Persons in Context: Developmental Processes (Human Development in Cultural and Historical Contexts)

HUMAN DEVELOPMENT IN CULTURAL AND HISTORICAL CONTEXTS

Persons in context

HUMAN DEVELOPMENT IN CULTURAL AND HISTORICAL CONTEXTS General Editor: Urie Bronfenbrenner Associate Editor: Glen H. Elder, Jr.

Adolescent mothers in later life Frank F. Furstenberg, Jr., J. Brooks-Gunn, and S. Philip Morgan

Persons in context Developmental processes Edited by NIALL BOLGER, AVSHALOM CASPI, GERALDINE DOWNEY, MARTHA MOOREHOUSE

The right of the University of Cambridge to print and sell all manner of books was granted by Henry VIII in 1534. The University has printed and published continuously since 1584.

CAMBRIDGE UNIVERSITY PRESS Cambridge New York New Rochelle

Melbourne

Sydney

CAMBRIDGE UNIVERSITY PRESS Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sao Paulo Cambridge University Press The Edinburgh Building, Cambridge CB2 2RU, UK Published in the United States of America by Cambridge University Press, New York www.cambridge.org Information on this title: www.cambridge.org/9780521355773 © Cambridge University Press 1988 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 1988 This digitally printed first paperback version 2007 A catalogue record for this publication is available from the British Library Library of Congress Cataloguing in Publication data Persons in context : developmental processes / edited by Niall Bolger . . . [et al.]. p. cm. - (Human development in cultural and historical contexts) Includes index. ISBN 0 521 35577 X 1. Child development. 2. Socialization. 3. Child psychology. 4. Child psychobiology. I. Bolger, Niall. II. Series. RJ131.P39 1988 303.3'2-dcl9 87-36765 ISBN-13 978-0-521-35577-3 hardback ISBN-10 0-521-35577-X hardback ISBN-13 978-0-521-03584-2 paperback ISBN-10 0-521-03584-8 paperback

Contents

Preface Contributors

page vii xii

1.

Development in context: research perspectives Niall Bolger, Avshalom Caspi, Geraldine Downey, and Martha Moorehouse

2.

Interacting systems in human development. Research paradigms: present and future Urie Bronfenbrenner

25

Children, families, and communities: ways of viewing their relationships to each other Jacqueline J. Goodnow

50

Human development and social change: an emerging perspective on the life course Glen H. Elder, Jr., and Avshalom Caspi

77

3.

4.

1

5.

Family process: loops, levels, and linkages Gerald R. Patterson

114

6.

On the constructive role of problem behavior in adolescence Rainer K. Silbereisen and Peter Noack

152

7.

The sociogenesis of self concepts Robert B. Cairns and Beverly D. Cairns

181

8.

Putting persons back into the context Daryl J. Bern

203

9.

How genotypes and environments combine: development and individual differences 217 Sandra Scarr

vi

Contents Author index Subject index

245 251

Preface

This volume is a product of a study group convened by the Society for Research in Child Development at Cornell University to assess the progress that social scientists have made in understanding the processes linking persons and contexts in the course of development. The contributors to this volume represent the disciplines of developmental, personality, and clinical psychology; behavioral genetics; and sociology. They are also identified with a wide range of methodological approaches including longitudinal studies, laboratory experiments, field observations, and the sequential analysis of social interactions. In addition, each contributor has made a distinctive theoretical and empirical contribution to an interactional perspective on human development. In preparation for the conference, the participants were asked to circulate working papers summarizing their approaches. To stimulate discussion during the conference, a less usual procedure was adopted. Each participant presented the viewpoint of another group member rather than a commentary on his or her own work. The successful dialogue that emerged in the course of the conference continues in this volume. The central theme of the volume is how to relate different environmental contexts to one another and to individuals in the course of development. In the first chapter, Niall Bolger, Avshalom Caspi, Geraldine Downey, and Martha Moorehouse provide a map of the territory to be explored as they enumerate the various linkages between different environmental influences and individual development. Environmental influences range from social conditions at the macro level, like a prosperous or depressed national economy, to the immediacy of a child's playful exchange with a parent at the micro level. Chapter 1 reviews different perspectives on the processes that operate within and between these different levels and links them to one another and to other recent advances in the study of development in context. The chapters that follow gradually shift from macro to micro levels of environmental analysis. Earlier chapters concentrate on the stable and changing vii

viii

Preface

aspects of the macrosocial environment and how they influence persons in their proximal environments. Later chapters focus on developmental processes operating at the proximal level, bringing into the foreground the daily lives of individuals, the activities they pursue, their interactions with one another, and their personal attributes. The volume concludes with a discussion of the contributions that individuals make to their own development as they select and shape the very environments they experience. This organizational framework allows the contributors to provide a variety of perspectives and empirical accounts of the essential features of the developing person, the changing environment, and the processes that link them across time. Urie Bronfenbrenner describes existing and emerging research paradigms and discusses their strengths and weaknesses as tools for analyzing development in context. He argues that researchers must make use of more complex conceptual schemes that have the power to explicate relationships among person, process, and context. Especially needed, in his view, are "person-process-context" models that can determine whether a developmental process, like the facilitation of the child's development by parent-child activities, varies as a function of both the context and the characteristics of the individuals involved. Equally important are designs that reveal the effects of stability and change in environmental systems by incorporating a temporal dimension in their analysis of the environment. Bronfenbrenner further reminds us that we need to be equally evenhanded in our treatment of the individual. Calling for new conceptions of developmental outcomes that are integrative and sufficiently complex, he outlines requirements for both a subjective, experiential component and an objective, behavioral component. Jacqueline Goodnow pursues this theme in chapter 3. She considers how children's access to different settings, actual and perceived, changes with development and is facilitated or constrained by the nature of their communities. She critically evaluates some widely held but untested assumptions about desirable relationships among children, their families, and their communities. These assumptions include the ideas that exposure to a wide variety of settings enhances development, continuity between settings is desirable, and the world is open for exploration and enjoyment by every individual. She then examines the developmental processes underlying the relationships among children, their families, and their communities and emphasizes models that are useful in conceptualizing the nature of access between people and settings. Several questions receive special attention here: Is access to information and to settings restricted or freely available within the larger community? How do families conceive of the world outside the home and how do they convey these beliefs to their children? What contributes to individual differences and developmental changes in children's perception of access to settings and social roles? In chapter 4, Glen Elder and Avshalom Caspi bring a life-course perspective

Preface

ix

to their analysis of human development in a changing society as they consider how historical contexts shape individual life patterns and how individual development is related to social transitions in the age-graded life course. Examining social changes in the Great Depression, they document the antecedents and lifecourse implications of problem behavior for children growing up in stressful times. Their model of control cycles in the evolving life course details some of the complex causal linkages between changing social conditions, dynamics of family relationships, personality characteristics of family members, and the later personality development and life patterns of these individuals. In particular, this work identifies factors that mediate and moderate the influence of stressful times on children's lives, features that are also examined and elaborated in chapter 5 by Patterson. Gerald Patterson is not only a scientist who studies self-sustaining destructive interactions but a social engineer whose clients expect him to intervene and deliberately produce discontinuity and change. His chapter provides a methodological framework that illustrates conceptual points about the development and maintenance of family interaction patterns. It also shows how analyzing the microdynamics of social interactions in the family can help describe and explain enduring patterns of antisocial behavior. This perspective on development raises issues regarding the factors that elicit specific instances of action and reaction, what Patterson has termed microsocial processes. His model incorporates variables outside the immediate family environment and connects these to family processes in order to account for behavioral change along a discernible trajectory. Patterson's data also support the notion, advanced by several other contributors to this volume, that children's difficulties in one context are likely to produce difficulties in other contexts. In this chapter he elucidates the mechanisms underlying these developmental progressions. Rainer Silbereisen and Peter Noack's conceptual framework, "development as action in context," denotes a model of development that entails two critical issues: (a) Development is seen as the outcome of a person's own intentional, goal-oriented action aimed at adjusting individual goals and potentials to contextual demands and opportunities; (b) such action produces not only change in the individual, but change in the context of development as well. The contextual changes thus induced continually provide opportunities for new action aimed at further development. In chapter 6, these authors apply this distinctive European "look" to a developmental phenomenon of worldwide concern: the interplay between person and context in the psychogenesis of substance use in adolescence. Using a multilevel assessment strategy that integrates field and observational data with a cohort-sequential design, Silbereisen and Noack examine the relationship between self-esteem, cathartic leisure-time settings, and substance use in early and middle adolescence. Robert Cairns and Beverly Cairns consider the question, "Why would a

x

Preface

woman who is destitute, elderly, ill, widowed, hungry, black, and living in abject circumstances in an urban ghetto, report high life satisfaction?" "Why not?" sums up their response. That is, why do we assume that convergence between self-evaluations and the evaluations of others is normative and adaptive? The authors present a strong case for an alternative view based on a developmental analysis of the functions of social cognitions concerning the self. Their core premise is that adaptation is not synonymous with accuracy. Instead, modestly inflated views of one's competence may serve to enhance both well-being and interpersonal adaptations as they provide room for plans, dreams, as well as recourse from discouragement. When, then, would correspondence occur? Again, when it is developmentally functional for the individual. In their chapter, the Cairnses summarize propositions concerning when and why reports from various sources converge or diverge and draw on empirical evidence from their unique multimethod, multimeasure, longitudinal study of childhood and adolescence. Daryl Bern argues that development cannot be understood fully unless one adopts a perspective that attends to the role of the person's differentiating psychological characteristics. A personologist, Bern argues that our fundamental scientific task is to convert observations of particular persons behaving in particular ways in particular situations into assertions that certain kinds of people will behave in certain kinds of ways in certain kinds of situations. In chapter 8, he presents a commentary on research paradigms in personality psychology that complements Bronfenbrenner's commentary on research paradigms in developmental psychology, and concludes by providing guidelines for renewed attention to personological concerns in developmental research. Putting the person back into the context, Bern argues for a morphogenic theory of response-style variables that could explicate how the person's intrapersonal patterning of styles changes and develops in interaction with particular kinds of environments over time. In the final chapter, Sandra Scarr also addresses the problem of individuality and development by posing the question, "How do organisms and environments combine to produce human development?" Scarr suggests that to understand development and individuality we need a theory about the ways in which people determine their own experiences. She proposes that the individual's genotype both determines the impact that environments will have on him or her and, increasingly during the course of development, drives the individual's own selection of environments. From infancy to adolescence, children become increasingly effective in creating environments that are compatible with their motivations, needs, and abilities. Moreover, Scarr takes issue with the proposition that genotype-environment interactions have major importance in explaining variability and proposes instead that genotype-environment correlations have

