various

Enterline (1998b)

1816–1992

2

Spline logit regression

Statistically insignificant

New autocracies, not democracies, are more dispute prone

1946–1992

3

Spline logit regression

Statistically insignificant

New autocracies, not democracies, are more dispute prone

1816–1992

4

Spline logit regression

Statistically insignificant

New autocracies, not democracies, are more war prone

Literature review

The pacifying impact of democraticness

Most of the democratic peace-related literature support the democratic peace propositions

Ray (1997)

19

20 Civil–Military Dynamics

regression of data from 1816 to 1992, Mousseau (1998) ascertains that “high democracies” (i.e., higher democracy score in a democratic dyad) are about three times more likely than “high autocracies” (i.e., higher autocracy score in a autocratic dyad) to resolve their MIDs through compromise. Since compromise is one of the central defining attributes of democratic political culture, Mousseau’s findings strengthen the normative argument for the democratic peace. With a focus on interstate rivalries during the period 1816–1992, Hensel et al. (2000) confirm and extend the robust and pacifying effects of democracy on both rivalry onset and rivalries between democracies. In response to the criticism related to transition to democracy (i.e., democratization), Enterline (1998b) employs spline logit regressions during the periods 1816–1992 and 1946–1992 and finds that new autocracies, not democracies, are more dispute- and war-prone. However, democratized dyads do not show statistical significance in these results. In other words, democratization does not appear to decrease the likelihood of international conflict in a noteworthy way. Ray (1997) and Maoz (1998) provide overall appraisals of the debates on the democratic peace. In response to realist- and culturally oriented criticisms, Maoz’s bivariate statistics show that the democratic peace proposition is supported. After reviewing the pros and cons of the democratic peace proposition, Ray (1997: 63) concludes that “the evidence is sufficiently robust to justify prudent efforts and policies designed to take advantage of the pacifying impact of democracy.”4 Since the preceding five articles stress the impact of democraticness rather than the other two neo-Kantian peace factors (i.e., economic interdependence and joint membership in international organizations), table 2.2 does not provide a point of comparison as related to these two. It should be noted, however, that most democratic peace studies continue to pay closer attention to the effect of regime type than the other two factors as related to war and peace. Notable exceptions are Oneal and Russett’s serial works as shown in table 2.1, which attempt to incorporate

Review of Democracy, Peace, and Other Things 21

all three neo-Kantian peace factors into various models. In short, the empirical results from the preceding tables indicate that the results are mixed and further research is essential. Debates surrounding the democratic peace obviously are not complete, so we may expect to find some other plausible causes of peace or conflict beyond those addressed in the studies so far. As Rosato (2003: 599) points out, “the causal logics that underpin democratic peace theory cannot explain why democracies remain at peace with one another because the mechanisms that make up these logics do not operate as stipulated by the theory’s proponents.” Innovative, even revolutionary, ideas might come from comparative politics, a field that students of international relations need to communicate with in a more sustained way. If we want to learn more about international conflict, crisis, and war, these two subfields must be joined together more effectively to produce scholarly innovation—this is the most promising path toward a “scientific revolution” (Kuhn 1962), if there is one to be had. In chapter 3, we explore the possibility that current scholarship on peace and conflict can benefit from such an endeavor by building a model of the foreign policy decision-making process that incorporates the four key elements noted at the outset: (a) civil–military relations, (b) military manpower system, (c) diplomatic channels, and (d) media openness.

This page intentionally left blank

Chapter 3

A New Look at International Conflict and Peace: Four Factors to Consider This chapter provides a theoretical exposition in the form of a foreign policy decision-making process model. The model traces how international conflict between two states can occur with an emphasis on the roles played by the following four factors: (a) civil–military relations, (b) military manpower system, (c) diplomatic channels, and (d) media openness. Conceptual model building will pay close attention to these four factors, which democratic peace studies have yet to consider. The first section explains the inspiration for (or intuition behind) the model. The second section focuses on a flow chart that shows how interstate conflict occurs in relation to the four factors. The third section provides two historical exemplars, while the fourth and final section offers some conclusions.

Intuition, or, How does It Work? Democratic peace studies generally are put forward without an explicit model of the foreign policy decision-making process to guide empirical analysis.1 Instead, the “institutional” model appears in an implicit sense; it is inferred that democracies in conflict with each other are less likely to pursue escalation

24 Civil–Military Dynamics

because decision-making takes significantly longer and therefore allows “cooler heads to prevail” (Maoz and Russett 1993). This chapter attempts to fill the theoretical gap by presenting a foreign policy decision-making process model that highlights the roles of leadership and communications in selecting war versus peace. Our model building is inspired by the substance and drama of international relations as depicted by films such as Hollywood’s Thirteen Days and the documentary Thirteen Days in October: Cuban Missile Crisis. These films, to cite just two of many examples available in popular culture, vividly depict the dynamic interactions between civilian and military leaders at the brink of war. With some dramatic license, the two films show how civilian and military leaders in the Kennedy Administration interacted with one another in the process of coping with the Missile Crisis in October 1962. The following dialogue from Thirteen Days exemplifies well these dynamic interactions: In his Oval Office on Thursday, October 18, President Kennedy asked, “General LeMay, do you truly believe [the air strike] is the best course of action?” General LeMay responded, “Mr. President, I believe it is the only course of action. America is in danger.” (Interactions between the president and the general continued.) General LeMay then remarked, “I think a blockade or a bunch of political talk would be considered by a lot of our friends and neutrals as a pretty weak response.” Thus, LeMay moved easily into a military frame of reference at or near the outset of the crisis. Overall, while civilian leaders such as John F. Kennedy (President), Adlai E. Stevenson (American Ambassador to the United Nations), and Robert F. Kennedy (Attorney General) preferred diplomatic solutions to military ones, military leaders like General Maxwell D. Taylor (Chairman of the Joint Chiefs of Staff) and General Curtis LeMay (Air Force Joint Chief) called for immediate military actions—including higher levels of mobilization. During the Missile Crisis, the likelihood of respective options seemed to oscillate as a function of the

Four Factors to Consider 25

power shifts between the preceding two groups. For example, it is interesting to consider the possibility that, by intentionally increasing the military alert status from DEFCON (Defense Condition) III to DEFCON II, military leaders may have been attempting to provoke hostilities with Cuba or even the USSR to push through their agenda. The Missile Crisis de-escalated through political communications, especially when Robert Kennedy and Anatoliy F. Dobrynin (Soviet Ambassador to the United States) reached a secret diplomatic agreement.2 Given the intuition about the usual inclinations of military and civilian leaders with respect to their preferred means of conflict management, the model developed here recognizes the interactive nature of the foreign policy decision-making process and is consistent with the vision offered by the democratic peace. A flow chart (i.e., figure 3.1, to be introduced momentarily) will be utilized to represent the intuition from the films and scholarly record. In sum, there will be more to leaders’ decision-making than what might be conveyed by variables restricted to national and international attributes such as democraticness, contiguity, and other factors found in the increasingly paradigmatic research designs on interstate dyads over the last two decades. Since the leader of a state makes a war decision, the cause and effect should be established on that basis: “selection of war or peace is a choice that is initiated, conducted, and concluded by individual leaders. . . . The choice of war or peace depends on the choices of individuals and not on compulsion by circumstance” (Bueno de Mesquita 1981: 5; Stoessinger 2001). In the end, national leaders are the ones who make war or peace, as put forward in the classic exposition from Waltz (1959; see applications in Stoessinger 2001). While the international system and national attributes might be regarded as constituting the permissive conditions for war, in the end, decisions are made by people rather than anthropomorphic entities. For that reason, civil–military relations, as opposed to national attributes such as regime type or the distribution of capabilities,

26 Civil–Military Dynamics

could turn out to be most important with regard to the process of conflict escalation or de-escalation. Since each foreign policy decision-making process includes civilian as well as military leaders, interactions between the two sets of officials are expected as a matter of course. It is essential, especially in times of crisis, to make decisions that include both political and military considerations.3 All other things being equal, it is assumed that conventional wisdom holds, namely, that “where you stand depends on where you sit” (Allison 1969, 1971; Gray 1975: 86; Allison and Zelikow 1999): While civilian leaders are anticipated to be dovish, military leaders are expected to be hawkish, at least in relation to each other (Allison et al. 1985).4 Students of civil–military relations have put forward the idea that, as military leaders increase their influence in the foreign policy decision-making process, the state becomes more likely to be aggressive and belligerent, which in turn may result in war (Schofield 2000: 132).5 Since military leaders are what Lasswell (1941, 1997) called “specialists on violence,” they have a lower aversion to interstate war than do civilian leaders. Thus, military decisionmakers who participate in foreign policy decision-making “are likely to urge or endorse the use of force and regard it as a proper and feasible step” (Morgan 1993: 246). Along the same line of reasoning, Brecher (1996: 220) contends that “[the military] in power are likely to employ violence or more severe violence, even if alternative techniques of crisis management are available.” However, not all students of civil–military relations agree with the preceding assertions about “hawkish” military leaders and “dovish” civilian leaders in decision-making processes. Modernization of military technology, information, and organization as well as gender balance make military decision-makers increasingly similar to civilian leaders, so that they tend to advise caution rather than adventurism (Perlmutter 1986; Feaver and Gelpi 1999, 2004; Schofield 2000). Several studies of American foreign policy go as far as to suggest the opposite

Four Factors to Consider 27

relationship, that is, civilian hawks and military doves (see Huntington 1957; Janowitz 1960; Betts 1977). Huntington (1957: 70), in a landmark study, The Soldier and the State, argues that “the tendency of the civilian politician is to court popular favor by curbing the arms budget and simultaneously pursuing an adventurous foreign policy. The military man opposed both tendencies. . . . Believing in the ultimate inevitability of war, [the professional military man] raises the strongest voice against immediate involvement in war.” Betts’ (1977) classic work, Soldiers, Statesmen, and Cold War Crises, maintains that military professionals are less likely than civilians to advocate the use of force because (a) they have a better appreciation for how “chancy” it is; and (b) if things go badly, their lives are on the line. Along similar lines, Andreski (1992: 103) argues that “military dictators (at least in modern times) have been notably pacific in external relations, while all the most aggressive and successfully imperialist polities have been ruled by civilians.” This follows intuition, in one way: Military dictators might be more likely to understand the risks entailed by war via their direct experience at either the tactical or the strategic level or perhaps even both. Most recently, Feaver and Gelpi (1999, 2004) report empirical findings that American military leaders were more dovish than civilian ones during the period from 1816 to 1992.6 Furthermore, in some countries, civilian leaders sometimes come from the military, usually after extensive careers; President Dwight D. Eisenhower, U.S. Secretary of State Colin Powell, and Israeli Prime Minister Ariel Sharon are good examples. The important question then becomes this one: Are they military leaders or civilian leaders? Perhaps one way to reconcile the two different arguments would be to unpack the basic logic of the “Powell Doctrine” which in various respects is compatible with both views. Powell, speaking for most American military professionals, advocated using force only in very restrictive circumstances, which is in line with the

28 Civil–Military Dynamics

arguments of hawkish civilian and dovish military. However, if it did become necessary to go to war, Powell would want to do so without restrictions on its scope and intensity (see Lane 1995; Korb 1997; Campbell 1998). Most of the available case studies in the context of civil– military relations and the propensity toward war focus on great powers. Through an examination of civil–military relations in Germany, France, and Russia before World War I, Van Evera (1984) and Snyder (1984) uncover “the cult of the offensive” in those states, with military leaders being more inclined than civilian ones to use military force. Snyder (1984: 140) argues that the cult of offensive was also played out in Japan after World War I. In this regard, Rousseau (1996: 69) sides with Snyder in asserting that, during the era of imperial Japan, the weak power of the civilian elites versus that of the military frequently facilitated international disputes. Some studies that focus on the United States, the leading power over the last century, reach similar conclusions about the tendencies in military leadership with respect to strategy and tactics. Vietnam is a case in point: Donovan (1970: 100) argues that the military “participated in what may not have been exactly a conspiracy but was at least a well-organized readiness—indeed, an inclination—to get into the [Vietnam] war.” Along the same lines, Allison’s (1969: 714; see also 1971) bureaucratic politics paradigm explains well the military leaders’ hawkish position during the Cuban Missile Crisis as follows: “To the Joint Chiefs of Staff the issue was clear. Now was the time to do the job for which they had prepared contingency plan. . . . As the President recalled on the day the crisis ended, ‘An invasion would have been a mistake—a wrong use of our power. But the military are mad. They wanted to do this.’ ” (Second and third emphases added.) While hawkish military leaders should be inclined to emphasize the importance of military strength to deter war and be ready to use military forces, dovish civilian leaders are expected to prefer cooperation or negotiations through political communications

Four Factors to Consider 29

such as diplomatic channels and media (Russett 1990–1991: 516). Thus, in terms of relative influence for political and military leaders, two possibilities must be assessed: political (military) stronger than military (political).

A Flow Chart of the Foreign Policy Decision-Making Process Model Figure 3.1 depicts the foreign policy decision-making process as a model that incorporates four factors that so far have been exogenous in the democratic peace. These factors, two of which focus on processes within states and two that pertain to dyadic communications, have the potential to affect the likelihood of MIDs: (a) civil–military relations at a general level, in terms of relative influence; (b) military manpower system as a more specific manifestation of civil–military relations; (c) diplomatic channels; and (d) media openness. While democratic peace scholarship has not yet taken into account the collective theoretical significance of these elements, the same criticism, of course, could be leveled at all of the other leading research programs in international relations. At this point the model will be presented, with specific attention to the role of the preceding factors in relation to interstate disputes. Figure 3.1 illustrates how two states, A and B, interact with each other through foreign policy decision-making processes. Whenever State A is engaged in a conflict with State B or vice versa, civilian and military leaders are convened to seek out plausible solutions. In the process, civilian leaders must interact with military ones to identify the best options. Interactions between military and civilian leaders are anticipated to lead to two possible courses of action, depending on which group’s influence becomes stronger. When the civilian elite’s presumably dovish view prevails in a conflict (i.e., if M  C in figure 3.1 for either State A or B), a state is presumed to be more willing to attempt to resolve its

30

Conflict between State A and State B

Decision process for State B

Decision process for State A

Peaceful settlement If M > C

Interactions between military & civilian leaders

Greater influence for military leaders

Military measures (mobilization)

Key: M = Influence of military leaders C = Influence of civilian leaders

If M < C

If M < C If successful

Greater influence for civilian leaders

Political communications: diplomatic channels media openness

Interactions between military & civilian leaders

Greater Influence for civilian leaders

If unsuccessful

Militarized interstate disputes

Figure 3.1. Foreign Policy Decision-Making Process Model

If M > C

Greater influence for military leaders

Military measures (mobilization)

Four Factors to Consider 31

crisis through political communications such as diplomatic channels or the media to reach a peaceful settlement. In other words, when dovish civilian leaders in both states A and B in a dyad interact, they are more likely to prefer nonmilitary solutions to military solutions. Various diplomatic contacts, through an ambassador, envoy, or minister, chargé d’affaires, or mediators, and the like provide State A with at least some opportunities to negotiate or communicate peacefully with State B, or vice versa (Singer and Small 1966; Small and Singer 1973). Since media openness refers to the ability to “exchange information swiftly enough and with enough volume to effect the news media coverage in [each] country and influence the decision-making context for the [concerned] leader[s]” (Van Belle and Oneal 2000: 55), it can also facilitate political communication between the two states. Peaceful settlement therefore becomes more likely as a function of the interactions between free presses of two opposing states. Failure of efforts to communicate, however, is expected to bring about a military approach instead—or even war as the time-honored pursuit of diplomacy by other means (Clausewitz 1976; Gochman and Maoz 1984: 587–589). This process is depicted in figure 3.1, where unsuccessful communication increases the degree of influence for military leaders and leads to military measures. When, by contrast, the military elite’s generally more hawkish standpoint becomes dominant in a conflict (i.e., if M  C in figure 3.1 for either State A or B), a state is more willing to resort to military measures and thus the dyad’s overall propensity toward a dispute is heightened. When the military is powerful enough to insist on flexing its “muscle” (as illustrated in the Hollywood movie, Thirteen Days), a higher likelihood of violence comes along with that posture. Military leaders in both states, put simply, are more likely to prefer military solutions. Furthermore, if State A or B can be ready immediately to deploy its soldiers under some kind of a conscription system, it becomes even more likely to initiate or become involved in a MID—hence the importance of civil–military dynamics along

32 Civil–Military Dynamics

dual dimensions, including the degree of civilian control and the acquisition of military personnel. In this course of events, a MID is the subsequent result—a consequence of the process of the foreign policy decision-making procedure from figure 3.1. It should be noted that, for ease of exposition, the above discussion occurs partially at the monadic or state level. However, it is clear that, when military influence increases in both states A and B, the likelihood of interstate dispute should reach its apex as a result of synergy. By contrast, when civilian supremacy maintains in both states A and B, the likelihood of peaceful settlement through political communication should be at its maximum. Put differently, we infer that, if both states in a dyad interact at the point of the sword, they are more likely to be at swords’ points due to their hawkish nature. This logic may also be applicable to a situation in which State A is militaristic while State B is not (or vice versa) (i.e., one of two states may have M  C and the other M  C). According to Dixon (1993, 1994), the probability of conflict is determined primarily by the more autocratic of the two states in a dyad. This state is inferred to be the “weak link” in the chain holding peace together. The less constrained state should be more inclined toward military means and therefore more likely to be responsible for an interstate dispute in such a heterogeneous dyad. In short, the model predicts either military or peaceful measures as a result of the profiles of two interacting states that are matched within the foreign policy decision-making process model. The model focuses on the causes of MIDs in general, including wars. Interstate wars are uncommon and it is not easy to generalize regularities and patterns in why and how they break out (Russett and Oneal 2001: 94).7 For example, James et al. (1999: 11–12) illustrate graphically how rarely international wars occur. Based on data from 1950 to 1986, they find a highly skewed distribution among the standard MID event categories: of the 16,322 dyad-year observations, 97 percent are at peace, 0.14 percent exhibit threat, 0.43 percent show a display of force,

Four Factors to Consider 33

2.05 percent include use of force, and 0.38 percent qualify as international wars. The relative and absolute scarcity of war makes any empirical generalization difficult. Yet, as Russett and Oneal (2001: 94) argue, “the influences and constraints that affect the occurrence of wars do not appear to differ much from those that are relevant to militarized disputes in general.”8 Since the causes of war in particular also seem to work for MIDs in general, we use the same more inclusive data on MIDs for the empirical analysis as Russett and Oneal (2001) (i.e., all dyads rather than politically relevant dyads).9 The model is parsimonious, so it does not catch all of the dynamics associated with each factor just noted; instead, it identifies highlights and patterns. One example is that the model may be expanded into two sequential stages of interaction. If M  C, then we expect interactions and communication, but still can get to the conflict stage if contact is not successful. The failure of political communications via diplomatic channels or media openness does not guarantee military measures in the real world—a point not yet conveyed. The model, however, does point toward a new direction in research on peace and war: It captures a simple but interactive foreign policy decision-making process between two states and is sufficient for present purposes, namely, to assess the importance of some neglected factors in explaining why leaders might decide in favor of military versus other options when foreign policy action is needed.

Two Historical Exemplars Israeli Response to Egyptian Military Actions, May 1967: Greater Influence from Civilian Leaders Greater civilian leaders’ influence in a foreign policy decisionmaking process can be inferred from the Israeli decisions in response to external threats from Egypt and other Arab states in the spring of 1967. After demanding the withdrawal of

34 Civil–Military Dynamics

U.N. peacekeeping forces from the Sinai on May 16, Egyptian President Gamal Abdel Nasser moved Egyptian military forces into the desert area bordering Israel. On May 22, Egypt blockaded the Straits of Tiran, which prevented Israeli shipping access to the Straits. This incident prompted an urgent meeting of the Israeli foreign policy decision-makers on May 23. They met first as the Ministerial Committee on Defense and then as the cabinet. These foreign policy decision-makers had to deal with two critical issues: (a) the aggressive military actions from Egypt, and (b) a request from U.S. President Lyndon Johnson, who pressed Israel to defer any military action for 48 hours to pursue other possible diplomatic alternatives at the international community level. From the beginning, Yitzak Rabin, the then Military Chief of Staff, argued that immediate military action was necessary to maintain the credibility of Israel’s deterrence policy. Abba Eban, the then Foreign Minister, asserted that military action should be put off in favor of diplomacy. He argued that Israel must consult with the United States before taking any action. More heated debates (or interactions) followed between civilian and military leaders, centering on Eban and Rabin. At the final stage, Rabin and the other military leaders were persuaded that a delay of 48 hours in the initiation of military action would not be as critical as they had thought. As a result, the cabinet members unanimously agreed to the Eban position of diplomatic consultation with the United States. In this case, along with the diplomatic efforts of the international mediator, U.S. President Lyndon Johnson, the civilian leaders’ influence in the foreign policy decision-making process was greater than the military (Hermann et al. 2001: 156).

The KAL 007 Shootdown, September 1983: Greater Influence from Military Leaders A fighter jet from the former Soviet Union shot down the Korean Air Lines (KAL) 007 on September 1, 1983, which

Four Factors to Consider 35

claimed the lives of 269 people. After detecting the intrusion of an unidentifiable aircraft into its airspace, an air defense control base in Sakhalin sent six interceptors. For a while, they could not find the aircraft. One of the Soviet interceptors located the KAL 007 sometime later, which gave the ground base and fighter pilot less response time (Pearson 1987: 65–80). After detecting the KAL 007, there were two options: forcing it to land or shooting it down. Under the rigid Soviet system, the standard organizational procedure regarding airspace intrusion was that the Soviet air defense authorities would be “inclined. . . . to shoot first and think about it later. No Soviet air defense officer [was] likely to be court-martialed for shooting at a foreign aircraft. He could even be shot for allowing a foreign plane to enter, and then make good its escape” (Harsch September 22, 1983: 23). The conventional practice was also ascertained by Soviet border law and internal instructions, which indicated that the Soviets would, “in the last resort, shoot down aircraft of any character overflying prohibited military zones if they fail to obey instructions or warning signals to land” (Maechling October 14, 1983: 16) regardless of civilian or military, deliberate or unintentional, presence. Since the fighter pilot could not force it to land, he was told to destroy the KAL 007 before it escaped from the Soviet airspace. More importantly, according to Dusko Doder (October 5, 1983: A1), a journalist who has reported extensively on Soviet and Eastern European affairs for the Washington Post, “the Soviet Far East command had been in direct telephone contact with top military officials in Moscow on several occasions prior to the downing of the plane . . . [but] the political leadership had not been consulted.” That is, the Soviet defense leaders made the decision on the basis of information provided by local commanders who were anxious about losing the intruding airplane as it moved out of the Soviet airspace. Although this case provides little information on the interaction between civilian and military leaders at the time of the shootdown, it is evident that the KAL 007 tragedy occurred due to greater

36 Civil–Military Dynamics

influence from Soviet military leaders who, quite notoriously, had louder voices in decision-making under the rigid Soviet political system during the Cold War era.

Where To? The flow chart of the foreign policy decision-making model in figure 3.1, to sum up, depicts dynamic interactions between civilian and military leaders who attempt to search for political or military solutions in response to an interstate conflict; states can interact in one of two ways: either political communications that produce a peaceful settlement or military measures that produce a MID. The model assumes that, all other things being equal, civilian leaders prefer political communications such as diplomatic channels or media openness to military measures, and military leaders prefer things the other way around (e.g., military mobilization versus negotiations) in coping with an interstate conflict. The model predicts different outcomes that depend on which leadership group, if either, becomes dominant in the interactive foreign policy decision-making process. To be more precise, the model assigns a higher likelihood to one outcome versus another as a function of relative influence between groups in decision-making. The two historical cases provide clear exemplars in line with the foreign policy decision-making process model depicted in figure 3.1. Since the democratic peace studies do not take into account the decision-making process depicted above, so far, collectively speaking, they seem to have produced an outcome like “the blind men and the elephant” (James 1988: 78–91). Democratic peace researchers touch an elephant’s “tail” (i.e., test the regime type factor) and then claim that an elephant is just like a snake—the dyadic democratic nature based largely on executive constraints of the Polity data set (see Gurr et al. 1989, 1991; Gleditsch and Ward 1997; Munck and Verkuilen 2002) is the essential cause of peace. Just as a blind man would need

Four Factors to Consider 37

to touch the whole body of an elephant to understand its overall size and makeup, democratic peace scholarship should begin to incorporate considerations such as civil–military relations, military manpower system, diplomatic channels, and media openness into the process of modeling. It is essential to investigate the politically important impact of intrastate factors beyond the type of government—otherwise, reductionism in terms of political structure will continue and ultimately serve to limit understanding of international conflict, crisis, and war.10 Of course, this line of reasoning applies not only to the democratic peace enterprise but also to any major finding that might encourage closure in terms of causal analysis. The democratic peace is used as the example here precisely owing to its prominence for such a sustained period of the field’s development. In fact, the democratic peace research enterprise, to date, would represent the scientific ideal as represented by its accumulation of findings through carefully replicated stages of research, that is, incremental progress toward greater knowledge. In the next chapter, we present the research design to test the four factors (i.e., civil–military relations, military manpower system, diplomatic channels, and media openness) described in the foreign policy decision-making process model depicted in figure 3.1.

This page intentionally left blank

Chapter 4

Research Design This chapter begins with a presentation of hypotheses. Measurement and data-related issues are then addressed. Logistic regression and neural network models provide the basis for empirical analysis in chapters 5 and 6.

Hypotheses Based on the foreign policy decision-making process model, as conceptualized in figure 3.1, four hypotheses are derived below. Students of civil–military relations have put forward, but not as yet tested with aggregate data, the notion that an asymmetrical relationship between civilian and military elites may contribute to various military disputes or acts of aggression in international relations. That is, as military leaders in both states A and B increase their influence in decision-making processes, the states become more likely to be aggressive and belligerent, which in turn may result in war (Hill 1978; Zuk and Thompson 1982; Diehl and Kingston 1987). This basic connection is put forward in various ways throughout the literature on international conflict processes. Morgan (1993: 246) argues that military decision-makers who participate in decision-making “are likely to urge or endorse the use of force and regard it as a proper and feasible step.” Based on International Crisis Behavior (ICB) Project data analysis, Brecher (1996: 220) reports that “[the military] in power are likely to employ violence or more severe violence, even if alternative techniques of crisis management are available.” Brecher’s conclusions are in line with Lasswell’s

40 Civil–Military Dynamics

(1941, 1997) famous phrase describing the military as “specialists on violence.” Based on his cross-national statistical analysis of comparative foreign policy during the ten-year period between 1959 and 1968, Geller (1985: xviii) finds that “nations in which the military has substantial influence on the policy process exhibit foreign behaviors that are more conflictual and less cooperative than nations in which the military lacks this input.” Barnett (1970: 369) identifies the power of the army in Germany as a crucial factor in causing World War I. During the era of imperial Japan, to cite what is arguably the archetypal case, the weak power of civilian elites versus the strong power of the military frequently facilitated international disputes (Rousseau 1996: 69). It is also likely that, whenever military influence increases within two states in a dispute, higher stakes will ensue through a tendency toward escalation. In other words, all other things being equal, as military influence rises, military measures are expected to come to the fore. Thus, the first hypothesis about civil–military relations is as follows:1 H1: As the relative influence of the military in civil–military relations increases for a dyad, the likelihood of involvement in MIDs also increases for that dyad.

Students of civil–military relations have also argued that a state’s system of military manpower may affect the likelihood of its involvement in international conflict. In particular, international disputes may be more likely to occur in the presence of conscripted forces because of their quicker and higher military readiness or preparedness than all-volunteer forces (Ross 1994). In this regard, Duindam (1999: 119) contends that, “if the members of the democratic community differ in their preferences for a war, a decision in favour of a war will be more probable in the case of a conscription force than in the case of an all-volunteer force.” In the same vein, in Triangulating Peace, Russett and Oneal (2001: 19) point out that Napoleon’s

Research Design 41

conscripted soldiers had the potential to overwhelm the volunteers of other states, so France threatened to become a hegemon. On the basis of cross-sectional data for 1980, Anderson et al. (1996: 199–200) ascertain as well that “warlike” states are more likely to employ conscription. From data of a sample of 143 countries for 1984, White (1989: 780) finds that “countries that use conscription may be more likely to become involved in wars because they maintain larger armed forces and the cost to the government of getting additional soldiers is reduced by conscription.” In other words, conscription reduces the relative costs entailed by the pursuit of a military option—the most basic means are already available. In particular, if the two states in a dyad employ conscription, the likelihood of interstate dispute should be enhanced greatly because of the opportunity for adventurism. Thus, the hypothesis about military manpower systems is as follows:2 H2: Dyads composed of states with (without) a conscription system are more (less) likely to become involved in MIDs.

Figure 3.1 also stressed the importance of diplomatic efforts in unfolding international disputes. Dougherty and Pfaltzgraff (2001: 79) summarize a wide range of research that confirms “a crucial role in the preservation of peace” for diplomacy.3 When the opinion of the civilian leaders prevails, various diplomatic channels, such as an ambassador, envoy, or minister, chargé d’affaires or mediators, are more likely to be utilized to reach a peaceful solution without military action. As Singer and Small (1966; Small and Singer 1973) point out—even before their nowclassic findings on regime type and war (Small and Singer 1976)—when two states exchange diplomatic representation with each other, this provides the foundation for peaceful negotiation and bargaining.4 If a leader seeks diplomatic assistance from a third-party state or an intermediary organization, all other things being equal, that may produce a better chance to resolve an international dispute without the use of military force.

42 Civil–Military Dynamics

More importantly, if numerous diplomatic channels are already present or open in two states in dyad before a conflict arises, the two may utilize their diplomats or international mediators more readily toward a peaceful solution. Contrarily, fewer diplomatic channels not only mean more difficulties with respect to facilitating political or diplomatic communication in the quest for a peaceful resolution, but also suggest a degree of political or diplomatic isolation in the diplomatic “amphitheater.” Thus, the hypothesis about diplomatic channels is as follows: H3: Dyads composed of states with a lower (higher) number of diplomatic missions are more (less) likely to become involved in MIDs.

Media openness is another principal indicator of the potential for effective political communication between two states. Media openness, with freedom of the press and other news media, gives voice to public opinion and empowers the public’s eyes and ears to detect governmental misconduct, most notably abuse of power (Graber 1986: 258). Thus media openness helps to prevent national leaders from exploiting international conflict for political gain, for example, in pursuit of a rally effect. Moreover, media openness may provide each state with an opportunity to reduce miscommunications and the attendant risk of escalation by providing open channels. In other words, open media structures the international flows of information that create or provide the means by which democracy works to reduce conflict.5 Since a more unrestricted information flow on preferences, capabilities, allies, and so on should reduce the chances of misperception and miscalculation, greater media openness may lead to a decrease in international disputes (Blainey 1988; Gerber and Green 1999; Stoessinger 2001). In this regard, Small and Singer (1976: 51) cite East and Gregg’s (1967) findings that “states with more authoritarian regimes—as reflected in their practices of press

Research Design 43

censorship—appeared to exhibit more foreign-conflict behavior than ‘freer’ nations.” Along the same lines, on the basis of global press freedom data from 1950 to 1992, Van Belle and Oneal (2000: 72) find that, “when the media is independent of governmental control and able to report events objectively, national leaders are constrained in resorting to force, reducing the likelihood that a state will become engaged in a militarized interstate dispute.” In particular, when both states in a dyad maintain a high degree of media openness, the likelihood of an interstate dispute should be reduced significantly. Thus, the hypothesis about media openness is as follows:6 H4: Dyads composed of states with (without) free or imperfectly free media openness are less (more) likely to become involved in MIDs.

Measurement and Data It should be noted that, to the best of our knowledge, so far there exists no readily available cross-sectional and time-series data collection that measures the degree of influence between military and civilian leaders in each state. Assessing each leadership group’s influence within a state is easier said than done (Whitten and Bienen 1996: 214). In his recent book, Civilian Control of the Military: The Changing Security Environment, Desch (1999: 3) points out that “civil–military relations is a very complicated issue. Analysts disagree about how to define and measure civil–military relations as the [in]dependent variable.” Whatever limited agreement exists is with respect to the idea that the concept is multidimensional and difficult to quantify in a meaningful way at the cross-national level. Given the fact that our research is a first attempt to test systematically the relationship between civil–military relations and MIDs, we utilize military expenditure to measure degrees of military versus civilian influence in each state. Rising

44 Civil–Military Dynamics

military power vis-à-vis civilian power is likely to increase military expenditure (Hill 1978; Zuk and Thompson 1982), which in turn may cause more interstate disputes. According to Diehl and Kingston (1987: 803), “this is often an assumption of many scholars and analysts, without empirical documentation. Although the belief that military influence increases with military spending has yet to be adequately documented, it remains a part of many scholars’ thinking.”7 For example, Goertz and Diehl (1986: 554) argue as follows: “Military allocations are thought to play a prominent role in national decisions for war. . . . High military allocations are indicative of this military preparation and perhaps also of the willingness to use military force. . . . High allocations could indicate the influence that military officials have in government decision making.” Thus, we argue that, since each nation’s military expenditure fluctuates according to the degree of the military’s influence, the civil–military relations measure reflects the dynamics of influence among civilian and military leaders.8 The utility of the military expenditure measurement is supported by Benjamin and Edinger’s (1971) 62 historical case studies in four states: France (1905–1954), Germany (1871– 1938), Japan (1894–1945), and the United States (1918– 1954). They claim that “military control is a function of specific patterns of interaction between civilian and military leaders involved in the [foreign policy] decision-making process and of the extent to which available resources [i.e., military expenditure] are committed to military purposes” (Benjamin and Edinger 1971: 18). (Emphasis in the original.) Thus, the degree of military versus civilian leaders’ influence is assumed to be a function of change in military expenditure.9 The civil–military relations variable is dichotomous. If both states in a dyad experience an increase in the annual growth rate in their military expenditure in a dyad-year greater than the average annual growth rate increase through their full range of dyad-years, the dyad is recorded as “1”;

Research Design 45

if not, the score is “0.” The average annual growth rate increase is used as a threshold for ascending military power. For example, if the annual growth rate increase of military expenditure in a dyad is 20 percent in 1990, and the average annual growth rate increase of military expenditure in a dyad is 15 percent for all dyad-years from 1886 to 1992 (i.e., the annual growth rate increase of 20 percent is greater than the average annual growth rate increase of 15 percent in the data set as a whole), the dyad is recorded as “1” for 1990.10 It should be noted that, although the civil–military relations variable utilizes military expenditure as the degree of influence, our operationalization does not take the numerical dollar value of military expenditure, but uses its increase or decrease (i.e., a dichotomy) beyond the threshold of the average annual growth rate increase to identify the shifting direction of military leaders’ influence.11 Military expenditure data come from Bennett and Stam’s (2000b) data set, EUGene (i.e., Expected Utility Generation and Data Management Program, version 2.10).12 Bennett and Stam’s military expenditure data originally come from the 1993 update to the Correlates of War (COW) National Capabilities data file, which is used widely in the study of international conflict. Since military expenditure is recorded as a nominal value (i.e., not inflation-adjusted), it is converted into a real value using Sahr’s Inflation Conversion Factors for 1700 to Estimated 2010.13 The Consumer Price Index for (CPI) 2000 is the base year. Military manpower system is a dichotomous variable. The variable is coded as “1” if both states in each dyad-year adopt a conscription system; it is “0” otherwise. Since students of international conflict, crisis, and war have paid little attention to the impact of military manpower system on MIDs, there has been no data collection over an extended spatial and temporal domain. Despite the data collection constraint, the following two sources, to our knowledge, are most comprehensive and

46 Civil–Military Dynamics

representative with respect to each state’s military manpower system: Horeman and Stolwijk (1998); Prasad and Smythe (1968).14 Singer and Small (1966) and Small and Singer’s (1973) diplomatic importance scores are used to assess diplomatic channels. They measure the number and rank of the diplomatic missions accredited and dispatched to each capital city. A normalized score ranging from “1” to “100” is assigned once per half-decade on that basis. As Small and Singer suggest, considering the generally slow rate of change in the normalized scores, the same score for each of the following years is used. In this chapter, each normalized score is rank-ordered from “1” to “4” by quartile to ameliorate the discrete dispersion between small and large normalized scores as well as to reduce statistical variations in model fitting. The dyad-year is recorded as a smaller value out of the two rank-ordered values (i.e., Dixon’s [1993, 1994] “weak link” approach, which in this context emphasizes the impact of fewer diplomatic channels).15 We utilize Van Belle’s (2000: 137–148) global press freedom data collection to measure media openness for each state. Van Belle provides a five-category coding scheme for the media openness of states included in the Polity III data set: 0 for no news media to code (e.g., Vanuatu); 1 for clearly free news media (e.g., United States, United Kingdom, and Australia); 2 for imperfectly free but relatively free news media (e.g., Finland and Mexico); 3 for restricted news media but not directly controlled by the government (e.g., Jordan and El Salvador 1956–1992); 4 for government controlled or strictly censored news media (e.g., China and North Korea). The media openness variable is dichotomous. The variable is coded as “1” if both states in each dyad-year have free or imperfectly free news media capable of functioning as an area of political competition or debate (i.e., categories 1 and 2); it is “0” otherwise (i.e., the news media are either restricted or controlled by the government, or no news media, categories 0, 3, and 4).16

Research Design 47

Building Logit and Neural Network Models We test the four hypotheses by replicating a standard and prominent research design from the democratic peace literature. Oneal and Russett’s (1999c) work has emerged as one of the most frequently replicated studies, so it will be used to provide the foundation for the analysis here.17 This will also establish the significance of the empirical findings relative to those of Russett and Oneal’s (2001) overall program of research, most notably Triangulating Peace: Democracy, Interdependence, and International Organization,18 as well as reduce bias that might inadvertently appear in the research program (Gartzke 1997: 13, 51). Oneal and Russett’s (1999c) research design is familiar to students of international conflict, so we briefly summarize only the three neo-Kantian peace factors: democraticness, economic interdependence, and joint membership in international organizations. Democraticness assumes the “weak link” (Dixon 1993, 1994): the score for the less democratic state in a dyad is taken to be the stronger determinant of how interactions will proceed. Hence, the more democratic that state is, the more constrained it will be from engaging in a dispute and therefore the more peaceful the dyad.19 Economic interdependence also assumes the “weak link”: the score for the less interdependent state in a dyad is taken to be the stronger determinant of interstate disputes. Hence, the more interdependent that state is, the more constrained it will be from engaging in a dispute and therefore the more peaceful the dyad. The international organization variable is measured by the number of joint memberships. Hence, the more joint memberships in intergovernmental organizations, the more constrained the two states will be from engaging in a dispute and therefore the more peaceful the dyad. The other five variables in the neo-Kantian or democratic peace model from Oneal and Russett (1999c) are (a) national capability ratio (i.e., to control for power preponderance),

48 Civil–Military Dynamics

(b) whether the members of each dyad are allied, (c) noncontiguous states, (d) geographic distance, and (e) whether each member of the dyad is a minor power. These five variables are expected to decrease the likelihood of MID involvement and generally have obtained statistical significance in previous studies. Collectively speaking, these variables can be characterized as realist in orientation. Each is connected in some way to capabilities. National capability ratio, great power and alliance status reflect the mindset of power politics in various forms (Vasquez 1993; James 2002). Geographic distance and contiguity measure the opportunity to fight—each represents the effects of a loss-of-strength gradient, with adversaries seen as less feasible targets, all other things being equal, when further removed from a state’s home base of power (Bueno de Mesquita 1981; James 1988). Since Oneal and Russett (1999c) present MID involvement rather than initiation as their dependent variable,20 we choose MID involvement as the dependent variable for purposes of comparison. All independent variables are lagged by one year, so they are not affected by a dispute to be explained. The following equation is based on the foreign policy decision-making process model in figure 3.1 and Oneal and Russett’s (1999c) democratic peace model: Yt  ␣  ␤1X1t1  ␤2X2t1  ␤3X3t1  ␤4X4t1  ␤5X5t1  ␤6X6t1  ␤7X7t1  ␤8X8t1  ␤9X9t1  ␤10X10t1  ␤11X11t1  ␤12X12t1  ␧

Here, Yt: interstate dispute involvement X1t1: civil–military relations X2t1: military manpower system X3t1: diplomatic channels X4t1: media openness X5t1: democraticness

Research Design 49

X6t1: economic interdependence X7t1: joint membership in international organizations X8t1: national capability ratio X9t1: allied states X10t1: noncontiguous states X11t1: geographic distance X12t1: only minor powers ␧: error term. The equation combines the two civil–military dynamic variables (i.e., civil–military relations and military manpower system) and the two political communication variables (i.e., diplomatic channels and media openness) with the eight variables from Oneal and Russett’s (1999c: 21) first equation. Table 4.1 presents the 12 hypotheses that appear in the equation. With a special emphasis on civil–military dynamics and political communications, this study purports to test the 12 preceding hypotheses about international disputes during the period from 1886 to 1992 at the dyadic level. Data-related constraints limit empirical testing of (a) the civil–military relations hypothesis plus Oneal and Russett’s (1999c) eight hypotheses to the period from 1886 to 1992; (b) the civil– military relations and military manpower system hypotheses to the period from 1886 to 1992; (c) the civil–military relations, military manpower system and diplomatic channels hypotheses to the period from 1886 to 1974; and (d) the civil–military relations, military manpower system, diplomatic channels, and media openness hypotheses to the period from 1950 to 1974.21 Table 4.2 summarizes the rules of operationalization for all of the variables. As in Oneal and Russett’s (1999c: 22) results, the logistic regression model is tested in four different ways during the various periods in the next section: (a) all dyads, using the generalized estimating equation (GEE) method,22 (b) all dyads, using the logistic model with peace years correction,23

Variable New hypotheses Civil–military relations (H1) Manpower system (H2) Diplomatic channels (H3) Media openness (H4) Replicating hypotheses from Oneal and Russett (1999c) Democraticness Economic interdependence International organizations National capability ratio Allied states Noncontiguous states Geographic distance Only minor powers

Hypothesis

As the relative influence of the military in civil–military relations increases for a dyad, the likelihood of involvement in militarized interstate disputes also increases for that dyad Dyads composed of states with (without) a conscription system are more (less) likely to become involved in militarized interstate disputes Dyads composed of states with a lower (higher) number of diplomatic missions are more (less) likely to become involved in militarized interstate disputes Dyads composed of states with (without) free or imperfectly free media openness are more (less) likely to become involved in militarized interstate disputes

The more democratic the less democratic state, the more constrained it will be from engaging in a dispute, and the more peaceful the dyad The more interdependent the less economically interdependent state, the more constrained it will be from engaging in a dispute, and the more peaceful the dyad The more joint memberships in international organizations, the more constrained two nation-states will be from engaging in a dispute, and the more peaceful the dyad The more preponderant the military capabilities, the more constrained the two states will be from engaging in a dispute, and the more peaceful the dyad If two states are militarily allied, they are less likely to engage in a dispute If two states are not contiguous, they are less likely to engage in a dispute The farther apart the two states, the less the militarized disputes If there is no major power involvement, the less the militarized disputes

50

Table 4.1. Four New and Eight Replicating Hypotheses

Table 4.2. Rules of Operationalization for Variables Variable New hypotheses Civil–military relations (H1) Manpower system (H2) Diplomatic channels (H3) Media openness (H4) Replicating hypotheses from Oneal and Russett (1999c) Democraticness Economic interdependence International organizations National capability ratio Allied states Noncontiguous states Geographic distance Only minor powers

Operationalization rule

1 if both states in a dyad-year have the increased above-average military expenditure, 0 otherwise 1 if both states in a dyad-year have conscripted soldiers, 0 otherwise Recording the smaller score between two scores from 1 to 4 in a dyad-year 1 if both states in a dyad-year have open media, 0 otherwise

Recording a smaller score from 10 to 10 in a dyad-year Recording a smaller score in percentage in a dyad-year Recording the total number of a joint membership in international organizations in a dyad-year Recording the natural logarithm of the ratio of the stronger state’s military capability to that of the weaker in a dyad-year 1 if both states in a dyad-year are allied, 0 otherwise 1 if both states in a dyad-year are noncontiguous, 0 otherwise Recording the natural logarithm of the geographic distance between states in a dyad-year 1 if both states in a dyad-year are minor powers, 0 otherwise

51

52 Civil–Military Dynamics

(c) all dyads during the pre–Cold War period, using the GEE method, and (d) politically relevant dyads, using the GEE method. Finally, Beck et al.’s (2000) Bayesian neural network model applies to the politically relevant dyads during the period from 1950 to 1974. Although this study is not a methodological probe, some explanation of neural network modeling is in order, to facilitate data analysis in the next chapter.24 Figure 4.1 compares simple model fittings among linear regression, logistic regression, and neural network, which show the statistical relationship between two variables, X and Y . Given the randomly scattered observations (i.e., the asterisks), each line represents a different model fitting: (a) the solid line for linear regression, (b) the dotted line for logistic regression, and (c) the bold solid line for neural network. It is apparent that the linear regression model fitting does not properly represent the statistical relationship between X and Y in the scatter plot since the scattered marks show a nonlinear pattern. The logistic

Neural network

Variable Y

Linear

Logit

0

Variable X

Figure 4.1. Linear, Logit, and Neural Network Model Fitting

Research Design 53

regression model fitting, which uses an S-shaped curve, could be closer to the seemingly nonlinear relationship between X and Y than the linear, but not accurate enough. The neural network model fitting reflects very closely the nonlinear relationship between X and Y in the scatter plot. That is, the neural network model is designed to fit or find nonlinear causal structure regardless of whether the value of the dependent variable is continuous or discrete (e.g., dichotomous). As in Beck et al.’s (2000: 24–26) formal presentation, the three simple model fittings can be extended to complex multidimensional expressions in the following three hierarchical steps for ␲i, the probability of an international conflict, and Xi  {1, X1i, X2i, . . ., Xki}, a vector of a constant term and k independent variables, where i  1, 2, . . ., N. The linear relationship between ␲i and Xi in a matrix expression can be written in ␲i  linear(Xi); the logistic relationship in ␲i  logit (linear(Xi)); and the neural network relationship in ␲i  logit(linear(logit(linear(Xi))). The only difference between logit and neural network model is to fit in the “closer” shape of the nonlinear curve as depicted in figure 4.1. In this sense, as Beck et al. (2000) claim, the neural network model is a methodological improvement over linear and logit models in terms of more accurate model fitting. It is also more consistent with the reality of theorizing in the social sciences, where interactions between and among multiple factors are likely to cause nonlinear effects to be nearer to the rule than the exception. Based on the foreign policy decision-making process model depicted in figure 3.1, this chapter has described the research design by presenting hypotheses, measurement and data, and logit and neural network modeling, which provides the basis for empirical analysis discussed in chapters 5 and 6.

This page intentionally left blank

Chapter 5

Accounting for Militarized Interstate Disputes This chapter consists of three sections, the first two of which convey the results of data analysis in accounting for MID involvement, and the third provides a case study. The first section presents a series of logistic regression models that preserve continuity with the general approach taken in the field. The second section follows up on the potentially pathbreaking method put forward by Beck et al. (2000) with respect to the application of neural networks to the study of international conflict, crisis, and war. Taken together, the empirical results from the two sections of the analysis should provide a good sense of whether civil–military dynamics and political communications are relevant to the new quest for international peace. The third and final section of this chapter addresses the most persistent anomaly in the findings, namely, the apparent connection of diplomatic channels with MID involvement. The role of diplomacy is reassessed through a contemporary case: the interactions of the United States with North Korea over the latter’s nuclear program during the first term of President George W. Bush.

Logistic Regression Analysis Tables 5.1 through 5.4 present the empirical results based on the logistic regression model1 in chapter 4.2 Given the data availability discussed earlier, each table consists of a somewhat different arrangement of variables and time period. Oneal and

56 Civil–Military Dynamics

Russett’s (1999c) eight variables appear as a “standard menu” in each table. Table 5.1 shows the effect of civil–military relations during the period from 1886 to 1992; table 5.2 focuses on the civil–military dynamics (i.e., civil–military relations and military manpower system) during the period from 1886 to 1992; table 5.3 conveys the civil–military dynamics and diplomatic channels during the period from 1886 to 1974; and table 5.4 includes civil–military relations and political communications (i.e., diplomatic channels and media openness) during the period from 1950 to 1974. Each table reveals theoretically stimulating results: Each of the four additional variables, to one extent or another, is associated with the likelihood of MID involvement. The following data analysis focuses mainly on the four new variables and the three neoKantian peace variables—the main theoretical interests of this research.3 Table 5.1 presents the empirical results with a special emphasis on the civil–military relations hypothesis for 1886 to 1992. As shown for GEE in the second column, the civil– military relations variable is statistically significant at the 0.001 level, indicating that, the more powerful the military leaders, the greater the likelihood of MID involvement. While statistically significant at the 0.001 level, the coefficient (0.0675) of the democraticness variable turns out to be a little smaller than the replicated one (0.0681), as does the coefficient of the economic interdependence variable (59.8950 versus 61.5110). The coefficient of the joint membership in international organizations variable is not statistically significant.4 The third column shows the results based on Beck et al. (1998) peace years correction model, which are very similar to those of the GEE model except for the joint membership in international organizations. The result for the international organizations hypothesis turns out to be counterintuitive and significant at the 0.001 level; joint membership in international organizations seems to increase the likelihood of MID

Accounting for MIDs 57

Table 5.1. Civil–Military Relations and Predicting MID Involvement, 1886–1992 Variable

Civil–military relations (IAAMEAB) Lower democracy (DEML) Trade/GDP (DEPENDL) International organizations (IGO) Capability ratio (CAPRATIO) Alliances (ALLIANCES) Noncontiguity (NONCONTIG) Log distance (DISTANCE) Only minor powers (MINORPWRS) Constant Chi2 P of Chi2 Log likelihood Pseudo-R2 N

1886–1992 Simplest, all dyads

1886–1992 Peace years correction

1886–1939 All dyads

1886–1992 Politically relevant dyads

0.2396***

0.2710***

0.5508***

0.0657 0.0675*** 0.0111 59.8950*** 17.7647 0.0003

0.0731 0.0609*** 0.0095 32.0792** 11.9300 0.0169^^^

0.1101 0.0484*** 0.0110 42.8286* 18.7344 0.0094

0.0695 0.0633*** 0.0115 37.3735** 14.6988 0.0053

0.0040 0.2489*** 0.0533 0.2852* 0.1724 2.0287*** 0.1910 0.4362*** 0.0602 1.8348*** 0.1755 2.2173*** 0.5031 1271.32 0.0000

0.0042 0.2003*** 0.0424 0.4341** 0.1604 1.5661*** 0.1576 0.3303*** 0.0512 1.6854*** 0.1353 1.7708*** 0.4270 1825.25 0.0000 5163.145 0.286 134,634

0.0068 0.4212*** 0.0664 0.1556 0.1985 1.2235*** 0.1922 0.3087*** 0.0616 1.8544*** 0.1969 2.6175*** 0.5125 454.53 0.0000

0.0041 0.2930*** 0.0533 0.3183* 0.1764 1.1381*** 0.1819 0.2282*** 0.0645 0.6550** 0.2169 1.8608*** 0.5378 173.79 0.0000

134,556

26,622

0.1147*

30,505

* p  .05; ** p  .01; *** p  .001, one-tailed tests; ^^^ p  .001, one-tailed test but wrong sign.

involvement.5 The peace years correction model reports a pseudo-R2 that tells how well the model performs in an overall sense—28.6 percent for MID involvement, which is very slightly higher than the replicated version, at 28.5 percent. The fourth column presents the results for the multipolar period after 1886 and before the Cold War, which are similar

58 Civil–Military Dynamics

to those of the GEE model. While the civil–military relations, democraticness, and economic interdependence hypotheses are supported, the international organizations hypothesis is not. The results in the fifth and final column of table 5.1, for politically relevant dyads, are similar to those of the GEE model. It is intriguing that the civil–military relations variable is statistically significant at the 0.05 level, but its significance level is weaker than those (i.e., at the 0.001 level) in columns 2, 3, and 4. Overall, the empirical results indicate not only that civilian control over the military matters in accounting for interstate disputes, but also that it appears to be a potential candidate for complementing Oneal and Russett’s (1999c) triangular peace. Table 5.2 presents the empirical results with a special emphasis on civil–military dynamics (i.e., civil–military relations and military manpower system) during the period from 1886 to 1992. As shown in the GEE results from the second column, the civil–military relations variable is statistically significant at the 0.001 level, indicating that, with the military ascendant, the likelihood of MID involvement becomes greater. The military manpower system variable is not supported. Both the democraticness and economic interdependence variables are statistically significant at the 0.001 level. The coefficient of the international organizations variable turns out to be statistically insignificant. The third column shows the results based on the Beck et al. (1998) peace years correction model. The civil–military relations and military manpower system variables are statistically significant at the 0.001 level. It can be inferred that the increased influence of military leaders over civilian ones as well as adopting a conscription system lead to the likelihood of MID involvement. The democraticness and economic interdependence variables are supported. The international organizations variable turns out to be counterintuitive and statistically significant at the 0.001 level; joint membership of international organizations seems to increase the likelihood of MID

Accounting for MIDs 59

Table 5.2. Civil–Military Dynamics and Predicting MID Involvement, 1886–1992 Variable

Civil–military relations (IAAMEAB) Military manpower system (CONSCRAB) Lower democracy (DEML) Trade/GDP (DEPENDL) International organizations (IGO) Capability ratio (CAPRATIO) Alliances (ALLIANCES) Noncontiguity (NONCONTIG) Log distance (DISTANCE) Only minor powers (MINORPWRS) Constant Chi2 P of Chi2 Log likelihood Pseudo-R2 N

1886–1992 Simplest, all dyads

1886–1992 Peace years correction

1886–1939 All dyads

1886–1992 Politically relevant dyads

0.2358***

0.2464***

0.5941***

0.1045

0.0683 0.0971

0.0769 0.3674***

0.1199 0.0565

0.0716 0.0597

0.1277 0.0695*** 0.0114 59.1483*** 19.0377 0.0026

0.1181 0.0599*** 0.0095 31.2549** 13.1821 0.0145^^^

0.1545 0.0599*** 0.0121 34.4351* 20.4320 0.0071

0.1338 0.0671*** 0.0119 35.4453* 15.8753 0.0074*

0.0042 0.2356*** 0.0571 0.2978 0.1814 2.0556*** 0.1936 0.4450*** 0.0622 1.7910*** 0.1745 2.0883*** 0.5369 1215.51 0.0000

0.0044 0.1821*** 0.0479 0.4350** 0.1682 1.5491*** 0.1640 0.3258*** 0.0556 1.5973*** 0.1341 1.8226*** 0.4767 1600.49 0.0000 4821.882 0.292 122,137

0.0070 0.4361*** 0.0727 0.2637 0.2110 1.1551*** 0.2059 0.2721*** 0.0654 1.6438*** 0.2016 2.6433*** 0.5577 344.03 0.0000

0.0043 0.2872*** 0.0564 0.3784* 0.1851 1.1640*** 0.1844 0.2370*** 0.0671 0.6076** 0.2183 1.7183** 0.5788 175.35 0.0000

122,071

18,452

27,658

* p  .05;** p  .01; *** p  .001, one-tailed tests; ^^^ p  .001, one-tailed test but wrong sign.

involvement. The peace years correction model reports slightly higher pseudo-R2 values: 29.2 percent versus 28.9 percent for the replicated version. The fourth column shows results from the multipolar period after 1886 and before the Cold War, which are similar to those of the GEE model. While the civil–military relations,

60 Civil–Military Dynamics

democraticness, and economic interdependence hypotheses are supported, the military manpower system and international organization hypotheses are not. The politically relevant dyads in the fifth and final column of table 5.2 produce interesting results. Both civil–military relations and military manpower system turn out to be statistically insignificant. It can be speculated that the behavior of politically relevant dyads is somewhat different from that of all kinds of dyads in general. One likely explanation for the lack of a connection with civil–military relations is that the hypothesis about hawkish military and dovish civilian leaders and MID involvement may not be applicable when restricted to this domain. As Huntington (1957), Janowitz (1960), and Betts (1977) point out, the hypothesized roles for civilian and military leaders within major powers (i.e., advanced industrial countries) might have been switched or fused alongside with political institutionalization and development of military professionalism. The basic logic of the “Powell Doctrine” also may be in play here. Powell, speaking for most American military professionals, advocated using force only in very restrictive circumstances, which is in line with the arguments about hawkish civilian and dovish military. However, if it did become necessary to go to war, Powell would want to do so without restrictions on its scope and intensity, which is compatible with Van Evera (1984) and Snyder’s (1984) “cult of the offensive” (see Lane 1995: 20–21; Korb 1997: 24–25; Campbell 1998). In addition, due to high military preparedness and responsiveness, major powers (e.g., United States under the voluntary system) might easily become involved in MIDs regardless of the type of military manpower system.6 It is also plausible that, since contiguous dyads are most dispute-prone, they are likely to respond to external threats regardless of the characteristics of civil– military relations or military manpower system. The three neoKantian peace hypotheses are supported at the 0.001, 0.05, and 0.05 level, respectively. Overall, the empirical results in

Accounting for MIDs 61

table 5.2 show that both civil–military relations and military manpower system variables to some extent are potential candidates for complementing Oneal and Russett’s (1999c) triangular peace. Table 5.3 presents the empirical results with a special focus on both civil–military dynamics (i.e., relative civil versus military influence and military manpower system) and diplomatic channels for 1886 to 1974. As shown for the GEE in the second column, the civil–military relations variable is statistically significant at the 0.001 level, indicating that, with a more powerful military, the likelihood of MID involvement becomes greater. The military manpower system variable fails to be statistically significant. Although the diplomatic channels variable is statistically significant at the 0.01 level, it turns out to be counterintuitive, indicating that more diplomatic channels mean a greater likelihood of MID involvement. It seems that political talks through diplomatic channels, such as an ambassador, envoy, or minister, chargé d’affaires or mediators, may be inclined toward “cheapness.” Or, instead, the presence of numerous diplomatic channels is likely to be a proxy of relevance and importance for members of each dyad. The number of diplomatic channels may be a proxy for, and collinear with, major powers. In other words, everything else being equal, minor powers are likely to have fewer diplomatic representations abroad. Major powers are more inclined to see their interests engaged in different areas, so they have greater capabilities to get involved in interstate disputes. Another possibility is that, as some students of international conflict point out, Singer and Small’s (1966; Small and Singer 1973) diplomatic importance scores might not be fully accurate. Democraticness and economic interdependence as well as international organizations are statistically significant at the 0.001, 0.001, and 0.01 level, respectively. Note that, in the two previous GEE models in the second column in tables 5.1 and 5.2, the coefficient of the international organizations variable

62 Table 5.3. Civil–Military Dynamics, Diplomatic Channels, and Predicting MID Involvement, 1886–1974 Variable

Civil–military relations (IAAMEAB) Military manpower system (CONSCRAB) Diplomatic channels (SANSDC) Lower democracy (DEML) Trade/GDP (DEPENDL) International organizations (IGO) Capability ratio (CAPRATIO) Alliances (ALLIANCES) Noncontiguity (NONCONTIG) Log distance (DISTANCE) Only minor powers (MINORPWRS) Constant Chi2 P of Chi2 Log likelihood Pseudo-R2 N

1886–1974 Simplest, all dyads

1886–1974 Peace years correction

1886–1939 All dyads

1886–1974 Politically relevant dyads

0.3021***

0.4320***

0.6291***

0.1507*

0.0883 0.0690

0.0969 0.3929**

0.1204 0.0464

0.0886 0.0756

0.1560 0.2637^^ 0.0998 0.0749*** 0.0129 80.0811*** 24.7603 0.0174**

0.1517 0.3878^^^ 0.0985 0.0636*** 0.0111 54.9760** 19.9008 0.0105*

0.1540 0.1583 0.1069 0.0617*** 0.0125 38.4309* 20.5667 0.0024

0.1536 0.4057^^^ 0.0977 0.0856*** 0.0130 53.3533** 20.7934 0.0188***

0.0057 0.2242*** 0.0714 0.2181 0.1876 1.6815*** 0.2184 0.3911*** 0.0722 1.7866*** 0.2159 2.4871*** 0.6332 812.79 0.0000

0.0059 0.1316* 0.0639 0.3075* 0.1869 1.3743*** 0.2091 0.2802*** 0.0728 1.5438*** 0.1852 2.5854*** 0.6377 1048.98 0.0000 3027.736 0.268 57,221

0.0073 0.4018*** 0.0800 0.2450 0.2068 1.1141*** 0.2097 0.2497*** 0.0669 1.5308*** 0.2421 3.0517*** 0.6328 306.74 0.0000

0.0055 0.2125*** 0.0689 0.3672* 0.1839 0.8865*** 0.1979 0.2089** 0.0751 0.2964 0.2691 2.3397*** 0.6470 156.07 0.0000

57,185

17,697

16,431

* p  .05; ** p  .01; *** p  .001, one-tailed tests; ^^ p  .01; ^^^ p.001, one-tailed test but wrong sign.

Accounting for MIDs 63

shows no consistent connection, so empirical results seem to vary with model specification. The third column shows the results based on the Beck et al. (1998) peace years correction model, which are very similar to those of the GEE model except for the military manpower system variable. Since the military manpower system variable turns out to be statistically significant at the 0.01 level, it appears that, when conscription exists in both states in a dyad, the likelihood of MID involvement increases. The pseudo-R2 shows that the model explains 26.8 percent of the variation in MID involvement, which is a little better than the replicated version, at 25.7 percent. The fourth column shows the results during the multipolar period after 1886 and before the Cold War. While the civil– military relations, democraticness, and economic interdependence hypotheses are supported, the military manpower system, diplomatic channels and international organizations hypotheses are not. Results for politically relevant dyads in the fifth and final column of table 5.3 are very similar to those of the GEE model. While the civil–military relations, diplomatic channels, democraticness, economic interdependence, and international organizations variables are supported in one way or another, the military manpower system variable is not. Overall, however, the empirical results show that both civil–military relations and military manpower system as well as diplomatic channels variables need to be incorporated into Oneal and Russett’s (1999c) triangular peace to better understand the nature of interstate disputes. Table 5.4 presents very stimulating empirical results with respect to both civil–military dynamics and political communications as well as the three neo-Kantian peace variables for 1950–1974. This data analysis reflects the complete foreign policy decision-making process model shown in figure 3.1. Based on the empirical testing during the period from 1950 to 1974, table 5.4 shows the importance of media

64

Table 5.4. Civil–Military Dynamics, Political Communications, and Predicting MID Involvement, 1950–1974 Variable Civil–military relations (IAAMEAB) Military manpower system (CONSCRAB) Diplomatic channels (SANSDC) Media Openness (AMOAB) Lower democracy (DEML) Trade/GDP (DEPENDL) International organizations (IGO) Capability ratio (CAPRATIO)

Replicated

0.0702** 0.0247 50.2864 53.2811 0.0382*** 0.0118 0.2303* 0.1257

Simplest, all dyads 0.1147 0.1210 0.2020 0.2848 0.5103^ 0.2407 1.7215** 0.5848 0.0135 0.0316 31.3038 46.1728 0.0505*** 0.0140 0.1752 0.1294

Replicated

0.0573*** 0.0187 14.2924 32.6656 0.0031 0.0134 0.1153 0.0936

Peace years correction 0.1359 0.1473 0.5263* 0.2391 0.4961^ 0.2277 1.3337** 0.4853 0.0049 0.0253 0.6085 26.4551 0.0211 0.0142 0.0680 0.1020

Replicated

Politically relevant dyads

0.0548** 0.0215 42.5348 40.4653 0.0338** 0.0116 0.3054** 0.1216

0.0522 0.1201 0.1384 0.2635 0.6602^^ 0.2274 1.4570** 0.5405 0.0077 0.0279 37.0212 41.8110 0.0519*** 0.0132 0.2512* 0.1224

Alliances (ALLIANCES) Noncontiguity (NONCONTIG) Log distance (DISTANCE) Only minor powers (MINORPWRS) Constant Chi2 P of Chi2 Log likelihood Pseudo-R2 N

0.3545 0.3631 2.2651*** 0.3784 0.6288*** 0.1353 2.4494*** 0.4396 0.0852 1.1911 529.36 0.0000

37,423

0.2055 0.3589 2.3332*** 0.3719 0.6126*** 0.1352 2.2666*** 0.4312 0.5202 1.1509 439.82 0.0000

37,423

0.3913 0.3340 1.6113*** 0.3017 0.3648** 0.1285 1.9112*** 0.3022 0.3055 1.1405 510.11 0.0000 1244.585 0.414 37,452

0.1658 0.3077 1.6713*** 0.3112 0.3051** 0.1259 1.5919*** 0.2988 1.2432 1.0776 474.67 0.0000 1209.663 0.430 37,452

0.6222* 0.3248 1.0356** 0.3646 0.3305** 0.1310 0.6779 0.4932 0.2761 1.1539 69.71 0.0000

8,379

0.4420 0.3010 1.0850** 0.3643 0.3272** 0.1366 0.4340 0.4757 0.3673 1.1535 78.96 0.0000

8,379

* p  .05; ** p  .01; *** p  .001, one-tailed tests; ^ p  .05; ^^ p  .01, one-tailed test but wrong sign.

65

66 Civil–Military Dynamics

openness, a neglected aspect of the concept of democracy, not democraticness as measured in the Polity data set, in accounting for the likelihood of MID involvement. As shown in the third column, where the GEE results appear, both the civil–military relations and military manpower system variables turn out to be statistically insignificant; the diplomatic channels variable is statistically significant at the 0.05 level, but counterintuitive; and the media openness variable is statistically significant at the 0.01 level. It seems that, as hypothesized, dyads composed of states with free or imperfectly free media openness are less likely to become involved in MIDs. While both democraticness and economic interdependence are not statistically significant, the international organization variable is statistically significant at the 0.001 level. The fifth column shows the results based on the Beck et al. (1998) peace years correction model. Among the most theoretically interesting variables, the military manpower system and media openness variables are statistically significant at the 0.05 and 0.01 level, respectively.7 The diplomatic channels variable is statistically significant at the 0.05 level, but in the direction opposite to what had been expected. None of the three neo-Kantian peace hypotheses is empirically supported. The peace years correction model reports a pseudo-R2 of 43 percent for MID involvement, once again a bit better than the replicated version, at 41.4 percent. The politically relevant dyads in the last column reveal interesting results. Civil–military dynamics (i.e., civil–military relations and military manpower system) turn out to be statistically insignificant. Political communications (i.e., diplomatic channels and media openness) are statistically significant, but the diplomatic channels variable is counterintuitive. Among the three neo-Kantian peace hypotheses, the one about international organizations is supported. Overall, the empirical results show that, irrespective of model specification and methods, the media openness variable is very consistent in accounting for interstate disputes. Democraticness based on the Polity

Accounting for MIDs 67

data fades away, but media openness comes to the forefront in explaining the likelihood of MID involvement. It should be emphasized that these findings do not falsify the democratic peace; instead, the results direct our attention to the necessity of sharpening (or refining) the research program because media openness rather than other structural and cultural aspects of democracy appears to have the most pacific effect on international conflict.8 After looking into the impact of media openness on MID involvement in the logistic regression model, some diagnostic tests for multicollinearity are in order. The independent variables must be evaluated carefully in this context because multicollinearity might have “washed out” the effects of the other important variables, including Oneal and Russett’s democraticness. For example, it could be argued that, because media openness is associated with democracy (i.e., democracies tend to have a more open media), they are just derivatives of democraticness. We do not agree with this view on two grounds. First of all, it should be noted that, with respect to conceptualization, measurement, and aggregation, our media openness is quite different from democraticness (as measured in the Polity data). While we introduce media openness as one of the most prominent attributes of democracy, Oneal and Russett and many others use the Polity III data set, which is built on the following five structural attributes of democracy: (a) competitiveness of participation, (b) regulation of participation, (c) competitiveness of executive recruitment, (d) openness of executive recruitment, and (e) constraints on the executive. Polity III’s coding decisions are not explicitly based on the extent of media openness (Gurr et al. 1989, 1991). At the aggregate level, Oneal and Russett’s version of democraticness treats the degree of constraint on the chief executive as a main determinant of the conventional “democracy minus autocracy” score.9 The “bottom line” is that our concepts, measurements, and data for media openness are not the same as democraticness based on Polity III. We attempt to present another key element of liberal democracy, while the

68 Civil–Military Dynamics

emphasis of Polity is on something closer to checks and balances on chief executive power, perhaps summed up as institutional democracy. Since each index is defined, measured, and aggregated differently, it can be incorporated into a model without immediate, theoretically based concerns. Second, two standard diagnostic tests discussed below show no indication of severe multicollinearity (see Gujarati 1995; SAS/STAT® User’s Guide, Version 7-1, Volume 5 1999; and Stata Reference Manual Set 2001). The first test is based on the R2 statistic. According to Menard (1995), when multicollinearity is suspected, the independent variable of concern should be regressed on all other independent variables in a standard ordinary least squares (OLS) regression to obtain R2 values. The R2 will indicate how much of the variance in that independent variable is explained by the others. If the R2 values reach the 0.80 benchmark, problematic multicollinearity exists. As shown in table 5.5, however, none of R2 values— for both simplest, all dyads, and peace years correction as well as politically relevant dyads—reaches beyond the 0.80 benchmark. Thus, no serious multicollinearity problem exists between independent variables. The highest R2 value for the simplest instance, all dyads, is 0.62 when joint membership in international organizations is regressed on the other 11 independent variables (including democraticness and media openness) in a standard OLS regression. The highest R2 value for peace years correction is 0.64 when joint membership in international organizations is regressed on the other 11 independent variables in a standard OLS regression. The highest R2 value for politically relevant dyads is 0.73 when joint membership in international organizations is regressed on the other 11 independent variables in a standard OLS regression. The second diagnostic test is based on eigenvalues and the condition index. (We will not discuss the meaning of eigenvalues here, for that would take us into topics in matrix algebra that are beyond the scope of this book.) The condition index is derived from the eigenvalues. The condition index is the square

Table 5.5. Multicollinearity Diagnostics, 1950–1974 Simplest, all dyads R2 Civil–military relations Military manpower system Diplomatic channels Media openness Lower democracy Trade/GDP International organizations Capability ratio Alliances Noncontiguity Log distance Only minor powers

0.0243 0.1019 0.3970 0.5268 0.5482 0.2146 0.6245 0.3434 0.2882 0.2234 0.4038 0.3430

Eigenvalue Condition index 2.9146 1.6979 1.5619 1.1931 0.9577 0.8190 0.7370 0.6855 0.5033 0.3876 0.2914 0.2510

1.0000 1.3102 1.3660 1.5630 1.7445 1.8865 1.9886 2.0620 2.4065 2.7420 3.1628 3.4077

Peace years correction R2 0.0250 0.1799 0.4109 0.5320 0.5511 0.2198 0.6380 0.3504 0.2997 0.2307 0.4297 0.3476

Eigenvalue Condition index 3.8644 2.4241 1.7008 1.5250 1.0721 0.9484 0.8291 0.7645 0.6897 0.6430 0.4571 0.3771

1.0000 1.2626 1.5074 1.5919 1.8986 2.0186 2.1589 2.2483 2.3671 2.4516 2.9077 3.2010

Politically relevant dyads R2 0.0410 0.0977 0.4806 0.6459 0.6646 0.2923 0.7286 0.4842 0.2966 0.5882 0.5508 0.6584

Eigenvalue Condition index 3.8067 2.3329 1.1412 0.9706 0.8648 0.7294 0.6878 0.4691 0.3621 0.2495 0.2097 0.1762

1.0000 1.2774 1.8264 1.9804 2.0981 2.2845 2.3526 2.8488 3.2423 3.9061 4.2603 4.6475

Note: Italics indicate the highest level for each diagnostic.