Preface

xi

greater effects on development and individuality. According to this view, the responses that individuals elicit and evoke from others and the actions they take to select and shape their environments are correlated with their genotypes. Indeed, although the search for aptitude-treatment interactions as well as genotype-environment interactions has produced uneven and unreliable results, evidence accumulating in the field of behavior genetics provides a good deal of support for Scarr's proposed model. In combination, the contributions to this volume provide a useful beginning for new considerations of how persons and contexts are linked in the course of development. They show that we need not decide between the study of development out of context and the study of context without development. They confirm the promise and importance of studying development in context. A successful group endeavor requires a gathering of open-minded, curious, and energetic conferees. This they were. It also required the support and assistance of Gerri Jones, who managed the logistics of the conference. Finally, we wish to acknowledge the support of the Society for Research in Child Development in bringing the conference and this volume to fruition. Niall Bolger Avshalom Caspi Geraldine Downey Martha Moorehouse

Contributors

Daryl J. Bern Department of Psychology Uris Hall Cornell University Ithaca, NY 14853

Beverly D. Cairns Department of Psychology Davie Hall University of North Carolina Chapel Hill, NC 27514

Avshalom Caspi (coeditor) Department of Psychology William James Hall Harvard University Cambridge, MA 02138 Geraldine Downey (coeditor) Institute for Social Research University of Michigan Ann Arbor, MI 48106 Glen H. Elder, Jr. Carolina Population Center University of North Carolina Chapel Hill, NC 27514 Jacqueline J. Goodnow School of Behavioural Sciences Macquarie University Sydney N.S.W. 2109 Australia Martha Moorehouse (coeditor) Department of Psychology University of California Santa Cruz, CA 95064

Robert B. Cairns Department of Psychology Davie Hall University of North Carolina Chapel Hill, NC 27514

Peter Noack Institute for Psychology Hardenbergstrasse 28 Technical University of Berlin D-1000 Berlin 12 Federal Republic of Germany

Niall Bolger (coeditor) Institute for Social Research University of Michigan Ann Arbor, MI 48106 Urie Bronfenbrenner Department of Human Development and Family Studies MVR Hall Cornell University Ithaca, NY 14853

xn

Contributors Gerald R. Patterson Oregon Social Learning Center 207 East 5th Avenue, Suite 202 Eugene, OR 97401 Sandra Scarr Department of Psychology Gilmer Hall University of Virginia Charlottesville, VA 22903

xiii Rainer K. Silbereisen Department of Psychology Otto-Behaghel-Strasse 10 University of Giessen D-6300 Giessen Federal Republic of Germany

Development in context: research perspectives

Niall Bolger, Avshalom Caspi, Geraldine Downey, and Martha Moorehouse

Development in context Individuals develop in an ever-changing environment, and students of human development must do more than simply acknowledge the interaction between the two. The contributors to this volume assume the task of translating this homily into an integrative perspective on how individuals and environments are linked in the course of development. In this introductory chapter, we outline the orienting principles that inform the study of development in context and that also unify the contributions to this volume. We begin with a representation of the environment in which development takes place and then examine the role of individuals as producers of their development. Representations of the environment The environment in which individuals develop consists of a complex system of physical, social, cultural, and historical factors that interact with each other and with the developing individual. Most psychologists agree that these factors are influential in regulating development, yet they persist in viewing the environment as an undifferentiated set of factors. The difficulty with this position can be appreciated best by considering where developmental theory would be if we failed to distinguish between cognitive, affective, and behavioral sources of individual change. Would a theory of development that details the structure and process of environmental sources of change add to our understanding of development? Several theorists have suggested as much. The most consistent advocate of a theory of development in context is Bronfenbrenner (1979, this volume). Following Lewin's (1943) field theory and a general systems approach (von Bertalanffy, 1969), Bronfenbrenner has devised an ecological model for the systematic analysis of environmental influences on development. 1

2

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

Bronfenbrenner conceptualizes developmental contexts as nested, hierarchical structures with influences ranging from those located in the distal, macrolevel setting (e.g., class, culture) to those located in the proximal setting (e.g., family, school, work). The macrosystem comprises the social, cultural, and political structures of society - including its laws, norms, and customs - that define the character, structure, and functioning of proximal settings. According to Bronfenbrenner, three types of proximal settings can be distinguished effectively: the exosystem, microsystem, and mesosystem. The exosystem consists of face-to-face settings in which the individual does not actively participate but that can affect or be affected by the developing person (e.g., the parent's workplace). The microsystem is that part of the environment that an individual is in contact with and can interact with directly in daily life during a certain period of time (e.g., the family, school, leisure setting). It contains those settings in which people can readily engage in face-to-face interaction, and these patterns of interaction, as they persist and evolve through time, constitute the vehicles of behavioral change and individual development. The mesosystem consists of relationships between major microsystem settings at a particular point in the individual's development (e.g., the home-school relationship as it affects children's behavior). This basic conception introduces more conceptual rigor to what is often viewed by psychologists as an amorphous and unclearly differentiated set of influences. It implies that individuals are not independent and that, over and above any commonalities at the biological level, they share experiences and influences owing to their membership in higher-level systems, such as families, organizations, communities, societies, and historical periods. Indeed, the surprises that emerge when we examine the course of development in different cultural settings and in different social strata continue to reinforce the point that developmental processes cannot be fully understood without information on the many levels of social organization in which human action and development are embedded (LeVine, 1981). Bronfenbrenner's ecological model provides a useful guide, offering several analytic challenges that are addressed in this volume. Our first task is to examine the linkages between different levels of the environment. Assuming a hierarchical organization of the environment, we can expect influences on development to originate at different levels of analysis. For example, all individuals are embedded in a proximal system that is subject to change over time, the family. And the family, in its functioning as a socializing entity, also depends on certain characteristics of the larger social system in which it is embedded. Psychologists, however, often avoid probing how these higherlevel structures influence lives. Of course, before we can understand developmental processes, we must understand each level in terms of its own principles of organization. Proximal

Development in context: research perspectives

3

settings, such as schools and families, have unique influences on development that cannot be explained solely in terms of the higher-level social structures in which they are embedded or in terms of the characteristics of the individuals in the immediate setting. Thus, the second challenge for developmentalists is to account for and describe the unique effects of these proximal settings. A third challenge is to determine how development is affected by the features of the immediate environment, which covary because they are part of the same macrosystem. For example, neighborhood and school quality, family structure, and the quality of parent-child interactions are likely to covary because of the common influence of a higher-level system in which they are embedded: social class. Considered individually, each of these factors may predict certain developmental outcomes. But these factors come as a package, and therefore we must examine the joint effects of multiple, simultaneous environments on development. It might be said at this point that we are staging a play that has all sets but no character parts. Of course, a realistic theory of development in context should lead us to expect active individuals who possess the capacity to influence their own development. A fourth challenge, then, is to determine how individuals shape the very settings that influence their development and psychological growth. A fifth challenge is to deal with the fact that both persons and environments change over time. Developmentalists have traditionally focused on changes over time in individuals and have neglected the changing configurations of the environments in which individuals are embedded. To study behavior as it typically develops in natural circumstances, however, we must examine the mutual interplay of individuals and their environments over time, a feature that Bronfenbrenner (this volume) has now explicitly recognized by adding a chronosystem to his ecological model. A sixth challenge is to assess the implications of the above issues for interventions aimed at enhancing human development.

Macrosocial influences on development We cannot begin to understand human development without considering opportunities and constraints in the social structure and the general culture pattern or without taking into account possible and probable sequences arising from social and historical change (Elder, 1975; Riley, 1985). As Mills (1959) noted: The biographies of men and women, the kinds of individuals they have become, cannot be understood without reference to [social and] historical structures in which the milieux of their everyday life are organized. . . . [They] carry meaning not only for individual ways

4

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

of being, but for the very character - the limits and possibilities of the human being, (p. 157) How are social structural aspects of the environment related to the psychological characteristics of individuals? Our understanding of the relationships between macrosocial phenomena and individual development has been impeded by a general failure to explicate and test theories of how these linkages occur. An adequate analysis involves attending to three analytic principles. First, we must adequately understand the nature of the social structure or system in question. As we shall see later in the chapter, complex social phenomena almost always have multiple dimensions or components, and we must identify those that are most relevant for understanding observed differences in individual functioning. Second, we must recognize that the effects of social structures or systems are transmitted through stimuli that impinge directly on the individual. Thus, a major task in the study of social structure and social change is to trace how macrosocial processes affect increasingly smaller social structures and ultimately those proximal settings that directly impinge on the individual. Third, we must understand how these proximal settings affect individual development. Unfortunately, few studies to date have incorporated these analytic principles (Elder, Caspi, & Burton, 1988). To do so, researchers will have to seek more interdisciplinary cooperation, especially because they are apt to find themselves in need of tools that are not adequately developed within a single discipline. The most impressive effort along these lines is that of Kohn and his associates, who have studied the influence of social class on individual behavior and personality for over two decades (Kohn, 1977; Kohn & Schooler, 1983). What has been particularly exciting about Kohn's work is the attempt to show that the influence of social class emanates from class-associated conditions of life, notably occupational conditions. By outlining a multidimensional model of stratification and specifying the structural imperatives of jobs that define the social and psychological realities of work, Kohn and his colleagues have shown that certain features of work (e.g., the amount of autonomy, the degree of routinization, and the amount of substantive complexity experienced on the job) give rise to values of either autonomy or conformity in workers. On the one hand, the greater the freedom experienced on the job and the more complex and challenging the work, the more likely is the worker to place a high value on self-direction. On the other hand, the more constraining and routine the work, the more likely is the worker to value conformity. Kohn has also expanded his examination of the consequences of the structural imperatives of work to a wider range of psychological variables, such as intellectual flexibility. In a series of studies, he has shown that substantively