69

70 Civil–Military Dynamics

root of the ratio of the largest eigenvalue to each individual eigenvalue. If the condition index exceeds 30, there is severe multicollinearity. Some believe that the condition index is the best available multicollinearity diagnostic (see Belsley et al. 1980). As shown in table 5.5, none of the condition indices— for both simplest, all dyads, and peace years correction as well as politically relevant dyads—is close to 30, indicating that multicollinearity should not be a problem.

Neural Network Analysis Neural network modeling is utilized to better identify evidence of magnitude (by slope) as well as causal structure (by shape) of the occurrence of MID involvement in a graphic presentation. Neural network analysis more effectively represents nonlinear relations among variables and can deal with correlations (i.e., interaction effects) between variables. As Russett and Oneal (2001: 39) point out, militarized disputes result from complex interactions. Individual causal relationships cannot be figured out well in isolation. Peace may result from various liberal factors’ interactions or overlap among civil– military relations, media openness, democracy, interdependence, and so on which are also influenced by several realist factors such as power preponderance. A neural network model is capable of fully embracing the interactive and nonlinear relationships among variables (see Lagazio and Russett 2004).10 Moreover, neural network models are designed to help when there is an extremely unequal distribution of response values, such as hundreds of thousands of times more “0”s (no dispute) than “1”s (dispute). While logit models may “wash out” many small effects with a few large ones, neural network models are “well suited to [such] data with complex, nonlinear, and contingent relationship[s]” (Beck et al. 2000: 22), as with international conflict, crisis, and war.

Accounting for MIDs 71

As in Beck et al.’s (2000) presentation, Bayesian “backpropagation” selects 12 nodes in the input layer, 4 nodes in the hidden layer, and 1 node in the output layer, and is applied to politically relevant dyads only during the period from 1950 to 1974.11 The model produces the optimum value of root mean square (RMS) error of 0.1974 in the training data set. Increasing the number of nodes in the hidden layer (i.e., what Beck et al. [2000] called “M”) does not improve the model’s performance.12 It should be noted that, unlike the logistic regression model, the neural network model does not provide a statistical significance test for each independent variable.13 As with logistic regression analysis, the neural network analysis focuses mainly on theoretically and empirically interesting variables such as the civil–military dynamics, political communications, and the three neo-Kantian peace variables.14 It should be kept in mind that only three variables (i.e., media openness, diplomatic channels, and international organizations) among the seven theoretically interesting ones show statistical significance in the last column in table 5.4, based on the logistic regression model for the politically relevant dyads during the period from 1950 to 1974. Figure 5.1 shows three-dimensional plots of the neural network model output (i.e., the quantitative effect of MID involvement) against two of the inputs (i.e., independent variables) during the period from 1950 to 1974. That is, the surface plots show how the predicted value of MID involvement varies in terms of two independent variables. Figure 5.1a shows that both civil–military relations (i.e., CMR) and military manpower system (i.e., CONSCRT) variables have a complex causal structure. The plot portrays a nonlinear, positive relationship between military leaders’ influence and MID involvement, and a seemingly negative relationship between employing conscription and MID involvement. It should be noted that, while the results in the final column in table 5.4 from the logistic regression model indicated no relationship with civil–military relations, neural networks seem to

72 0.10

Bayes1

Bayes1

0.22

0.00

0.08 0.00

1.00

0.04 0.00

CMR

DIPCHAN MEDOPEN

CONSCRT 1.00

1.00

1.00

(a)

4.00

(b)

0.24

Bayes1

Bayes1

0.12

2.00

10.00

0.03 0.00

0.01 0.00

DEMO

JMIO

CONTIGUITY

ECOINT 0.14

102.00

10.00

1.00

(c)

(d)

Bayes1

0.00

0.01 0.00

CAPRATIO

1.87

0.08 0.00

DISTANCE

MAJPOWER

ALLIANCE 1.00 (e)

Bayes1

0.26

0.17

9.41

7.67

1.00

(f)

Figure 5.1. Three-Dimensional Plots of the Network Output against Two Inputs, 1950–1974. The likelihood of MID involvement against (a) civil–military relations and conscription, (b) diplomatic channels and media openness, (c) democraticness and economic interdependence, (d) international organizations and contiguity, (e) capability ratio and alliance, and (f) distance and major power.

Accounting for MIDs 73

be closer to the intuitive causal structure (i.e., military influence leads to an increase of the likelihood of MID involvement) based on the interactive foreign policy decision-making process model shown in figure 3.1. Figure 5.1b shows that both diplomatic channels (i.e., DIPCHAN) and media openness (i.e., MEDOPEN) variables are connected to the occurrence of MID involvement. The plot portrays a nonlinear relationship between political channels (i.e., media openness and diplomatic channels) and MID involvement. It is evident that media openness has a noteworthy negative impact on the occurrence of MID involvement. The relationship between diplomatic channels and MID involvement is nonlinear. As with the logistic regression model, which produced a counterintuitive result, neural networks suggest that an increase in diplomatic channels may contribute to MID involvement. Figure 5.1c shows that, while the democraticness variable (i.e., DEMO) seems to have a minor effect on MID involvement, the economic interdependence variable (i.e., ECOINT) leads to a greater decrease of MID involvement. A seemingly linear relationship between democraticness and MID involvement as well as a nonlinear, negative relationship between economic interdependence and MID involvement are portrayed. Figure 5.1d shows that, while the international organizations variable (i.e., JMIO) contributes to lessening the MID involvement, the contiguity variable (i.e., CONTIGUITY) is associated with MID involvement. A nonlinear, negative relationship between international organizations and MID involvement as well as a seemingly linear, positive relationship between contiguous dyads and MID involvement are portrayed. Figure 5.2 shows sensitivity plots (i.e., marginal effects) for each independent variable during the period from 1950 to 1974.15 Each sensitivity plot offers a visual representation of a nonlinear relationship for an independent variable. As shown in figure 5.1a, the ascendance of military leaders’ power features a nonlinear, positive relationship. It appears

74 0.13

0.13

0.00

0.00 0.00

(a)

1.00

0.00

0.22

0.22

0.00

0.00 1.00

(c)

4.00

0.00

0.13

0.13

0.00

0.00 –10.00

10.00 (e)

(b)

(d)

0.00

1.00

1.00

0.14 (f)

Figure 5.2. Marginal Effects of Each Independent Variable, 1950–1974. (a) civil– military relations, (b) conscription, (c) diplomatic channels, (d) media openness, (e) democraticness (f) economic interdependence, (g) international organizations, and (h) contiguity, (i) alliance, (j) capability ratio, (k) distance, and (l) major powers.

75 0.25

0.25

0.00

0.00 2.00

(g)

102.00

0.00

0.17

0.17

0.00

0.00 0.00

(i)

0.00

1.00

0.30

(h)

(j)

1.00

7.67

0.30

0.00

0.00 1.87

9.41

0.00

(k)

1.00 (l)

Figure 5.2. Continued

76 Civil–Military Dynamics

that greater military influence on the foreign policy decisionmaking process contributes to an increase in MID involvement. Military manpower system seems to be nonlinear as well. It appears that, unlike the hypothesis, employing conscripted soldiers contributes to a decrease in MID involvement, but the effect seems very weak. Diplomatic channels turn out to be nonlinear and positive. A 27.3 percent increase in diplomatic channels (from 2.176 to 2.77), for example, increases MID involvement by 36.95 percent (from 0.098629 to 0.135073).16 The plot is counterintuitive, meaning that an increase in diplomatic channels contributes to an increase in MID involvement. These results are very similar to (or consistent with) those from the logit models. Despite the various model specifications, methods, and time period, the consistent results produce two conjectures, assuming that the data are accurate: (a) political talk through diplomatic channels tends to be cheap; and (b) the presence of numerous diplomatic channels might lead states toward aggressive foreign policy as a by-product of status, that is, an unfortunate interaction effect between military prowess and national interests. If our speculation is on target, diplomatic channels could be used as a positive indicator of the likelihood of MID involvement. The media openness variable once again confirms a seemingly nonlinear, negative relationship, where a 50 percent increase in media openness from 0.0 to 0.5, for example, decreases the predicted value of MIDs by 29.79 percent (from 0.098629 to 0.069244). In other words, as shown in the results from the logit model, if media openness in a dyad is high enough, MID involvement becomes less likely. It appears counterintuitive that democraticness seems to have a positive impact on the occurrence of MID involvement. This result concurs with Henderson’s (2002) interesting but relatively isolated findings. Using Oneal and Russett’s (1997) data set, he reports that “the presence of democracy is hardly a guarantor of peace—and under certain conditions, it increases the probability of war” (191). Economic interdependence seems

Accounting for MIDs 77

to show a complex causal structure: It starts to dramatically decrease MID involvement and then to very slightly increase MID involvement. Joint membership in international organizations reveals a nonlinear, negative relationship. An increase in joint membership in international organizations seems to decrease MID involvement where a 60.75 percent increase in international organizations (from 33.282001 to 53.5), for example, decreases the occurrence of MIDs by 38.97 percent (from 0.095155 to 0.058077). The contiguity variable reveals a seemingly linear, positive relationship. It is apparent that, if two states are contiguous, they are more likely to engage in MIDs. Table 5.6, with respect to statistical significance, summarizes the across-the-board results from both logit and neural network models from the first two sections in this chapter. Both civil–military dynamics (i.e., civil–military relations and military manpower system) and political communications (i.e., diplomatic channels and media openness), in one way or another, show statistical significance across the different model specifications, methods, and time periods. The civil–military relations variable mostly turns out to be intuitive while the military manpower variable is supported only with Beck et al.’s (1998) peace years correction model. The diplomatic channels variable appears counterintuitive while the media openness variable is supported in a consistent manner. Most importantly, the empirical results indicate that our media openness variable, not the democraticness variable based mainly on the degree of constraint on the chief executive in the Polity data set, seems like the most defining attribute of democracy in accounting for MID involvement. The neo-Kantian peace variables, on average, are not supported in the most complete foreign policy decision-making process model during the period from 1950 to 1974. Civil–military dynamics also loses its statistical significance. But political communications are sustained. Overall, it appears that the four variables throw light on the new quest for peace in studies of international conflict, crisis, and war.

78

Table 5.6. A New Quest for International Peace?: A Summary of Statistical Significance in MID Involvement Variable

Logistic regression 1886–1992

1886–1992

1886–1974

Neural network 1950–1974

1950–1974

[1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] New _ _ Civil–military relations (H1) *** *** *** * *** *** *** _ *** *** *** * n/a n/a n/a n/a _ *** _ _ _ ** _ _ _ * Manpower system (H2)

[3]

[4]

n/a n/a

_ _

Diplomatic channels (H3) Media openness (H4)

^^ **

n/a n/a n/a n/a n/a n/a n/a n/a ^^^ ^^^ _ ^^^ ^ n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a **

^ **

n/a n/a

Kantian peace Democraticness *** *** *** *** *** *** *** *** *** *** *** *** _ Economic interdependence *** ** * ** *** ** * * *** ** * ** _ International organizations _ ^^^ _ _ _ ^^^ _ * ** * _ *** ***

_ _ _

n/a _ n/a _ n/a ***

[4] Intuitive Somewhat counterintuitive Counterintuitive Intuitive Counterintuitive Somewhat intuitive Mostly intuitive

* p  .05; ** p  .01; *** p  .001, one-tailed tests; ^ p  .05; ^^ p  .01; ^^^ p  .001, one-tailed test but wrong sign. _: Not statistically significant. n/a: Not applicable. [1]: Simplest, all dyads. [2]: Peace years correction. [3]: All dyads. [4]: Politically relevant dyads.

Accounting for MIDs 79

A Case Study: Bush Administration Diplomacy and the North Korean Nuclear Program This section17 delves into the counterintuitive findings about diplomatic channels and interstate disputes in the previous two sections. It appeared that, the more diplomatic channels exist, the more conflictual things become. Diplomatic activities may not facilitate peaceful settlements between two disputing states, instead, they aggravate or complicate a situation. Here, at odds with its primary purpose, diplomacy seems problematic at best. Diplomacy surrounding the North Korean nuclear weapons development may provide some insights about this puzzling result.

Background The origin of North Korea’s (the Democratic People’s Republic of Korea, or DPRK) nuclear crisis dates back to the late 1970s. The North’s nuclear programs started with the Soviet-supplied 2MW IRT-research reactor and 0.1MW critical assembly located at Yongbyon. On December 12, 1985, the North signed the Nuclear Nonproliferation Treaty (NPT) that required the North to sign a safeguards agreement with the International Atomic Energy Agency (IAEA) within 18 months of negotiation. The deadline was extended to another 18 months when IAEA learned that it had mistakenly sent the wrong safeguards agreement document to the North. On September 1989, James A. Baker, U.S. Secretary of State, and Eduard A. Shevardnadze, Soviet Foreign Minister, expressed concerns both that the North had not signed the safeguards agreement and that it might have a nuclear weapons program. In October 1989 and February 1990, the North demanded the removal of U.S. nuclear weapons stationed in South Korea before it would sign the agreement. The United States and the

80 Civil–Military Dynamics

South rejected this demand on the grounds of North Korea’s large conventional forces. On July 21, 1990, the North proposed, as a precondition to IAEA on-site inspections, to hold direct talks with the United States to discuss the withdrawal of U.S. nuclear weapons from the South and nonuse and nonthreat of nuclear weapons against the North. Besides, the North pointed out that, if it was to be subjected to inspections, then so should the South. By attempting to bypass IAEA and deal directly with the United States, the North launched a new diplomatic tactic. On September 27, 1991, the then U.S. President George Bush announced that the United States would remove all shortrange land- and sea-based tactical nuclear weapons from the South. On January 21, 1992, U.S.–North Korean high-level political meetings were held at the U.S. Mission to the United Nations in New York where the North agreed to sign the agreement. On January 30, 1992, North Korea signed the IAEA safeguard agreement, but it did not accept international inspections of its nuclear facilities on the spot due to the long process of ratification by the legislature. On April 9, 1992, the North Korean parliament ratified the agreement. Six IAEA inspections were performed in the North’s nuclear facilities with its less than enthusiastic cooperation until late 1994. Accusing the IAEA of spying on it for the United States, along with experiencing uneasiness about allowing two suspected sites for the nuclear weapons program to be inspected, the North suggested its withdrawal from the NPT on March 12 and November 30, 1993. Further, the North demanded a package deal involving an improvement in bilateral relations from the United States on December 3, 1993. After futile diplomatic efforts, the North submitted a letter officially relinquishing its IAEA membership on June 13, 1994. But after former U.S. President Jimmy Carter met with the North Korean President Kim Il-Sung on June 16, 1994, some optimistic news came out. More

Accounting for MIDs 81

diplomatic rounds between the North and the United States followed, and the two parties signed an accord, the Agreed Framework (AF) on October 21, 1994. The AF specified the actions that the two states would take to resolve the North Korean nuclear problems. In addition, the AF stated that (a) a U.S.-led international consortium would help the North replace its graphite-moderated reactors with two 1,000MW light-water reactors and the international consortium would compensate the North for the freeze on its graphite-moderated reactors by supplying 500,000 tons of heavy-fuel oil annually until the new reactors came online; (b) the two states would make efforts to normalize their political and economic relations by reducing investment and trade barriers; (c) the two states would strive toward establishing a nuclear-weaponsfree-zone on the Korean Peninsula; and (d) the North would help strengthen the nonproliferation regime by remaining a member of the NPT—it would also allow the IAEA to implement the safeguards agreement and monitor the freeze on its nuclear facilities.18 Although there were some disagreements over the AF, the North and the United States continued to meet from time to time in search of a peaceful resolution during the Clinton administration. Despite the fact that the North might be reneging on the AF, the diplomatic channels stayed open for negotiations. The peaceful diplomacy between the two was culminated in October 2000 when the Secretary of State Madeleine Albright met the North Korean President Kim Jong-Il in Pyongyang (Yun 2003).

Bush Administration Diplomacy For the first year of the George W. Bush Administration, the North Korea nuclear issue did not approach the top of the agenda. Dialogue with the North slowed down because the new administration took some time to review the U.S. policy

82 Civil–Military Dynamics

toward North Korea. The new administration did not seem to propose significant changes to Clinton’s approach, which had emphasized a diplomatic resolution. Although Republicans, including some Bush aides, engaged in harsh rhetoric about the North, after three months of review, the president announced that his administration would stick with the basic outlines of the existing policy in the form of the AF. Japan and South Korea had pressed the new administration to uphold the AF. This lobbying, combined with the Secretary of State Colin Powell’s successful work in fending off the more conservative Bush advisors, proved very influential in bringing about this decision (Diamond 2001). Although the government reaffirmed its full compliance with the AF that provided two 1,000MW Light Water Reactors (LWRs) in return for North Korean restraint of nuclear development, the Bush Administration found domestic opposition to fulfillment of those requirements difficult to bear. From the beginning of the administration, some Members of the Congress and commentators in academe and the media argued repeatedly that one of the two LWRs should be replaced with a thermal power station. The reason that some officials and Members of the Congress opposed the provision of LWRs is that nuclear weapon grade plutonium could be extracted from them. A hidden reason behind this objection also might have been the increasing cost of 500,000 metric tons of heavy oil given to North Korea since 1995, slated to continue until the LWRs are completed. While expected to be finished by 2003–2004, due to financial problems, allegedly caused by the United States, analysts expect the LWRs to be finished around the end of the decade. Republican partisans, of course, did not want to fund a regime that they believed was hostile to the United States. The South Korean government, which bears 70 percent of the construction costs for the two LWRs, maintained its opposition to their replacement with thermal power stations because (a) that would violate the most critical agreement between the United

Accounting for MIDs 83

States and DPRK; (b) it would further delay the project and bring additional costs; and (c) it would be nearly impossible for North Korea to extract plutonium of nuclear weapons grade because, although extraction remains theoretically possible, it would not obtain the extremely sophisticated reprocessing technology. North Korea also was opposed to such a change in the AF. Although no sign of significant change in the policy came out of the Bush Administration right away, even reviewing the agreement was enough to upset the North. On March 17, 2001, the North Korean Central Broadcasting Station issued the following warning: “If the Bush administration feels it burdensome and troublesome to perform the Geneva Agreed Framework, we don’t need to be indefinitely bound by an agreement that is not honored. We will go on our way in case the agreement is not honored.” Rodong Sinmun, the statecontrolled newspaper of the North, observed that “North Korea would take ‘countermeasures’ if the United States does not perform its obligations under the agreement. We will also demand compensation for the delay in construction of the LWRs” (North Korea Handbook 2003). At the end of 2001, hope remained for continuation of bilateral talks, although the North was suspicious of the renewed dialogue. Crisis-escalating discourse between the United States and North Korea continued into the New Year. Another year of tense relations for the North and the United States unfolded in 2002. The September 11 terrorist attacks on U.S. soil had transformed the Bush Administration foreign policy. The Bush team decided to deal with unfriendly regimes more decisively and, if necessary, unilaterally and forcefully. The watershed event of the year occurred on January 29, when President Bush, in the State of the Union address, accused North Korea of being one of the three members of the “Axis of Evil” that threatened United States and even world security. In this highly controversial speech, Bush described North Korea as “a regime arming with missiles and weapons of mass

84 Civil–Military Dynamics

destruction, while starving its citizens. . . . The United States of America will not permit the world’s most dangerous regimes to threaten us with the world’s most dangerous weapons” (Gross 2002). This speech sent shock waves around the world, as leaders awaited to see what it might mean in practice. Shortly after the speech, both the State Department and the U.S. Ambassador to South Korea, Thomas C. Hubbard, insisted that the president’s statement did not mean a policy shift. The United States, according to these sources, remained fully open to resuming bilateral talks with North Korea without any preconditions. North Korea, however, reacted harshly to the Axis of Evil description and responded to the speech directly. Ominously, Pyongyang turned up the rhetoric: “The option to strike impudently advocated by the United States is not its monopoly” (as quoted in Gross 2002). Thus, with Bush’s speech and Pyongyang’s reaction to it, limited hopes for a renewed diplomatic exchange between the United States and DPRK disappeared, at least for the foreseeable future. Considerable public reaction ensued in South Korea and Japan in particular. Various political groups accused the United State of destroying the North–South dialogue and threatening peace in East Asia. Although the State Department and Secretary of State Colin Powell himself asserted on some occasions that the United States was ready to resume a dialogue with North Korea at “any time, any place, or anywhere without preconditions,” this did not convince the North Koreans (Arms Control Agency 2003). A memorandum from President Bush stated that he would not certify North Korea’s compliance with the AF; however, for national security considerations, he waived legal obstacles to continuing financial support for the AF (Arms Control Agency 2003). Continuation of that support under such hostile conditions, however, did not bring the North back to the negotiation table either. Once again, the United States warned North Korea in August 2002 to comply as soon as possible with IAEA

Accounting for MIDs 85

safeguarding procedures. The North replied that it would not do so for at least three more years. Developments in the fall of 2002 escalated tensions between the United States and DPRK, and led to the ongoing confrontation that continues at this time of writing. In October 2002, the Assistant Secretary of State for East Asian and Pacific Affairs James Kelly visited North Korea and presented U.S. concerns about North Korea’s nuclear and—although self-delayed at the time—ballistic missile programs, export of missile components, conventional force posture, human rights violations, and overall humanitarian situation. Kelly informed North Korea that the way to improve bilateral relations might be a “comprehensive settlement” addressing these issues. North Korea called this approach “high handed and arrogant” and maintained its noncooperative stance. More importantly, the United States announced on October 16, 2002 that North Korea had admitted to the existence of its clandestine program to enrich uranium for nuclear weapons after Kelly had stated U.S. awareness of the program. Such a serious violation of the AF by North Korea raised immediate and intense reaction around the world. The Korean Peninsula Energy Development Organization (KEDO) announced on November 21, 2002 that it had suspended the oil deliveries, and the IAEA asked for clarification from the North about its new nuclear program. North Korea rejected those demands and announced that, due to the cut in oil supply, it would reopen the frozen nuclear reactors to produce electricity. In December 2002, North Korea cut all seals on IAEA surveillance equipment on its nuclear facilities and materials and ordered the inspectors out of the country. Confrontational behavior by North Korea continued with the announcement that it would withdraw from the NPT on January 10, 2003. In February 2003, the United States confirmed that North Korea had restarted a nuclear reactor previously frozen by the AF. The North also conducted two missile tests in February and March 2003 (Arms Control

86 Civil–Military Dynamics

Agency 2003). Perhaps most ominously, North Korea sent a fighter into South Korean airspace and shadowed a U.S. reconnaissance plane (Gross 2003). Meantime, the North Korean nuclear crisis became worse in that the Bush administration refused to hold any direct talks with the North. But with both China’s active diplomatic efforts and international pressure, the three party talks among the United States, North Korea, and China were held in Beijing in April 2003. The North proposed a package deal with a step-by-step plan. North Korea would take corresponding measures to meet the U.S. nuclear concerns while the United States would take a series of steps to re-supply heavy oil and food assistance, to continue construction of light-water reactors, and to give the North a written security assurance. Pyongyang demanded direct face-to-face talks with Washington and insisted that the dismantling of its nuclear program would be the last step. The United States flatly refused the proposal and wanted to include South Korea and Japan in a future meeting. On August 17–29, 2003, February 25–28, 2004, and June 23–26, 2004, six-party talks that included the North, United States, South Korea, China, Japan, and Russia ended without resolution. Russia was included upon the North’s request for its own sake. The talks could be considered a step forward because of the extended form of dialogue among concerned parties, but they did not represent diplomatic “giveand-take.” In particular, both the North and the United States repeated their principal positions. While the North pursued political and economic assistance prior to dismantling its nuclear facilities, the United States sought a complete, verifiable, and irreversible dismantlement (CVID) of the North’s nuclear program prior to any assistance (Hwang September 8, 2003, February 23, 2004; Gross 2004). The benefit of these talks to the United States was that China and Russia endorsed for the first time the U.S. goal of

Accounting for MIDs 87

fully dismantling North Korea’s nuclear programs through diplomatic routes, which could become the institutionalization of the process in the future. This increased hopes that China, in the future, might bridge the U.S. and North Korean positions. The diplomatic landscape, however, has became more complex not only because too many parties could cause confusion and dilute responsibility, but also since each state could prioritize its own interest first (Gross 2004). It appears that, “so far, [three] rounds of talks involving North and South Korea, China, the United States, Japan and Russia have failed to make any inroads towards an end to the nuclear crisis” (CNN.com April 21, 2004; see also CNN.com September 13, 2004).

Why Keep Diplomacy Alive? While some limited achievements, such as avoidance of war, can be detected from the diplomatic record between George W. Bush and Kim Jung-Il over the last four years, this case is more notable for the problems it reveals. The case bears not so much on the number of diplomatic channels available as on something that arguably is even more fundamental: The use of whatever nodes of communication do happen to exist in a time of trouble. Diplomacy, in this case, has not produced much more than the absence of war in the dyad. Three features, two of which have become more salient with time, combine to undermine the quest for peace through diplomatic channels between the United States and North Korea. First, diplomacy is deflected from problem solving by the presence of multiple audiences that clearly influence its progression, and not necessarily in helpful ways. The sometimes-inconsistent behavior of the United States toward North Korea,19 for example, can be traced to varying salience in the president’s perceived need to placate hard-liners in both Congress and the general public within his natural base of

88 Civil–Military Dynamics

supporters. Even a dictatorship such as North Korea is not able to focus completely; evidence suggests that on one or more occasions, Pyongyang used diplomatic venues to advertise nearby state of its own power status via nuclear weapons acquisition. Thus, either internal or external audiences, much more likely as matters progress to be watching diplomacy in real time (as opposed to later, after some delay in reporting), become part of the negotiating behavior itself and tend to contaminate rather than improve political communications between adversaries. Second, diplomacy is virtually by definition a fragile thing, and exogenous shocks can introduce new dimensions that may have harmful consequences. The 9/11 attacks and subsequent designation of North Korea by President Bush as one of the world’s “evil” states did not help matters. Intense shocks such as 9/11 arguably are more troublesome now than ever before because of their rapid and worldwide transmission as news events. Participants in diplomatic bargaining, especially when the situation is as delicate as that faced in recent years by the United States regarding the North Korean nuclear crisis, are at risk of immediate complications from major events that happen to unfold at or around the same time as their talks. Third, and finally, U.S. interactions with North Korea tend to support the idea that diplomacy’s link to MIDs may be at least in part a by-product of its correlation with great power status. Difficult negotiations, such as those over the North Korean nuclear problem, draw in great powers out of nearnecessity. Thus, a wide range of diplomatic connections, which great powers will feature, also will be linked to involvement in the most difficult and escalation-prone negotiations. While the preceding case illustration may not fully explain the anomalous findings about diplomacy and dispute involvement, it does suggest some causal mechanisms that are not built into the model of foreign policy decision-making process. Perhaps, in the end, not all forms of political communication

Accounting for MIDs 89

are an unmixed blessing. While media openness seems uniformly pacific in its effects, the same cannot be said for the volume of diplomatic channels in operation. Chapter six will take the analysis a step further by focusing on fatal MIDs.

This page intentionally left blank

Chapter 6

Accounting for Fatal Militarized Interstate Disputes Whether military fatalities can restrain belligerent foreign policies is an interesting issue for both theory and practice of international relations.1 Vigilant about potential loss of precious life, the public may be expected to raise its voice against military actions (Gartner and Segura 1998). In this regard, Luttwak (1996: 36) argues that “the prospect of high casualties, which can rapidly undermine domestic support for any military operation, is the key political constraint when decisions must be made on which forces to deploy in a crisis, and at what levels.” Sophisticated data analysis suggests that to be more than just speculation (Gartzke 2001). Oneal et al. (2003) use distributed-lag models to determine that democracy, economic interdependence, and joint membership in international organizations decrease the likelihood of fatal MIDs. This chapter examines military fatalities in MIDs as related to the newly introduced variables from the foreign policy decision-making process model conceptualized in figure 3.1 in chapter 3. The analysis will provide a further test for both civil–military dynamics and political communications as signposts in the new quest for peace. At the outset, one expectation rises above the other: Since public opinion is channeled through mass media to a large extent, we expect that, among the new variables in this research project, media openness will play

92 Civil–Military Dynamics

the most important role in predicting the relative likelihood of fatal MIDs. We adopt in this chapter the research design developed in chapter 4, but replace MID involvement with fatal MIDs as the dependent variable. This practice follows the compelling study by Oneal et al. (2003): The causal mechanism for MIDs is also applied to fatal MIDs in this study. We use Maoz’s dyadic MID dataset version 1.1 (updated January 2001) to identify the fatality level for each dyad in a given year. “Fatal MIDs” is a dichotomous dependent variable, that is, “1” (with at least one solider killed) versus “0” (no soldiers killed). The empirical results will unfold in two stages: logistic regression and then neural network. The focus on fatalities increases the likelihood that empirical analysis directly addresses the violent militarized disputes of greatest concern to foreign policy decisionmakers and scholars alike (Oneal et al. 2003: 12).

Logistic Regression Analysis Tables 6.1 and 6.2 present the empirical results based on the GEE logistic regression model2 from chapter 4.3 Each table consists of four different models based on data availability and time period. Oneal and Russett’s (1999c) eight variables appear as a “standard menu” in each column. Table 6.1 shows the effects of civil-military dynamics (i.e., civil–military relations and military manpower system) and political communications (i.e., diplomatic channels and media openness) based on all dyads; table 6.2 focuses on the civil–military dynamics and political communications based on politically relevant dyads only. Each table reveals theoretically intriguing results: The newly introduced variables, except for military manpower system, are associated, to one extent or another, with the likelihood of fatal MIDs. The following data analysis focuses mainly on the four new variables and the three neo-Kantian

Accounting for Fatal MIDs

93

peace variables—the main theoretical interests of this research.4 Table 6.1 reports the empirical results during the period from 1886 to 1992, using all dyads. As shown for GEE in the Table 6.1. Civil–Military Dynamics, Political Communications, and Predicting Fatal MIDs: Simplest, All Dyads Variable

1886–1992

Civil–military 0.4072 relations (IAAMEAB) 0.2982 Military manpower system (CONSCRAB) Diplomatic channels (SANSDC) Media openness (AMOAB) Lower democracy 0.0074 (DEML) 0.0340 Trade/GDP 548.5289** 188.4627 (DEPENDL) International 0.0202 organizations (IGO) 0.0208 Capability ratio 0.4615*** (CAPRATIO) 0.1421 Alliances 0.6833 (ALLIANCES) 0.7009 Noncontiguity 4.1397*** (NONCONTIG) 0.5948 Log distance 0.3899*** (DISTANCE) 0.1185 Only minor powers 1.2755* (MINORPWRS) 0.5540 Constant 5.9645*** 1.0504 Chi2 369.89 P of Chi2 0.0000 N 134,505

1886–1992

1886–1974

1950–1974

0.5089*

0.6239*

0.9393*

0.2979 0.0374

0.3304 0.1786

0.4244 0.5103

0.4186

0.6679 0.8846^ 0.4466

0.5962 1.1775^ 0.6326 22.7126*** 0.4789 0.0890^^^ 0.0248 1116.7450*** 361.1004 0.0309^

0.0113 0.0371 613.3767** 203.4084 0.0209 0.0219 0.4571*** 0.1418 0.7955 0.7280 4.0551*** 0.6076 0.4469*** 0.1235 1.3463** 0.5498 5.5015*** 1.0804 369.44 0.0000 122,026

0.0178 0.0289 942.8616** 317.6998 0.0203 0.0254 0.6120*** 0.2011 1.4976* 0.7205 3.9783*** 0.7271 0.6176*** 0.1268 1.0526 0.6853 5.5274*** 1.5838 228.18 0.0000 57,155

0.0168 0.3602 0.2211 1.5240** 0.5691 4.1862*** 0.7736 0.5527** 0.1979 1.0895 0.7289 6.5481*** 1.8910 7730.76 0.0000 37,405

* p  .05; ** p  .01; *** p  .001, one-tailed tests; ^ p  .05; ^^^ p  .001, one-tailed test but wrong sign.

94

Table 6.2. Civil–Military Dynamics, Political Communications, and Predicting Fatal MIDs: Politically Relevant Dyads Variable

1886–1992

Civil–military 0.4078 relations (IAAMEAB) 0.2983 Military manpower system (CONSCRAB) Diplomatic channels (SANSDC) Media openness (AMOAB) Lower democracy 0.0063 0.0348 (DEML) Trade/GDP 520.6034** (DEPENDL) 192.1641 International organizations 0.0161 (IGO) 0.0206 Capability ratio 0.4981*** (CAPRATIO) 0.1462 Alliances 0.7745 (ALLIANCES) 0.6765 Noncontiguity 2.2492*** (NONCONTIG) 0.4354 Log distance 0.2592* (DISTANCE) 0.1152 Only minor powers 0.4744 (MINORPWRS) 0.5623 Constant 4.4773*** 0.9383 107.87 Chi2 P of Chi2 0.0000 N 30,453

1886–1992

1886–1974

0.5105*

0.6708*

0.2988 0.0160

0.3297 0.1887

0.4166

0.6384 0.9057^ 0.4139

0.0093 0.0383 586.7418** 209.846 0.0161 0.0213 0.5005*** 0.1439 0.9132 0.6898 2.2063*** 0.4326 0.3189** 0.1218 0.5045 0.5634 3.9576*** 1.0366 115.48 0.0000 27,612

0.0169 0.0298 897.0855** 302.4596 0.0154 0.0242 0.6408*** 0.1980 1.5481* 0.6796 2.2806*** 0.5294 0.5033*** 0.1303 0.3116 0.6935 4.1992** 1.3925 80.42 0.0000 16,400

1950–1974 0.9897* 0.4430 0.4113 0.5485 1.3209^ 0.5901 19.9836*** 0.5204 0.0970^^^ 0.0264 1020.9320*** 330.8902 0.0167 0.0157 0.4526* 0.2240 1.6204*** 0.5106 1.8516** 0.7325 0.3868* 0.2024 0.0662 0.8116 4.7507*** 1.7980 4254.74 0.0000 8,360

* p  .05; ** p  .01; *** p  .001, one-tailed tests; ^ p  .05; ^^^ p  .001, one-tailed test but wrong sign.

Accounting for Fatal MIDs

95

second column, the civil–military relations variable is not statistically significant. While the economic interdependence hypothesis is supported, the democraticness and international organizations are not. It seems that, among the three neoKantian peace variables, economic interdependence alone decreases the likelihood of fatal MIDs. The third column focuses on the effects of civil–military dynamics (i.e., civil–military relations and military manpower system) during the period from 1886 to 1992, using all dyads. While the civil–military relations hypothesis is supported, the military manpower system hypothesis is not. When military influence increases, fatal MIDs are more likely. The democraticness hypothesis is not supported; the economic interdependence hypothesis is supported at the 0.01 level of statistical significance; and international organizations is not supported. The fourth column presents the results for civil–military dynamics and diplomatic channels during the period from 1886 to 1974, using all dyads. The civil–military relations coefficient is statistically significant at the 0.05 level; military manpower system is not statistically significant; and diplomatic channels is statistically significant at the 0.05 level, but counterintuitive. It appears that, the more the diplomatic channels exist, the more the fatal MIDs are to be expected, although it could be argued instead that diplomatic connections only indicate widespread interests. The economic interdependence hypothesis is supported, while democraticness and international organizations are not. The results in the last column of table 6.1 reflect the complete foreign policy decision-making process model portrayed in figure 3.1. The study period spans a relatively shorter period, from 1950 to 1974, owing to data availability. All dyads are analyzed. Civil–military relations is statistically significant at the 0.05 level; military manpower system is not statistically significant; diplomatic channels are statistically significant at the 0.05 level, but counterintuitive; and media openness is statistically significant at the 0.001 level.

96 Civil–Military Dynamics

Military leaders’ influence seems to encourage fatal MIDs, which may reflect consequences of the Powell Doctrine as emulated at a cross-national level. The U.S. Secretary of State Powell has contended that he would want to deploy force without restrictions on its scope and intensity, which would be likely to lead to lethal disputes (see Lane 1995: 20–21; Korb 1997: 24–25; Campbell 1998). The military manpower system variable never comes close to statistical significance in the last three columns. It appears that fatal, bloody disputes, most of which may involve high-level national security interests, are independent of different manpower systems. Diplomatic channels are likely to increase fatal MIDs. It can be inferred that diplomatic channels may be linked to states’ expectation of, and preparation for, lethal disputes. It seems that, as expected, the more open the mass media, the lower the likelihood of fatal MIDs. The so-called “casualty hypothesis” or “body-bag syndrome” holds that, with the possibility of military fatalities, the public will recoil or withdraw its support. This connection “is often mentioned by politicians and in the media as if it were an evident and established fact of life” (Everts 2001: 18). It is intriguing that the democraticness variable shows statistical significance at the 0.001 level, but with a counterintuitive connection. It appears that, within a dyad, the more democratic the less democratic state of the two happens to be, the more likely it is that fatalities will occur in a MID. Indeed, this finding, to some extent, is similar to those of democratization studies that report that transition to democracy seems to bring about the highest likelihood of international conflict (Mansfield and Snyder 1995a, 1995b, 1996). Since the study period, from 1950 to 1974, coincides with the second wave of democratization after World War II (Huntington 1991, 1997; Shin 1994), the transition to democracy might lead to an increase in fatal MIDs. This anomaly may be explained in another way. The democraticness variable may not be negatively sensitive to fatalities.

Accounting for Fatal MIDs

97

Although Feaver and Gelpi’s (2004: 7) survey-based study focuses on the United States only, it makes a case that “the belief, widely accepted by policymakers, civilian elites, and military officers, that the U.S. public is especially casualty phobic (meaning that public support for a mission will evaporate at the first sign of casualties) is a myth.”5 It may be argued that, since the four different model specifications and time periods do not support the prominent democratic peace hypothesis, regime type (as measured in the Polity data) has nothing to do with the decrease in fatal MIDs. Economic interdependence is statistically significant at the 0.001 level and international organization is statistically significant at the 0.05 level, but once again counterintuitive. It seems that joint membership in international organizations increases the fatal MIDs—a baffling result.6 Overall, the empirical results show that the civil–military relations, diplomatic channels, and media openness variables offer added value to Oneal and Russett’s (1999c) triangular peace. Table 6.2 reports the empirical results during the period from 1886 to 1992, based on politically relevant dyads only. The findings turn out to be very similar to those using all dyads in table 6.1. As a matter of fact, the statistical significance levels of the theoretically interesting variables in table 6.2 concur with those in table 6.1. Given the fact that the results from MID involvement as the dependent variable have shown a discrepancy between all dyads and politically relevant dyads in various studies, those from fatal MIDs shed a new light on peace studies. It seems that, when bloody battles occur, every international dispute becomes politically relevant—even in the absence of major power involvement or geographical contiguity. As shown for GEE in the second column, the civil–military relations variable is not statistically significant. While the economic interdependence hypothesis is supported, the democraticness and international organizations variables do not come out as expected. It seems that, once again, only economic

98 Civil–Military Dynamics

interdependence among the three neo-Kantian peace variables plays a consistent role in constraining fatal MIDs. The third column focuses on the impact of civil–military dynamics (i.e., civil–military relations and military manpower system) during the period from 1886 to 1992, using politically relevant dyads. While the civil–military relations hypothesis is supported, the military manpower system is not. When military power is greater in decision-making, the likelihood of fatal MIDs increases. The democraticness hypothesis is not supported; economic interdependence is supported at the 0.01 level of statistical significance; and international organizations is not supported. The fourth column presents the results for civil–military dynamics and diplomatic channels during the period from 1886 to 1974, using politically relevant dyads. Civil–military relations is statistically significant at the 0.05 level; military manpower system is not statistically significant; and diplomatic channels is statistically significant at the 0.05 level, but counterintuitive. It appears that, the more diplomatic channels there are, the more the fatalities a MID will produce. Political talks using an ambassador, envoy, or minister, and the like may turn out to be “cheap” and not even helpful—a normatively disappointing result that may be reflected in the unfolding example of U.S. negotiations with North Korea over weapons of mass destruction. Thus it seems that, while we do not measure whether such connections were actually used, diplomatic channels in a world of anarchy may be used in preparation for more aggressive, belligerent actions in the future—a result friendly to realism but also discouraging in a more general sense. While the democraticness and international organizations hypotheses are not supported, the economic interdependence hypothesis is confirmed. This is worth noting because two of the three neo-Kantian peace factors, including democraticness, seem to recede. The results in the last column of table 6.2 reflect the complete foreign policy decision-making process model shown in

Accounting for Fatal MIDs

99

figure 3.1, using politically relevant dyads. The civil–military relations coefficient is statistically significant at the 0.05 level; military manpower system is not statistically significant; diplomatic channels is statistically significant at the 0.05 level, but counterintuitive; and media openness is statistically significant at the 0.001 level. It should be noted that, while democraticness loses its centrality, media openness, once again, gains a strong foothold as a key element of democracy as related to political communication. It is worth noting that the democraticness variable does not hold up even with politically relevant dyads; it achieves statistical significance at the 0.001 level, but is counterintuitive. It seems that the more democratic the less democratic state in a dyad happens to be, the more the fatal MIDs we can expect. Economic interdependence is statistically significant at the 0.001 level and international organization is not statistically significant. Overall, the empirical results continue to suggest that the civil–military relations, diplomatic channels, and media openness variables are worthy of further consideration in extending the neo-Kantian peace of Oneal and Russett.

Neural Network Analysis As discussed in the second section in chapter 5, neural network modeling is utilized to better identify evidence of causal structure for the occurrence of fatal MIDs. The surprising results from the logit analysis discussed in the first section in this chapter (see footnote 6) can be reassessed effectively through neural network analysis. This technique more accurately represents nonlinear relations among variables and can deal with high correlations (i.e., interaction effects) between variables. As in Beck et al.’s (2000) presentation, Bayesian “backpropagation” selects 12 nodes in the input layer, 4 nodes in the

100 Civil–Military Dynamics

hidden layer, and 1 node in the output layer, and is applied to politically relevant dyads only during the period from 1950 to 1974.7 The model produces the optimal value of root mean square (RMS) error of 0.0563 in the training data set. Increasing the number of nodes in the hidden layer (i.e., what Beck et al. [2000] called “M”) does not improve the model’s performance.8 As with logistic regression analysis, the neural network analysis focuses mainly on civil–military dynamics, political communications, and three neo-Kantian peace variables.9 It should be kept in mind that only five variables (i.e., civil–military relations, diplomatic channels, media openness, democraticness, and economic interdependence) among the seven theoretically interesting ones show statistical significance in the last column in table 6.2, based on the logistic regression model for politically relevant dyads during the period from 1950 to 1974. Figure 6.1 shows three-dimensional plots of the neural network model output (i.e., the quantitative effect of fatal MIDs) against two of the inputs (i.e., independent variables) during the period from 1950 to 1974. That is, the surface plots show how the predicted value of fatal MIDs varies in terms of two independent variables. Figure 6.1a shows the casual structure of the civil–military relations (i.e., CMR) and military manpower system (i.e., CONSCRT) variables. The plot portrays a positive relationship between military leaders’ influence and fatal MIDs, and seemingly also for employing conscription and fatal MIDs. Figure 6.1b shows that both diplomatic channels (i.e., DIPCHAN) and media openness (i.e., MEDOPEN) variables are associated with the occurrence of fatal MIDs. The plot portrays a complex, nonlinear relationship between political channels (i.e., media openness and diplomatic channels) and fatal MIDs. It seems that media openness has a mixed impact on the occurrence of fatal MIDs. The relationship between diplomatic channels and MID involvement is nonlinear as well. As with

101 0.02

0.00 CMR

0.01 0.00 CONSCRT

1.00

Bayes1

Bayes1

0.03

0.01 0.00

1.00

1.00 DIPCHAN MEDOPEN

(a)

1.00

4.00

(b)

0.03

Bayes1

Bayes1

0.04

2.00

10.00 DEMO

0.02 0.00 ECOINT

0.14

10.00

JMIO

0.00 0.00 CONTIGUITY 1.00

102.00

(d)

(c)

0.03

0.00

–0.00 0.00

CAPRATIO ALLIANCE 1.00 (e)

7.67

Bayes1

Bayes1

0.02

1.87

0.00 0.00

DISTANCE MAJPOWER 1.00

9.41

(f)

Figure 6.1. Three-Dimensional Plots of the Network Output against Two Inputs, 1950–1974. The likelihood of fatal MIDs against (a) civil–military relations and conscription, (b) diplomatic channels and media openness, (c) democraticness and economic interdependence, (d) international organizations and contiguity, (e) capability ratio and alliance, and (f) distance and major power.

102 Civil–Military Dynamics

the logistic regression model, which produced a counterintuitive result, neural networks suggest that an increase in diplomatic channels may contribute to fatal MIDs. Figure 6.1c shows that, while the democraticness variable (i.e., DEMO) seems to have a marginal effect on fatal MIDs, the economic interdependence variable (i.e., ECOINT) leads to a great increase in fatalities from MIDs. A marginal linear relationship between democraticness and fatal MIDs, as well as a positive relationship between economic interdependence and fatal MIDs are portrayed. In contrast to the results from the logistic regression model, economic interdependence seems to heighten fatal MIDs. Figure 6.1d shows that, while the international organizations variable (i.e., JMIO) contributes to increasing, and then lessening, fatal MIDs, the contiguity variable (i.e., CONTIGUITY) is connected to an increase in fatal MIDs. A complicated, nonlinear, mixed relationship between international organizations and fatal MIDs, as well as a seemingly linear, positive relationship between contiguous dyads and fatal MIDs, is portrayed. Figure 6.2 shows sensitivity plots (i.e., marginal effects) for each independent variable during the period from 1950 to 1974.10 Each sensitivity plot offers a visual representation of a nonlinear relationship for an independent variable. As shown in figure 6.1a, the ascendance of military leaders’ power features a seemingly linear, positive relationship. It appears that greater military influence on a foreign policy decision-making process contributes to an increase in fatal MIDs. Military manpower system with conscripts seems to have a positive but statistically meaningless impact on fatal MIDs. Diplomatic channels turn out to be linear and positive. A 26.67 percent increase in diplomatic channels (from 2.175 to 2.755), for example, increases fatal MIDs by 13.79 percent (from 0.021637 to 0.02462).11 The plot is counterintuitive, meaning that an increase in diplomatic channels contributes to an increase in fatal MIDs. These results are very similar to

103 0.02

0.02

0.00

0.00 0.00

1.00

0.00

0.04

0.04

0.00

0.00 1.00

4.00

0.00

(c) 0.04

0.00

0.00 10.00 (e)

1.00 (d)

0.04

–10.00

1.00 (b)

(a)

0.00

0.14 (f)

Figure 6.2. Marginal Effects of Each Independent Variable, 1950–1974. (a) civil– military relations, (b) conscription, (c) diplomatic channels, (d) media openness, (e) democraticness, (f) economic interdependence, (g) international organizations, (h) contiguity, (i) alliance, (j) capability ratio, (k) distance, and (l) major powers.

104 0.04

0.04

0.00

0.00 2.00

102.00

0.00

(g) 0.03

0.03

–0.00

–0.00 0.00

1.00 (h)

1.00

0.00

(i)

7.67 (j)

0.03

0.03

0.00

0.00 1.87

9.41

0.00

(k)

1.00 (l)

Figure 6.2. Continued

Accounting for Fatal MIDs

105

(or consistent with) those from the logit models. The media openness variable reveals a seemingly nonlinear, somewhat mixed relationship. In other words, when media openness in a dyad increases, fatal MIDs become more likely, and then somewhat less likely toward the end. It appears counterintuitive that democraticness should have even a slightly positive impact on the occurrence of fatal MIDs. Economic interdependence also seems to show a generally positive causal structure: It starts to dramatically increase fatal MIDs and then becomes static toward the end. This startling result should be taken with multiple grains of salt, perhaps, given that it is inconsistent with both our logit analysis and neural network analysis of all MIDs. Joint membership in international organizations reveals a complicated, nonlinear, negative relationship. An increase in joint membership in international organizations seems to increase, and then to decrease, the chance of fatal MIDs. The contiguity variable, as hypothesized, reveals a seemingly linear, positive relationship. It is apparent that, if two states are contiguous, they are more likely to produce fatal MIDs. Table 6.3, with respect to statistical significance, summarizes the across-the-board results from both logit and neural network models from the first and second sections in this chapter. The empirical results from both logit and neural network models indicate that media openness, not democraticness based mainly on the degree of executive constraints as quantified in the Polity data, appears to be the most defining attribute of democracy in accounting for fatal MIDs. The fascinating findings about media openness throughout this chapter and in chapter 5 encourage further investigation. Two important and unresolved issues arise from the preceding analysis of the neo-Kantian peace elements in connection with fatal MIDs. First, the empirical results from the first and second sections in this chapter, in general, are at odds with those of Oneal et al. (2003) and less encouraging to the

106

Table 6.3. A New Quest for International Peace?: A Summary of Statistical Significance in Fatal MIDs Variable

Logistic regression Simplest, all dyads

Neural network Politically relevant dyads

Politically relevant dyads

[1]

[2]

[3]

[4]

[1]

[2]

[3]

[4]

[4]

New Civil–military relations (H1) Manpower system (H2) Diplomatic channels (H3) Media openness (H4)

n/a n/a n/a

* _ n/a n/a

* _ ^ n/a

* _ ^ ***

_ n/a n/a n/a

* _ n/a n/a

* _ ^ n/a

* _ ^ ***

Intuitive Intuitive Counterintuitive Mixed

Kantian peace Democraticness Economic interdependence International organizations

_ ** _

_ ** _

_ ** _

^^^ *** ^

_ ** _

_ ** _

_ ** _

^^^ *** _

Counterintuitive Counterintuitive Mixed

* p  .05; ** p  .01; *** p  .001, one-tailed tests; ^ p  .05; ^^^ p  .001, one-tailed test but wrong sign. _: Not statistically significant. n/a: Not applicable. [1]: 1886–1992. [2]: 1886–1992. [3]: 1886–1974. [4]: 1950–1974.

Accounting for Fatal MIDs

107

triangular peace.12 Second, according to the neural network models, economic interdependence is likely to increase fatal MIDs, which is not congruent with the earlier logit results. Given the preceding two points, it is fair to say that more questions have been raised than answered with respect to fatal MIDs.

This page intentionally left blank

Chapter 7

A Key Issue of Measurement: Military Expenditure and Civil–Military Relations This chapter explores a key measurement issue with respect to civil–military relations. The issue is singled out for attention because of its potential for controversy: The degree of civilian versus military control is perhaps the least effectively measured entity among the concepts introduced by the foreign policy decision-making model. In the previous chapters, we have measured the degree of civilian versus military influence through change in military expenditure. However, this might be regarded instead as an indicator of the severity of threats to national security as perceived by national leaders—in other words, more serious dangers could produce higher military expenditures. This chapter looks into that salient possibility by treating military expenditure and other variables, such as military manpower system, as components of an endogenized indicator of civil–military relations. The results obtained here will be important to the overall project, in which civil versus military influence is put forward as a key causal factor in relation to the propensity of a state to become involved in external disputes. This chapter continues with two additional sections. The following section operationalizes civil–military relations in two steps. The second section reports the empirical implications— in effect, an answer to whether an endogenized measurement of civil–military relations really matters.

110 Civil–Military Dynamics

Measuring Civil–Military Relations in Two Steps A General Hypothesis As discussed in the previous chapters, we side with the idea that, as military leaders in a dyad increase their influence in the decision-making processes, the two states become more likely to be aggressive and belligerent, which in turn may result in interstate disputes. Thus, the general hypothesis about civil–military relations is as follows: H1: As the relative influence of the military increases for the states in a dyad, the likelihood of involvement in MIDs also increases for that dyad.

To test the general hypothesis about civil–military relations, this study follows the two steps of model building depicted in figure 7.1. In the first step, the civil–military relations model is built, incorporating societal militarism (i.e., Militarism in year t), previous military influence in civil–military relations (i.e., CMR in years t  1, t  2, and t  3), conscription (i.e., Conscription in year t), and severity of threats to national security (i.e., Threats in year t) on the far left-hand side in figure 7.1. The first step of model building depicts the impact of those six factors on civil–military relations (i.e., CMR in year t). This first step will produce predicted values for civil–military relations that will be used in the second step. In the second step from figure 7.1, the neo-Kantian peace model is combined with the civil–military relations model by incorporating a year-lagged term of the predicted values for civil–military relations (i.e., PVCMR in year t  1).1 As indicated by the direction of the arrows, step two puts forward nine factors (i.e., PVCMR, Democraticness, Interdependence, Joint membership in international organizations, Capability ratio, Alliance, Noncontiguity, Geographic distance, Only minor powers in year t  1) as responsible for international

2nd step

1st step

Militarism in year t

Democraticness in year t – 1

Interdependence in year t – 1

CMRa in year t – 1 CMR in year t – 2

Capability ratio in year t – 1 CMR in year t

PVCMRb in year t – 1

MID involvement in year t Alliance in year t – 1

CMR in year t – 3 Conscription in year t

Joint membership in year t – 1

Only minor powers in year t – 1

Geographic distance in year t – 1

Noncontiguity in year t – 1

Threats in year t

Notes:

a

Civil–military relations

b Predicted values for civil–military relations

: Causal direction : Taking one year-lagged term

111

Figure 7.1. Two-Step Model Building Procedure for Civil–Military Relations and Neo-Kantian Peace

112 Civil–Military Dynamics

conflict at the dyadic level (i.e., MID involvement in year t). (Each of these factors will be presented in more detail momentarily.) The combined model attempts to answer the question of what factors account for dyads with MIDs, more specifically, whether civil–military relations measured in the two steps really matter. The arrows indicate the causal direction of each variable and, in particular, the arrow with a dot at the left end means taking a year-lagged term of predicted values of civil–military relations (i.e., from CMR in year t to PVCMR in year t  1). More detailed discussion follows below.

The First Step: Measuring Military Influence in Civil–Military Relations In the first step, we develop an endogenized indicator of civil–military relations. The first of the two steps is to obtain predicted values for civil–military relations in numerical terms by taking into account societal militarism, past influence of military leaders, military conscription system, and severity of threats to national security. Thus, the first step elaborates our definition of civil–military relations by introducing dimensions such as societal militarism, past influence of military leaders, military manpower system, and international threats. Sarkesian’s (1992) classic model of civil–military relations provides the background for our theoretical elaboration; a brief exegesis of it, along with our modifications, appears in the appendix to this book.2 The three elements in the domestic political–social system and the one element from international relations will serve as important determinants of military influence in civil–military relations for each state. Thus, the equation for the first step in figure 7.1 is as follows: CMRt  ␣  ␣1Militarismt  ␣2CMRt1  ␣3CMRt2  ␣4CMRt3  ␣5Conscriptiont  ␣6Threatst  1

(1)

A Key Issue of Measurement

113

Here, CMRt: military influence in civil–military relations in year t Militarismt: societal militarism in year t CMRt1: past military influence in civil–military relations in year t  1 CMRt2: past military influence in civil–military relation in year t  2 CMRt3: past military influence in civil–military relations in year t  3 Conscriptiont: military conscription system in year t Threatst: severity of international threats in year t 1: error term. Military influence in civil–military relations in year t (CMRt), as the dependent variable in Equation 1, measures military relative to civilian influence in decision-making. As mentioned in chapter 4, there is no quantitative indicator available as a “bottom line” for civil–military relations on a crossnational, time-series basis. Thus, we utilize military expenditure to measure the degree of military influence in each state. The operationalization is based on the inference that rising military vis-à-vis civilian power is likely to result in increased military expenditures (Hill 1978; Zuk and Thompson 1982). Niskanen’s (1968) model of the size-maximizing bureaucrat provides the intuition here—the relative influence of military bureaucrats should determine shifts in the total budget of their sector within the government. Goertz and Diehl (1986: 554) sum up this line of reasoning: “High [military] allocations could indicate the influence that military officials have in government decision making.” In a word, military influence is assumed to be a function of change in military expenditure. Current military leaders’ influence (CMRt), therefore, is calculated as follows: First, an annual growth rate for military expenditure is calculated for each state in a dyad for a given year. Second, the smaller value of the percentage expenditure between the two states in each dyad-year becomes the recorded

114 Civil–Military Dynamics

value.3 It should be noted that, although the civil–military relations variable utilizes military expenditure as the degree of influence, the operationalization does not rely on the number of dollars of military expenditure, but instead uses its increase or decrease in annual growth rate to identify the direction of change for military leaders’ relative influence (Hartley and Russett 1992: 909).4 As noted in chapter 4, the data come from Bennett and Stam’s (2000b) data set (EUGene, version 2.10). As shown on the left- and right-hand sides of the dotted vertical line in figure 7.1 (i.e., under the first step), we postulate that current military influence in civil–military relations in year t (CMRt) in Equation 1 is a function of the four elements, expressed as independent variables: (a) current societal militarism; (b) military leaders’ previous activities (i.e., directed toward policy influence) for the last three years; (c) current military service system (i.e., conscripted or voluntary); and (d) current severity of international threats (i.e., level of hostility). Societal militarism is an important element in determining military influence in civil–military relations in that it governs or embraces its militaristic characteristics. The basic inference is that, when states maintain more soldiers at the ready, military leaders have more of an opportunity to exercise power over civilians and have the potential to become more militaristic as a by-product (Vagts 1959; Donovan 1970; Souva and Prins 2001). In particular, Rosecrance (1986) contends that some states are poised for military action, while others are directed toward wealth accumulation. In other words, more soldiers translate into a greater likelihood of military preferences being realized in policy outcomes. Societal militarism is calculated as follows: First, the number of soldiers as a percentage of population is determined for each state. Second, the weak link assumption is used once again to derive a score for the dyad-year. Like the military expenditure data, the data for this variable are obtained from Bennett and Stam’s data set, EUGene, which includes military personnel and total population in the form of the National Capabilities data file.

A Key Issue of Measurement

115

Military budgets are allocated, pretty much in general, according to long-term plans. In other words, current military influence in civil–military relations can be attributed to previous influence exerted by military leaders in decision-making for a given state. To account for these past effects, we take three lagged terms of current military influence in civilian versus military interactions (i.e., CMRt  1, CMRt  2, and CMRt  3).5 Past values of military expenditure (i.e., influence) for the previous three years are used to predict the present values of military expenditure (i.e., relative influence). Once again, the predicted values are not numerical dollars, but the annual percentage change. Two basic kinds of military manpower system exist: conscripted and voluntary. Since a conscription system effectively expands military society more than would a voluntary system of service, all other things being equal, it can be expected to reinforce military influence.6 This argument derives from the standard references regarding bureaucratic politics and infighting (i.e., “where you stand depends upon where you sit” [Gray 1975: 86]), which emphasize budget maximization (Niskanen 1968). Influence correlates almost directly with spending, that is, “size does matter.” In that context, conscription serves as a catalyst to military spending that rivals or exceeds even the most extravagant expenditures on hardware. Moreover, nothing makes the case for expansion of the military more effectively than the use of force itself. Conscripted forces also provide quicker and higher military readiness or preparedness than all-volunteer forces (Ross 1994). In this regard, Duindam (1999: 119) also contends that, “if the members of the democratic community differ in their preferences for a war, a decision in favour of a war will be more probable in the case of a conscription force than in the case of an all-volunteer force.” Thus, military leaders can claim that options involving deployment or use of force in a foreign policy situation are relatively more available under conscription, which in turn contributes to military influence. It is worth noting that the danger of conscripted soldiers is latent

116 Civil–Military Dynamics

within the theoretical discussion from Russett and Oneal (2001: 19), perhaps the most prominent advocates of neo-Kantianism: Napoleon Bonaparte, drawing on popular enthusiasm, created a large nationalist army drawn from the mass of the French citizenry, rather than from professional soldiers and mercenaries. This nationalist army, combined with Napoleon’s military genius, had the potential to overwhelm the old, more aristocracybased armies of other states. Consequently, France threatened to become the dominant state, a hegemon, that could reduce all others on the continent to a more or less subservient status.

In Russett and Oneal’s sense, conscripted soldiers run against the logic of the democratic peace in that they seem to lead to international conflict. Conscription is a dichotomous variable. The variable is coded as “1” if both states in a dyad adopt a conscription system for the active duty military personnel; it is “0” otherwise. As noted in chapter 4, no conscription data have been collected over an extended spatial and temporal domain, so our data are based mainly on the sources described in that chapter. It can be argued that the military leaders’ influence (CMRt), measured via the annual growth rate of military expenditure, is at least to some degree an indicator of the severity of threats to security as perceived by national leaders. In other words, the more serious the threats, the more the military expenditure to be expected. If so, it becomes plausible that, due to international tension and/or perceived security threats, states with high military expenditures are more likely to get into international disputes. Salient examples are Britain and France during the period from 1938 to 1939, and the United States during the period from 1940 to 1941, with escalated expenditures in reaction to monitoring external threats from the Axis powers.7 For this reason, we choose to control the possible severity of international threats (i.e., states do [or do not] spend more in the

A Key Issue of Measurement

117

face of threats) in the model. We utilize Maoz’s (1999) standard five categories for the level of hostility in each dyad-year: 1 for no militarized action; 2 for threat to use force; 3 for display of force; 4 for use of force; and 5 for war.8 The ordinality of the level of hostility ranges from least (i.e., level 1) to most severe threats (i.e., level 5) to national security. The control variable for threats (i.e., Threatst) is measured as the highest level of hostility reached by either member of a dyad in a year in the five categories. After the military leaders’ influence (CMRt) is regressed on societal militarism (i.e., Militarismt), military leaders’ activities to exert influence for the last three years (i.e., CMRt1, CMRt2, and CMRt3), military manpower system (i.e., Conscriptiont), and the severity of international threats (i.e., Threatst), the GEE Gaussian regression produces the following results for Equation 1, along with predicted values for relative military influence in civil–military relations:9 CMRt  0.0774  0.0006Militarismt  0.0251CMRt1 (0.0041) (0.0026) (0.0035)  0.0347CMRt20.0633CMRt3 (0.0041) (0.0035)  0.0064Conscriptiont  0.0044 Threatst  1 (0.0017) (0.0038)10

(1
)

The coefficient for Militarismt shows that, contrary to expectations, societal militarism—operationalized as the total percentage of soldiers in a population—has a negative relationship with military influence in civil–military relations, although it is statistically insignificant. As expected, however, military activities during the previous three years and the presence of conscripted soldiers are associated with greater military influence in civil–military relations, measured in terms of military expenditure in the annual growth rate. All of the coefficients for previous military activities (i.e., CMRt1, CMRt2, and CMRt3) and conscription (i.e., Conscriptiont) are statistically significant at the 0.001 level.

118 Civil–Military Dynamics

Our control variable for international threats (i.e., Threatst) is not statistically significant, indicating that military leaders’ decision to increase military expenditure (CMRt) is not influenced by the severity of current threats to national security. Thus, we conclude that our operationalization, which uses the annual growth rate of military expenditure, properly captures our intuition about “hawkish” military leaders and “dovish” civilian leaders rather than international threats. Predicted values for military influence in civilian versus military interactions are used, in the next section, for the second step.

The Second Step: Building a Combined Model of Civil–Military Relations and the Neo-Kantian Peace As discussed above, the predicted values from Equation (1
) for military influence in civil–military relations are incorporated into Oneal and Russett’s (1999c) neo-Kantian peace model to aim for a better explanation for the likelihood of MIDs. The anticipated causal mechanism for military influence in civil–military relations is as follows: Rising military vis-à-vis civilian power is likely to increase military expenditure, which in turn may cause more involvement in MIDs. Table 7.1 presents both the hypotheses related to MIDs that will be tested and auxiliary hypotheses related to the current military influence in civil–military relations in year t, which produced the predicted value of military influence in civil–military relations that will be used in Equation 2. It also describes how to operationalize each variable and its data sources. While the top part of table 7.1 shows seven new hypotheses, and operationalization and data sources (the first hypothesis is the general one, the next six are auxiliary for the “indirect” relationship), the bottom part presents eight replicating hypotheses, and operationalization and data sources from Oneal and Russett. Equation 2 completes the research design that is shown on the right-hand side of the dotted vertical line in figure 7.1.

119 Table 7.1. Hypotheses and Operationalization Variable

Hypothesis

Operationalization (and data source)

As the relative influence of the military increases for the states in a dyad, the likelihood of involvement in militarized interstate disputes also increases for that dyad

The smaller value of the annual growth rates for military expenditure in dyad-year t (data from the EUGene, version 2.10)

Societal militarism (H1–1)

As the relative influence of the societal militarism increases for the states in a dyad, the likelihood of military influence increases for that dyad

The smaller value of the number of soldiers as a percentage of population in dyad-year t (data from the EUGene, version 2.10)

Civil–military relations1 (H1–2)

As the relative influence of the military in civil–military relations increases for the states in a dyad at time t  1, the likelihood of military influence increases for that dyad

The smaller value of the annual growth rates for military expenditure in dyad-year t  1 (data from the EUGene, version 2.10)

Civil–military relations2 (H1–3)

As the relative influence of the military in civil–military relations increases for the states in a dyad at time t  2, the likelihood of military influence increases for that dyad

The smaller value of the annual growth rates for military expenditure in dyad-year t  2 (data from the EUGene, version 2.10)

Civil–military relations3 (H1–4)

As the relative influence of the military in civil–military relations increases for the states in a dyad at time t  3, the likelihood of military influence increases for that dyad

The smaller value of the annual growth rates for military expenditure in dyad-year t  3 (data from the EUGene, version 2.10)

Conscription (H1–5)

Dyads composed of states with (without) a conscription system are more (less) likely to increase military influence

1 if both states adopted conscripted soldiers in dyad-year t, 0 otherwise (data from various surveys and statistical books)

International threats (H1–6)

As international threats are perceived to be more serious, the likelihood of military influence increases for that dyad (i.e., the more the military expenditure)

The higher value of the level of hostility in dyad-year t (data from Maoz’s dyadic MIDs dataset, version 1.1)

New hypotheses Civil–military relations (H1)

120 Table 7.1. Continued Variable

Hypothesis

Replicating hypotheses from Oneal and Russett (1999c) Democraticness The more democratic the less democratic state, the more constrained it will be from engaging in a dispute, and the more peaceful the dyad

Operationalization (and data source)

The smaller value of the democracy scores (data from Oneal and Russett’s [1999c] collection)

Economic The more interdependent the less interdependence economically dependent state, the more constrained it will be from engaging in a dispute, and the more peaceful the dyad

The smaller value of the trade-to-GDP ratios (data from Oneal and Russett’s [1999c] collection)

International organizations

The more joint memberships in international organizations, the more constrained two nation-states will be from engaging in a dispute, and the more peaceful the dyad

Number of international organization memberships shared (data from Oneal and Russett’s [1999c] collection)

Alliance

If two states are militarily allied, they are less likely to be engaged in a dispute

1 if both states are linked by defense treaty, neutrality pact, or entente (data from Oneal and Russett’s [1999c] collection)

National capability ratio

The more preponderant the military capabilities, the more constrained the two states will be from engaging in a dispute, and the more peaceful the dyad

Logarithm of ratio of higher to lower power capability (data from Oneal and Russett’s [1999c] collection)

Noncontiguity

If two states are not contiguous, they are less likely to be engaged in a dispute

1 if both states are not contiguous by land border or less than 150 miles of water, 0 otherwise (data from Oneal and Russett’s [1999c] collection)

Geographic distance

The farther apart the two states are, the less likely they are to be involved in a militarized dispute

Logarithm of dyadic distance in miles between capitals or major ports (data from Oneal and Russett’s [1999c] collection)

Only minor powers

If there is no major power involvement, militarized disputes are less likely

1 if minor powers involved, 0 otherwise (data from Oneal and Russett’s [1999c] collection)

121

Since Oneal and Russett present MID involvement rather than initiation as their dependent variable,11 we choose MID involvement for purposes of comparison. Due to the fact that the dependent variable, Disputet, is dichotomous, a logistic regression model is used. As is the usual practice in the neoKantian peace research program, all independent variables are lagged by one year. Thus, they are not affected by the degree of conflict present in the dyad-year to be explained. Disputet  ␤  ␤1PVCMRt1  ␤2Demot1  ␤3EcoIntt1  ␤4JMIOt1  ␤5CapRatiot1  ␤6Alliancet1  ␤7NonContiguityt1  ␤8Distancet1  ␤9MinPowert1  ␧2

(2)

Here, Disputet: MID involvement PVCMRt1: predicted value of military influence in civil– military relations Demot1: democraticness EcoIntt1: economic interdependence JMIOt1: joint membership in international organizations CapRatiot1: national capability ratio Alliancet1: allied states NonContiguityt1: noncontiguous states Distancet1: geographical distance MinPowert1: only minor powers 2: error term.