Development in context: research perspectives

5

complex work positively affects the intellectual flexibility of workers and contributes to the development of autonomy values. A core feature of this research program on job conditions is the general premise that work influences personality and behavior through a type of learninggeneralization process by which job influences are translated to other domains of life, such as marriage, parenting, and leisure. Men should thus function differently in the father role according to whether they have experienced job conditions that foster self-direction. This hypothesis has been successfully tested in both the United States and Poland. In both societies, family social position appears to influence the valuation of self-directed values for children through the occupational self-direction of parents. Parental values, then, tend to be extensions of the modes of behavior that are functional for parents in their occupational spheres. This transformation of work-generated values into parental practices constitutes an important link between the occupational and family contexts of socialization. The model and tests are not complete, but they have given us an unparalleled account of how social structural conditions are linked to aspects of psychosocial functioning in different contexts. To extend Kohn's model to research on maternal employment as it affects children's development would seem to be a natural priority. Instead, maternal employment has simply been added to the list of what Bronfenbrenner has termed "social address" variables as researchers continue to compare children whose mothers hold jobs outside the home with those whose mothers do not. The assumption here is that a mother's work-related absence, rather than her work conditions, is the critical variable influencing child development. Moreover, even when features of the mother's work status have been considered, her personal satisfactions have been emphasized and the structural imperatives of her work conditions have been ignored. Moorehouse (1986) has attempted to identify some of the influential aspects of maternal employment in a study of first graders and their families. This study related two aspects of the mother's employment - her work hours and the continuity of her work history - to parent-child activities and to the child's school performance and social adjustment. The results suggest that discontinuity in employment, especially an increase to full-time hours, is associated with more negative social and cognitive school outcomes, but only when joint mother-child activities are infrequent. The findings concerning discontinuity in employment are important for several reasons. Previous studies of maternal employment have sampled primarily women working full-time and on a stable basis. However, since discontinuity rather than continuity is the more typical pattern of employment for women with

6

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

school-aged and younger children, this approach may have led us to misidentify some of the common effects of maternal employment on children's development (Moen, 1985). More important, this study highlights the shortcomings of developmental research on social structure that focuses exclusively on men's class position and their job conditions. The social class position of women has traditionally been defined in relation to their husbands' occupational status. This definition is valid for women who are homemaking exclusively, but it is inappropriate for the majority of women who are now in the labor force. Indeed, a pressing issue for research on social structure and socialization is to examine how features of work are translated into the experiences of both fathers and mothers and how the combined effects of these experiences are transmitted to the next generation through family socialization practices. These limitations notwithstanding, Kohn's important contribution has been to stress a more general problem with the use of structural concepts, such as social class, that psychologists clumsily borrow from sociology. Much has been written in developmental psychology about the effects of social class as measured by a single composite index combining education, occupation, and income. Aggregating variables as such, however, tends to obscure what may often be differential effects of these component variables. Moreover, Kohn's research program provides an important reminder that the characteristics associated with different social addresses are quite wide ranging. Much has been written about differences in the quality and quantity of parentchild interactions across social addresses. The Gurins (Gurin & Gurin, 1970; Gurin, Gurin, Lav, & Beattie, 1969) have argued that it is more difficult for working-class parents to transmit certain values to their children owing to classbased constraints that may limit their ability to ensure their children are rewarded consistently for certain efforts. In addition, the environmental characteristics associated with different social class backgrounds range from the quality of housing conditions to the nature of toys and the number of books available to the child. Thus it may be that these characteristics exert a direct influence on the perceptual, cognitive, and motivational development of children by providing a substrate of stimulation beyond that produced through interpersonal interactions (e.g., Wachs, 1979; Wohlwill, 1983). Structural conditions, then, may impinge on the proximal settings of development in multiple ways and along a variety of dimensions. Indeed, as Elder and Caspi emphasize in their discussion of social change and human development in chapter 4, the crucial issue remains that of specifying the components and psychological processes linking structural conditions to individual development. To accomplish this task we need a more refined view of the proximal environments that impinge directly on the individual. Some advances along these lines have been made by recent research on schools and families.

Development in context: research perspectives

1

Proximal influences on development The number of proximal settings in a complex society is almost limitless. Here we focus on two settings, school and family, to illustrate problems and promises in research linking the proximal environment to individual development. The school as a developmental context. School socialization is by and large the result of direct instruction, buttressed by a system of social reinforcement, expectancy effects, and social comparison processes. But the quality of education obtained by children depends on more than just teacher-student interactions in the classroom. These interactions are structured by the type of school children attend and by the overarching educational policies of the counties and states in which the schools are located. In contrast to the early reports of Coleman (1966) and later Jencks and his colleagues (1972), recent research suggests that the school does matter, and what matters most is the quality of cultural and social organization, the attitudes, values, and mores. In an impressive study of 12 secondary schools (Rutter, Maughan, Mortimore, & Ouston, 1979), data from interviews and questionnaires completed by students, classroom observations, and official records were used to derive a "school process" score dealing with items such as style of classroom management, patterns of rewards and punishments, and student opportunities for responsibility and participation. The differences between schools were systematically and strongly associated with student outcomes in these schools, including disruptive behavior, truancy, damage to school property, and examination failure. It remains unclear just how the variables making up the "school process" score combine to create a school environment that fosters particular patterns of student behavior and academic progress, although Rutter and his colleagues suggest that these qualities determine the nature of the values and norms of behavior set by the school and thus affect students' motivation to behave in certain ways. As Rutter's work demonstrates, proximal environments have unique influences on development that do not emanate either from macrolevel social structures or from the additive characteristics of individuals in the school setting. School effects cannot be attributed entirely to ecological factors because publicly funded schools in the same catchment area differ with respect to school process variables. Moreover, school effects cannot be accounted for by selection factors whereby some schools admit a higher proportion of disruptive students. Rutter and his colleagues tackled this latter question by asking whether student characteristics at intake correlate as strongly with the school process variables as do student characteristics at the time of leaving school. The finding that student intake measures were only weakly correlated with school variables suggests that

8

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

school functioning and teacher performance were not shaped by the characteristics of the students admitted to the school. Rather, the direction of causal influence was from school factors to student behavior. A related body of research has examined the developmental implications, particularly in the area of social mobility, of the ways schools structure educational opportunities (Hauser, Sewell, & Alwin, 1976; Rosenbaum, 1976). In a case study of a school in a primarily white working-class town, Rosenbaum (1978) has shown that students' uninformed self-selection into educational tracks during high school constrains their long-term educational achievement and social mobility. Students who selected themselves into noncollege tracks jeopardized their chances of attending college not only because of the different training they received but also because of an implicit assumption that their performance on school tests was less credible than that of college-track students. This assumption was reflected in the lower weighting their grades received in calculating class rank. Rosenbaum's study illustrates that the social structure of proximal settings influences individual lives and also suggests that lack of access to information can affect developmental outcomes, a point elaborated by Goodnow in chapter 3. An issue that has unfortunately been neglected in research on schools as contexts for development concerns the relationship between school characteristics and other community and family factors, and how these jointly influence what happens in a given setting, such as the classroom. As Goodnow makes clear in this volume, we need to broaden our examination of person-environment interactions and assess the joint effects of the multiple settings in which the individual is an active participant. Linkages between developmental contexts. By applying person-environment models to the context of multiple environments, researchers can address fundamental questions about the extent to which settings containing the developing person conflict with or complement each other, and about the role of compatibility between settings in the process of development. Some of these issues are addressed in Hansen's (1986) study of the combined effects of family and classroom interactional styles on children's school performance. Family interaction patterns were classified as cohesive, coercive, or laissez-faire using definitions analogous to Baumrind's (1966) distinction between authoritative, authoritarian, and permissive styles. Teacher reports provided data for a similar classification of classroom styles. To measure academic performance, children's current grade point averages (GPA) were compared with their averages for the previous two years. Not surprisingly, the findings indicate that family style is the best predictor of GPA for any given year. However, it was the fit between classroom and family styles that predicted changes in GPA. A match predicted an increase in GPA whereas a mismatch was associated with a

Development in context: research perspectives

9

decrease in performance. This work suggests the importance of congruence between developmental contexts, but it does not take into account the characteristics of the children who occupy these settings. Promising work along these lines is found in Epstein's (1983) assessment of the match or mismatch between family, school, and personality characteristics and its effects on students' academic development. This longitudinal study followed adolescents from their last year of middle school through their first year of high school in order to examine the joint effects of two setting factors (the extent to which the formal organization of the school emphasizes student participation and the extent of decision-making opportunities for students at home) and a person factor (student self-direction and independence) on students' academic performance. The findings provide evidence for the benefits of the "fit" and for disadvantages of the "lack of fit" of persons and environments (French, Rodgers, & Cobb, 1974; Pervin, 1968; Thomas & Chess, 1980). For example, each type of student is rewarded in high-participatory schools, but those students who come from high-participatory families and who are also initially high in independence have the highest grades. In addition, highly independent students who are mismatched with both socializing environments (i.e., they come from low-participatory families and also attend low-participatory schools) are among those with the lowest grades. The findings are especially interesting because they show how research designs that incorporate multiple developmental settings can reveal the power of compensatory environmental experiences. Thus, we find that school environments are most important for highly independent students who come from families that do not emphasize participation in decision making. Epstein's study also illustrates some directions that research on home-school relations might profitably take and provides provocative findings from a promising research paradigm: the mesosystem component of Bronfenbrenner's (1979) ecological model that emphasizes the relationships between different developmental settings. Unfortunately, studies documenting the joint effects of processes occurring within and between two or more settings in which the developing person is an active participant are still in short supply (see chapter 3). The neglect of these issues is especially disappointing given the prevalence of "reconstituted" families and the attendant number of residential and school moves that today's children can expect to experience during the course of development. The family as a developmental context. The family is, of course, the context most closely associated with developmental processes. It is the first socialization context that most of us experience and it is the place where we develop our initial sense of self. Two advances in recent years have renewed interest in research on the family. First, the operational model of the parent-

10

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

child relationship has been expanded from dyadic to polyadic forms. Second, findings from human behavior genetics studies have converged on the conclusion that nonshared environmental influences represent a critical source of environmental variance for producing individual differences. The recent growth of family systems theories in developmental psychology, as well as the development of conjoint family therapies for alleviating children's psychiatric problems, has strengthened the interest in interactional processes and relationships within the family (Minuchin, 1985). The basic concept of systems approaches is that the family constitutes a social group in which the functioning of the whole is qualitatively different from the sum of its parts. This is so because properties of the family as a whole derive from the properties of the relationships between individuals in the family and not just from the characteristics of the individuals as separate persons (Hinde, 1979; Hinde & Stevenson-Hinde, 1988). Thus, the nature of the home environment and family dynamics can best be understood in terms of a synthetic framework that links family members in a network of interactions and personal relationships (Belsky, 1984). The first systematic studies to incorporate this framework documented the role of the father in the mother's treatment of the child (Clarke-Stewart, 1978; Parke, 1979). Another such study demonstrated that a child's classroom performance varies systematically as a function of whether or not the teacher had taught the child's older sibling, and if so how well that older sibling had performed (Seaver, 1973). Other recent examples are provided by the sibling rivalry or jealousy phenomenon, whereby children with particularly warm and close relationships with their mothers are more likely to develop negative reactions toward their younger sibling (Dunn & Kendrick, 1982), and by the sibling deidentification phenomenon whereby children in the same family react in ways that emphasize differences from their siblings (Schachter, 1982; Tesser, 1980). Findings from the literature on marital and parent-child relationships in a wide range of domains, including developmental psychology, family sociology, and clinical psychology, have also drawn attention to the interdependence of family relationships. For example, in a study of families with school-aged children, Brody, Pillegrini, and Sigel (1986) found different patterns of parent-child interaction in a problem-solving task among couples varying in marital quality. In the case of low marital quality, mothers tended to impose structure on the interaction, and fathers were more intrusive and provided less feedback about their children's problem-solving efforts. Other studies investigating different age periods and different domains of development support the general conclusion that positive marital adjustment facilitates parental sensitivity to the child (e.g., Feldman, Nash, & Aschenbrenner, 1983; Goldberg & Easterbrooks, 1984). The interdependence of family members is further illustrated in a study of the