Empirical Implications: Do Civil–Military Relations Still Matter? Table 7.2 presents the empirical results based on the logistic regression model during the period from 1886 to 1992, as in Oneal and Russett. It includes predicted military influence in civil–military relations and the eight variables from Oneal and

122 Civil–Military Dynamics

Russett. The main focus is on estimating the impact of the degree of influence for civilian versus military leaders on the likelihood of MID involvement. Table 7.2 consists of both replications and further empirical results and includes several statistical specifications that parallel those that predominate in the neo-Kantian peace literature. All of the replications in each unshaded column show very similar results to Oneal and Russett with respect to both direction and statistical significance.12 The first column lists the nine independent variables. Based on the GEE logistic regression, the second, unshaded column shows the replications of Oneal and Russett’s model and the third, shaded column reports the results from Equation 2 for the simplest run— all dyads from 1886 to 1992. The fourth, unshaded column, which is based on Beck et al.’s (1998) peace years correction method, shows the replications of Oneal and Russett’s model. The fifth, shaded column reports the results from Equation 2 for all dyads during the period from 1886 to 1992. Based on GEE logistic regression, the sixth, unshaded column shows the replications of Oneal and Russett’s model and the seventh, shaded column reports the results from Equation 2 for all dyads during the multipolar period after 1886 and before the Cold War (i.e., from 1886 to 1939). Finally, based on GEE logistic regression, the eighth, unshaded column shows the replications of Oneal and Russett’s model, and the ninth, shaded column reports the results from Equation 2 for politically relevant dyads only during the period from 1886 to 1992. As will become apparent, the bottom line to be derived from the table is this: It reveals statistical significance for military influence in civil–military relations in accounting for MID involvement. In a word, when military influence in civil– military dynamics increases, the likelihood of MID involvement becomes greater. As shown for GEE in the second column of the table, the civil–military relations variable is statistically significant at the 0.001 level, indicating that, with greater overall military

Table 7.2. Do Civil–Military Relations Still Matter? Variable

Civil–military relations (PSAGRME) Lower democracy (DEML) Trade/GDP (DEPENDL) International organizations (IGO) Capability ratio (CAPRATIO) Alliances (ALLIANCES) Noncontiguity (NONCONTIG)

1886–1992 Simplest, all dyads

1886–1992 Peace years correction

10.0493***

1886–1939 All dyads

14.5357***

23.6575***

1886–1992 Politically relevant dyads 5.4456**

0.0643*** 0.0119 71.1420*** 20.5816 0.0037

2.3469 0.0626*** 0.0118 73.9270*** 21.3162 0.0045

0.0476*** 0.0097 42.5550*** 12.9179 0.0122^^

2.4017 0.0429*** 0.0094 45.4603*** 13.1892 0.0099^^

4.4275 0.0664*** 0.0537*** 0.0120 0.0120 47.5744* 50.0953* 23.0807 23.6584 0.0162^ 0.0109

0.0607*** 0.0124 42.1343** 16.8875 0.0085*

2.0727 0.0597*** 0.0122 43.7819** 17.4218 0.0092*

0.0043 0.2380*** 0.0587 0.2933 0.1838 2.0626*** 0.1985

0.0043 0.2376*** 0.0594 0.2867 0.1830 2.0548*** 0.1982

0.0041 0.1954*** 0.0469 0.4256** 0.1592 1.4701*** 0.1593

0.0042 0.1966*** 0.0474 0.4213** 0.1582 1.4586*** 0.1590

0.0078 0.0080 0.4042*** 0.4154*** 0.0750 0.0731 0.2497 0.2761 0.2085 0.2116 1.1500*** 1.1683*** 0.2206 0.2211

0.0043 0.2866*** 0.0579 0.3986* 0.1916 1.1943*** 0.1957

0.0044 0.2901*** 0.0585 0.3977* 0.1905 1.1779*** 0.1958

123

124

Log distance (DISTANCE) Only minor powers (MINORPWRS) Constant Chi2 P of Chi2 Log likelihood Pseudo-R2 N

0.4679*** 0.0631 1.8531***

0.4661*** 0.0639 1.8317***

0.1807 1.7432*** 0.5441 1189.96 0.0000

0.1797 0.9522 0.5848 1193.57 0.0000

111,389

111,389

0.3633*** 0.0559 1.6812*** 0.1316 0.9914* 0.4849 1725.86 0.0000 4381.839 0.298 111,475

0.3709*** 0.0543 1.6605*** 0.1305 0.2440 0.5233 1747.75 0.0000 4347.414 0.303 111,475

0.2395*** 0.2259*** 0.0728 0.0727 1.5541*** 1.5594***

0.2589*** 0.0681 0.6743**

0.2597*** 0.0695 0.6590**

0.2100 0.2104 2.6893*** 0.9237 0.6644 0.7729 268.60 288.06 0.0000 0.0000

0.2348 1.4581** 0.5938 154.14 0.0000

0.2340 0.9907 0.6346 157.67 0.0000

14,607

24,921

24,921

* p  .05; ** p  .01; *** p  .001, one-tailed tests. ^ p  .05; ^^ p  .01, one-tailed tests but wrong sign.

14,607

125

influence, MID involvement increases in likelihood. Statistically significant at the 0.001 level, the coefficient (0.0626) of the democraticness variable turns out to be about the same as the replicated one (0.0643). These results, which preserve the significance of democracy, also show that civil–military relations matter. Consistent with what classical liberals argued long ago (Oneal and Russett 1997), the coefficient of the economic interdependence variable (73.9270 versus 71.1420) appears somewhat stronger as the research design becomes more complete. The coefficient of the international organizations variable is not statistically significant. While the alliance hypothesis is not supported by the GEEbased results, the propositions about national capabilities, noncontiguity, geographic distance, and only minor powers present are confirmed. As students of power preponderance argue, it seems that an asymmetry in national capabilities is less likely to produce MID involvement. Allied status seems to have no effect on the likelihood of MID involvement for a dyad. As expected, non-neighboring dyads are relatively less likely to engage in MIDs. A geographic obstacle (i.e., long distance) between members of a dyad appears to decrease the likelihood of MID involvement. And minor power dyads are less likely to become involved in MIDs than those that include at least one major power. Columns four and five in table 7.2 show the results based on Beck et al.’s peace years correction model. The purpose of this analysis is to see whether a different statistical method produces consistent results. To wit, it does. The civil–military relations hypothesis once again is supported strongly at the 0.001 level. The three neo-Kantian peace variables show statistical significance at the 0.001, 0.001, and 0.01 levels, respectively. However, among the latter, the connection involving international organizations and MIDs turns out to be counterintuitive. Joint membership in international organizations seems to increase the likelihood of MIDs. As expected, the other control variables retain statistical significance. The peace

126 Civil–Military Dynamics

years correction model reports a pseudo-R2 that tells how well the model performs in an overall sense—30.3 percent for MID involvement, which is similar to, but slightly better than the replicated version, at 29.8 percent. The sixth and seventh columns of table 7.2 present the results for the multipolar period after 1886 and before the Cold War, which are similar to those of the GEE model. These columns test realist criticisms that Cold War tensions are the actual forces behind the democratic peace (Gartzke 1998; James et al. 1999). Civil–military relations still show a strong statistical significance at the 0.001 level. While the democraticness and economic interdependence hypotheses are supported, the international organizations hypothesis, once again, is not. The control variables are also supported—except for alliances. The results in the eighth and ninth columns of table 7.2 are for politically relevant dyads only, that is, when either major powers become involved in dyadic disputes or where members of the dyad are geographically contiguous. Politically relevant dyads make up most of the dispute-prone pairs, to which students of security relations denote importance for the stability of the international system (Russett and Oneal 2001). It is interesting that all of the hypotheses are supported. The civil– military relations variable, once again, is statistically significant at the 0.01 level. The three neo-Kantian peace variables—this time including international organizations—pass the statistical significance test, as do the control variables. In sum, the endogenized indicator of civil–military relations based on the two steps has been tested with different statistical methods, sampling, and time periods. It seems that, as civilian supremacy obtains, MID involvement becomes less likely. Thus, as shown in the previous chapters, the newly introduced factor in the present study—civil–military relations—is a good candidate for policy-relevant research in the new quest for a more peaceful world in the millennium and beyond. It should be emphasized that, according to the empirical results, the

A Key Issue of Measurement

127

severity of threats to national security appears to be unconnected with power struggles between civilian and military leaders. Thus, controlling for national threats could not have altered the persistent empirical implications of civil–military relations for MID involvement in chapter 5 as well as fatal MIDs in chapter 6.

This page intentionally left blank

Chapter 8

A Closer Look at Media Openness: From Institutional Democracy to Media Openness? Among the four factors introduced in this book, media openness has emerged as preeminent at the dyadic level in its association with avoiding MIDs. The effects of media openness supersede those from the other components of the foreign policy decision-making model as well as Oneal and Russett’s (1999c) three neo-Kantian peace variables (i.e., democraticness, economic interdependence, and joint membership in international organizations). These persistent and important findings encourage more in-depth assessment of media as a component of democracy in direct comparison with Polity-based democracy. In particular, the institutional elements from the quasistandard Polity Index might be re-examined when compared with media openness with regard to explanation of MIDs and related phenomena. Thus, the present chapter presents (a) more refined theoretical linkages between media and conflict and (b) additional empirical tests for the longer period from 1950 to 1992, which corresponds approximately to the Cold War era.1 Three additional sections appear below. The second section discusses the role of media openness with respect to democracy and interstate disputes and then presents two theoretical linkages between media and conflict. Hypotheses, measurement,

130 Civil–Military Dynamics

and model building appear in the third section. The fourth section focuses on empirical evaluation of media openness in inhibiting MIDs and fatal MIDs.

Media Openness, Democracy, and Interstate Disputes Contemporary democratic or neo-Kantian peace scholarship has not emphasized the role of media in international conflict. Rather, its pacifying potential has been alluded to in relatively few studies. Studies by Babst (1964: 10, 1972: 55), now regarded as classics, establish that freedom of speech in general and of the press in particular are crucial characteristics of freely elected governments in independent states. Small and Singer’s (1976: 51) study of regime types and war-proneness acknowledges the benign impact of media by citing East and Gregg’s (1967) finding that “states with more authoritarian regimes—as reflected in their practice of press censorship— appeared to exhibit more foreign-conflict behavior than ‘freer’ nations.” In the context of presenting his composite index of political democracy, Bollen (1980) considers press freedom as one of three key indicators that represents civil liberties in liberal democracy. Seeing transparency of information flow as a key component of the democratic nature of a polity, Starr (1992, 1997) argues that freedom of speech and the press are essential elements that inhibit liberal democracies from fighting each other. Data analysis by Van Belle (1997) and Van Belle and Oneal (2000) provides evidence in favor of the idea that there is a causal mechanism between press freedom and the avoidance of international conflict.2 Taken together, these studies suggest that enhancing media openness could be a neglected policy goal in the quest for a more peaceful world. Despite the potential importance of media openness, research on the democratic or neo-Kantian peace generally has paid closer and more sustained attention to other structural and cultural

Media Openness 131

elements of democracy (Starr 1992, 1997; see also Gurr et al. 1989, 1991; Gleditsch and Ward 1997). The centrality of the Polity data sets has encouraged democratic peace studies to measure democracy along some dimensions while not incorporating others. Put differently, while Polity clearly is the best data set on national attributes, some aspects of politics may have been downplayed as a result of its common use in research on the democratic peace. The Polity data sets include five important attributes of democracy— competitiveness of participation, regulation of participation, competitiveness of executive recruitment, openness of executive recruitment, and constraints on the executive—but do not take into account aspects of media openness. In this regard, Starr (1992: 52) provides a useful insight: “while constraints are important, they do not give us the whole story. There is still something about being a democracy that affects the war relationship between states.” In other words, students of the democratic peace do not yet have a full and firm grasp of what exactly it is about democracies that keeps them from engaging in serious armed conflicts with each other. More importantly, among the five democratic attributes, the degree of constraint on the chief executive, at the aggregate level, emerges as the main determinant of the standard “democracy,” “autocracy,” and “democracy minus autocracy” scores (Gleditsch and Ward 1997). As a result, “using the data summarily to classify modern polities as democracies directs attention away from the actual data that have been collected on authority patterns” (Gleditsch and Ward 1997: 381). In their recent and excellent review, Munck and Verkuilen (2002: 26) also observe that “not only is virtually no theoretical justification [offered for generating the democracy, autocracy and combined polity scores], but it also is open to criticism due to the index’s problems of conceptual logic.” While other efforts have been made to measure democracy, and it is understood that “the careful development of measures constitutes the

132 Civil–Military Dynamics

foundation for efforts at drawing causal inferences and is a critical task in itself” (Munch and Verkuilen 2002: 31),3 development of a compelling index for democracy remains a great challenge. One aspect of that challenge is especially salient for this project: Do all of the plausible attributes of democracy contribute equally to decreasing the likelihood of international conflict, crisis, and war? The answer implicit within the democratic peace studies would seem to be “yes,” with an emphasis on synergy among these attributes. Perhaps, however, some elements of democracy are more important than others and it is worth trying to pinpoint a main, driving force. To paraphrase, what is the most defining attribute of democracy that stands out in reducing the dyadic tendency toward war? Based on the Polity data sets, democratic peace studies have emphasized various structural and cultural factors, especially the degree of constraint imposed on the chief executive, in identifying the main determinants of democracy (Morgan and Campbell 1991; Gleditsch and Ward 1997). Although we are not necessarily in dispute with the conventional approaches, another potentially important aspect of democracy, media openness, is explored in this chapter. Conceptualizing and operationalizing democracy only on the basis of structural and cultural factors that do not include media openness, especially in relation to the degree of constraint on the chief executive, may be likened to shooting a bull’s-eye on the wrong target. As the experience of the French Revolution shows, a valid and complete treatment of democracy should focus more on the liberties of ordinary people, from the right to vote all the way through to civil rights. As Lincoln said, democracy is “the government of the people, by the people and for the people.” More is involved here than the mechanics of voting. Although the two issues are not unconnected, ensuring citizens’ rights would seem like a more central concern with respect to placing inhibitions on interstate disputes than simply constraining the chief executive’s authority

Media Openness 133

in democracies. A discussion about how to define and measure democracy, which includes these and other components, will provide theoretical reasoning in favor of an emphasis on civil liberties, especially media openness. Although the meaning of democracy can vary in terms of kinds, levels, and degrees, two basic categories stand out: (a) electoral (or minimalist) democracy and (b) liberal (or procedural) democracy (Huntington 1991; Diamond 1996, 1999; Collier and Levitsky 1997).4 Electoral democracy embraces Dahl’s (1971) concept of polyarchy and requires a minimal level of political rights so that political competition and participation are guaranteed. These traits, in turn, allow the opposition to have some realistic chance of winning an election and taking office. Thus, electoral democracy takes the form of regular, free and fair elections to some kind of national legislature (Diamond 1996). Liberal democracy goes further. It guarantees not only the political rights of electoral competition and participation but also civil liberties. It does not allow privileged social classes, perhaps based on membership in the military establishment or some other group, to determine political outcomes. Instead, liberal democracy entails a self-constrained executive power through constitutional checks and balances. It thereby ensures individual freedom and pluralism. Students of comparative politics argue that a liberal (procedural) definition of democracy is superior to an electoral (minimal) democracy (Huntington 1991; Diamond 1996, 1999; Collier and Levitsky 1997). Civil liberties, in sum, are crucial in defining liberal democracy. Civil liberties include not only the right to vote and contest for office, but also to speak and publish dissenting views, form and join organizations and obtain access to alternative sources of information. Diamond (1999: 8) emphatically stresses that, without civil liberties, no kind of democracy can be truly meaningful. Media openness—the modern meaning, in practice, of freedom of expression—is one of the most important elements of civil liberties. Moreover, media openness emerges

134 Civil–Military Dynamics

as the crucial characteristic of liberal democracy that can contribute to dampening the likelihood of international conflict. In his recent review, Ray (1998: 32) acknowledges the superiority of liberal democracy and observes that even “advocates of the democratic peace proposition have an admittedly procedural concept of democracy, focusing on . . . civil rights, freedom of the press, etc.” According to Diamond (1999: 9), the Polity data set “acknowledges civil liberties as a major component of democracy but, because of the paucity of data, does not incorporate them.”5 Studies of international conflict need to include aspects of liberal democracy beyond the electoral or minimal level. Thus, our study introduces media openness as one of the most important features of liberal (i.e., procedural) democracy. How does media openness constrain the likelihood of international conflict? Two theoretical linkages exist between media and interstate conflict: (a) reporting and (b) framing. The primary function of mass media is reporting news. In doing so, they facilitate the free flow of information between news contributors and news readers. As gatekeepers, the media allow the sources to participate in foreign policy debates at the domestic and international levels, which also leads to communication across international borders: “relations of different centers of state power are represented simultaneously to each other and to mass audience—both within their societies and worldwide— through media” (Shaw 2000: 30). Thus, we live in the era of transnational mass media. Moreover, this is true in a limited way even for the beginning of the era under consideration, with print media and radio gradually giving way to television and the Internet as the main conduits for information. Students of media politics define the primary reporting function as source indexing. Reporters interview key policymakers from the White House, State Department, Defense Department, Capitol Hill, and the like, and then convey news stories that summarize what they have been told. In addition,

Media Openness 135

the media’s so-called “power indexing behavior” reveals the views of authoritative sources who are most likely to project and influence the outcome of particular foreign policy issues (Kennamer 1992; Bennett 1994; Nacos et al. 2000a; Zaller and Chiu 2000; Everts 2002). The source or power indexing provides each state with an essential supply of information and even a good deal of intelligence gathering in the first place, since it serves as a driving force behind the media’s increasingly transnational character. In other words, “states survey each other” (Shaw 2000: 30). When international conflict arises, foreign policy decisionmakers, such as presidents or prime ministers, often turn to major newspapers or television as quickly as they can. In particular, when a variety of mass media is regarded as reliable and trustworthy sources or as representing political transparency, decision-makers are more willing to use these sources for primary intelligence information. Along those lines, La Balme’s (2001) and Isernia’s (2001) case studies of international conflict conclude that French and Italian decision-makers relied heavily on mass media for information (Nacos et al. 2000b: 5). Thus, whether the media are sufficiently open and reliable may contribute to the free flow of information on foreign policy activities across borders and, in turn, enhance each government’s credibility and validity. In the end, a high degree of media openness tends to increase confidence and prospects for peaceful settlements between states. The very openness of media may impact upon how intentions are played out. Even if intentions on one side or the other might be hostile, a limited ability to act in secret could help to discourage escalation due to a natural desire to avoid being labeled as the aggressor, which in turn could stimulate efforts by third parties to engage in balancing. Put differently, source or power indexing not only encourages the international free flow of information, but also facilitates the development of trust and even reduces uncertainty across borders. As a result, states are less willing to go to war.6

136 Civil–Military Dynamics

Information flows also facilitate political communication across borders. It has been argued that foreign policy decisionmakers often try to promote their future policies through mass media. Such framing or political indexing occurs when the media report news that decision-makers want to have heard (Entman 2000; Zaller and Chiu 2000). So policy-makers frequently do news management in response to international crises. The journalist Marvin Kalb defines this as “press politics, reflecting the inseparability of foreign policy from its management in the news” (Bennett 1994: 171). In particular, through use of the media, presidents or prime ministers who realize the potentially high cost of military action send political signals to each other in the quest for political solutions (and possibly to achieve other objectives as well). Thus, mass media, in effect, serve as political tools or messengers for these leaders. Ironically, this seemingly manipulative behavior can help to build trust because a potential adversary will have reason to believe that the government is trying to communicate in a public and verifiable way—this linkage, of course, is expected to be much stronger when bidirectional. As with reporting, the framing function of the media is likely to decrease international conflict in that it may ameliorate or unblock communications across borders. Given that one of the most important means of conflict resolution is to facilitate or influence political communications that lessen uncertainty, misunderstanding, and misperception, media framing can lead to a decrease in international strife. In short, our theoretical reasoning entails a special focus on information flow and communications between foreign policy decision-makers, such as presidents or prime ministers, as information senders or receivers via the media, across borders. When reporting conveys intelligence information to decisionmakers, and framing initiates communications across borders, that enters into the foreign policy leadership’s decision-making equation. Thus, we expect that the preceding factors at the international level are likely to produce a diminished level of conflict. In a word, if a higher degree of media openness

Media Openness 137

accompanies interactions between the decision-makers of states, the result could be a more peaceful world.

Testing the Role of Media Openness The hypothesis and operationalization for media openness stay the same as those introduced in chapter 4. Dyads composed of states with (without) free or imperfectly free media openness are less (more) likely to become involved in MIDs. Utilizing Van Belle’s (2000: 137–148) global press freedom data collection to measure media openness for each state, the media openness variable is dichotomous and coded as before: “1” if both states in each dyad-year have free or imperfectly free news media capable of functioning as an area of political competition or debate (i.e., categories 1 and 2); it is “0” otherwise (i.e., the news media are either restricted or controlled by the government, or no news media, categories 0, 3, and 4). As in chapter 4, we also choose to test the media openness hypothesis by replicating Oneal and Russett’s (1999c) research design. Oneal and Russett’s three neo-Kantian peace variables are democracy, economic interdependence, and joint membership in international organizations. Each of them is anticipated to decrease the likelihood of MIDs. The other five control variables in the model, as per the norm, are (a) national capability ratio (i.e., to control for power preponderance); (b) whether the members of each dyad are allied; (c) noncontiguity; (d) geographic distance; and (e) whether each member of the dyad is a minor power. These five variables are expected to decrease the likelihood of MIDs. With a special emphasis on media openness, this chapter purports to test the nine preceding hypotheses about MIDs during the period from 1950 to 1992 at the dyadic level.7 Our dependent variable is MID involvement and we use the GEE logistic regression. For the GEE logistic regression model, all independent variables are lagged by one year, so they are not

138 Civil–Military Dynamics

affected by a dispute to be explained. The extended model in Equation 1 is a combination of the media openness variable and the eight variables from Oneal and Russett’s (1999c: 21) democratic peace model: Yt  ␣  ␤1X1t1  ␤2X2t1  ␤3X3t1  ␤4X4t1  ␤5X5t1  ␤6X6t1  ␤7X7t1  ␤8X8t1  ␤9X9t1  ␧

(1)

Here, Yt: MID involvement X1t1: media openness X2t1: democracy X3t1: economic interdependence X4t1: joint membership in international organizations X5t1: national capability ratio X6t1: allied states X7t1: noncontiguous states X8t1: geographic distance X9t1: only minor powers ␧: error term. This research design is distinguished from Oneal and Russett’s (1999c) and many other treatments by the presence of X1t1: media openness. In doing so, this chapter attempts to compare quite directly the performance of our media openness variable and Oneal and Russett’s democracy variable in the extended model.8 For the sake of theoretical and empirical differentiation, hereafter we use the term media openness to refer to our single indicator of democracy centering on media openness, and institutional democracy to refer to Oneal and Russett’s aggregate indicator of democracy based on the five institutional attributes from the Polity data set. We will report the effect of media openness on fatal MIDs as well. Media influence on military fatalities is theorized as a two-step flow of communication, first from the public to the

Media Openness 139

mass media or vice versa, and then to foreign policy decisionmakers such as presidents and prime ministers (see RisseKappen 1991). We expect that a high degree of media openness is likely to decrease military fatalities. The so-called “casualty hypothesis” or “body-bag syndrome” holds that, with the possibility of military fatalities, the public will recoil or withdraw its support. This connection “is often mentioned by politicians and in the media as if it were an evident and established fact of life” (Everts 2001: 18). In the same vein, Luttwak (1996: 36) argues that “the prospect of high casualties, which can rapidly undermine domestic support for any military operation, is the key political constraint when decisions must be made on which forces to deploy in a crisis, and at what levels.” For example, the decision of the NATO allies to rely on air power alone in the recent conflict in Kosovo (1999) illustrates the importance of media and public opinion about military fatalities (Everts 2001: 1; see also Everts 2002: 158–181). Oneal et al. (2003) suggest that, since democracies are more sensitive to public opinion than nondemocracies, they are less likely to engage in disputes that become lethal. In a sense, Oneal et al.’s statistical results, which show that democratic dyads are less disposed toward fatal MIDs, would seem to justify a continuing focus on democratization as the potential solution to international disharmonies and disputes.9 Since Oneal and Russett (1999c) rely on MID involvement rather than fatal MIDs as their dependent variable, we use Maoz’s dyadic MID data set version 1.1 (updated January 2001) to identify the fatality level for each dyad in a given year.10 “Fatal MIDs” is a dichotomous dependent variable, that is, “1” (with at least one solider killed) versus “0” (no soldiers killed). This practice follows Oneal et al. (2003): The causal mechanism for MIDs is also applied to fatal MIDs in this chapter. In sum, with a special emphasis on media openness, this chapter purports to test the preceding model of both MID involvement and fatal MIDs during the period from 1950 to 1992 at the dyadic level.

140

Empirical Implications: The Power of Media Openness Table 8.1 shows the empirical results based on our extended logistic regression model from Equation 1.11 We employ a onetailed test for each variable since the hypotheses are directional.12 Table 8.1 is composed of nine columns. Column 1 lists the nine variables. Columns 2–5 report the results when the dependent variable is MID involvement and columns 6–9 when it is fatal MIDs. The shaded columns show the empirical results for Equation 1 during the period from 1950 to 1992 and the unshaded columns (except for the first one) are replications of Oneal and Russett’s (1999c: 22) neo-Kantian peace model. The replicated results (except for the economic interdependence variable13) in the unshaded columns 2 and 4, in general, concur with the reported results of Oneal and Russett with respect to direction of signs and statistical significance. The replicated results in the unshaded columns 6 and 8, however, are at some distance from our expectation. When the dependent variable is fatal MIDs, all of the three neo-Kantian variables, surprisingly enough, turn out to be statistically insignificant. For present purposes, the discussion that follows will focus mainly on the impact of media openness and the three neo-Kantian peace variables. The third, shaded column presents the empirical results for all dyads when the dependent variable is MID involvement. As hypothesized, the coefficient for the media openness variable turns out to be statistically significant at the 0.001 level, indicating that dyads composed of states with free or imperfectly free media openness are less likely to become involved in MIDs. Surprisingly, none of the three neo-Kantian peace variables is statistically significant. It appears that, once the media openness variable is incorporated, the three neo-Kantian peace variables (including institutional democracy [measured as a composite indicator of the five institutional attributes of the Polity data set]) lose their effect on the likelihood of MID

Table 8.1. Media Openness or Institutional Democracy, 1950–1992? Variable

MID involvement All dyads

Politically relevant dyads

Fatal MIDs All dyads

Politically relevant dyads

* p  .05; ** p  .01; *** p  .001, one-tailed tests. ^ p  .05; ^^ p  .01, one-tailed tests but wrong sign.

141

Media openness 1.5579*** 1.1794*** 2.7337*** 2.5898*** (AMOAB) 0.3271 0.3163 0.5216 0.4851 Institutional 0.0633*** 0.0153 0.0439** 0.0075 0.0110 0.0550^ 0.0033 0.0682^^ democracy 0.0163 0.0166 0.0156 0.0169 0.0317 0.0266 0.0299 0.0258 (DEML) Trade/GDP 46.7964 31.5443 32.8056 26.9002 263.3825 207.1352 252.2876 211.4247 28.5656 22.5118 22.8697 19.7082 163.8017 145.7483 157.4701 139.0917 (DEPENDL) International 0.0052 0.0017 0.0082 0.0056 0.0064 0.0072 0.0068 0.0073 organizations (IGO) 0.0059 0.0060 0.0059 0.0060 0.0182 0.0159 0.0165 0.0147 Capability ratio 0.1544* 0.1526* 0.2180** 0.2198** 0.0984 0.0869 0.1809 0.1934* 0.0816 0.1058 0.1050 0.1136 0.1094 (CAPRATIO) 0.0770 0.0759 0.0833 Alliances 0.4496* 0.4066* 0.4732* 0.4460* 0.3500 0.2387 0.6697 0.5873 0.2376 0.6205 0.5923 0.5757 0.5516 (ALLIANCES) 0.2455 0.2389 0.2432 Noncontiguity 2.4255*** 2.4481*** 1.3110*** 1.4173*** 3.3832*** 3.3933*** 1.6954*** 1.8363*** 0.3125 0.4861 0.4732 0.4810 0.4712 (NONCONTIG) 0.2703 0.2641 0.3178 Log distance 0.5472*** 0.5224*** 0.2583** 0.2374* 0.4444*** 0.4076*** 0.2459* 0.1842 0.1051 0.1285 0.1304 0.1311 0.1375 (DISTANCE) 0.0962 0.0957 0.1016 Only minor powers 1.8332*** 1.9275*** 0.4454 0.5875 0.9009* 1.0669* 0.3055 0.1154 0.3648 0.5415 0.4926 0.6196 0.5776 (MINORPWRS) 0.2831 0.2689 0.3776 Constant 1.5205* 1.4888* 1.5073* 1.5829* 4.5574*** 4.4845*** 3.5339** 3.6940*** 0.8012 0.7874 0.8541 0.8634 1.1031 1.1069 1.1724 1.2091 882.63 869.66 100.57 101.84 524.75 514.16 91.55 96.63 Chi2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 P of Chi2 N 109,710 109,710 20,666 20,666 102,228 102,228 19,594 19,594

142 Civil–Military Dynamics

involvement. In other words, media openness seems to represent an essential element within neo-Kantianism, namely, the central role of free and open communication in preserving peace.14 The fifth, shaded column presents the empirical results for politically relevant dyads only, which are similar to those for all dyads in column 3 (except for the only minor powers variable). Once again, the media openness hypothesis is supported. It appears that media also matter in the case of politically relevant dyads, which students of conflict processes look into most commonly due to their relatively dispute-prone nature. None of the three neo-Kantian peace variables is statistically significant. These findings create new and potentially important theoretical implications for the democratic peace proposition. The results suggest that a single indicator of democracy focusing on media openness directly captures effects on the actual process of foreign policy decision-making. Given the fact that the leader of a state initiates (or makes a decision about involvement in) war, a cause and effect relationship should be established on that basis. In this sense, as discussed earlier, media openness through reporting and framing is factored directly into the decision-makers’ choice of peace versus war. In other words, while institutional democracy, economic interdependence, and joint membership in international organizations could be expected to have some effects on institutional constraints and norms faced by decision-makers, mass media directly influence their behavior. Thus, the results imply that, to obtain a more peaceful world, we need to promote a higher degree of media openness—or, put differently, the free exchange of ideas emerges as the core of neo-Kantianism. The seventh, shaded column presents empirical results for all dyads when the dependent variable is fatal MIDs.15 As expected, the coefficient of the media openness variable turns out to be statistically significant at the 0.001 level. Thus, with greater media openness, fatal MIDs become less likely. Given the assumption that the media effectively work together with

Media Openness 143

the public and exert pressure on policy-makers regarding the most salient issue, military fatalities, our findings appear very plausible. Surprisingly, the coefficient for institutional democracy is statistically significant at the 0.05 level, but in the wrong direction. It seems that states in a democratic dyad (measured in the Polity) are more likely to experience fatal MIDs. In addition, both economic interdependence and joint membership in international organizations are not statistically significant. The ninth, shaded column presents empirical results for politically relevant dyads only. The media openness hypothesis is supported. By contrast, the institutional democracy hypothesis must confront counterintuitive results, and the other two neo-Kantian variables, interdependence and international organizations, are not statistically significant. It is possible that, with large samples, even a small effect can be statistically significant. Thus, it becomes increasingly important to estimate the substantive effects of variables as the sample size increases. Table 8.2 shows the substantive significance of the four theoretically interesting variables in the extended model: (a) media openness, (b) institutional democracy, (c) economic interdependence, and (d) joint membership in international organizations. It is apparent that, while the three Table 8.2. Substantive Effects of MID Involvement and Fatal MIDs, 1950–1992* Variable

MID involvement (%) All dyads

Media openness equals 1 (AMOAB) Institutional democracy increased by 1 standard deviation (DEML) Economic interdependence increased by 1 standard deviation (DEPENDL) International organizations increased by 1 standard deviation (IGO)

PRDs**

Fatal MIDs (%) All dyads

PRDs*

79 14

68 7

93 76

93 91

9

5

45

31

3

12

12

14

* The baseline values are as follows: 0 for no media openness, mean for continuous variables, 1 for contiguity, 0 for non-alliance, and 0 for only minor powers. ** Politically relevant dyads.

144 Civil–Military Dynamics

neo-Kantian variables reduce the likelihood of conflict one way or another (except for the international organizations variable for MID involvement and the institutional democracy variable for fatal MIDs), media openness dramatically decreases it, regardless of different model specifications and samplings. As compared with a typical dyad, the risk that open media dyads will become involved in a dispute is decreased by 79 and 68 percent. The likelihood that the open media dyads will face fatal MIDs is decreased by 93 and 93 percent. It should be noted that the substantive significance for fatal MIDs is higher than that for MID involvement. This implies that the body-bag syndrome does exist. With an increase of one standard deviation for institutional democracy, the likelihood that a dyad will become involved in a dispute is decreased only by 14 and 7 percent. Given the same magnitude of change, the risk that a dyad will experience fatal MIDs is increased (not decreased) by 76 and 91 percent. In short, the empirical results in table 8.2 confirm the importance of media openness in a neo-Kantian world.16 In sum, we have explored a potentially significant aspect of democratic governance, media openness, with respect to decreasing the likelihood of international conflict. Cross-sectional, time-series data analysis shows that media openness is the most defining element of democracy with respect to MID involvement and fatal MIDs during the period from 1950 to 1992. Media openness clearly complements the three neo-Kantian peace factors—institutional democracy (measured by five institutional attributes from the Polity data set), economic interdependence, and joint membership in international organizations—in the quest for a more peaceful world, at least as understood in terms of interstate dyads. Based on these findings, we speculate that, while the three neo-Kantian peace variables produce a noteworthy impact, media—by reporting and framing—provide a concrete and direct effect via foreign policy decision-makers such as presidents or prime ministers who decide on war or peace.

Media Openness 145

Our findings complement neo-Kantianism and sharpen the democratic peace debates by suggesting that media openness, rather than an aggregate indicator of other institutional, cultural, and economic attributes of a regime, has the most pacifying effect on international conflict. In a word, it turns out that the pen (i.e., media openness) is more important than many other things (e.g., institutional democracy) in restraining the sword within democratic dyads.