Development in context: research perspectives

11

effects of a depressed adult on the psychological well-being of other adult family members. Coyne and his colleagues (1987) found that over 40% of adults living with a depressed person in the midst of an episode were sufficiently distressed to require intervention. In contrast, these elevated rates of distress were not found among adults living with a depressed person who was not currently in the midst of a depressive episode. Moreover, these elevated rates of depression were primarily mediated by the formidable objective and subjective burden placed on the family by the depressed person's behavior. In conjunction with findings regarding the impaired parenting of depressed women (e.g., Zahn-Waxler, 1981), these findings illustrate the influence of individual problems on the functioning of other family members through family relationships. Beyond the family, recent research has also documented the implications of social network membership for individual development. For example, Kessler and McLeod (1984) have proposed that women's increased vulnerability to depression may be due to their greater sensitivity to negative life events in the lives of other network members. There is suggestive evidence that women are more likely than men to hear about the life events of others and then to assume more of the burden associated with the latter's problems and crises. Emerging from these new microsystem models and findings is a more dynamic and differentiated concept of the immediate environment. Instead of studying the child in isolation from his or her family context, researchers are increasingly adopting family-based dyads and triads as their fundamental units of analysis (Hinde & Stevenson-Hinde, 1988). Such studies provide persuasive evidence that many developmental outcomes depend on properties of higher-order relationships rather than on properties of family members considered individually (Maccoby, 1982). This dependence is well illustrated by Patterson's (1982) model of violence among family members, suggesting that the density of coercive exchanges in the family covaries with intensity. He has labeled this process the anarchy progression. Given that one family member increases his or her coerciveness beyond the normal range, other members become increasingly at risk of being hit. In addition, as more dyads within the family become more coercive, the risk for physical assault increases commensurately. A view of the proximal environment as a system of interdependent members and relationships also has critical implications for intervention strategies. Consider the highly aggressive child. Traditional treatments have asked how we can change the characteristics of the child in order to reduce aggression. These individual-centered approaches have met with minimal success, however. Therefore Patterson (1982) asked a different question: What is the aggressive child's microsystem (i.e., family) like, and how can we change it? The success rate of

12

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

their interventions indicates the relevance of Bronfenbrenner's concern in chapter 2 that child development and personal functioning be viewed in a systemic and ecological perspective. Clearly, a conceptualization of the environment as an organized system of parents and children interacting in multiple and changing environments demands an eclecticism that is open to new conceptual and methodological approaches (Applebaum & McCall, 1983). The extant evidence suggests that an emphasis on family relationships, as opposed to the structural characteristics of the family, may deliver new advances in our understanding of the family as a developmental context. Some of the most important findings about environmental influences have also emerged from behavior genetics research (Plomin, 1986; Plomin & Daniels, 1987). This research converges on the conclusion that, for many variables of interest, the correlations between siblings in the same family are not much higher than those between nonrelated individuals. Scarr and Grajek (1982) make explicit the implications of these findings: Upper middle-class brothers who attend the same school and whose parents take them to the same plays, sporting events, music lessons, and therapists, and use similar child rearing practices on them are little more similar in personality measures than they are to working class or farm boys, whose lives are totally different. Now, perhaps this is an exaggeration of the known facts, but not by much. Given the low correlations of biological siblings and the near zero correlations of adopted siblings, it is evident that most of the variance in personality arises in environmental differences among siblings, not in the differences among families, (p. 361) These findings are rich in implications for research and theory. In terms of offering a practical guide to research, they suggest that family studies should observe more than one child per family in order to identify environmental factors that make children in a family so different from one another. Traditional environmental research has attempted to relate measures of the family environment to measures of the behavior of one child per family, but the yield from these efforts has been disappointing (see Maccoby & Martin, 1983, for an assessment). It may make more sense to focus on environmental sources of differences between children in the same family, for these differences in experience (and perhaps perception) appear to be the factors that drive behavioral development. The importance of nonshared environments also suggests a need for a theoretical reconceptualization of environmental influences on development. As Scarr points out in chapter 9, individual differences in adoption and twin research arise from genetic and environmental differences within families, not from environmental differences between families. Our psychological theories, however, address only between-family differences. To develop more powerful accounts of environmental influences we will have to identify experiences that are not shared

Development in context: research perspectives

13

by family members and then relate these nonshared environmental factors to differences in siblings' behavior. Thus far, only a few relevant studies have been reported (e.g., Daniels, Dunn, Furstenberg, & Plomin, 1985). These studies suggest that siblings in the same family experience different environments and that differential parental treatment is related to differences in siblings' adjustment. But if, as this evidence suggests, most of the environmental variance that affects behavioral development is of the nonshared variety, the foremost problem is to determine whether nonshared environmental variance is systematic or stochastic, and, if it is systematic, to ascertain what gives rise to the different environments children experience. Are the relationships between nonshared environments and behavior mediated by the individual? As we shall see in the next section, a good deal of evidence has accumulated for child-directed socialization effects. The individual's effects on the environment An important process linking individuals to their environments is provided by the individual's effects on the environment. Indeed, the relative weights assigned to Lewin's formulation, B = f(PE), have undergone a great deal of change in the last several decades. We began with an oversocialized conception of human nature that assigned priority to the environment, B = f(P <— E). This conception, evident in historical and cultural concepts of childhood and parenthood, implied that parents were the agents of socialization and children the objects of their inculcation efforts. The parent was charged with teaching the child those values, attitudes, and behavioral traits that were considered appropriate and essential for effective functioning as a future adult. It was a perspective that unified sociological and psychological efforts to address the enduring question of how society is possible (Wrong, 1961). But the unidirectional model of parent-child, and, more broadly, personenvironment relations proved too simplistic and imprecise and gave way to increasing appreciation of reciprocal effects in development, owing in large part to Bell's (1968) classic reinterpretation of parent-child effects. Bell provided the important reminder that although parents affect their children, children also stimulate their parents to behave in certain ways, B = f(P <-» E). New questions were asked, new methodologies appeared, and we learned that children's attributes, not only their parents' attitudes, accounted for how they were being raised. We are now developing a new explanatory language that grants the individual autonomous powers of self-direction, B = f(P —» E). Individuals are active organisms that select, modify, and even create the environments in which they

14

Nm Bolger, A. Caspi, G. Downey, and M. Moorehouse

develop. This agentic conceptualization of development cannot claim originality (see Silbereisen & Noack, this volume), but it has set in motion a paradigmatic reversal: To document the influence of individuals on their environments requires casting the individual in the role of the independent variable and the environment in the role of the dependent variable. This paradigmatic reversal has unified different subdisciplines of psychology. Personologists since the time of Allport (1937) have defended the notion that individuals actively seek out situations that are compatible with their personalities and avoid others that are incompatible, and the strategy for understanding individuals in terms of the social worlds they create has revived a good deal of contemporary research on interactional psychology (see Bern, this volume, for an extended discussion of research paradigms in personality). The most lucid formulation of the "choice of situation" model has been proposed by Snyder (1984, 1987). Snyder and his colleagues have sought to understand the processes by which individuals actively choose to put themselves in certain situations and avoid others, as well as the consequences of these choices. These theorists have proposed that the process of choosing situations reflects basic features of one's identity, including conception of self, beliefs and attitudes, and characteristic dispositions. The choice-of-situation paradigm has successfully addressed a number of research questions, such as the origins of correspondence between attitudes and behavior. For example, by choosing to enter and spend time in social situations that dispose them to perform the actions implied by their personal attitudes, individuals may generate correspondence between their attitudes and behavior (Snyder & Kendzierski, 1982). Individuals' active construction of their social worlds also helps account for consistencies in behavior. For example, when extroverts preferentially seek out social situations, they thereby select themselves into environments that further nourish and sustain their sociability. Similarly, individuals with a high need for order tend to avoid novel situations, thus ensuring greater regularity in their lives (Diener, Larson, & Emmons, 1984). Indeed, the most striking and enduring differences between individuals are to be found not by studying their responses to the same situation, but by analyzing their construction of new situations (Buss, 1984). Analyzing behavior in terms of presented stimuli offers a misleading picture because individuals more often create the stimuli to which they respond (Wachtel, 1973). For example, individuals work quite hard to ensure the stability of their self-conceptions in new situations. They preferentially search for evidence that confirms propositions they are testing, whether these are about objects, other people, or themselves. Moreover, they act in ways that generate behavioral disconfirmation for hypotheses that challenge or threaten central aspects of their self-concept (Darley & Fazio, 1980; Swann, 1983).

Development in context: research perspectives

15

Some of these issues have now been examined in a series of longitudinal studies. Using data from the Berkeley Guidance Study, Caspi, Elder, and Bern (1987, 1988) have suggested that problem behaviors are maintained across time, in diverse situations, and in new relationships because the individual's dispositions systematically select him or her into particular environments that, in turn, reinforce and sustain those dispositions. Their life-course studies have shown that explosive, undercontrolled boys are likely to lose out in formal education, thereby limiting their future career opportunities and selecting themselves into low-status jobs that further provoke a pattern of striking out explosively against the world. Similarly, withdrawn children are likely to exert different selective efforts than nonwithdrawn children and respond to social demands in a manner that elaborates further environmental experiences. Unable to initiate action, males with a history of social withdrawal delayed marriage and an occupational career, thereby creating a structurally incompatible nexus of roles at midlife that produced family and work strains. Parallel to these advances in personality and social psychology, behavioral geneticists have advanced a theory of development in which experience is directed by genotypes. Genotypic differences are proposed to affect phenotypic differences through three kinds of genotype-environment effects (Plomin, DeFries, & Loehlin, 1977): a passive kind, whereby the genetically related parents provide a rearing environment that is correlated with the genotype of the child; an evocative kind, in which the child receives or evokes responses from the environment that are influenced by his or her genotype; and an active kind, whereby people seek out environments that are compatible with their genotypes. Scarr (this volume; Scarr & McCartney, 1983) has used this model to describe developmental processes, arguing that the relative importance of the three kinds of genotype-environment effects change over the course of development. Genotype-environment effects of the evocative sort persist through life as we elicit responses from others on the basis of our attributes. These responses, in turn, reinforce and sustain particular dispositions. Moreover, from infancy to adolescence, passive genotype-environment effects wane as the importance of active genotype-environment effects increases. Thus, the child may behave in ways that help promote continuity in social settings and relationships by virtue of his or her choices and preferences. Such processes occur throughout development and are especially evident in the maintenance of friendships and in activity choices (see also Cairns & Cairns, and Patterson, this volume). This compelling developmental account has successfully addressed findings that neither purely genetic nor environmental models can effectively explain, such as the declining similarities across time of DZ twins and siblings relative to MZ twins. The power of this model is especially well illustrated by Lytton's (1980) study of interaction sequences between identical and fraternal twins and

16

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

their parents. This unique research design examines how differences in environments arise by describing the behavioral initiations and responses of children and their parents in everyday situations. Lytton finds that the greater similarity in the treatment of identical than fraternal twins arises from the evocations of more similar treatment by the former, suggesting that the environments children experience are at least, in part, the result of their own characteristics. Understanding how individuals influence their environments can add to our understanding of how environments influence individuals. By choosing and influencing their environments, individuals determine which settings will have an opportunity to influence their development. Thus, we must go beyond questions about environmental influences to questions about what particular individuals have done to mold, shape, and create the very settings that then, in turn, influence and constrain their subsequent development. Only then will the mutual interplay of individuals and their environments be truly apparent. Linking persons and context across time The failure to examine the mutual interplay of individuals and their environments is especially apparent in longitudinal studies, the lifeblood of developmental psychology (McCall, 1977). In many ways we have been quite deficient in our conception of the problem of continuity or lawfulness in development. The task is usually conceived as that of predicting, on the basis of individuals' initial status on some characteristic, what their later status should be. But the interactional model we have proposed suggests that early status is only one component of the equation. Other components must include the nature of the social system in which development occurs. As Cairns and Cairns suggest in chapter 7, methods that assess only early status on a given characteristic and leave out information about the nature of the social system in which various developmental transformations take place are clearly inadequate. Indeed, one of the most important sources of support for individual-difference predictability occurs in the social system that is maintained across different periods in development (Cairns & Hood, 1983). Continuity in social networks may promote stability both because of the similarity in demands made by the social environment and because of the similarity in how others relate to the individual, which, in turn, may affect how individuals view and define themselves. Conversely, drastic changes in the social system should promote concomitant changes in social accommodation and facilitate changes in individual behavior. Patterson (this volume) has pursued these issues in his research on aggressive boys, showing how family interactions can create and sustain destructive and aversive patterns of behavior. But there is little information on how social and

Development in context: research perspectives

17

ecological factors outside the family sustain such behavior, although we are beginning to see some change on this front with the reemergence of research on children's social networks. Social network investigations reported recently by developmental psychologists suggest that the main problems of aggressive expression stem from the aggressive child's lack of social skills. Deficiencies in social understanding contribute, in turn, to the child's rejection by peers, which may exacerbate the child's problem (Dodge, 1980; Dodge & Frame, 1982; Dodge, Schlundt, Schocken, & Delugach, 1983). Important and compelling as these findings are, they cannot tell the whole story since longitudinal studies point repeatedly to the importance of ties to deviant peers in delinquency and indicate that aggressive children are not disaffiliate%d (West & Farrington, 1973). In a thoughtful critique, Cairns (1983) suggests that the recent emphasis by developmental researchers on identifying isolated and rejected children reflects a bias toward analyzing social networks at the level of the individual rather than the group. Moreover, a psychometric rather than a sociometric approach to the study of peer groups may overemphasize individual determinants of behavior at the expense of a more thorough probing of environmental circumstances. An instructive example of this observation is provided by research showing that children subjected to deprivation or disadvantage in the early years often go on to show chronic psychological and psychiatric disorders. Usually this is taken to mean that lasting damage has occurred. When environmental circumstances are examined, however, it is apparent that, to an important extent, the persistence of sequelae is a function of persistence of environmental circumstances. Rutter, Quinton, and Liddle (1983), for example, have found that institutionreared women are likely to show poor parenting skills. One of the mechanisms that underlies the long-term effects of this early experience, however, resides in the constancy of environmental forces. Institutional rearing was associated with an increased rate of behavioral disturbance, which, in turn, was associated with a greater likelihood of dropout from education and dismissal from jobs. This, in turn, led to work of low social status or unemployment. Adult social conditions had a pronounced effect on women's parenting quality, but the conditions of earlier child rearing also played a part in determining whether women experienced poor social conditions in adulthood. Clearly, behavioral connections across time cannot be fully understood without information on environments across time. The role of environmental experiences in maintaining or diverting life trajectories is nicely illustrated by Frank Furstenberg's longitudinal study of adolescent mothers (Furstenberg, Brooks-Gunn, & Morgan, 1987). The data come from a sample of some 300 women from Baltimore who were first interviewed in the mid-1960s, shortly after they became pregnant. The earlier, initial phase of

18

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

this research, carried out over a five-year period, had culminated in Unplanned Parenthood (Furstenberg, 1976), an assessment of life-course variations in the transition to adolescent motherhood. Seventeen years later, Furstenberg and his colleagues revisited these families. The children of the teenage mothers were now approaching adolescence. The study examines the processes linking teenage motherhood to outcomes later in adulthood and also breaks fresh ground by exploring issues of continuity and change across generations in the lives of the teenage mothers as well as their offspring. Contrary to what many observers might have predicted, a disadvantaged life course was not a certain outcome among teenage mothers. Furstenberg and his colleagues are able to explain some of the variability in outcomes at the 17-year follow-up by examining the occurrence, timing, and sequencing of multiple events in the mother's life course subsequent to childbirth. The most viable escape route from disadvantage involved education. Teen mothers who completed high school were half as likely to be receiving public assistance and twice as likely to be economically secure in adulthood. A second component of success can be traced to the young mothers' ability to restrict further childbearing. Women who restricted their family size in the five-year period after the birth of their first child were three and a half times more likely to be economically secure in 1984. Fertility control not only improved a woman's chances of avoiding welfare, it also slightly improved her chances of remarriage. Indeed, a third important route to economic security is a stable marriage and independence from the family of origin. Although not all mothers in the study did, adolescent mothers can and do find a number of different routes for recovery and for repair of damaged life chances. The teenage mothers' decisions and choices also influenced the environment in which their children were raised. In various domains, the children of teenage mothers displayed more symptoms of maladjustment than did teenagers whose mothers were older when they had their first child. Across these various domains there was also a great deal of variability in children's development, variability that can be explained, in part, by the timing and sequencing of maternal events and decisions. Using academic success as one child outcome, Furstenberg and his colleagues find that high fertility in the years immediately following the birth of the first child was a negative predictor of academic success, suggesting that rapid subsequent fertility reduces time and resources spent on the first child. Conversely, a trajectory of academic failure could be attenuated by lower fertility rates. Indeed, notable connections exist between the life course of mothers and their children, and between decisions made by the mother subsequent to her early entry into motherhood and the fate of her child. Furstenberg's work provides an important reminder that we continue to underutilize longitudinal designs by restricting our attention to individual patterns of

Development in context: research perspectives

19

constancy and change while ignoring the changing configurations of the environments in which individuals are embedded. It is incumbent upon us to confront the developmental dynamics of successive environmental events in order to better understand the effect of early experience on later adjustment and behavior. Individuals, environments, and interventions The study of development in context also raises several issues pertinent to intervention. If, as we have suggested, researchers are often misguided by a view of the environment as an undifferentiated set of influences and by a view of the person as a passive recipient of these influences, intervention planners are at risk of falling prey to even more disastrous errors. Interventions must be planned and implemented with some appreciation of the linkages between individuals and their ever-changing environment. It is especially troubling that many intervention programs continue to be characterized by an excessively individualistic bias, attempting to modify behavior with little regard for contextual demands (Sarason, 1982; Wachtel & Wachtel, 1986). For example, theory and research on perceived control have emphasized the negative effects of the loss of perceived control on individual functioning and, by implication, point to the benefit to be derived from increasing people's sense of perceived control. Applied research, however, has challenged the desirability of implementing control-enhancing interventions with all groups, for there exists the possibility that such interventions, inappropriately matched with environmental opportunities, will raise an individual's sense of control to unrealistically high levels (Schulz, 1980). In some circumstances, the net impact of the intervention is likely to be negative. It is necessary to pay close attention to the context in which individual control takes place, and it is critical to consider the opportunities and constraints afforded in the larger social structure. A related issue concerns the failure to recognize emergent group constraints: Interventions that have a specified effect on single individuals may not have the same effect if carried out on all persons within a given group. Consider, for example, the issue of job search as an adaptive strategy for unemployed persons in economically depressed regions. The standard pep talk emphasizes that if the unemployed seek hard enough they will be gainfully reemployed. However, for any given person, this course of action would be effective only if one could assume that most unemployed persons would not search vigorously for work. Such advice is clearly ineffective as a general remedy for unemployment in economically depressed regions or at particular points in the business cycle when there are many fewer jobs than persons seeking employment. Interventions based on acontextual models of development are especially prone to difficulties because they fail to take into account emergent group constraints.

20

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

A further source of difficulty in planning interventions stems from the interdependent and systemic nature of developmental contexts. Indeed, interventions aimed at a specific developmental setting or designed to alter a particular domain of an individual's life are likely to have unintended consequences in other developmental settings and in other life domains. The Seattle-Denver Income Maintenance Experiment illustrates what can happen in such cases. The goal of this group-level intervention was to provide families with sufficient income to move off welfare. Its unintended consequence was an increase in the rate of divorce. Relative to a welfare control group, the rate of marital dissolution increased significantly in groups with an income support level 25 percent above the poverty line. Tuma, Hannan, and Groeneveld (1979) suggest that income maintenance programs (which guarantee income regardless of family structural change) not only raise family income, but they also decrease dependence on marriage by providing new financial alternatives. A change in a single aspect of the environment can thus have multiple and opposing effects on families and individuals. Turning from the environment to the individual, one finds that human agency also poses problems for those who wish to change behavior. Many program evaluators have been chastened by their experience with individuals and organizations that obstruct and distort the intervention process in the service of their own goals (Cook & Campbell, 1979). Similarly, in the area of government policy implementation, the goal-oriented behaviors of privileged individuals and groups often seem impossible to thwart. Thus, middle-class families capitalize on housing and educational policies designed for the poor, and the wealthy inevitably find their tax loopholes. It is significant that in the interventionoriented discipline of economics there has emerged a powerful critique - the rational expectations model - which argues that it is impossible to improve welfare through government intervention because policy makers can never outguess the shrewdest and most well-placed economic actors (see Kantor, 1979). In view of these problems, we suggest that interventions should not be initiated without some conceptual or theoretical analysis of the interdependent nature of developmental settings. Nor should they be implemented without an appreciation of the active role that individuals play in shaping their own development. Only by attending to these issues can intervention planners prevent their wellintentioned efforts from yielding unintended and potentially disastrous consequences. The chapters that follow provide no definitive answers to questions about the processes entailed in the lifelong transactions between individuals and their everchanging environment. They do delineate, however, both by illustration and implication, a promising highroad in developmental science: the systematic study of human development in its human context.