This page intentionally left blank

Chapter 9

The Quest for Peace Findings Although world politics in the twenty-first century appears at some distance from the Hobbesian world of “every state against every state” (Hobbes 1991, edited by Tuck), significant parts of it remain outside of the advanced, developed world’s zone of mutual peace. Chaos and violence in the developing world continue to be depressingly familiar features of everyday life. Globalization, empowerment of individuals, and rapid change within and beyond the states system add up to an increasingly turbulent world (Rosenau 1990). The overall direction of change continues to be difficult to anticipate, with so many factors coming into play. Fortunately, academics and those in policy circles have not been idle in their efforts to obtain ideas for coping with this challenging world. In particular, students of the democratic peace have made a significant contribution to the quest for a better world during the past decades. Based primarily on Kant’s Perpetual Peace (1795), the main implication of their argument is that the presence of more democracies is the most promising path to a world with less turbulence and more cooperation (Oneal and Russett 1999c; Russett and Oneal 2001). However, as Rosato (2003) points out, current scholarship still falls short of a complete answer to the central question: What are the key determinants of a lasting world peace? Built on both a theoretical model and state-of-the-art statistical techniques, this study has attempted to complement the neo-Kantian revolution by delving into potentially important aspects of political systems in the ongoing quest for a more

148 Civil–Military Dynamics

peaceful world. The foreign policy decision-making process model introduced in chapter 3 provides a theoretical foundation for the further study of international conflict, crisis, and war, and effectively complements the neo-Kantian peace vision of the world. A range of empirical methods and aggregation of data are employed in the preceding chapters of this book to test implications from the decision-making process model. In particular, utilizing the most sophisticated statistical techniques available, logistic regression, and neural networks (i.e., artificial intelligence), analysis of interstate dyads produces compelling results. More precisely, empirical findings from both logit and neural network models during the period from 1886 to 1992, which are based on the interactive foreign policy decisionmaking process model with a special emphasis on the four new factors—civil versus military influence, military manpower system, diplomatic channels, and media openness—reveal the last of these as a key element in accounting for MIDs. Thus, media openness, in particular, should be incorporated into Russett and Oneal’s (2001) triangulation of the neo-Kantian peace. The other new findings, potentially important but not as statistically compelling as those related to media openness, can be summed up as follows: (a) the more powerful the military leaders are in civil–military relations, the more likely the dyads are to become involved in MIDs; (b) under a conscription system, dyads are more likely to engage in MIDs; and (c) an increase in diplomatic channels for a dyad may increase involvement in MIDs. Empirical results establish that civil–military dynamics matter in ameliorating interstate disputes. It appears that maintaining civilian control over the military, measured in the ebb and flow of expenditure, may help to discourage involvement in MIDs. The key issue of a measurement for civil– military relations based on military expenditure is addressed in chapter 7, where empirical results show that the operationalization is valid, at least in an applied sense—and perhaps even

The Quest for Peace 149

beyond a “reasonable doubt.” When measurement takes account of international security threats as an exogenous variable that might increase military expenditure, validity is maintained. The military manpower system, furthermore, tends to be critical for foreign policy decision-making at the brink of serious interstate conflict. It seems that (subject, to some degree, to the choice of statistical model) conscription is associated with involvement in MIDs.1 Empirical results also reveal that political communications, in one way or another, can be crucial factors in accounting for world politics. Diplomatic channels produce a significant but unanticipated result—a greater likelihood of MIDs.2 This result seems to reflect the realist perspective: Under anarchy, states may pursue their national interests with broken pro-mises and cheap talk. Thus, the results seem to call for a more trustworthy diplomacy in the quest for peace. Media openness, by contrast, reveals normatively desirable effects as related to MIDs. It appears that, rather than institutional or perhaps other cultural attributes, media openness is the most compelling aspect of democraticness to consider in relation to a more peaceful world, at least for interstate dyads during the period from 1950 to 1974 for which a complete model so far can be evaluated. Moreover, this connection outweighs the contribution of democraticness as measured in the Polity data. These results, it should be noted, stand regardless of different statistical techniques. Chapter 8 presents more systematic probe into the role of media openness as a follow-up to its impressive performance throughout the preceding stages of data analysis. Theoretical linkages and empirical results for the extended period from 1950 to 1992 confirm the previous logic and findings: In sum, media openness matters more than institutional democracy in pacifying interstate disputes. This finding does not contradict the neo-Kantian peace, but it does suggest that, rather than other structural, economical, and institutional aspects of democracy, media openness should be brought to the fore in a new quest for international peace.

150 Civil–Military Dynamics

This study, in spite of its emphasis on civil–military relations, has not addressed an important question: How to know explicitly when the military side is (or is becoming) dominant in decision-making, especially in a democracy. Democracies, by definition, might be better in maintaining civilian control over the military than nondemocracies (such as military regimes). An interaction term between civil–military relations and “dichotomous” democraticness to detect civilian supremacy in democracies might be considered for further statistical models.3 Other possible measures of relative civil–military influence, such as statutory factors (as identified in footnote 8 in chapter 4), should be explored. Since conscripted soldiers may not need to be deployed during peacetime, the latter also can lead to nonconscription. In future aggregate analysis, simultaneous equations should be utilized to assess mutual causality between conscription and MIDs (or peace). The counterintuitive results with respect to diplomatic channels also demand more exhaustive research. Although we discuss this issue in relation to Bush Administration diplomacy surrounding the North Korean nuclear crisis in chapter 5, further detailed case studies should follow to “zero in” on causal mechanisms. Given the fact that media openness has emerged as a main determinant of ameliorating conflict over such a short study period, its positive impact should be explored within a longer time frame for the purpose of generalization. It would also be interesting to test the foreign policy decision-making process model using other data, such as that offered by the International Crisis Behavior (ICB) Project data collection (Brecher and Wilkenfeld 1997). Such testing would serve either to increase confidence in, or justify concerns about, the more controversial findings in this book.

Policy Implications Important policy implications may be derived from the empirical findings. Most notably, democratization centering

The Quest for Peace 151

on institutional aspects such as executive constraints, especially in isolation from other kinds of changes, may not be the best public (or foreign) policy for the United States to emphasize as a system leader for the foreseeable future. Instead, the United States should redirect its public (or foreign) policy toward internal civil–military dynamics, such as civil–military relations and military manpower systems, and political communications, like diplomatic channels and media openness, in the quest for a more peaceful world. As civilian influence increases, MIDs become less likely. Adopting a voluntary manpower system may also contribute to lessening MID involvement. As both logit and neural network analyses demonstrate in chapters 5 and 6, a greater number of diplomatic channels seems to increase involvement in MIDs, which demands a more honorable diplomacy. (Or, perhaps instead, the result says more about the centrality of great powers in international conflict than anything else.) Increasing media openness can facilitate political communication and signaling and also decrease mistrust and misperception between states in conflict. Thus, the role of the four factors introduced in this book should be emphasized in the further quest for knowledge about how to achieve a more peaceful world. In sum, civil–military dynamics and political communications have been neglected for too long among students of international conflict, crisis, and war—they can and should be combined together in the further quest for knowledge about peace. Transforming regimes—a straightforward goal that some might derive from the democratic peace—is a challenging task.4 Making changes in civil–military relations, redirecting military manpower systems, cultivating more honorable diplomacy, and utilizing media openness, by comparison, look like more feasible options. Put differently, overturning nondemocratic regimes by either overt or covert force is hard to carry out, but promoting those four factors is relatively easy to do, at least by comparison. While providing nondemocratic regimes with military and economic aid, the United States also might concentrate on encouraging these states to restructure

152 Civil–Military Dynamics

civil–military dynamics and engage in more sustained political communications, which ultimately would enhance prospects for peace at the dyadic level and possibly beyond. Conditions continue to be difficult in Iraq, recently the target of an invasion by the United States, at this time of writing. Consider Landy’s (2004: 60) insightful summary of the situation, written at some point within the last year, but equally relevant to the situation now, a month before the scheduled transition in sovereignty from the United States to the people of Iraq: The war was surprisingly easy, but the peace has proven aggravating and deadly. The American public has been troubled both by the steady stream of American deaths and the sullen ingratitude shown by many segments of the Iraqi populace. This disappointment has been fanned by the media, both national and foreign, which set absurdly high expectations for what would happen in Iraq after the military victory was won, as if this was the first war which ever ended with widespread looting or that a people deprived of political freedom for more than a generation would immediately behave like model citizens.

This summary of Iraq after the fall of Saddam is useful in bringing out themes connected to media openness, which has emerged as the most empirically powerful component of the model of foreign policy decision-making. To begin, evidence is strong that, aside from the small number of beneficiaries from Saddam Hussein’s nightmarish dictatorship, Iraqis are better off in the new era precisely because they hope for something better. So, why then is there so much resentment, if not open resistance, with regard to the relatively short-term presence of the United States? The answer, at least in part, may rest with media experiences. Iraqi citizens lived for decades under poisonous propaganda from the Saddam Hussein regime and continue to receive media coverage that is generally discouraging to U.S. efforts and, perhaps most importantly,

The Quest for Peace 153

plays on prejudice against Israel. Landy refers to Iraqis as something other than “model citizens”; the analysis in our book suggests that neither model citizens individually nor model states collectively can be expected in the absence of open and clear communication, which includes accurate reporting of events as opposed to deliberately distorted accounts through self-serving, state-controlled outlets. Perhaps the most uncomfortable implications from the data analysis in this book come from the ability to anticipate extreme difficulty ensuing from situations in which regime change is pursued in a revolutionary and violent manner through military force as opposed to an evolutionary approach that takes advantage of improving technology. With time, even Saddam Hussein’s iron grip on his state might have been weakened through an infusion of communications through the Internet and other means that are increasingly difficult for dictators to control. This line of reasoning is consistent with Rosenau’s (1990) emphasis upon a micro-revolution through worldwide empowerment of individuals. If such things could happen in the German Democratic Republic—an imposed entity that was none of the three things mentioned in its name—perhaps the potential exists elsewhere as well. By definition, we can never know now what changes might have occurred in Iraq in the absence of an invasion. We should be careful, of course, not to push our results too far. The pacifying power of media openness in interstate dyads does not prove that the U.S. invasion of Iraq in 2003 was the wrong thing to do. Instead, the findings in this book suggest that, on balance, with more widespread improvements in communication technology, the world’s prospects for seeing an end to dictatorship have never been brighter than they are right now. Landy’s summary also alerts us to the fact that the impact of media is not always favorable. On a worldwide rather than dyadic basis, the impact of media on the development of Iraq right now, in a normative sense, does not seem positive.

154 Civil–Military Dynamics

The intense desire for breaking news in a war zone breeds competition for stories and a sensationalist tendency to emphasize the worst of everything. The hard work of stabilizing Iraq and building a decent life for its people is not facilitated in the tabloid environment that seems in place at this time of writing. Such unfortunate side effects, of course, should not be blamed on the media but instead put at the doorstep of the real culprit— war itself.

Some Final Thoughts To conclude, both scholars and foreign policy decision-makers need to be reminded that, two years after Perpetual Peace (1795), Kant also pointed out in The Metaphysics of Morals (1797) that “his league of peace would grow to the point where it was ‘ungovernable’. . . . would return to the original state of war, rational institutions notwithstanding” (quoted in Owen 2001: 968). If the prediction is believable, it is imperative to identify factors that can either prevent the demise of the would-be perpetual peace or at least serve as knowledge that complements Russett and Oneal’s (2001) triangulation of the peace. The four factors introduced in this book—civil–military relations, military manpower system, diplomatic channels, and media openness, as represented in the foreign policy decision-making process model—are potential candidates for engineering or maintaining a more peaceful world in the millennium and beyond.

Appendix A decade ago, Sarkesian (1992) presented a conceptual model about a political and social system of civil–military relations. Based on an ideal, Weberian type of civilian control over the military in a liberal society, most notably the United States, his model consists of three key elements: (a) a political–social system; (b) interactions between civilian and military elites; and (c) military society. Figure A.1 shows relationships for the domestic political–social system adapted from Sarkesian’s system-oriented model. The diagram on the left-hand side, corresponding to the state as a whole, portrays Sakesian’s three elements of civil–military relations in regular font style. Using arrows (i.e., →), our interpretations are added in italics into his model; the four arrows that appear, along with the italicized components, do not imply any causal hierarchy. As shown in figure A.1, our extension of Sarkesian’s system of relationships is that (a) the political–social system can be reified as societal militarism; (b) interactions between civilian and military elites produce, as a resultant, a relative degree of military influence in civil–military relations; and (c) military society can be conceived of, at least to some degree, in terms of the military manpower system. State as a whole (1) Societal militarism

Political–social system Civilian elites

Military elites

International relations (threats)

Military society Cooperation or conflict (i.e., dispute)?

(2) Civil–military relations

(3) Military manpower system

Figure A.1. Relationships for the Domestic Political–Social System: Societal Militarism, Civil–Military Relations, and Military Manpower System Source: Adapted from Sam C. Sarkesian, “Civil–military Relations in a Liberal Society: The United States in a New Security Era,” Paper presented at the Ohio State University Mershon Center Conference on Civil–Military Relations, Columbus, OH, December 4–6, 1992; cited in Don M. Snider and Miranda A. Carlton-Carew (eds.), “The Current State of U.S. Civil–Military Relations: An Introduction,“ U.S. Civil–Military Relations: In Crisis or Transition? (Washington, DC: The Center for Strategic & International Studies, 1995).

156 Appendix Figure A.1 also includes an extension of Sarkesian’s model to the interstate level that embraces aspects of the international environment such as international tension and/or perceived security threat. It is inferred that, in the end, foreign policy in each state is affected significantly by the interaction of the three conceptual factors as well as international threats. This combination, in turn, will lead to either international cooperation or conflict in the international arena. The large arrow (i.e., ) between the left- and righthand boxes indicates the causal direction between the domestic attributes and foreign policy outcomes in the realm of international relations. In sum, the figure represents a projection of Sarkesian’s model into the international domain and provides the theoretical foundation for our model building.

Notes Chapter 1 Introduction 1. Recent democratic peace studies have extended to the neo-Kantian or liberal peace, which encompasses trade as well as international organizations. For the purposes of this study, the democratic peace includes the neo-Kantian and liberal peace as well. Examples from the most recent wave of democratic peace studies, which incorporate an increasing range of political and economic variables, include Dixon (1993, 1994, 1998), Russett (1993, 1998), Hermann and Kegley (1995, 2001), Kegley and Hermann (1996, 1997), Hewitt and Wilkenfeld (1996), Chan (1997), Elman (1997), Oneal and Ray (1997), Oneal and Russett (1997, 1999a, 1999b, 1999c, 2000, 2001), Ray (1997, 1998), Enterline (1998a, 1998b), Russett et al. (1998), Russett and Starr (2000), Russett and Oneal (2001), Prins (2002), Oneal et al. (2003), and Choi and James (2003). 2. It is fair to say that, in recent decades, analysis of pairs of states on an annual basis has emerged as the paradigmatic focal point for rigorous studies of war and peace, with MID-related variables at the center of research. A MID is defined as “a set of interactions between or among states involving threats to use military force, displays of military force, or actual uses of military force” (Gochman and Maoz 1984: 587). Although many disputes occur, which are not brought on by crises (i.e., there are low-level MIDs), the focus here is on disputes linked to crises. See also Jones et al. (1996) and Sarkees (2000) on concept formation and data as related to MIDs. 3. Maoz and Russett (1992: 247), e.g., adopt essentially the same reasoning as Ray (1998: 36) in a relatively early attempt to test the effects of five additional factors on international disputes—wealth, economic growth, contiguity, common alliance bonds, and political stability. They find that, except for political stability, the factors do not seem to be connected to the lack of conflict between democracies. The current project, which focuses on civil–military dynamics and political communications, should be seen as a logical continuation of that effort. 4. See, e.g., Farber and Gowa (1995, 1997), Gartzke (1997, 1998, 2000, 2001), Wolfson et al. (1998), Gowa (1999), James et al. (1999, 2000), Mousseau and Shi (1999), Cederman (2001), and Snyder (2002). 5. Prominent literature in this context includes Lasswell (1941, 1997), Huntington (1957), Janowitz (1960), Betts (1977), Snyder (1984),

158 Notes Van Evera (1984), Goertz and Diehl (1986), Diehl and Kingston (1987), White (1989), Ross (1994), Anderson et al. (1996), Desch (1999), Duindam (1999), Feaver (1999), Feaver and Gelpi (1999, 2004), Gibson and Snider (1999), Looney and Frederiksen (2000), Schofield (2000), and Hendrickson (2002). 6. Prominent literature in this context includes Singer and Small (1966), Small and Singer (1973), Graber (1986), Van Belle (1997, 2000), and Van Belle and Oneal (2000). 7. As compared with their significance in the real world of international relations, fatal MIDs so far have received relatively little attention in the research program on the democratic peace.

Chapter 2 A Review of Democracy, Peace, and Other Things 1. To mention just a few excellent reviews of the democratic peace, Morgan (1993), Chan (1997), Maoz (1998), Ray (1998), and Russett and Starr (2000) are representative and offer generally favorable judgments about its prospects as an ongoing program of research. 2. The definitive treatment of economic concepts for application in other social sciences appears in Sandler (2001). 3. It should be noted that, as per the norm in the field, Oneal and Russett’s democracy variable does not necessarily reflect the concept in its most encompassing form. Its main determinant, among the five institutional attributes as defined in the Polity data, is executive constraints. Oneal and Russett (1999c: 12, footnote 30) acknowledge this drawback of the Polity III data set when observing that their “democracy minus autocracy” score does not reflect “the relative importance of its components [which are] unstable over time.” Thus, questions related to both internal and external validity continue for democracy as the central concept within neo-Kantianism. 4. Ray’s article, it should be noted, appeared in the most prestigious journal on democracy, the Journal of Democracy.

Chapter 3 A New Look at International Conflict and Peace: Four Factors to Consider 1. One recent exception is Bueno de Mesquita et al. (2003). As noted in chapter 2, this model, however, does not address the specific concerns of the present study, such as civil–military dynamics and political communications, which will be introduced momentarily.

Notes 159

2. For scholarly analysis of the Cuban Missile Crisis, see Allison (1969, 1971), Kahan and Long (1972), Bendor and Hammond (1992), and McKeown (2001). 3. Gochman and Maoz (1984: 587), in a study focusing on MIDs in particular, acknowledge the importance of decision-making in an indirect way: “Since our primary concern is with the evolution of conflict as a product or consequence of government decision making, we require that actions taken or threatened be attributable to responsible state authorities.” (Emphasis added.) 4. This difference is expected to hold in a relative sense within each state. The assertion is that for a given State A, civilian leaders will be less inclined than military leaders toward the use of military options. Both sets of leaders in State A could be more (or less) hawkish than those in another State B. The point of comparison is between leaders within a given state. 5. See also Vagts (1956, 1959), Benjamin and Edinger (1971), and Whitten and Bienen (1996) for additional historical material on the balance between civil and military influences on foreign policy. 6. Compelling studies of civilian “hawks” and military “doves” in the American foreign policy decision-making process (e.g., U.S. joint chiefs of staff are quite reticent to resort to force, perhaps because of the “Vietnam” syndrome or other reasons) in recent years include Feaver and Gelpi (1999, 2004) and Hendrickson (2002). We are aware not only that civil–military relations in some advanced industrial countries (including the United States) might show an attenuated connection at present, but also that there is a gray area in drawing the line between military and civilian leaders. Since our main research interest is civil–military relations across states using cross-sectional, time-series data, we emphasize the overall tendency toward “hawkish” military leaders and “dovish” civilian leaders as collectivities. Put differently, while most contemporary works address a single case (e.g., American foreign policy), our study, in a sense, attempts to broaden the temporal and spatial domain of those studies by including all states. 7. Exceptional book-length expositions that identify such patterns and develop generalizations include Midlarsky (1975, 2000), Organski and Kugler (1980), Bueno de Mesquita (1981), Doran (1991), Vasquez (1993, 2000), Geller and Singer (1998), Powell (1999), Russett and Oneal (2001), Lemke (2002), and Bueno de Mesquita et al. (2003). 8. For an example of the dissenting view that emphasizes selection effects and differences in causation across types of conflict, see Reed (2002). 9. It should be noted that some MIDs may result from indirect choices by leaders. The recent “spy plane incident” between the United States and China, e.g., occurred when, apparently by accident, a Chinese fighter struck a U.S. aircraft.

160 Notes 10. As a reinforcement to the idea conveyed by the fable of the blind men and the elephant, consider this hypothetical situation: Suppose that you go to a dealer to buy a new car, and a salesman claims that, if you drive one of the new 2006 models, you will be less likely to get involved in a car accident. This raises suspicion about the salesman’s credibility because the chance of a car accident is likely to depend much more on your driving skills or habits rather than on the model of a car. The new model can at most enhance your safety, but the main cause of accidents can be attributed to careless or reckless driving. In this story, the new model of the car plays a role analogous to the regime type in contemporary peace research, while your driving skills would correspond to civil–military dynamics and political communications.

Chapter 4 Research Design 1. It should also be noted that, as Diamond (1999: 11) and Feaver and Gelpi (2004: 8) point out, civilian supremacy in civil–military relations is one of the most important attributes of liberal democracy. So it becomes a priority to delve into civil–military relations in the context of the foreign policy decision-making process. 2. It should be noted that, in this study, the presence of conscription refers to active duty soldiers who have already been conscripted and trained for combat missions before an international conflict begins. One might claim that states adopt conscription as a result of anticipating conflict in the future. So the reverse causal direction—that militarized disputes lead to conscription—is highly plausible. It can be hypothesized that peace could lead to no conscription and, in turn, more peace. Militarized disputes between two states might lead one of them to institute the draft in expectation that they would soon be at war, making the initial dispute or crisis the causal agent. Neither the United States nor Britain, however, switched from a voluntary system to conscription in either the 1991 Iraq or the 2003 wars. Other exceptions could be noted. In addition, some states prefer conscripts as a reflection of civic duty rather than military opportunism (see Anderson and Honegger 1982). Thus, we postulate H2 alone in this study since our main research interest is to look into international conflict with respect to the decision-making process model. In passing, Kant argued that dispute propensity is increased by a standing volunteer army that is trained, and primed for combat, as opposed to a “citizen army” that must be drafted, trained, and then sent into the field. Kant’s (1957: 5–6) standing volunteer army in all likelihood referred to a collectivity of “foreign” mercenary soldiers during the

Notes 161

3.

4. 5.

6.

medieval period, which is different from a contemporary national army made up of volunteers. For more details, see Choi and James (2003). It should be noted that, unlike Dougherty and Pfaltzgraff, Morgenthau’s (1985: 569–575) standard realist treatment downplays the importance of the diplomat’s role. He argued that an increase in summit meetings, development of advanced communication technology, and the like have contributed to the decline of diplomacy in the modern era. The present study is perhaps closer to the position associated with the democratic peace in its assertion that open channels of interstate communication can play a key role in conflict management and even resolution. Exceptions, such as the notorious case of the Japanese before Pearl Harbor, are so visible precisely because they are quite rare. Few students of international conflict, crisis, and war have systematically investigated the likely importance of media openness as a key attribute of democracy. Van Belle and Oneal’s (2000) study is a pioneering effort to introduce media openness into the ongoing literature devoted to the democratic peace at the dyadic level. See also East and Gregg (1967), Babst (1972), Small and Singer (1976), Bollen (1980, 1991, 1993), Mueller (1992), and Van Belle (1997). Ironically, the relative neglect of media openness, to a large extent, is associated with the evolution of the otherwise remarkably comprehensive Polity data sets that have become a widely used source of cross-national, time-series data on the authority characteristics of modern polities since the 1970s (see Gurr et al. 1989, 1991). The centrality of the Polity data sets has enabled democratic peace studies to measure democracy along some dimensions while not incorporating others. Put differently, while Polity clearly is the best data set on national attributes, some aspects of politics may have been downplayed as a result of its common use in research on the democratic peace. Gleditsch and Ward (1997: 381) argue that “using the [Polity] data summarily to classify modern polities as democracies directs attention away from the actual data that have been collected on authority patterns.” According to Diamond (1999: 9), the Polity data set “acknowledges [media openness] as a major component of democracy but, because of the paucity of data, does not incorporate [it].” In sum, in this study, we distinguish institutional elements of democracy (that are best captured by the Polity data sets) from liberal elements of democracy (that would be captured by press freedom data, to be introduced momentarily). It should be noted that Fearon’s (1994) and Smith’s (1998) exegesis of audience costs, Schultz’s (1998) analysis of political opposition, and Bueno de Mesquita and Siverson’s (1995) account of leadership survival shed light on a parallel institutional constraint at the domestic level. Although these studies do not deal with how the audience receives its

162 Notes information, it is implicit that media may allow opposition voices to be heard. This, in turn, should make a given leader at least a bit more cautious out of fear of political consequences; in particular, a free press makes it harder to bluff and not get caught. Citizens then serve as the sanctioning mechanism for bluffing. As Fearon (1994: 581) observes, “since the mid-nineteenth-century, mass publics informed by mass media in many cases” constrain foreign policy decision-makers’ behavior (emphasis added). Fearon’s formal model, however, does not specify how media works in practice. Rather, mass media are treated implicitly as an intervening variable between the public and the leaders. So the media may be a proxy for the leadership’s concerns about political costs from using military force (see also Mueller 1992; Owen 1997; Van Belle 2000). While we do not disagree with identification of this role for media as a domestic constraint, our focus is on other functions at the international level. In particular, communication and information exchanged between states should allow for fewer misperceptions and higher trust. We acknowledge that the international effects of media openness and domestic effects of the media as a standard “democratic constraint” (i.e., the media allows voters to monitor and hold accountable leaders) may be symbiotic. 7. Diehl and Kingston (1987) find no direct impact, however, for a state’s military spending on its propensity to initiate conflict. 8. There are at least four other viable ways of assessing civil–military relations or degrees of military versus civilian influence in each state: The first is to identify whether a state adopts a constitutional clause on civilian supremacy. A second means is to categorize each state’s regime type based on the leader’s military background. A third approach is to utilize a survey questionnaire on civil–military relations. The fourth and final way is based on case studies about who prevails when civilian and military leaders’ preferences diverge. Although these four methods have various merits, they are not as useful as the military expenditure measurement, which seeks to assess military leaders’ influence on foreign policy decision-making. For example, since the statutory clause in a constitution does not necessarily guarantee civilian supremacy in less developed countries, it fails to capture degrees of military versus civilian influence within such states. Since the military regime type assumes strong leadership by a single ruler, it does not reflect what the model defines as the dynamic interactions between civilian and military leaders in each state. Lastly, surveys and case studies are too costly and time-consuming at this moment. For further details about how to operationalize civil–military relations, see the authoritative study by Desch (1999: 1–7). 9. Two qualifications should be noted about this measurement: First, our research interest is in military leaders’ influence on the foreign policy

Notes 163

10.

11.

12.

13. 14.

decision-making process, not overall militarism. Although militarism might be represented more effectively by counting men under arms, this is not the focus of our conceptual model. Second, we acknowledge that changes in expenditure might be due to military adaptation of new technologies, retirement of obsolete weapons, or new external threats. It is beyond the scope of this book to consider the possibilities mentioned earlier. In recognition that the operationalization of military expenditure is both a new measurement and is subject to a range of interpretations, we have also designed and implemented two further measurements. One is calculated as the combined annual growth rate for military expenditure in each dyad. For example, if State A’s military expenditure shows a 20% increase and State B’s a 5% decrease in 1990 as compared to 1989, the dyad is recorded as a 15 overall percentage increase in 1990. The other measurement is dichotomous. If both states in a dyad experience an annual growth rate increase in their military expenditure at a time t as compared to t1, the dyad is recorded as “1”; if not, it is coded as “0.” For example, if State A’s military expenditure shows a 20% increase and State B’s shows a 10 % increase (i.e., both states have a positive annual growth rate increase) in 1990 as compared to 1989, the dyad is recorded as “1” in 1990, indicating a relative increase in military over civilian influence. Results based on the two additional measurements are similar to those reported here and can be obtained upon request from the authors. It can be argued that military influence, measured via military expenditure, is to some degree an indicator of the seriousness of threats to national security as perceived by leaders. We will look into this possibility in chapter 7. It should be noted that, making general use of the numerical dollar amount of military expenditure, students of arms races “assume that military spending is an important measure of a nation’s future military capabilities and therefore also intentions and commitment” (Hartley and Russett 1992: 909). The webpage address is http://www.eugenesoftware.org/ (last accessed on June 16, 2004). For some thoughtful ideas about the range of application for the military expenditure data, see Diehl and Crescenzi (1998: 116). See http://www.orst.edu/dept/pol_sci/fac/sahr/sahr.htm (last accessed on June 16, 2004) for an explanation of this method. The following sources have been used for cross-checking and clarifying military manpower data with respect to reliability and validity: International Institute for Strategic Studies (1970–1971 through 2000/01), Anderson and Honegger (1982), Keegan (1979, 1983), Stockholm International

164 Notes

15.

16.

17.

18.

19.

20.

21.

Peace Research Institute (1985), Pope (1987), and Schumacher et al. (1989). See also the Inter-University Consortium for Political and Social Research (ICPSR) Study Number 5025 and the ICPSR Study Number 5026. The web addresses are http://www.icpsr.umich.edu:8080/ ABSTRACTS/05025.xml?formatICPSR (last accessed on June 16, 2004) and http://www.icpsr.umich.edu:8080/ABSTRACTS/05026.xml? formatICPSR (last accessed on June 16, 2004), respectively. The data are available from 1948 to 1994. For more details about this data, see Van Belle (2000: 137–148). We are very grateful to Doug Van Belle, who kindly provided us with the data. It should be noted that Oneal and Russett’s (1999c) article is, in turn, a replication and extension of their earlier prominent article on democracy, independence, and peace (Oneal and Russett 1997). Oneal and Russett (1999c), however, add another important neo-Kantian peace element: international organizations. Furthermore, Oneal and Russett extend the study period and examine what neorealists might call systemlevel effects on militarized interstate disputes. While Oneal and Russett’s studies are not the only one available, their clarity and prominence combine to make this a reasonable choice in providing a basis for further research. While Oneal and Russett (1999c) include all dyads in their analyses, Russett and Oneal’s (2001) recent book is limited to the politically relevant dyads only. Thus, for broader theoretical and empirical application, we choose to replicate Oneal and Russett (1999c). It should be re-emphasized that Oneal and Russett’s version of democraticness treats the degree of constraint on the chief executive as a main determinant of the conventional “democracy minus autocracy” scores. In footnote 30, Oneal and Russett (1999c: 12) acknowledge this measurement drawback of the democraticness variable from the Polity III data set when observing that their democracy minus autocracy score does not reflect “the relative importance of its components [which are] unstable over time.” In this sense, Oneal and Russett’s democraticness variable represents an important aspect of democracy, especially executive constraints, rather than the concept writ large. For more detailed discussions, see Gleditsch and Ward (1997) and Munck and Verkuilen (2002). According to Oneal and Russett (1999c: 23), after having tested both MID involvement and initiation, they find that both dependent variable measurements have “produced nearly identical results.” Oneal and Russett (1999c) report the MID involvement results only in their article. The data-related constraints are due to diplomatic channels, whose data are available only up to 1974 at this moment, and media openness,

Notes 165

whose data exist from 1948 to 1994 only. We have searched for an updated version by accessing the Scientific Study of International Politics (SSIP) e-list, [email protected], but have not found any work in progress on diplomatic channels. 22. See Diggle et al. (1994: 142–145, 151–152) and Zorn (2001) for explanation of this technique. 23. On use of peace years correction, see Beck and Katz (1995), Beck et al. (1998), and Beck (2001). 24. Beck et al. (2000) have introduced the Bayesian neural network model to improve quantitative studies of international conflict, crisis, and war where the logistic regression model has prevailed despite its analytical limitations. On the neural network model and its applications in political science, see Cimbala (1987), Hertz et al. (1991), Hudson (1991), Huntley (1991), Kimber (1991), Schrodt (1991, 2000), MacKay (1992a, 1992b, 1992c, 1992d), Neal (1992, 1993, 1996), Fausett (1994), Werbos (1994), Bishop (1995), Domany et al. (1996), Ripley (1996), Zeng (1999, 2000), Bearce (2000), Pinto and Neal (2001), and NeuroBook II in http://www.spss.com/neuro/.

Chapter 5 Accounting for Militarized Interstate Disputes 1. As a preliminary check on the results, bivariate analysis using Spearman’s rank-order correlations shows positive relationships for civil–military relations (i.e., ␳  0.011 at p  0.001, N  134,634), military manpower system (i.e., ␳  0.0205 at p  0.001, N  122,137), and diplomatic channels (i.e., ␳  0.0508 at p  0.001, N  57,221) with involvement in MIDs. A negative relationship appears for media openness (i.e., ␳  0.0435 at p  0.001, N  37,452) with involvement in MIDs. It should be noted that, since each number of observations is large, the correlation coefficients are small, but still statistically significant. The p-levels are at best advisory, of course, since the results are based on a population rather than a sample of cases. 2. Stata Statistical Software (version 7.0) is used for the replications and empirical tests. All of the replications are very close to Oneal and Russett’s (1999c: 22) results and can be obtained upon request from the authors. 3. The empirical results of the other five realist-oriented variables (i.e., national capability ratio, alliance, noncontiguity, geographic distance, and only minor powers) in the following tables, in general, concur with those of Oneal and Russett (1999c: 22).

166 Notes 4. As mentioned earlier, all of the replications can be obtained upon request from the authors. 5. This result concurs with that reported by Oneal and Russett (1999c: 22). 6. Yet it must be noted that the civil–military relations variable of politically relevant dyads in table 5.1 shows statistical significance at the 0.05 level. 7. It is worth noting that the military manpower system variable shows statistical significance under Beck et al.’s peace years correction model in all four tables. 8. Students of comparative politics have argued that liberal democracy based on civil liberties including media openness is superior to electoral democracy based on voting rights and chief executive constraints. This line of reasoning is already well developed in Mueller’s (1992) qualitative analysis of the development of democracy. 9. It is interesting to note that some highly autocratic states do have a free press. Our data reveal that, while Nigeria, Uganda, Swaziland, and the like fall below 6 on the conventional democracy minus autocracy score, their media openness is reported as free or imperfectly free and therefore capable of functioning as an area of political competition or debate. 10. Thus, the multicollinearity issue among independent variables including democraticness and media openness in the previous logistic regression models is not applicable to neural network models by definition. 11. Neural Connection® 2.1 is the software used for the neural network modeling; its application is explained at greater detail in Beck et al. (2000). Neural Connection® 2.1 is a Windows® version developed by SPSS Inc. Despite its superb performance, one drawback should be mentioned: the upper limit on data is 15,000 observations. This limits the neural network modeling to the politically relevant dyads (N  8,386) only. For more detailed information about the neural network analysis, see Neural Connection® 2.0: User’s Guide (1997), Neural Connection® 2.0: Applications Guide (1997), and Neural Connection® 2.1: Update (1999). 12. As in Beck et al. (2000), we have experimented with the issue of insample and out-of-sample. For example, we randomly divided the data into a training set (80%) to fit the model, and a validation set (10%) as well as a test set (10%) to confirm the performance (i.e., goodness of fit) of the trained model. The results, where the RMS error is 0.1972 in the training set, 0.1939 in the validation set, and 0.2028 in the testing set, are very similar to those reported in this study and can be obtained upon request from the authors. In this research, we choose to report the results based on the training data during the period from 1950 to 1974 as a replacement for that from the randomized one to facilitate comparison with those from the previous logistic regression models, which cover 1950–1974.