Development in context: research perspectives

21

References Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt, Rinehart, and Winston. Applebaum, M. I., & McCall, R. B. (1983). Design and analysis in developmental psychology. In W. Kessen (Ed.), P. H. Mussen (Series Ed.), Handbook of child psychology: Vol. 1. History, theory, and methods. New York: Wiley. Baumrind, D. (1966). Effects of authoritative parental control on child behavior. Child Development, 37, 887-907. Bell, R. Q. (1968). A reinterpretation of the direction of effects in studies of socialization. Psychological Review, 75, 81-95. Belsky, J. (1984). The determinants of parenting: A process model. Child Development, 55, 83-96. Brody, G., Pillegrini, A. D., & Sigel, I. E. (1986). Marital quality and mother-child and fatherchild interactions with school-aged children. Developmental Psychology, 22, 291-6. Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA: Harvard University Press. Buss, D. M. (1984). Toward a psychology of person-environment (PE) correlation: The role of spouse selection. Journal of Personality and Social Psychology, 47, 361-77. Cairns, R. B. (1983). Sociometry, psychometry, and social structures: A commentary on six studies of popular, rejected, and neglected children. Merrill-Palmer Quarterly, 29, 429-44. Cairns, R. B., & Hood, K. E. (1983). Continuity in social development: A comparative perspective on individual difference prediction. In P. B. Baltes & O. G. Brim, Jr. (Eds.), Life-span development and behavior (Vol. 5). New York: Academic Press. Caspi, A., Elder, G. H., Jr., & Bern, D. J. (1987). Moving against the world: Life-course patterns of explosive children. Developmental Psychology, 23, 308-13. Caspi, A., Elder, G. H., Jr., & Bern, D. J. (1988). Moving away from the world: Life-course patterns of shy children. Developmental Psychology. Clarke-Stewart, A. (1978). And daddy makes three: The father's impact on mother and young child. Child Development, 49, 466-78. Coleman, J. S. (1966). Equality of educational opportunity. Washington, DC: U.S. Government Printing Office. Cook, T. D., & Campbell, D. T. (1979) Quasi-experimentation: Design and analysis issues for field settings. Boston: Houghton-Mifflin. Coyne, J. C , Kessler, R. C , Tal, M., Turnbull, J., & Wortman, C. B. (1987). Living with a depressed person. Journal of Consulting and Clinical Psychology, 55, 347-52. Daniels, D., Dunn, J., Furstenberg, F. F., Jr., & Plomin, R. (1985). Environmental differences within the family and adjustment differences within pairs of adolescent siblings. Child Development, 56, 164-1 A. Darley, J., & Fazio, R. H. (1980). Expectancy confirmation processes arising in the social interaction sequence. American Psychologist, 35, 867-81. Diener, E., Larsen, R. J., & Emmons, R. A. (1984). Person x situation interactions: Choice of situations and congruence response models. Journal of Personality and Social Psychology^, 580-92. Dodge, K. A. (1980). Social cognition and children's aggressive behavior. Child Development, 51, 162-70. Dodge, K. A., & Frame, C M . (1982). Social cognitive biases and deficits in aggressive boys. Child Development, 54, 1386-99. Dodge, K. A., Schlundt, D. G., Schocken, I., & Delugach, J. D. (1983). Social competence and children's sociometric status: The role of peer group entry strategies. Merrill-Palmer Quarterly, 29, 309-36.

22

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

Dunn, J., & Kendrick, C. (1982). Siblings: Love, envy, and understanding. Cambridge: Harvard University Press. Elder, G. H., Jr. (1975). Age differentiation and the life course. Annual Review of Sociology (Vol. 1). Palo Alto, CA: Annual Reviews. Elder, G. H., Jr., Caspi, A., & Burton, L. M. (1988). Adolescent transitions in developmental perspective: Sociological and historical insights. In M. Gunnar (Ed.), Minnesota symposia on child psychology (Vol. 21). Hillsdale, NJ: Erlbaum. Epstein, J. L. (1983). Longitudinal effects of family-school-person interactions on student outcomes. In Research in sociology of education and socialization (Vol. 4). Greenwich, CT: JAI Press. Feldman, S. S., Nash, S. C , & Aschenbrenner, B. G. (1983). Antecedents of fathering. Child Development, 54, 1628-36. French, J. P., Rodgers, W., & Cobb, S. (1974). Adjustment as person-environment fit. In G. V. Coelho, D. M. Hamburg, & J. E., Adams (Eds.), Coping and adaptation. New York: Basic Books. Furstenberg, F. F. (1976). Unplanned parenthood. New York: Free Press. Furstenberg, F. F., Brooks-Gunn, J., & Morgan, S. P. (1987). Adolescent mothers in later life. New York: Cambridge University Press. Goldberg, W., & Easterbrooks, M. A. (1984). Role of marital quality in toddler development. Developmental Psychology, 20, 504-14. Gurin, G., & Gurin, P. (1970). Expectancy theory in the study of poverty. Journal of Social Issues, 26, 83-104. Gurin, P., Gurin, G., Lav, R. C , & Beattie, M. (1969). Internal-external control in the motivational dynamics of Negro youth. Journal of Social Issues, 25, 29-53. Hansen, D. A. (1986). Family-school articulations: The effects of interaction rule mismatch. American Educational Research Journal, 23, 643-59. Hauser, R. M., Sewell, W. M., & Alwin, D. F. (1976). High school effects on achievement. In W. Sewell (Ed.), Schooling and achievement in America. New York: Academic Press. Hinde, R. A. (1979). Towards understanding relationships. New York: Academic Press. Hinde, R. A., & Stevenson-Hinde, J. (Eds). (1988). Relationships within families. New York: Oxford University Press. Hoffman, L. W. (1984). Work, family, and the socialization of the child. In R. D. Parke (Ed.), Review of child development research (Vol. 7). Chicago: University of Chicago Press. Jencks, C , Smith, M., Acland, H., Bane, M. J., Cohen, D., Gintis, H., Heyns, B., & Michelson, S. (1972). Inequality: A reassessment of the effects of family and schooling in America. New York: Basic Books. Kantor, B. (1979). Rational expectations and economic thought. Journal of Economic Literature, 17, 1422-6. Kessler, R. C , & Mcleod, J. D. (1984). Sex differences in vulnerability to undesirable life events. American Sociological Review, 49, 620-31. Kohn, M. (1977). Class and conformity. Chicago: University of Chicago Press. Kohn, M., & Schooler, C. (1983). Work and personality: An inquiry into social stratification. Norwood, NJ: Ablex. LeVine, R. A. (1981). Anthropology and child development. New Directions in Child Development, 8, 70-86. Lewin, K. (1943). Defining the "field at a given time." Psychological Review, 50, 292-310. Lytton, H. (1980). Parent-child interaction: The socialization process observed in twin and singleton families. New York: Plenum. McCall, R. R. (1977). Challenge to a science of developmental psychology. Child Development, 48, 333-43.

Development in context: research perspectives

23

Maccoby, E. E. (1982). Organization and relationships in development. In W. A. Collins (Ed.), Minnesota symposia on child psychology (Vol. 15). Hillsdale, NJ: Erlbaum. Maccoby, E. E., & Martin, J. A. (1983). Socialization in the context of the family: Parent-child interaction. In E. M. Hetherington (Ed.), P. H. Mussen (Series Ed.), Handbook of child psychology: Vol. 4. Socialization, personality and social development. New York: Wiley. Mills, C. W. (1959). The sociological imagination. New York: Oxford University Press. Minuchin, P. (1985). Families and individual development: Provocations from the field of family therapy. Child Development, 56, 289-302. Moen, P. (1985). Continuities and discontinuities in women's labor force activity. In G. H. Elder, Jr., (Ed.), Life course dynamics: Transitions and trajectories, 1968-1980. Ithaca, NY: Cornell University Press. Moorehouse, M. (1986). The relationships among continuity in maternal employment, parent-child communicative activities, and the child's school competence. Unpublished doctoral dissertation. Cornell University, Ithaca, NY. Parke, R. D. (1979). Perspectives on father-infant interaction. In J. Osofsky (Ed.), Handbook of infant development. New York: Wiley. Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castallia. Pervin, L. A. (1968). Performance and satisfaction as a function of individual-environment fit. Psychological Bulletin, 69, 56-68. Plomin, R. (1986). Development, genetics, and psychology. Hillsdale, NJ: Erlbaum. Plomin, R., & Daniels, D. (1987). Why are children in the same family so different from one another? Behavioral and Brain Sciences, 10, 1-16. Plomin, R., DeFries, J. C , & Loehlin, J. C. (1977). Genotype-environment interaction and correlation in the analysis of human behavior. Psychological Bulletin, 84, 309-22. Riley, M. W. (1985). Age strata in social systems. In R. H. Binstock & E. Shanas (Eds.), Handbook of aging and the social sciences. New York: Van Nostrand Reinhold. Rosenbaum, J. E. (1976). Making inequality: The hidden curriculum of high school tracking. New York: Wiley. Rosenbaum, J. E. (1978). The structure of opportunity in school. Social Forces, 57, 236-56. Rutter. M., Maughan, B., Mortimore, P., & Ouston, J. (1979). Fifteen thousand hours: Secondary schools and their effects on children. Cambridge, MA: Harvard University Press. Rutter, M., Quinton, D., & Liddle, C. (1983). Parenting in two generations: Looking backwards and looking forwards. In N. Madge (Ed.), Families at risk. London: Heinemann. Sarason, S. B. (1982). An asocial psychology and a misdirected clinical psychology. American Psychologist, 36, 827-36. Scarr, S., & Grajek, S. (1982). Similarities and differences among siblings. In M. E. Lamb & B. Sutton-Smith (Eds.), Sibling relationships: Their nature and significance across the lifespan. Hillsdale, NJ: Erlbaum. Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of genotype-environment correlations. Child Development, 54, 424-35. Schachter, F. F. (1982). Sibling deidentification and split-parent identification. In M. E. Lamb & B. Sutton-Smith (Eds.), Sibling relationships: Their nature and significance across the lifespan. Hillsdale, NJ: Erlbaum. Schulz, R. (1980). Aging and control. In J. Garber & M. E. P. Seligman (Eds.), Human helplessness: Theory and applications. New York: Academic Press. Seaver, W. B. (1973). Effects of naturally induced teacher expectancies. Journal of Personality and Social Psychology, 28, 333-42. Snyder, M. (1984). When beliefs create reality. In L. Berkowitz (Ed.), Advances in experimental social psychology. New York: Academic Press. Snyder, M. (1987). Public appearances/Private realities. New York: Freeman.