Notes 167

13. The lack of statistical significance tests in neural network models is not addressed by Beck et al. (2000). Given the usual focus in international conflict research on a population rather than a sample, this should not be viewed as an important issue. For a more detailed discussion, see Zeng (1999, 2000). 14. As Beck et al. (2000) point out, interpretation and inference in neural network models are extremely complex. As a result, the coefficients are almost impossible to interpret directly, which creates the need to employ graphical presentations. The empirical results of the other five realist variables will not be discussed, but their patterns, in general, concur with the findings from existing democratic peace literature. 15. Following Oneal and Russett’s (1999c) baseline probability, we assume not only that each dyad has military leaders’ influence, a conscription system, no media openness, no major power involvement, is contiguous, and is not allied, but also that each continuous variable is equal to its mean value except that the economic interdependence variable is set at its median value. 16. The interpretation is based on the difference between the left-hand vertical line and the right-hand vertical line along the predicted line in the cross-section of the sensitivity plot. 17. Parts of this section are based primarily upon James and Özdamar (2004). 18. Center for Nonproliferation Studies, IAEA-North Korea: Nuclear Safeguards and Inspections 1977–1989, 90, 91, 92, 93, and 94. Downloaded from http://cns.miis.edu/research/korea/nuc/iaea7789.htm. 19. It is worth noting that Yun (2003), a former senior adviser to assistant secretaries of state Winston Lord and Stanley Roth, contends that, considering the insecure North Korean mentality and the communist penchant for secrets, the North shows “a certain predictability in the pattern of negotiations—a rigid dance, full of bluster. . . . North Korean negotiators must demonstrate to higher-ups that nothing is left on the table, which usually means using every form of tactical diversion to irritate. If one manages to endure the entire ‘ritual’ and show flexibility at the right time, an in-kind response is usually forthcoming” (13–14).

Chapter 6 Accounting for Fatal Militarized Interstate Disputes 1. Fatal MIDs are defined as MIDs with at least one solider killed in a dyad. Maoz’s dyadic MID dataset version 1.1 (updated January 2001) from ftp://spirit.tau.ac.il/zeevmaoz/dyadmid.html is used to identify the

168 Notes

2.

3.

4.

5.

6.

fatality level of each dyad. The missing and duplicated observations per dyad are deleted. Notably, research on fatal MIDs is less in the “spotlight” than studies of levels of hostility in dyadic disputes per se. Oneal et al.’s (2003) work appear to be a first, provocative introduction of lethal dispute analysis into the democratic peace literature. This relatively recent turn of interest may be because of missing observations; Maoz (1999: 3) warns that, “due to considerable missing information, this [fatality level] variable should be used with great deal of caution.” As a preliminary check on the results, bivariate analysis using Spearman’s rank-order correlations shows positive relationships for civil–military relations (i.e., ␳  0.0049 at p  0.0697, N  134,583), military manpower system (i.e., ␳  0.0063 at p  0.0273, N  122,092), and diplomatic channels (i.e., ␳  0.0104 at p  0.0127, N  57,191) with fatal MIDs. A negative relationship appears for media openness (i.e., ␳  0.0150 at p  0.0036, N  37,435) with fatal MIDs. It should be noted that, since each number of observations is large, the correlation coefficients are small, but still statistically significant. The p-levels are at best advisory, of course, since the results are based on a population rather than a sample of cases. Stata Statistical Software (version 7.0) is used for the replications and empirical tests. All of the replications can be obtained upon request from the authors. Surprisingly enough, regardless of model specification and time period, the democraticness variable does not reach statistical significance in the anticipated direction. The empirical results of the other five realist variables (i.e., national capability ratio, alliance, noncontiguity, geographic distance, and only minor powers) in the following tables, in general, follow intuition as described in the democratic peace research program. We also speculate that, since there are relatively few MIDs between democratic states, and the fatality threshold we use is very low (i.e., one battle death), the results may be highly sensitive to a few cases in which jointdemocratic disputes generated a very small number of fatalities. Thus, the results may reflect an operationalization that makes no distinction between the few people who might be killed in a joint-democratic dispute and the hundreds or thousands who might die in a dispute involving either nondemocratic states or a mixed dyad. There may be an alternative way of looking at the results in the last column. One may suspect endogeneity problems among independent variables to some extent in that, besides military manpower system, there are two more unanticipated, reversed coefficient signs. Both democraticness and international organizations show positive signs, indicating an increase

Notes 169

7. 8.

9.

10.

11.

12.

in fatal MIDs. The GEE logistic regression model is somewhat vulnerable to endogeneity problems. This issue will be dealt with in detail through the neural network models discussed in the next section. As discussed earlier, the upper cap on total observations limits the neural network modeling to the politically relevant dyads (N  8,369) only. As in Beck et al. (2000), we have experimented with the issue of insample and out-of-sample for fatal MIDs. For example, we randomly divided the data into a training set (80%) to fit the model, and a validation set (10%) as well as a test set (10%) to confirm the performance (i.e., goodness of fit) of the trained model. The results, where the RMS error is 0.0528 in the training set, 0.0762 in the validation set, and 0.0599 in the testing set, are very similar to those reported in this study and can be obtained upon request from the authors. In this research, we choose to report the results based on the training data during the period from 1950 to 1974, as a replacement for that from the randomized one, to facilitate comparison with those from the previous logistic regression models that cover 1950–1974. The empirical results of the other five realist-oriented variables will not be discussed, but their statistical significance, in general, is similar to that attained in previous democratic peace studies. Following Oneal and Russett’s (1999c) baseline probability, we assume not only that each dyad has military leaders’ influence, a conscription system, no media openness, no major power involvement, is contiguous, and is not allied, but also that each continuous variable is equal to its mean value except that economic interdependence is set at its median value. As noted in the previous set of figures from chapter 5, the interpretation is based on the difference between the left-hand vertical line and the right-hand vertical line along the predicted line in the cross-section of the sensitivity plot. It should be noted that, while Oneal et al. (2003) ran distributed-lag models, we employed logit and neural network models. Given the methodological discrepancy, further research is in order to assess the robustness of each set of results.

Chapter 7 A Key Issue of Measurement: Military Expenditure and Civil–Military Relations An earlier version of this chapter appeared in Armed Forces & Society as “Civil–Military Relations in a Neo-Kantian World, 1886–1992,” 2004, 30(2): 227–254.

170 Notes 1. The independent variable is lagged by one year, so it is not affected by the degree of international conflict present in the dyad-year to be explained in the second step. 2. Sarkesian’s model is cited in Snider and Carlton-Carew (1995). 3. As noted in the previous chapters, Dixon (1993, 1994) standardized this operationalization by calling it the “weak link” approach when applied to regime type within dyads. 4. It is assumed that, in a situation of conflict, civilians will support military spending to defend the state and its people. Our two-step method does not distinguish different roles and responsibility between civilian and military leaders and we leave further refinement to future study. 5. After four years, the lagged term for civil–military relations becomes statistically insignificant. With no theoretical reason to include a different number of terms, the choice of three is justified on statistical grounds. 6. Several controversies exist: for example, Kier (1997) discusses the tactical limits of a drafted army in England and France, and Anderson and Honegger (1982) provide an excellent collection of the pros and cons of each type of military manpower system. 7. These historical instances bring to mind the Richardson (1960) process model of arms races. The basic dynamic is that rivals procure armaments in response to each other, with some allowance for fatigue as well as the likelihood that the process will culminate in warfare. The Richardson model can be extended to encompass foreign policy across the board in terms of an action–reaction model. In events data studies, it is interesting to note that the previous behavior of an actor, along with the behavior directed toward it by another actor at the previous iteration, come close to being sufficient in terms of anticipating what will happen next (Wilkenfeld et al. 1980). Such iterated processes, however, are not theoretically salient to the current project, which probes for more subtle relationships involving civil–military relations and political communications in relation to foreign policy. 8. It should be noted that the level of hostility is originally based on MIDs data assembled from the Correlates of War project (see Gochman and Maoz 1984). 9. The GEE is one of the most commonly employed statistical methods among students of international conflict, crisis, and war. Stata Statistical Software (version 7.0) is used for empirical tests. Our model is corrected for the first-year autocorrelation and heteroskedasticity. The Wald test statistic of 471.37 is statistically significant at the 0.001 level. For explanation of GEE in practice, see Zorn (2001). 10. The numbers in parentheses are standard errors. 11. According to Oneal and Russett, after testing MID involvement and initiation, they find that both dependent variable measurements have

Notes 171

“produced nearly identical results.” Oneal and Russett report only the MID involvement results in their article. See Oneal and Russett (1999c: 23). 12. See table 1 in Oneal and Russett (1999c: 22).

Chapter 8 A Closer Look at Media Openness: From Institutional Democracy to Media Openness? 1. The study period in the previous chapters had spanned from 1950 to 1974 due to data restrictions on other variables that do not apply here. We acknowledge that missing cases could explain, for example, an unexpected result in this chapter, namely, the weak findings for interdependence. 2. An impressive extension of the research enterprise on media and international relations appears in Van Belle et al. (2004). This study’s comparative analysis of overseas development aid programs by the five largest and developed donor states reveals an important and substantive impact for media and encourages further research on a wider range of dependent variables. 3. Prominent examples include Bollen (1980), Alvarez et al. (1996), Gasiorowski (1996), and Freedom House (2000). 4. According to Collier and Levitsky (1997), there are hundreds of subtypes of democracy. 5. Despite the absence of civil liberties as an indicator, the Polity data sets should be commended for going beyond, at least to some extent, the definition of electoral (i.e., minimal) democracy by incorporating institutional constraints on the exercise of executive power. 6. As Fearon (1995) points out, private information is not the only important condition for war. One must also have incentives to misrepresent. A lower level of private information may lead to war, but depending on incentives it may not. For example, most wars are fought between contiguous states, or states in close proximity despite the fact that they know each other very well. This is the classic story of World War I. Familiarity sometimes breeds contempt rather than understanding. 7. It should be noted that, unlike the democracy hypothesis, the media openness hypothesis does not adopt the weak link assumption, which requires at least an ordinal level measurement. Although it is tempting to treat the media openness variable as a scale, Van Belle (2000: 140) notes that “categorical coding used for this analysis does not produce a five-point interval scale and the degree to which it can be used as an ordinal scale is unclear.” Stata Statistical Software (version 7.0) is used for empirical tests. 8. This strategy is suggested in Starr’s (1992: 46) seminal work from a decade ago: “There is varying empirical support for any of the possible

172 Notes

9.

10. 11.

12. 13. 14.

15.

16.

explanations for peace in democratic–democratic dyads. Which of these would seem to be the most promising? One way to deal with alternative explanations would be to engage in empirical analysis of the possible alternatives.” An excellent example is Maoz and Russett (1993), which compares institutional and normative measures of democracy in the same model. It should be noted that, in opposition to the body-bag syndrome, Feaver and Gelpi (2004) argue that the public is not casualty-averse, despite what political and military leaders might believe. Maoz’s dyadic MID data set’ version 1.1 (updated January 2001) is found at ftp://spirit.tau.ac.il/zeevmaoz/dyadmid.html. Spearman’s rho reports a correlation of 0.0279 between media openness and MID involvement, and 0.0131 between media openness and fatal MIDs. This also follows Oneal and Russett (1999c). The economic interdependence variable for all dyads barely misses statistical significance at the 0.05 level, with p  0.101. The overlap between our media openness and institutional democracy might be a suspected reason for statistical insignificance of the three neo-Kantian peace variables. The presence of multicollinearity causes greater standard errors that, in turn, encroach on statistical significance tests. However, as columns 2 and 3 in table 8.1 make clear, inflation of standard error is not a concern. The standard error for institutional democracy is slightly inflated, while somewhat deflated for economic interdependence, and slightly inflated for international organizations. By contrast, the magnitude of each coefficient turns out to be much smaller than before the introduction of media openness into the model. The statistical significance tests for the three neo-Kantian peace variables fail not because of multicollinearity, but because of their relative magnitude. Given the fact that multicollinearity has nothing to do with deflating coefficient magnitude, these results suggest that multicollinearity poses no problem. Spearman’s rho reports a correlation of 0.60 between media openness and institutional democracy with the MID involvement data set and 0.59 with the fatal MIDs data set. We have also tested for a monadic effect of the media openness hypothesis, which is not expected to hold because of the inherently dyadic nature of the causal argument. The monadic hypothesis, as anticipated, is not supported; statistical results can be obtained upon request from the authors.

Notes 173

Chapter 9 The Quest for Peace 1. Beck et al.’s (1998) peace years correction models show its statistical significance in a consistent manner in table 5.6. 2. Such unexpected results in quantitative research should not be treated as a total surprise. Since the research design and analysis are coherent and rigorous, the counterintuitive results provide us with an opportunity to think about causal inference like a social scientist, that is, “with skepticism and a concern for alternative explanations that may have been overlooked. Casual inference thus becomes a process whereby each conclusion becomes the occasion for further research to refine and test it. Through successive approximations [researchers] try to come closer and closer to accurate causal inference” (King et al. 1994: 32–33, emphasis in the original). We leave this task to ourselves as well as others. 3. Since Oneal and Russett’s (1999c) democraticness variable is continuous, our replication will not address this important issue for the time being, with further data analysis planned for a later study. 4. It should be noted that the principal academic advocates of the democratic peace have not used their statistical findings as a basis to call for efforts toward forcible regime change led by the United States.

This page intentionally left blank

References Allison, Graham T. 1969. “Conceptual Models and the Cuban Missile Crisis.” American Political Science Review 63(3): 689–718. ———. 1971. Essence of Decision: Explaining the Cuban Missile Crisis. Boston, MA: Little, Brown. ——— and Philip Zelikow. 1999. Essence of Decision: Explaining the Cuban Missile Crisis, 2nd edn. New York: Longman. ———, Albert Carnesale, and Joseph S. Nye, Jr. 1985. Hawks, Doves, and Owls: An Agenda for avoiding Nuclear War. New York: W.W. Norton & Company. Alvarez, Michael, José Antonio Cheibub, Fernando Limongi, and Adam Przeworski. 1996. “Classifying Political Regimes.” Studies in Comparative International Development 31(2): 1–37. Anderson, Gary M., Dennis Halcoussis, and Robert D. Tollison. 1996. “Drafting the Competition: Labor Unions and Military Conscription.” Defence and Peace Economics 7(3): 189–202. Anderson, Martin and Barbara Honegger (Eds.). 1982. The Military Draft: Selected Readings on Conscription. Stanford University, Stanford, CA: Hoover Institution Press. Andreski, Stanislav. 1992. Wars, Revolutions, Dictatorships: Studies of Historical and Contemporary Problems from a Comparative Viewpoint. Portland, OR: Frank Cass. Arms Control Agency. June 2003. Fact Sheets: Chronology of US–North Korean Nuclear and Missile Diplomacy. Downloaded from http://www.armscontrol. org/factsheets/dprkchron.asp. Babst, Dean V. 1964. “Elective Governments—A Force for Peace.” The Wisconsin Sociologist 3(1): 9–14. ———. 1972. “A Force for Peace.” Industrial Research 14:55–58. Barbieri, Katherine. 1998. “International Trade and Conflict: The Debatable Relationship.” Paper presented at the 35th Annual Convention of the International Studies Association, Minneapolis, MN, March 18–21. Barbieri, Katherine. 2002. The Liberal Illusion: Does Trade Promote Peace? Ann Arbor, MI: University of Michigan Press. Barnett, Correlli. 1970. Britain and Her Army, 1509–1970: A Military, Political, and Social Survey. New York, NY: William Morrow & Company. Bearce, David H. 2000. “Economic Sanctions and Neural Networks: Forecasting Effectiveness and Reconsidering Cooperation.” In Diana Richards (Ed.), Political Complexity: Nonlinear Models of Politics. Ann Arbor, MI: University of Michigan Press. 269–295.

176 References Beck, Nathaniel. 2001. “Time-Series-Cross-Section Data: What have We learned in the Past Few Years?” In Nelson W. Polsby (Ed.), Annual Review of Political Science 4: 271–293. ——— and Jonathan N. Katz. 1995. “What to do (and not to do) with Time-Series Cross-Section Data.” American Political Science Review 89(3): 634–647. ——— and Richard Tucker. 1998. “Taking Time Seriously: Time-Series-CrossSection Analysis with a Binary Dependent Variable.” American Journal of Political Science 42(4): 1260–1288. ———, Gary King, and Langche Zeng. 2000. “Improving Quantitative Studies of International Conflict: A Conjecture.” American Political Science Review 94(1): 21–35. Belsley, D. A., E. Kuh, and R. E. Welsch. 1980. Regression Diagnostics: Identifying Influential Data and Sources of Collinearity. New York: John Wiley & Sons. Bendor, Jonathan and Thomas H. Hammond. 1992. “Rethinking Allison’s Models.” American Political Science Review 86(2): 301–322. Benjamin, Roger W. and Lewis J. Edinger. 1971. “Conditions for Military Control over Foreign Policy Decisions in Major States: A Historical Exploration.” Journal of Conflict Resolution 15(1): 5–31. Bennett, D. Scott and Allan Stam. 2000a. “A Universal Test of an Expected Utility Theory of War.” International Studies Quarterly 44(3): 451–481. ———. 2000b. “EUGene: A Conceptual Manual.” International Interactions 26(2): 179–204. Bennett, W. Lance. 1994. “The Media and the Foreign Policy Process.” In David A. Deese (Ed.), The New Politics of American Foreign Policy. New York: St. Martin’s Press. 168–188. Betts, Richard K. 1977. Soldiers, Statesmen, and Cold War Crises. Cambridge, MA: Harvard University Press. Bishop, Christopher. 1995. Neural Networks for Pattern Recognition. Oxford: Clarendon Press. Blainey, Geoffrey. 1988. The Causes of War, 3rd edn. New York: The Free Press. Bollen, Kenneth A. 1980. “Issues in the Comparative Measurement of Political Democracy.” American Sociological Review 45(3): 370–390. ———. 1991. “Political Democracy: Conceptual and Measurement Traps.” In Alex Inkeles (Ed.), On Measuring Democracy: Its Consequences and Concomitants. New Brunswick, NJ: Transaction Publishers. 3–20. ———. 1993. “Liberal Democracy: Validity and Method Factors in CrossNational Measures.” American Journal of Political Science 37(4): 1207–1230. Brecher, Michael. 1996. “Crisis Escalation: Model and Findings.” International Political Science Review 17(2): 215–230.

References 177

———. 1999. “International Studies in the Twentieth Century and Beyond: Flawed Dichotomies, Synthesis, Cumulation: ISA Presidential Address.” International Studies Quarterly 43(2): 213–264. ——— and Jonathan Wilkenfeld. 1997. A Study of Crisis. Ann Arbor, MI: University of Michigan Press. Bueno de Mesquita, Bruce. 1981. The War Trap. New Haven and London: Yale University Press. ——— and Randolph M. Siverson. 1995. “War and the Survival of Political Leaders: A Comparative Analysis of Regime Type and Accountability.” American Political Science Review 89(4): 841–855. ———, James D. Morrow, Randolph M. Siverson, and Alastair Smith. 1999. “An Institutional Explanation of the Democratic Peace.” American Political Science Review 93(4): 791–807. ———, Alastair Smith, Randolph M. Siverson, and James D. Morrow. 2003. The Logic of Political Survival. Cambridge, MA: The MIT Press. Bunge, Mario. 1996. Finding Philosophy in Social Science. New Haven and London: Yale University Press. Campbell, Kenneth J. 1998. “Once Burned, Twice Cautious: Explaining the Weinberger–Powell Doctrine.” Armed Forces & Society 24(3): 357–374. Cederman, Lars-Erik. 2001. “Back to Kant: Reinterpreting the Democratic Peace as a Macrohistorial Learning Process.” American Political Science Review 95(1): 15–31. Center for Nonproliferation Studies, IAEA-North Korea: Nuclear Safeguards and Inspections 1977–1989, 90, 91, 92, 93, and 94. Downloaded from http://cns.miis.edu/research/korea/nuc/iaea7789.htm. Chan, Steve. 1997. “In Search of Democratic Peace: Problems and Promise.” Mershon International Studies Review 41(1): 59–91. Choi, Seung-Whan and Patrick James. 2003. “No Professional Soldiers, No Militarized Interstate Disputes?: A New Question for Neo-Kantianism.” Journal of Conflict Resolution 47(6): 796–816. ———. 2004. “Civil–Military Relations in a Neo-Kantian World, 1886–1992.” Armed Forces & Society 30(2): 227–254. Cimbala, Stephen J. (Ed.) 1987. Artificial Intelligence and National Security. Massachusetts and Toronto: Lexington Books. Clausewitz, Carl Von. 1976. In Michael Howard and Peter Paret (Eds., Trans.), On War. Princeton, NJ: Princeton University Press. CNN.com. April 21, 2004. “Kim ‘flexible’ on Nuke Standoff.” Downloaded from http://edition.cnn.com/2004/WORLD/asiapcf/04/21/china.kim/ index. html. CNN.com. September 13, 2004. “N. Korea ‘wants Six-Party Talks’.” Downloaded from http://edition.cnn.com/2004/WORLD/asipac/09/13/ nkorea.talks/.

178 References Cohen, Raymond. 1994. “Pacific Unions: A Reappraisal of the Theory that ‘Democracies do not go to War with Each Other’.” Review of International Studies 20(3): 207–223. Collier, David and Steven Levitsky. 1997. “Democracy with Adjectives: Conceptual Innovation in Comparative Research.” World Politics 49(3): 430–451. Dahl, Robert. 1971. Polyarchy. New Haven, CT: Yale University Press. Desch, Michael C. 1999. Civilian Control of the Military: The Changing Security Environment. Baltimore and London: Johns Hopkins University Press. Diamond, John. June 8, 2001. “On Foreign Policy, Bush Moving to Clinton Views.” Chicago Tribune. 1. Diamond, Larry. 1996. “Is the Third Wave Over?” Journal of Democracy 7(3): 20–37. ———. 1999. Developing Democracy toward Consolidation. Baltimore and London: Johns Hopkins University Press. Diehl, Paul F. and Mark J. C. Crescenzi. 1998. “Reconfiguring the Arms Race—War Debate.” Journal of Peace Research 35(1): 111–118. Diehl, Paul F. and Jean Kingston. 1987. “Messenger or Message?: Military Buildups and the Initiation of Conflict.” Journal of Politics 49(3): 801–813. Diggle, Peter J., Kung-Yee Liang, and Scott L. Zeger. 1994. Analysis of Longitudinal Data. New York: Oxford University Press. Dixon, William J. 1993. “Democracy and the Management of International Conflict.” Journal of Conflict Resolution 37(1): 42–68. ———. 1994. “Democracy and the Peaceful Settlement of International Conflict.” American Political Science Review 88(1): 14–32. ———. 1998. “Dyads, Disputes and the Democracy Peace.” In Murray Wolfson (Ed.), The Political Economy of War and Peace. Boston: Kluwer Academic Publishers. 103–126. Doder, Dusko. October 5, 1983. “Soviets said to remove Air Officers; Civil Sources cite Defense Failure in Korean Incursion.” The Washington Post. A1. Domany, E., J. L. van Hemmen, and K. Schulten (Eds.). 1996. Models of Neural Networks III: Application, Generalization, and Representation. New York: Springer-Verlag. Donovan, James A. 1970. Militarism, U.S.A. New York: Charles Scribner’s Sons. Doran, Charles F. 1991. Systems in Crisis: New Imperatives of High Politics at Century’s End. Cambridge and New York: Cambridge University Press. Dougherty, James E. and Robert L. Pfaltzgraff, Jr. (Eds.). 2001. Contending Theories of International Relations: A Comprehensive Survey, 5th edn. New York, San Francisco, and Boston: Longman.

References 179

Duindam, Simon. 1999. Military Conscription: An Economic Analysis of the Labour Component in the Armed Forces. Physica-Verlag: A Spring-Verlag Company. East, Maurice A. and Phillip M. Gregg. 1967. “Factors influencing Cooperation and Conflict in the International System.” International Studies Quarterly 11(3): 244–269. Elman, Miriam Fendius (Ed.). 1997. Paths to Peace: Is Democracy the Answer? Cambridge, MA and London, England: MIT Press. Enterline, Andrew J. 1998a. “Regime Changes, Neighbourhoods, and International Conflict, 1816–1992.” Journal of Conflict Resolution 42(6): 804–829. ———. 1998b. “Regime Changes and Interstate Conflict, 1816–1992.” Political Research Quarterly 51(2): 385–409. Entman, Robert M. 2000. “Declarations of Independence: The Growth of Media Power after the Cold War.” In Brigitte L. Nacos, Robert Y. Shapiro, and Pierangelo Isernia (Eds.), Decisionmaking in a Glass House: Mass Media, Public Opinion, and American and European Foreign Policy in the 21st Century. Lanham, Boulder, New York, and Oxford: Rowman & Littlefield Publishers, Inc. 11–26. Everts, Philip. 2001. “Introduction.” In Philip Everts and Pierangelo Isernia (Eds.), Public Opinion and the International Use of Force. London and New York: Routledge. 1–27. ———. 2002. Democracy and Military Force. New York: Palgrave. Farber, Henry S. and Joanne Gowa. 1995. “Polities and Peace.” International Security 20(2): 123–146. ———. 1997. “Common Interests or Common Polities? Reinterpreting the Democratic Peace.” Journal of Politics 59(2): 393–417. Fausett, Laurene. 1994. Fundamentals of Neural Networks: Architectures, Algorithms, and Applications. Englewood Cliffs: Prentice Hall. Fearon, James D. 1994. “Domestic Political Audiences and the Escalation of International Disputes.” American Political Science Review 88(3): 577–592. ———. 1995. “Rationalist Explanations for War.” International Organization 49(3): 379–414. Feaver, Peter D. 1999. “Civil–Military Relations.” In Nelson W. Polsby (Ed.), Annual Review of Political Science 2: 211–241. ——— and Christopher Gelpi. 1999. “Civilian Hawks and Military Doves: The Civil–Military Gap and the Use of Force, 1816–1992.” Paper presented to the TISS Conference on the Gap between the Military and Civilian Society. Feaver, Peter D. 2004. Choosing Your Battles: American Civil–Military Relations and the Use of Force. Princeton and Oxford: Princeton University Press.

180 References Freedom House. 2000. Annual Survey of Freedom Country Scores, 1972–73 to 1999–00. Data downloaded from http://freedomhouse.org/ratings/ index.htm. Gallarotti, Giulio M. 2001. “The Limits of International Organization: Systemic Failure in the Management of International Relations.” In Lisa L. Martin and Beth A. Simmons (Eds.), International Institutions: An International Organization Reader. Cambridge, MA and London, England: The MIT Press. 365–402. Gartner, Scott Sigmund and Gary M. Segura. 1998. “War, Casualties, and Public Opinion.” Journal of Conflict Resolution 42(3): 278–300. Gartzke, Erik Andrew. 1997. The Logic of the Democratic Peace. Ph.D. Dissertation. University of Iowa. ———. 1998. “Kant We All Just get Along? Opportunity, Willingness, and the Origins of the Democratic Peace.” American Journal of Political Science 42(1): 1–27. ———. 2000. “Preferences and the Democratic Peace.” International Studies Quarterly 44(2): 191–212. ———. 2001. “Democracy and the Preparation for War: Does Regime Type affect States’ Anticipation of Casualties?” International Studies Quarterly 45(3): 467–484. Gasiorowski, Mark J. 1996. “An Overview of the Political Regime Change Dataset.” Comparative Political Studies 29(4): 469–483. Geller, Daniel S. 1985. Domestic Factors in Foreign Policy: A CrossNational Statistical Analysis. Cambridge, MA: Schenkman Books, Inc. ——— and J. David Singer. 1998. Nations at War: A Scientific Study of International Conflict. Cambridge, UK: Cambridge University Press. Gerber, Alan and Donald Green. 1999. “Misperceptions about Perceptual Bias.” In Nelson W. Polsby (Ed.), Annual Review of Political Science 2: 189–210. Gibson, Christopher P. and Don M. Snider. 1999. “Civil–Military Relations and the Potential to Influence: A Look at the National Security DecisionMaking Process.” Armed Forces & Society 25(2): 193–218. Gleditsch, Kristian S. and Michael D. Ward. 1997. “Double Take: A Reexamination of Democracy and Autocracy in Modern Polities.” Journal of Conflict Resolution 41(3): 361–383. Gochman, Charles S. and Zeev Maoz. 1984. “Military Interstate Disputes, 1816–1976: Procedures, Patterns, and Insights.” Journal of Conflict Resolution 28(4): 585–615. Goertz, Gary and Paul F. Diehl. 1986. “Measuring Military Allocations: A Comparison of Different Approaches.” Journal of Conflict Resolution 30(3): 553–581. Gowa, Joanne. 1999. Ballots and Bullets: The Elusive Democratic Peace. Princeton: Princeton University Press.

References 181

Graber, Doris A. 1986. “Press Freedom and the General Welfare.” Political Science Quarterly 101(2): 257–275. Gray, Colin S. 1975. “Hawks and Doves: Values and Policy.” Journal of Political and Military Sociology 3(1): 85–94. Green, Donald P., Soon Yeon Kim, and David H. Yoon. 2001. “Dirty Pool.” International Organization 55(2): 441–468. Gross, Donald G. 2002. “Riding the Roller Coaster.” Comparative Connections: An E-Journal on East Asian Bilateral Relations 1st Quarter, CSIS. ———. 2003. “Tensions Escalate in Korea as the US Targets Iraq.” Comparative Connections: An E-Journal on East Asian Bilateral Relations 1st Quarter, CSIS. ———. 2004. “In the Eye of the Beholder: Impasse or Progress in the SixParty Talks?” Comparative Connections: An E-Journal on East Asian Bilateral Relations 1st Quarter, CSIS. Gujarati, Damodar N. 1995. Basic Econometrics, 3rd edn. New York: McGraw-Hill, Inc. Gurr, Ted Robert, Keith Jaggers, and Will H. Moore. 1989. Polity II Codebook. ———. 1990. “Transformation of the Western State: The Growth of Democracy, Autocracy and State Power since 1800.” Studies in Comparative International Development 25: 73–108. ———. 1991. “The Transformation of the Western State: The Growth of Democracy, Autocracy, and State Power Since 1800.” In Alex Inkeles (Ed.), On Measuring Democracy: Its Consequences and Concomitants. New Brunswick, NJ and London: Transaction Publishers. Harsch, Joseph C. September 22, 1983. “KAL Flight 7: Long-term Consequences.” The Christian Science Monitor. 23. Hartley, Thomas and Bruce Russett. 1992. “Public Opinion and the Common Defense: Who Governs Military Spending in the United States.” American Political Science Review 86(4): 905–915. Henderson, Errol. 2002. Democracy and War: The End of an Illusion? Boulder and London: Lynne Rienner Publishers. Hendrickson, Ryan C. 2002. “Clinton’s Military Strikes in 1998: Diversionary Uses of Force?” Armed Forces & Society 28(2): 309–332. Hensel, Paul R., Gary Goertz, and Paul F. Diehl. 2000. “The Democratic Peace and Rivalries.” Journal of Politics 62(4): 1173–1188. Hermann, Charles F., Janice Gross Stein, Bengt Sundelius, and Stephen G. Walker. 2001. “Resolve, Accept, or Avoid: Effects of Group Conflict on Foreign Policy Decisions.” International Studies Review 3(2): 133–168. Hermann, Margaret G. and Charles W. Kegley, Jr. 1995. “Rethinking Democracy and International Peace: Perspectives from Political Psychology.” International Studies Quarterly 39(4): 511–533.

182 References Hermann, Margaret G. and Charles W. Kegley, Jr. 2001. “Democracies and Intervention: A Danger Zone in the Democratic Peace?” Journal of Peace Research 38(2): 237–245. Hertz, John, Anders Krogh, and Richard G. Palmer. 1991. Introduction to the Theory of Neural Computation. Redwood City: Addison-Wesley Publishing Company. Hewitt, J. Joseph and Jonathan Wilkenfeld. 1996. “Democracies in International Crises.” International Interactions 22(2): 123–142. Hill, Kim Quaile. 1978. “Domestic Politics, International Linkages, and Military Expenditures.” Studies in Comparative International Development 13(1): 38–59. Hobbes, Thomas. 1991. In Richard Tuck (Ed.), Leviathan. Cambridge and New York: Cambridge University Press. Horeman, Bart and Marc Stolwijk (Eds.). 1998. Refusing to Bear Arms: A World Survey of Conscription and Conscientious Objection to Military Service. London: War Resister’s International. Hudson, Valerie M. (Ed.). 1991. Artificial Intelligence and International Politics. Boulder, CO: Westview Press. Huntington, Samuel P. 1957. The Soldier and the State: The Theory and Politics of Civil-Military Relations. Cambridge, MA and London, England: Belknap Press of Harvard University Press. ———. 1991. The Third Wave: Democratization in the Late Twentieth Century. Norman: University of Oklahoma Press. ———. 1997. “After Twenty Years: The Future of the Third Wave.” Journal of Democracy 8(4): 3–12. Huntley, David G. 1991. “Neural Nets: An Approach to the Forecasting of Time Series.” Social Science Computer Review 9(1): 27–38. Hwang, Balbina Y. September 8, 2003. “The Six-Party Talks: Much Ado About Nothing?” Washington, DC: Heritage Foundation Research on Asia and the Pacific, Web Memo #333. ———. February 23, 2004. “The Six-Party Talks: Keeping Diplomacy Alive.” Washington, DC: Heritage Foundation Research on Asia and the Pacific, Web Memo #430. International Institute for Strategic Studies. 1970–1971 through 2000/01. The Military Balance 1970–1971 through 2000/01. London: Oxford University Press. Inter-University Consortium for Political and Social Research (ICPSR) Study Number 5025: The Diplomatic Missions Received by Each International System Member, 1817–1970. Inter-University Consortium for Political and Social Research (ICPSR) Study Number 5026: Diplomatic Exchange Data, 1815–1970.

References 183

Isernia, Pierangelo. 2001. “Italian Public Opinion and the International Use of Force.” In Philip Everts and Pierangelo Isernia (Eds.), Public Opinion and the International Use of Force. London and New York: Routledge. 86–115. Jaggers, Keith and Ted Robert Gurr. 1995. “Tracking Democracy’s Third Wave with the Polity III Data.” Journal of Peace Research 32(4): 469–482. James, Patrick. 1988. Crisis and War. Kingston and Montreal: McGill-Gueen’s University Press. ———. 2002. International Relations and Scientific Progress: Structural Realism Reconsidered. Columbus: Ohio State University Press. ——— and Glenn E. Mitchell, II. 1995. “Targets of Covert Pressure: The Hidden Victims of the Democratic Peace.” International Interactions 21(1): 85–107. ——— and Özgür Özdamar. 2004. “The United States and North Korea: Avoiding a Worst Case Scenario.” Unpublished manuscript. ———, Eric Solberg, and Murray Wolfson. 1999. “An Identified Systemic Model of the Democracy-Peace Nexus.” Defence and Peace Economics 10(1): 1–37. ———. 2000. “Democracy and Peace: Reply to Oneal and Russett.” Defence and Peace Economics 11(2): 215–229. Janowitz, Morris. 1960. The Professional Soldier: A Social and Political Portrait. New York: The Free Press. Jones, Daniel M., Stuart A. Bremer, and J. David Singer. 1996. “Militarized Interstate Disputes, 1816–1992: Rationale, Coding Rules, and Empirical Patterns.” Conflict Management and Peace Science 15(2): 163–213. Kahan, Jerome H. and Anne K. Long. 1972. “The Cuban Missile Crisis: A Study of its Strategic Context.” Political Science Quarterly 87(4): 564–590. Kant, Immanuel. 1957 (1795). Perpetual Peace with Introduction by Lewis White Beck (Ed.). New York: The Liberal Arts Press. ———. 1991 (1797). The Metaphysics of Morals. Introduction, translation, and notes by Mary Gregor. Cambridge and New York: Cambridge University Press. Keegan, John. 1979. World Armies. New York: Facts on File. ———. 1983. World Armies, 2nd edn. Detroit, MI: Gale Research Company. Kegley, Jr., Charles W. and Margaret G. Hermann. 1996. “How Democracies Use Intervention: A Neglected Dimension in Studies of the Democratic Peace.” Journal of Peace Research 33(3): 309–320.