24

N. Bolger, A. Caspi, G. Downey, and M. Moorehouse

Snyder, M., & Kendzierski, D. (1982). Choosing social situations: Investigating the origins of correspondence between attitudes and behavior. Journal of Personality, 50, 280-95. Swann, W. B., Jr. (1983). Self-verification: Bringing social reality into harmony with the self. In J. Suls & A. G. Greenwald (Eds.), Psychological perspectives on the self (Vol. 2). Hillsdale, NJ: Erlbaum. Tesser, A. (1980). Self-esteem maintenance in family dynamics. Journal of Personality and Social Psychology, 39, 77-91. Thomas, A., & Chess, S. (1980). The dynamics of psychological development. New York: Brunner/Mazel. Tuma, N. B., Hannan, M. T., & Groeneveld, L. P. (1979). Dynamic analysis of event histories. American Journal of Sociology, 84, 820-54. von Bertalanffy, L. (1969). General systems theory. New York: Braziller. Wachs, T. D. (1979). Proximal experience and early cognitive-intellectual development. MerrillPalmer Quarterly, 25, 3-41. Wachtel, E. F., & Wachtel, P. L. (1986). Family dynamics and individual psychotherapy. New York: Guilford. Wachtel, P. L. (1973). Psychodynamics, behavior therapy, and the implacable experimenter: An inquiry into the consistency of personality. Journal of Abnormal Psychology, 83 324-34. West, D. J., & Farrington, D. P. (1973). Who becomes delinquent? London: Heinemann. Wohlwill, J. F. (1983). Physical and social environment as factors in development. In D. Magnusson & V. L. Allen (Eds.), Human development: An interactional perspective. New York: Academic Press. Wrong, D. (1961). The oversocialized conception of man in modern sociology. American Sociological Review, 26, 183-93. Zahn-Waxier, C. (1981). The social-emotional development of young children with a manic-depressive parent. Paper presented at Review of the Board of Scientific Counselors, NIMH, Bethesda, MD.

Interacting systems in human development. Research paradigms: present and future

Urie Bronfenbrenner

In this chapter I analyze the interacting systems incorporated explicitly or implicitly in existing research paradigms for the study of human development in context. My purpose here is threefold: first, to describe systematically the nature and scope of the models currently employed; second, to identify the strengths and shortcomings of each; and third, to explore possibilities for improving on existing paradigms by combining their constructive elements in more comprehensive conceptual and operational designs. Especially, given the third objective, it seems appropriate to examine existing paradigms by proceeding from the more simple and segmental to the more complex and developed. But in order to carry out this'task, one must establish the criteria of complexity and completeness. Even before that, however, one must ask, and answer, a more basic question: Along what parameters do research paradigms of development in context vary? A conceptual framework A useful framework for identifying such general dimensions and providing some criteria of complexity and completeness can be obtained by making a single and provocative substitution in Kurt Lewin's classical formula: B = f(PE) [Behavior is a joint function of person and environment.] The substitution involves replacing the term B (standing for Behavior) by the term D (standing for Development). In short, D — f(PE) [Development is a joint function of person and environment.] The provocative aspect of this substitution arises from the somewhat slippery question it indirectly poses of the conceptual difference between the two left-

25

26

Urie Bronfenbrenner

hand terms.1 As we shall see in due course, this distinction turns out to have critical significance both for theory and for research design. For the moment, however, let us focus our attention on some more obvious implications of the reformulated formula. To begin with, it explicitly identifies three global dimensions along which developmental paradigms can vary and implies two others of equal importance. The first domain of variation is the nature of possible developmental outcomes. Within a given paradigm, what characteristics of the person are regarded, at least by researchers, as susceptible to development? How are these characteristics conceptualized, explicitly or implicitly, and how do various paradigms differ in this respect? The second sphere of variability also deals with properties of the individual, but from a different perspective; namely, how do the characteristics of the person at Time 1 influence the development of subsequent characteristics at Time 2? What theoretical possibilities for continuity or change are implied in the models employed, and how do the several models vary in this regard? The third area of potential variability lies in the conceptualization of the environment made explicit or implied in various paradigms. What properties of the external world incorporated in the model are envisioned as capable of affecting the course and consequence of human development? Moreover, within each of these three domains, variation may occur not only with respect to content, the kinds of characteristics that are distinguished within the domain, but also in structure, the way in which these characteristics are organized and related to each other. As might be expected, this structural aspect of a given domain represents one determinant of the degree of complexity of a given paradigm. The possibility of interrelation also defines a fourth dimension of variation implied in our reformulated Lewinian equation. Since that equation defines development as a joint function of person and environment, the question arises, what forms can such joint influence take? As we shall see, despite theoretical assertions to the contrary, most developmental investigations employ analytic models that assume only additive effects; that is, the influences emanating from the person and the environment are treated as operating independently of each other, with the net effect estimated from an algebraic sum of the various factors included in the model. Yet, in those instances in which more complex schemas have been applied, they have typically revealed important interactive effects. 1 This is an issue that Lewin himself never directly addressed, or - perhaps to put it more precisely it is an issue that he finessed by defining psychology as an ahistorical science, as explained later in the chapter.

Interacting systems in human development research paradigms

27

That is, particular environmental conditions have been shown to produce different developmental consequences, depending on the personal characteristics of individuals living in that environment; conversely, the same personal qualities at Time 1 may lead to different psychological consequences at Time 2, depending on the environmental conditions to which the individual has been exposed. Moreover, existing models that do allow for such combined effects vary in the nature and scope of the possibilities that are envisioned. Examples of each of these phenomena are cited below. A fifth domain of difference is also implied in the Lewinian equation as redefined. That domain is the mysterious realm designated by the letter/. The letter stands for ''function," which implies that the relation of Development to variously combined characteristics of the Person and the Environment is not one of mere statistical association but involves a set of processes through which the course and consequences of development are determined. There is good reason to call this fifth domain of variation "mysterious." Not only do some of the processes turn out to be quite convoluted - involving feedback mechanisms, sequential stages, and alternative paths of direct and indirect influence - but some paradigms fail to provide any indication whatsoever of what the processes might be. In other words, the mechanisms through which development is brought about are not specified in the model. One can only speculate about what the investigators may have had in mind, not excluding the possibility of a tabula rasa. Lewin (1931, 1935) had a term for models of this kind. He referred to them as Aristotelian, or class-theoretical. In such formulations, phenomena are "explained" by the categories to which they are assigned, as with Aristotle's four elements (earth, air, fire, and water). In opposition to such static concepts, Lewin argued for "field-theoretical" or "Galileian" paradigms that specify the particular processes through which the observed phenomenon is brought about. Typically in the historical development of a given science, class-theoretical categories precede field-theoretical formulations, but are gradually replaced by the latter. Lewin wished to accelerate the process. However, although half a century has gone by since the publication of Lewin's seminal essay, Aristotelian concepts and research models still abound in the scientific study of human development. Moreover, I suggest that, given the incomplete state of our knowledge, such paradigms may often be the strategies of choice for exploring uncharted domains. Like the surveyor's grid, they provide a useful frame for describing the new terrain. I also contend, however, that today, when much research is conducted in areas for which usable maps have already been constructed, the continued widespread use of class-theoretical models yields results that are not only likely to be redundant but also highly susceptible to misleading

28

Urie Bronfenbrenner

interpretations. This risk arises when class-theoretical paradigms are employed, as they frequently are, for field-theoretical purposes, that is, when the investigator goes beyond purely descriptive information to draw conclusions about causal processes that are not specifically assessed in the research design. Some illustrations of such unwarranted inferences are offered below. I now turn to a systematic analysis of existing and potential research paradigms in terms of the five domains outlined above. Class-theoretical models in research on human development Because, in our revised Lewinian formula, development is defined as an outcome of the factors and functions specified in the right-hand side of the equation, I shall proceed in the posited logical sequence, moving from right to left. I shall also make use of Lewin's key distinction to consider first those paradigms that employ the more simple, class-theoretical models; that is, in terms of the revised Lewinian formula, those in which the/term has been left undefined. We shall also discover that, in many of these designs, this is not the only omission. In some, only E is present; development is viewed solely as a product of environmental factors. In others, only P appears; development is examined only as a function of the characteristics of the individual at an earlier age. The social address model This class-theoretical model is an all-time favorite. Today, to an even greater extent than when it first appeared more than a century ago (Schwabe & Bartholomai, 1870), it is the paradigm most frequently employed for the study of human development in context. It is also an appropriate first choice on our dimension of simple to complex. Indeed, from this perspective, it verges on the simplistic, since it involves little more than the comparison of children or adults growing up in different geographical or social locations. Accordingly, in their historical review, Bronfenbrenner and Crouter have referred to this type of research design as a social address model. Among the most common social "addresses" appearing in the research literature are the following: social class, rural versus urban, different nationalities or ethnic groups, and, more recently, what Bronfenbrenner and Crouter have called the "new demography" - oneversus two-parent families, home care versus day care, mother's employment status, how many times remarried, or - perhaps soon - how many hours the father spends in child care and household tasks, or homes with and without computers. As already indicated, the structure of the social address model is rudimentary in several respects. First, not only is it limited to the environment, but even that