184 References Kegley, Jr., Charles W. and Margaret G. Hermann. 1997. “A Peace Dividend? Democracies’ Military Interventions and Their External Political Consequences.” Cooperation and Conflict 32(4): 339–369. Kennamer, J. David. 1992. “Public Opinion, the Press, and Public Policy: An Introduction.” In J. David Kennamer (Ed.), Public Opinion, the Press, and Public Policy. Westport, CT and London: Praeger. 1–17. Keohane, Robert O. and Joseph S. Nye. 1977. Power and Interdependence: World Politics in Transition. Boston, MA: Little, Brown. Kier, Elizabeth. 1997. Imagining War: French and British Military Doctrine Between the Wars. Princeton: Princeton University Press. Kimber, Richard. 1991. “Artificial Intelligence and the Study of Democracy.” Social Science Computer Review 9(3): 381–398. King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry: Scientific Inference in Quantitative Research. Princeton: Princeton University Press. King, Gary and Langche Zeng. 2001. “Explaining Rare Events in International Relations.” International Organization 55(3): 693–715. Korb, Lawrence J. 1997. “The Use of Force.” Brookings Review 15(2): 24–25. Kuhn, Thomas S. 1962. The Structure of Scientific Revolutions. Chicago, IL: University of Chicago Press. La Balme, Natalie. 2001. “The French and the Use of Force: Public Perceptions and Their Impact on the Policy-Making Process.” In Philip Everts and Pierangelo Isernia (Eds.), Public Opinion and the International Use of Force. London and New York: Routledge. 186–204. Lagazio, Monica and Bruce Russett. 2004. “A Neural Network Analysis of Militarized International Disputes, 1885–1992: Temporal Stability and Causal Complexity”. In Paul Diehl (Ed.), The Scourge of War: New Extensions on an Old Problem. Ann Arbor, MI: University of Michigan Press. Landy, Marc. 2004. “The Bully Pulpit and the War on Terror.” In George C. Edwards III and Philip John Davies (Eds.), New Challenges for the American Presidency. New York, NY: Pearson Longman. Lane, Charles. October 16, 1995. “A Soldier’s Story: National Insecurity.” The New Republic 213(16): 20–21. Lasswell, Harold D. 1941. “The Garrison State.” American Journal of Sociology 46(4): 455–468. ———. 1997. Essays on the Garrison State. New Brunswick, NJ and London: Transaction Publishers. Lemke, Douglas. 2002. Regions of War and Peace. Cambridge: Cambridge University Press. Looney, Robert E. and Peter C. Frederiksen. 2000. “The Effect of Declining Military Influence on Defense Budgets in Latin America.” Armed Forces & Society 26(3): 437–449. Luttwak, Edward N. 1996. “A Post-Heroic Military Policy.” Foreign Affairs 75(4): 33–44.

References 185

MacKay, David J. C. 1992a. Bayesian Methods for Adaptive Models. Ph.D. Dissertation. California Institute of Technology. ———. 1992b. “Bayesian Interpolation.” Neural Computation 4: 415–447. ———. 1992c. “A Practical Bayesian Framework for Backpropagation Networks.” Neural Computation 4: 448–472. ———. 1992d. “The Evidence Framework applied to Classification Networks.” Neural Computation 4: 720–736. Maechling, Charles, Jr. October 14, 1983. “Superpower Football; KAL Flight 7—the Legal Aftermath.” Christian Science Monitor. 16. Mansfield, Edward D. and Jack Snyder. 1995a. “Democratization and the Danger of War.” International Security 20(1): 5–38. ———. 1995b. “Democratization and War.” Foreign Affairs 74(3): 79–97. ———. 1996. “The Effects of Democratization on War (The Authors Reply).” International Security 20(4): 196–207. Maoz, Zeev. 1998. “Realist and Cultural Critiques of the Democratic Peace: A Theoretical and Empirical Re-Assessment.” International Interactions 24(1): 3–89. ———. 1999. Dyadic Militarized Interstate Disputes (DYMID1.1) Dataset— Version 1.1 (i.e., Dyadic MID Codebook) in ftp://spirit.tau.ac.il/ zeevmaoz/dyadmid.html. 1–9. ——— and Bruce Russett. 1992. “Alliance, Contiguity, Wealth, and Political Stability: Is the Lack of Conflict among Democracies a Statistical Artifact?” International Interactions 17(3): 245–267. ———. 1993. “Normative and Structural Causes of Democratic Peace, 1946–1986.” American Political Science Review 87(3): 624–638. McKeown, Timothy J. 2001. “Plans and Routines, Bureaucratic Bargaining, and the Cuban Missile Crisis.” Journal of Politics 63(4): 1163–1190. Menard, Scott. 1995. Applied Logistic Regression Analysis. Thousand Oaks, London, and New Delhi: Sage Publications. Midlarsky, Manus I. 1975. On War: Political Violence in the International System. New York: The Free Press. ——— (Ed.). 2000. Handbook of War Studies II. Ann Arbor, MI: University of Michigan Press. Mitchell, Sara McLaughlin. 2002. “A Kantian System: Democracy and Third-Party Conflict Resolution.” American Journal of Political Science 46(4): 749–759. Morgan, Patrick M. 1993. “Disarmament.” In Joel Krieger (Ed.), The Oxford Companion to Politics of the World. New York and Oxford: Oxford University Press. 246–247. Morgan, T. Clifton. 1993. “Democracy and War: Reflections on the Literature.” International Interactions 18(3): 197–203. Morgan, T. Clifton and Sally Howard Campbell. 1991. “Domestic Structure, Decisional Constraints, and War: So Why Kant Democracies Fight?” Journal of Conflict Resolution 35(2): 187–211.

186 References Morgenthau, Hans J. 1985. Politics among Nations: The Struggle for Power and Peace, 6th edn. Revised by Kenneth W. Thompson. New York, St. Louis, and San Francisco: McGraw-Hill, Inc. Mousseau, Michael. 1998. “Democracy and Compromise in Militarized Interstate Conflicts, 1816–1992.” Journal of Conflict Resolution 42(2): 210–230. ——— and Yuhang Shi. 1999. “A Test for Reverse Causality in the Democratic Peace Relationship.” Journal of Peace Research 36(6): 639–663. Mueller, John. 1992. “Democracy and Ralph’s Pretty Good Grocery: Elections, Equality and Minimal Human Being.” American Journal of Political Science 36(4): 983–1003. Munck, Gerardo and Jay Verkuilen. 2002. “Conceptualizing and Measuring Democracy: Evaluating Alternative Indices.” Comparative Political Studies 35(1): 5–34. Nacos, Brigitte L., Robert Y. Shapiro, Natasha Hritzuk, and Bruce Chadwick. 2000a. “New Issues and the Media: American and German News Coverage of the Global-Warming Debate.” In Brigitte L. Nacos, Robert Y. Shapiro, and Pierangelo Isernia (Eds.), Decisionmaking in a Glass House: Mass Media, Public Opinion, and American and European Foreign Policy in the 21st Century. Lanham, Boulder, New York, and Oxford: Rowman & Littlefield Publishers, Inc. 41–59. Nacos, Brigitte L., Robert Y. Shapiro, and Pierangelo Isernia. 2000b. “Old or New Ball Game?: Mass Media, Public Opinion, and Foreign Policy in the Post-Cold War World.” In Brigitte L. Nacos, Robert Y. Shapiro, and Pierangelo Isernia (Eds.), Decisionmaking in a Glass House: Mass Media, Public Opinion, and American and European Foreign Policy in the 21st Century. Lanham, Boulder, New York, and Oxford: Rowman & Littlefield Publishers, Inc. 1–7. Neal, Radford M. 1992. “Bayesian Training of Backpropagation Networks by the Hybrid Monte Carlo Method.” Technical Report No. CRG-TR92–1. Connectionist Research Group. Department of Computer Science. University of Toronto. ———. 1993. “Probabilistic Inference using Markov Chain Monte Carlo Methods.” Technical Report No. CRG-TR-93–1. Department of Computer Science. University of Toronto. ———. 1996. Bayesian Learning for Neural Networks. New York: SpringerVerlag. Neural Connection® 2.0: Applications Guide. 1997. By SPSS Inc. Neural Connection® 2.0: User’s Guide. 1997. By SPSS Inc. Neural Connection® 2.1: Update. 1999. By SPSS Inc. NeuroBook II in http://www.spss.com/neuro/.

References 187

Niskanen, William A. 1968. “The Peculiar Economics of Bureaucracy.” American Economic Review 58(2): 293–305. North Korea Handbook. 2003. Seoul, Korea: Yonhap News Agency. Oneal, John R. and James Lee Ray. 1997. “New Tests of the Democratic Peace: Controlling for Economic Interdependence, 1950–85.” Political Research Quarterly 5(4): 751–775. ——— and Bruce Russett. 1997. “The Classical Liberals were Right: Democracy, Interdependence, and Conflict, 1950–1985.” International Studies Quarterly 41(2): 267–294. ———. 1999a. “Assessing the Liberal Peace with Alternative Specifications: Trade still reduces Conflict.” Journal of Peace Research 36(4): 423–442. ———. 1999b. “Is the Liberal Peace Just an Artifact of Cold War Interests? Assessing Recent Critiques.” International Interactions 25(3): 213–241. ———. 1999c. “The Kantian Peace: The Pacific Benefits of Democracy, Interdependence, and International Organizations, 1885–1992.” World Politics 52(1): 1–37. ———. 2000. “Comment: Why ‘an Identified Systemic Model of the Democracy-Peace Nexus’ does not Persuade.” Defence and Peace Economics 11(2): 197–214. ———. 2001. “Clear and Clean: The Fixed Effects of the Liberal Peace.” International Organization 55(2): 469–485. ——— and Michael Berbaum. 2003. “Causes of Peace: Democracy, Interdependence, and International Organizations, 1885–1992.” International Studies Quarterly 47(3): 371–393. Organski, A. F. K. and Jacek Kugler. 1980. The War Ledger. Chicago: University of Chicago Press. Owen, John M. 1997. Liberal Peace, Liberal War: American Politics and International Security. Ithaca: Cornell University Press. ———. 2001. Book Review of Triangulating Peace: Democracy, Interdependence, and International Organizations. By Bruce Russett and John R. Oneal. International Affairs 77(4): 967–978. Pearson, David E. 1987. KAL 007: The Cover-Up. New York: Summit Books. Perlmutter, Amos. 1986. “The Military and Politics in Modern Times: A Decade Later.” Journal of Strategic Studies 9: 5–15. Pinto, Ruxandra L. and Radford M. Neal. 2001. “Improving Markov Chain Monte Carlo Estimators by Coupling to an Approximating Chain.” Technical Report No. 0101. Department of Statistics. University of Toronto. Pollins, Brian M. 1989a. “Conflict, Cooperation, and Commerce: The Effect of International Political Interactions on Bilateral Trade Flows.” American Journal of Political Science 33(3): 737–761.

188 References Pollins, Brian M. 1989b. “Does Trade Still follow the Flag?” American Political Science Review 83(2): 405–480. Pope, Barbara H. (Ed.). 1987. World Defense Forces: A Compendium of Current Military Information for All Countries of the World. Santa Barbara, CA and Oxford, England: ABC-CLIO. Powell, Robert. 1999. In the Shadow of Power: States and Strategies in International Politics. Princeton, NJ: Princeton University Press. Prasad, Devi and Tony Smythe (Eds.). 1968. Conscription: A World Survey: Compulsory Military Service and Resistance to It. London: War Resisters’ International. Prins, Brandon C. 2002. “Democratic Politics and Dispute Challenges: Examining the Effects of Regime Type on Conflict Reciprocation, 1816–1992.” Paper prepared for delivery at the Annual Meeting of the American Political Science Association, Washington, DC. Ray, James Lee. 1997. “The Democratic Path to Peace.” Journal of Democracy 8(2): 49–64. ———. 1998. “Does Democracy cause Peace?” In Nelson W. Polsby (Ed.), Annual Review of Political Science 1: 27–46. Reed, William. 2002. “Selection Effects and World Politics Research.” International Interactions 28(1): 1–3. Richardson, Lewis Fry. 1960. Arms and Insecurity: A Mathematical Study of the Causes and Origins of War. Pittsburgh: Boxwood Press. Ripley, B. D. 1996. Pattern Recognition and Neural Networks. Cambridge: Cambridge University Press. Risse-Kappen, Thomas. 1991. “Public Opinion, Domestic Structure, and Foreign Policy in Liberal Democracies.” World Politics 43(4): 479–512. Rosato, Sebastian. 2003. “The Flawed Logic of Democratic Peace Theory.” American Political Science Review 97(4): 585–602. Rosecrance, Richard. 1986. The Rise of the Trading State: Commerce and Conquest in the Modern World. New York: Basic Books, Inc., Publishers. Rosenau, James N. 1990. Turbulence in World Politics: A Theory of Change and Continuity. Princeton: Princeton University Press. Ross, Thomas W. 1994. “Raising an Army: A Positive Theory of Military Recruitment.” Journal of Law and Economics 37(1): 109–131. Rousseau, David Laurent. 1996. Domestic Political Institutions and the Evolution of International Conflict. Ph.D. Dissertation. University of Michigan. Russett, Bruce. 1990–1991. “Doves, Hawks, and U.S. Public Opinion.” Political Science Quarterly 105(4): 515–538. ———. 1993. Grasping the Democratic Peace: Principles for a Post-Cold War World. Princeton: Princeton University Press. ———. 1998. “A Neo-Kantian Perspective: Democracy, Interdependence, and International Organizations in building Security Communities.” In

References 189

Emanuel Adler and Michael Barnett (Eds.), Security Communities. Cambridge, UK: Cambridge University Press. 368–394. ——— and John R. Oneal. 2001. Triangulating Peace: Democracy, Interdependence, and International Organizations. New York and London: W.W. Norton & Company. ——— and David R. Davis. 1998. “The Third Leg of the Kantian Tripod for Peace: International Organizations and Militarized Disputes, 1950–85.” International Organization 52(3): 441–467. Russett, Bruce M. and Harvey Starr. 2000. “From Democratic Peace to Kantian Peace: Democracy and Conflict in the International System.” In Manus I. Midlarsky (Ed.), Handbook of War Studies II. Ann Arbor, MI: University of Michigan Press. 93–128. Sandler, Todd. 2001. Economic Concepts for the Social Sciences. Cambridge and New York: Cambridge University Press. Sarkees, Meredith Reid. 2000. “The Correlates of War Data on War: An Update to 1997.” Conflict Management and Peace Science 18(1): 123–144. Sarkesian, Sam C. 1992. “Civil–Military Relations in a Liberal Society: The United States in a New Security Era.” Paper presented at the OSU Mershon Center Conference on Civil–Military Relations, Columbus, OH, December 4–6. SAS/STAT® User’s Guide, Version 7–1. 1999. V.5. Schofield, Julian. 2000. “Militarized Decision-Making for War in Pakistan: 1947–1971.” Armed Forces & Society 27(1): 131–148. Schrodt, Philip A. 1991. “Prediction of Interstate Conflict Outcomes using a Neural Network.” Social Science Computer Review 9(3): 359–380. ———. 2000. “Pattern Recognition of International Crises using Hidden Markov Models.” In Diana Richards (Ed.), Political Complexity: Nonlinear Models of Politics. Ann Arbor, MI: University of Michigan Press. 296–328. Schultz, Kenneth A. 1998. “Domestic Opposition and Signaling in International Crises.” American Political Science Review 92(4): 829–844. Schumacher, Rose, Gail K. Sevrens, Timothy S. O’Donnell, Lee Torrence, and Kate Carney (Eds.). 1989. World Defense Forces: A Compendium of Current Military Information for All Countries of the World, 2nd edn. Santa Barbara, CA and Oxford, England: ABC-CLIO. Shaw, Martin. 2000. “Media and Public Sphere without Borders?: News Coverage and Power from Kurdistan to Kosovo.” In Brigitte L. Nacos, Robert Y. Shapiro, and Pierangelo Isernia (Eds.), Decisionmaking in a Glass House: Mass Media, Public Opinion, and American and European Foreign Policy in the 21st Century. Lanham, Boulder, New York, and Oxford: Rowman & Littlefield Publishers, Inc. 27–40. Shin, Doh C. 1994. “On the Third Wave of Democratization: A Synthesis and Evaluation of Recent Theory and Research.” World Politics 47(1): 135–170.

190 References Singer, J. David and Melvin Small. 1966. “The Composition and Status Ordering of the International System: 1815–1940.” World Politics 18(2): 236–282. Small, Melvin and J. David Singer. 1973. “The Diplomatic Importance of States, 1816–1970: An Extension and Refinement of the Indicator.” World Politics 25(4): 577–599. ———. 1976. “The War-Proneness of Democratic Regimes, 1816–1965.” Jerusalem Journal of International Relations 1(4): 50–69. Smith, Alastair. 1998. “International Crises and Domestic Politics.” American Political Science Review 92(3): 623–638. Snider, Don M. and Miranda A. Carlton-Carew. 1995. “The Current State of U.S. Civil–Military Relations: An Introduction.” In Don M. Snider and Miranda A. Carlton-Carew (Eds.), U.S. Civil–Military Relations: In Crisis or Transition? Washington, DC: The Center for Strategic & International Studies. Snyder, Jack. 1984. “Civil–Military Relations and the Cult of the Offensive, 1914 and 1984.” International Security 9(1): 108–146. ———. 2002. “Anarchy and Culture: Insights from the Anthropology of War.” International Organization 56(1): 7–45. Souva, Mark and Brandon C. Prins. 2001. “International Trade and Militarized Dispute Initiation, Reciprocation, and Revision, 1950–1992.” Paper presented at the Annual Meeting of the International Studies Association, Chicago, IL. Starr, Harvey. 1992. “Why don’t Democracies fight One Another?: Evaluating the Theory-Findings Feedback Loop.” Jerusalem Journal of International Relations 14(4): 41–59. ———. 1997. “Democracy and Integration: Why Democracies don’t fight Each Other.” Journal of Peace Research 34(2): 153–162. Stata Reference Manual Set. 2001. Stata Statistical Software (Version 7.0). Stockholm International Peace Research Institute. 1985. World Armaments and Disarmament: SIPRI Yearbook 1985. London and Philadelphia: Taylor & Francis. Stoessinger, John G. 2001. Why Nations Go to War, 8th edn. Boston and New York: Bedford/St. Martin’s. Vagts, Alfred. 1956. Defense and Diplomacy: The Soldier and the Conduct of Foreign Relations. New York: King’s Crown Press. ———. 1959. A History of Militarism: Civilian and Military. New York: The Free Press. Van Belle, Douglas A. 1997. “Press Freedom and the Democratic Peace.” Journal of Peace Research 34(4): 405–414.

References 191

———. 2000. Press Freedom and Global Politics. Westport and London: Praeger. ——— and John Oneal. 2000. “Press Freedom and Militarized Disputes.” In Douglas A. Van Belle (Ed.), Press Freedom and Global Politics. Westport, CT and London: Praeger. 47–75. ———, Jean-Sebastien Rioux, and David M. Potter. 2004. Media, Bureaucracies, and Foreign Aid: A Comparative Analysis of United States, the United Kingdom, Canada, France and Japan. New York: Palgrave Macmillan. Van Evera, Stephen. 1984. “The Cult of the Offensive and the Origins of the First World War.” International Security 9(1): 58–107. Vasquez, John A. 1993. The War Puzzle. Cambridge and Victoria: Cambridge University Press. ——— (Ed.). 2000. What do We Know about War? New York: Rowman & Littlefield Publishers, Inc. Waltz, Kenneth N. 1959. Man, the State, and War: A Theoretical Analysis. New York: Columbia University Press. ———. 1979. Theory of International Politics. Reading: Addison-Wesley. Werbos, Paul John. 1994. The Roots of Backpropagation: From Ordered Derivatives to Neural Networks and Political Forecasting. New York, Chichester, Brisbane, Toronto, and Singapore: John Wiley & Sons, Inc. White, Michael D. 1989. “Conscription and the Size of Armed Forces.” Social Science Quarterly 70(3): 772–781. Whitten, Guy D. and Henry S. Bienen. 1996. “Political Violence and Time in Power.” Armed Forces & Society 23(2): 209–234. Wilkenfeld, Jonathan, Gerald W. Hopple, Paul J. Rossa, and Stephen J. Andriole. 1980. Foreign Policy Behavior: The Interstate Behavior Analysis Model. Beverly Hills: Sage Publications. Wolfson, Murray, Patrick James, and Eric Solberg. 1998. “In a World of Cannibals Everyone Votes for War: Democracy and Peace Reconsidered.” In Murray Wolfson (Ed.), The Political Economy of War and Peace. Boston, Dordrecht, and London: Kluwer Academic Publishers. 155–176. Yun, Philip W. 2003. “Facing the Boogeyman: A Korean American Diplomat Recounts His Trips to North Korea.” The Korean Society Quarterly 4(1): 6–14. Zaller, John and Dennis Chiu. 2000. “Government’s Little Helper: U.S. Press Coverage of Foreign Policy Crises, 1946–1999.” In Brigitte L. Nacos, Robert Y. Shapiro, and Pierangelo Isernia (Eds.), Decisionmaking in a Glass House: Mass Media, Public Opinion, and American and European Foreign Policy in the 21st Century. Lanham, Boulder, New York, and Oxford: Rowman & Littlefield Publishers, Inc. 61–84.

192 References Zeng, Langche. 1999. “Prediction and Classification with Neural Network Models.” Sociological Methods & Research 27(4): 499–524. ———. 2000. “Neural Network Models for Political Analysis.” In Diana Richards (Ed.), Political Complexity: Nonlinear Models of Politics. Ann Arbor, MI: University of Michigan Press. 239–268. Zorn, Christopher J. W. 2001. “Generalized Estimating Equation Models for Correlated Data: A Review with Applications.” American Journal of Political Science 45(2): 470–490. Zuk, Gary and William R. Thompson. 1982. “The Post-Coup Military Spending Question: A Pooled Cross-Sectional Time Series Analysis.” American Political Science Review 76(1): 60–74.

Index Afghanistan, 17 alliances, 9, 48, 57, 59, 62, 65, 69, 72, 74, 93–94, 101, 103, 110–111, 120–121, 123, 125–126, 141, 143, 157n, 165n, 168n anarchy, 98, 149 autocracy, 11, 20, 67, 131, 164n, 166n Axis of Evil, 84 Beck, Nathaniel, 11–12, 52–53, 55–56, 58, 63, 66, 70–71, 77, 99–100, 122, 125, 166n, 167n, 169n, 173n Betts, Richard K., 27, 60 Bonaparte, Napoleon, 40, 116 Brecher, Michael, 3, 9, 26, 39, 150 Bueno de Mesquita, Bruce, 13, 25, 161–162n Bush, George W., 5, 55, 79–88, 150 Bush Administration diplomacy, 81–87 Carter, Jimmy, 80 China, 46, 86–87 conflict over U.S. spy plane, 159n Choi and James, 10, 157n, 161n civil liberties, 130, 133–134, 166n civilians civilian hawks, 27, 159 influence, 113, 151, 162–163n supremacy, 32, 150, 160 see also leadership, civilian civil–military dynamics, 2–3, 5, 31, 49, 55–56, 58, 61, 63, 71, 77, 91–92, 100, 148, 151–152, 157n, 158n, 160n

civil–military relations, 4, 10, 23, 25–26, 28–29, 37, 39–40, 43–45, 48–50, 56, 58–61, 63, 66, 71–72, 77, 95, 97–100, 109–115, 117–118, 121–122, 126, 148, 150–151, 154–155 and neo-Kantian peace, 118–127 measurement of military influence, 112–118 Clinton, Bill, 81–82 Cold War era, 9, 16, 36, 129 communications, 1–3, 5, 8, 24–25, 28–31, 33, 36, 42, 49, 56, 63–64, 66, 77, 88, 91–94, 100, 136, 149, 151–153 conflict management, 25, 161 conscription, 1, 31, 40–41, 45, 50–51, 58, 63, 71–72, 74, 100–103, 110–117, 120, 149–150 contiguity, 25, 49–51, 57, 59–60, 62, 65, 69, 72–74, 77, 93–94, 97, 101–103, 105, 110–111, 120–121, 123, 125–126, 137–138, 141, 143 covert intervention, 7, 10 Cuban Missile Crisis, 24–25, 28 cultural norms, 8 decision-making, 3–7, 21, 23–24, 26, 29–30, 32, 34, 36–37, 39, 48, 53, 63, 91, 95, 98, 102, 110, 148 democracy, 6, 8, 10–11, 16–17, 20, 57, 59, 62, 64, 66–70, 76–77, 91, 93–94, 96, 99, 105, 120, 123, 125, 129–134, 137–138, 140–145, 149–150

194 Index democracy—continued high democracies, 19–20 institutional attributes of, 11 institutional democracy, 138, 140–145, 149 democratic peace, 1–4, 7, 9–10, 13–14, 16, 19–21, 23, 29, 36–37, 47–48, 97, 116, 126, 131–134, 138, 142, 145, 151 and its critics, 7–14 recent evolution of, 14–21 studies, 1, 4, 9, 13–14, 19–20, 23, 36, 130, 132 democraticness, 7, 10–11, 14–17, 19, 25, 47–48, 50–51, 56, 58, 60–61, 63, 66–68, 73–74, 76–78, 95–98, 100, 102–103, 105–106, 110–111, 120, 122, 125–126, 149–150 democratization, 4, 7, 9–10, 20, 96, 139, 150 dictatorships, 8, 27, 88, 153 diplomacy, 5, 31, 34, 41, 55, 79, 81, 87–88, 149–151 importance of, 46, 61 diplomatic channels, 1, 3–5, 13, 21, 23, 29–30, 33, 36–37, 42, 48–51, 55–56, 61–64, 66, 69, 71–74, 76–79, 81, 87, 89, 92–93, 95–96, 98–103, 106, 148–151, 154 distributed-lags, 12, 15–16, 18 Doder, Dusko, 35 doves, 26–29, 31, 60, 118, 159 see also hawks; militarism, military doves Duindam, Simon, 40, 115

95, 97–103, 105–107, 120, 121, 125–126, 129, 137–138, 140, 142–144 Egypt, 4, 33–34 empirical implications, 109, 122, 128, 140 empirical results, 5–6, 10, 15, 18–19, 21, 55–56, 58, 60–61, 63, 77, 92–93, 97, 99, 105, 121–122, 140, 142–144, 148–149 Enterline, Andrew J., 19–20 European Union (EU), 9, 45, 114, 119

economic interdependence, 1, 8, 15, 17, 20, 47, 49–51, 56, 58, 60–61, 63, 66, 72–74, 76, 78,

hawks, 26, 28, 32, 60, 118, 159n see also civilians, civilian hawks; doves

Fearon, James D., 13–14, 161–162n Feaver, Peter D. and Christopher Gelpi, 26–27, 97 fixed effects model, 16 foreign policy, 2–7, 10, 12–13, 16, 18, 21, 23–27, 29, 32–34, 36–37, 39–40, 44, 48, 53, 63, 73, 76, 88, 91–92, 95, 98, 102, 115, 134–136, 139, 142, 144, 148–150, 152, 154, 156 foreign policy decision-making process model, 23–29 flowchart of, 29, 30, 31–33 historical examples, 33–36 foreign policy outcome, 12, 156 frame of reference, 24 Gallarotti, Giulio M., 18 GEE model, 56, 58–59, 61, 63, 126 geographic distance, 48–51, 110–111, 121, 125, 137 global press freedom, 43, 46, 137 globalization, 147 Graber, Doris A., 42

Index

Hollywood depiction of international relations, 24, 31 human rights, 85 Hussein, Saddam, 17, 152–153 International Atomic Energy Agency (IAEA), 79–81, 84–85 international conflict, 1–3, 6, 10–11, 13, 18, 20, 23, 37, 39–40, 42, 45, 47, 61, 67, 70, 77, 96, 116, 130–132, 134–136, 144–145, 148, 151 International Crisis Behavior (ICB) Project, 39, 150 international disputes, 12, 14, 28, 41, 49, 97, 157 international organizations, 8, 18, 49, 51, 56, 58, 61, 63, 68, 72–74, 77, 95, 97–98, 101, 103, 105, 110, 121, 125–126, 129, 137–138, 143–145 joint membership in, 7–8, 14, 18, 20, 47, 49–51, 56, 58, 68, 77, 91, 97, 105, 110–111, 120–121, 125, 129, 138, 142–144 international peace, 6, 55, 78, 106, 149 international threat, 112–114, 116–118, 120, 156 Iraq, 17, 152–154, 160 invasion of, 152–153 Israel, 4, 27, 33–34, 153 response to Egyptian military actions, 33–34 Jefferson, Thomas, vi, 3 Johnson, Lyndon, 34 Kennedy, John F., 24 Kennedy, Robert F., 24–25

195

Korea see North Korea Korean Air Lines flight 007 (KAL 007), 4, 34–36 Kosovo, media coverage of conflict in, 139 Landy, Marc, 152–153 Lasswell, Harold D., 26, 39, 157 leadership, 24, 28, 36, 136, 161–162 civilian, 24–31, 33–34, 36, 41, 43–44, 118, 159n degree of influence, 43–46 military, 13, 24–31, 34–36, 39, 44, 58, 60, 71, 73, 96, 100, 102, 110, 112–118, 122, 159n liberal peace, 16, 157n Lincoln, Abraham, 132 logistic regression model, 49, 55, 67, 71, 73, 92, 102, 137, 139 logit model, 11, 53, 70, 76, 105 Luttwak, Edward N., 91, 139 major powers, 50, 60–61, 72, 74, 97, 101, 103, 121, 125 Mansfield, Edward D. and Jack Snyder, 10, 96 Maoz, Zeev, 8, 19–20, 24, 31, 92, 117, 120, 139 mass media, 31, 42, 46, 91, 96, 134–139, 142 media openness, 3–4, 6–7, 21, 23, 29–31, 33, 36–37, 42–43, 46, 48–50, 66–67, 70, 72–74, 76–77, 89, 92, 95, 97, 99–101, 103, 105, 129–132, 134–139, 142–145, 148–153 and democracy/interstate disputes, 129–136

196 Index media openness—continued power of, 139–145 role of, 137–139 Menard, Scott, 68 militarism military doves, 27, 159 military expenditure, 5–6, 43–45, 51, 109, 113–120, 148–149 military fatalities, 91, 96, 138–139, 143 military influence, 32, 40, 44, 76, 102, 109–110, 112–115, 117–120, 121–122, 148, 150 military preparedness, 60 societal militarism, 110, 112–114, 117, 119, 155 Militarized Interstate Disputes (MIDs), 2, 5, 19–20, 29, 31–33, 36, 40–43, 45, 48, 55–89, 110–112, 118, 120, 121–122, 125–128, 137–144, 148–151 logistic regression analysis of, 55–70 MID involvement, 5, 48, 55–58, 60–61, 63, 66–67, 70–73, 76–77, 92, 97, 100, 111–112, 121–122, 125–127, 137–141, 143–144 neural network analysis of, 70–78 Militarized Interstate Disputes (MIDs), fatal, 5, 89, 91–92, 95–100, 102, 105, 130, 138–139, 141–143, 144 logistic regression analysis of, 92–99 neural network analysis of, 99–107 military manpower system, 3, 29, 37, 41, 45, 48–49, 56, 58, 60–61, 63–64, 66, 69, 71,

76–77, 92, 95–96, 98–100, 102, 115, 117, 149, 151, 154–155 misperception, 13, 42, 136, 151 Morgan, T. Clifton, 26, 39, 132 multicollinearity, 67–70 multipolarity, 57, 59, 63, 122, 126 national capability ratio, 48–51, 120–121, 138 negotiation, 1, 7, 28, 36, 79, 81, 84, 88, 98 see also diplomacy neo-Kantian peace, 2, 9, 18, 20–21, 47, 63, 66, 71, 77, 98–100, 105, 110, 118, 122, 125–126, 129–130, 137, 141–142, 144, 148–149 neural network analysis, 5, 12, 70, 99, 105 noncontiguous, 49–51, 121, 138 North Korea (DPRK), 46, 55, 79–88, 98, 150 Agreed Framework (AF), 81–85 diplomacy with U.S., 79–89 Light Water Reactors (LWRs), 82–83 nuclear program, 5, 55, 79, 85–87 Oneal, John R. and Bruce Russett, 8, 14–16, 20, 47–51, 58, 61, 63, 67, 76, 92, 97, 118, 120–122, 126, 129, 137–139, 140, 147 only minor powers, 49–51, 57, 59, 62, 65, 69, 93–94, 111, 121, 125, 138, 140–142, 143 peace see democratic peace; international peace; liberal peace; neo-Kantian peace

Index

peace years correction model, 17, 56–59, 63, 66, 77, 125–126 Perpetual Peace (Kant), 7, 147, 154 political communication, 1–3, 8, 25, 28, 30–33, 36, 42, 49, 64, 66, 77, 88, 91–94, 99–100, 136, 149, 151 political opposition, 161 Polity, 11, 16, 36, 46, 66–68, 77, 97, 105, 129–132, 134, 138, 140, 143–144, 149 Powell, Colin, 27–28, 60, 82, 84, 96 Powell Doctrine, 27, 60 power shift, 25 public opinion, 42, 91, 139 rational choice theory, 3–4, 9, 13–14 Ray, James Lee, 2, 8, 10, 19–20, 133 realism, 9–10, 98 Rosato, Sebastian, 2, 21, 147 Russett, Bruce and John R. Oneal, 1–2, 5, 8, 10, 17, 32–33, 40, 47, 116, 126, 147–148, 154 Sarkesian, Sam C., 112, 155–156, 170

197

security, 5, 9–10, 14, 43, 83–84, 86, 96, 109–110, 112, 116–118, 126–127, 149, 155–156 sensitivity, 8, 17, 73, 102 September 11, 83 Singer, J. David and Melvin Small, 31, 41, 46, 61 Snyder, Jack, 10, 28, 60, 96 South Korea, 79, 82, 84, 86–87 Starr, Harvey, 8, 130–131 Triangulating Peace (Russett and Oneal), 1–2, 47 United Nations, 24 United States, 2, 8, 16–17, 25, 28, 34, 44, 46, 55, 60, 79–88, 97, 116, 151–152, 155 Van Belle, Douglas A., 43, 46, 130, 137 volunteer army, 160–161n vulnerability, 8, 17 Waltz, Kenneth N., 10, 25 World War I, 9, 28, 40, 96 World War II, 9, 96