Interacting systems in human development research paradigms

29

environment is little more than a name. "One looks only at the social address that is, the environmental label - with no attention to what the environment is like, what people are living there, what they are doing, or how the activities taking place could affect the child" (Bronfenbrenner & Crouter, 1983, pp. 3823). In other words, as far as process is concerned, some feature of the environment existing at the social address, a feature that remains unspecified and uninvestigated, is presumed to bring about the observed developmental outcome through a mechanism that remains unidentified and unexamined. Second, a pure social address model involves yet another omission. In terms of the Lewinian equation, not only the/is absent, but also the P. This omission carries with it an often unrecognized assumption; namely, the developmental impact of a given social address is presumed to be the same for all persons living at that address, irrespective of their biological or psychological characteristics. The personal attributes model What kinds of paradigms do incorporate characteristics of the developing organism? Again we begin with the most rudimentary case, the analogue of the social address model in the personal domain. Like its counterpart, it involves the comparison of psychological characteristics in later life among individuals or groups, this time distinguished by particular personal characteristics present earlier in life. Since it is convenient to be able to refer to designs of this type by name, I call them personal attribute models. Thus far, the prior characteristics studied in this kind of paradigm have been primarily those identified by a biological or physical feature, such as age, sex, or - less frequently - the physiological state of the organism, such as premature birth, or, later on in life, pubertal changes and menarche, and physical body type. What about studies of the sequelae of prior psychological characteristics? There are many investigations of this kind, but almost all have been limited in scope, confined to examining continuity and change in the same characteristic over time, for example, constancy in IQ, temperament, or other personality traits. Like their environmental counterparts, personal attribute models neither specify nor investigate the processes through which characteristics observed early in life can affect the course of later development. Whenever some predictability is found, however, it is typically assumed, consistent with the truncated research paradigm being applied, that the observed relationship reflects the influence of biological factors. For example, changes with age are interpreted as products of maturation, the constancy of the IQ is viewed as an entirely genetic phenomenon, and the predictability of subsequent psychological development from prematurity or other physical characteristics of the newborn is presumed to

30

Urie Bronfenbrenner

reflect the influence of early physiological deficit or disorganization. However, such interpretations may themselves be premature. At a more general level, the principal shortcoming of the personal attributes model repeats in reverse the overgeneralizing error of its environmental counterpart. Like the social address paradigm, it is based on a broad and often unexamined assumption, in this instance the tacit presupposition that particular personal characteristics present early in life will have the same consequences for later development irrespective of the environments in which such later development takes place. It is clear that, like its environmental counterpart, the personal attributes model can be criticized for being incomplete, and even simplistic. It may therefore seem paradoxical, if not outright dim-witted, for any researcher to assert, as I do now, that the model has been underused and deserves a brighter scientific future. I hasten to add that what I have in mind is not the application of the design in its isolated form, but as an essential element in a more complex and comprehensive paradigm. Specifically, I suggest that underlying the personal attributes model is a valid and powerful principle of psychological development, well known in clinical and educational practice, but all too seldom subjected to systematic scientific investigation, namely, that particular psychological characteristics of the person at Time 1 can be extremely important in influencing the development of subsequent psychological characteristics at Time 2. Moreover, these subsequent characteristics may appear, at least on the surface, to be quite different from those manifested by the same person in the earlier period. Before we can discuss these possibilities in more concrete terms, however, we need to consider a set of research paradigms still in the class-theoretical mode, but at a somewhat higher level of structural complexity. Sociological niche model In the first six decades of their application, social address models were limited to the study of the developmental "effects" of various types of social addresses taken one at a time. Then, an advance not in developmental theory but in purely statistical design led to a new analytic schema. R. A. Fisher's development of analysis of variance (1925) permitted the assessment of the joint effects of two or more factors simultaneously in the form of interaction effects. This meant that one could analyze statistically the psychological characteristics of persons living at the intersection of two or more social addresses. In more concrete, substantive terms, it was now possible to identify certain social locations particularly favorable, or unfavorable, for psychological development (for example, single-parent mothers of low income and low education, with two or more children). I refer to such locations of differential risk as sociological niches.

Interacting systems in human development research paradigms

31

Person-context model Our next paradigm introduces a new element that involves no change in its formal (still class-theoretical) design but is of key importance for developmental theory; namely, in addition to multiple social addresses, the design also includes groups differing in their personal characteristics (for example, male versus female, before versus after puberty). In other words, categories of social address and personal attribute are crisscrossed with each other. Moving from the operational to the conceptual level, this means that the paradigm now incorporates a distinctive feature of Lewin's equation, the possibility, indeed the likelihood, that various combinations of environmental and personal characteristics can produce developmental effects that cannot be predicted from knowledge about either of these domains of influence examined independently of the other. Despite the fact that the paradigm is hardly complex from a theoretical or methodological perspective, it has all too seldom been applied in developmental research. As a result, very little is known about the extent to which the same environments can have different effects on human beings with differing personal characteristics. A striking finding demonstrates by contradiction the paucity of our knowledge in this sphere. The modest but significant association between family background factors in childhood and subsequent educational and occupational achievement in adulthood has been documented many times. Yet, it is only recently that Scarr and McAvay (1984) reported an important qualification with respect to this often-cited relationship. Exploiting the methodological leverage provided by a longitudinal study of brothers and sisters brought up in adoptive and biologically related families, the investigators demonstrated that, within biological families, such family background characteristics are much more predictive for sons than for daughters.2 The reason for this asymmetry is as yet unknown. Nor can it be determined from the kinds of research paradigms we have discussed thus far. For neither the social address model, the personal attributes model, the sociological niche model, nor their combination in the person-context model include within the paradigm itself any explicit definition, conceptual or operational, of the processes through which properties of the person or the environment, alone or in combination, function to produce a particular developmental outcome. As noted at the outset, all of these research designs reflect a class-theoretical mode of thought in which t h e / i n the Lewinian equation is left unspecified. Before examining paradigms that fill the gap, I wish, once again, to emphasize the scientific utility and unexploited potential of the class-theoretical designs 2 This sex or gender difference appears to have some generality across both temporal and social dimensions. In two independent longitudinal studies completed more than two decades ago (Kagan & Moss, 1962; Schaefer & Bayley, 1963), the investigators found that parental treatment in early childhood predicted subsequent behavior in adolescence significantly better for boys than for girls.

32

Urie Bronfenbrenner

we have just discussed, in particular, the more comprehensive person-context model. Such designs are especially useful in exploratory studies for the purpose of identifying aspects of the person, environment, or their combination that offer the most promise for more theoretically oriented investigations of process. Some especially promising possibilities in this regard are discussed below in the context of more complex process paradigms that also incorporate a person-context interface. Process paradigms in research on human development What processes do researchers assume to be operating in producing developmental change? Applying our empirical criterion, we find operational definitions for mechanisms of different types functioning at different levels. The first, and most common, pertain to the effects of proximal environmental and organismic influences on human development. Such proximal influences emanate either from within the person, or from physical features, objects, and persons in the immediate face-to-face setting. Bronfenbrenner (1976, 1979) has used the term microsystem to refer to this innermost region of person-environment interaction. Microsystem process models Up to the present time, four types of proximal influences on human development have received major research attention: 1. Studies of the genetic transmission of psychological characteristics. The re-

2.

3.

4.

search models here are extensions of Galton's classic nature-nurture paradigm (1876) involving the comparison of groups differing in the degree of consanguineous relationship. Investigations of the subsequent effects of the physical and physiological state of the organism in early childhood. The most common examples include prematurity, birth complications, temperament, and physical handicaps. Curiously, the influence of childhood illnesses on later development has seldom been investigated. Analyses of interpersonal interactions, relations, and attitudes, especially within the family. It is of some interest, and considerable consequence, that developmental research on family interaction using techniques of direct observation has by and large been limited to the periods of infancy and early childhood. By contrast, studies at older age levels have relied overwhelmingly on verbal reports by the subjects themselves or their acquaintances. As will become apparent from what follows, this separation by method has had the unfortunate consequence of delaying the development and application of several especially promising theoretical and operational paradigms for investigating psychological growth at all stages of the life course. Research on the developmental effects of the immediate physical environment. These investigations have focused on two types of variables: the availability of objects that enable and invite particular types of activity, such as toys or reading

Interacting systems in human development research paradigms

33

materials; and the structure of the setting with respect to barriers and pathways restricting or directing movement and activity. The first two of these four microsystem models clearly envisage a causal process emanating primarily from characteristics of the organism itself. The remaining two designs focus mainly on events in the child's immediate surroundings. Until the late 1960s and early 70s, each type of process was assumed to be entirely unidirectional with either the organism or environment seen as determining the behavior and development of the child. But, especially after the publication of Bell's influential paper (1968), "Reinterpretation of the Direction of Effects in Studies of Socialization," the research paradigms were modified to allow for and to analyze the possibility in each case of influences operating in both directions. It is of interest that the more recent studies in this domain, employing more rigorous designs, reveal that, despite the prominent role of child-initiated developmental trajectories in early infancy, the balance of power and developmental influence in exchanges between parent and child appears to lie more with the former than the latter, primarily as a function of the parent's greater knowledge and ability to structure the nature of the child's experience, even in the parent's absence (Bronfenbrenner & Crouter, 1983). As already mentioned, all four of these process models are limited to what I have referred to as the microsystem, the immediate environment of the developing person, within which direct manipulation and face-to-face communication are possible. What about causal processes originating farther afield? Process-context models Research paradigms for investigating causal processes affecting psychological development from outside the microsystem have been slow to emerge in the behavioral sciences. One reason for this delayed growth may lie in the history of psychology as a discipline. The systematic empirical study of psychological processes began in the laboratory. Moreover, the basic prototype was not the laboratory of the biological sciences, but that of physics, in which the processes sought for and analyzed were viewed as universal; for example, the law of gravity was assumed to be everywhere the same (at least before Einstein). Given this provenience, developmental psychologists, especially the more "scientific" among them, have not come easily to the thought that the processes they were observing under controlled conditions might operate differently for people in different situations. To be sure, it was both obvious and understandable that individuals and groups differed in levels of psychological functioning. But the mechanisms through which these varying levels were attained were presumed to be the same and to operate with equal effectiveness, wherever applied.

34

Urie Bronfenbrenner

The possibility that this might not actually be the case was difficult to recognize, mainly because it was not provided for in the operational models traditionally employed in developmental research. One of the earliest, and, in my judgment, theoretically most significant set of innovations appeared in a doctoral dissertation conducted at Harvard by Tulkin in 1970. His aim was to study the influence of socioeconomic status on socialization processes and outcomes in a sample of 10-month-old female infants and their mothers. The most noteworthy feature of the investigation, from the perspective of the present analysis, was the research design that evolved over the course of the investigation. In the end, it both provided for and demonstrated the influence of the environment not only on developmental outcomes, but also on the nature and effectiveness of the processes producing these outcomes. For example, Tulkin (1970, 1977) began by demonstrating that middle-class mothers engaged in more reciprocal interaction with their infants, and regarded both infants and themselves as more capable. He then went on to show that the correlation between these behaviors and attitudes was appreciably higher for the middle-class group. Moreover, the mother's involvement in reciprocal interaction when her infant was 10 months old predicted her child's performance in school six years later on tests of mental ability and language skill. As before, these correlations were stronger for the middleclass families, ranging in the high .50s to lower .60s. Perhaps even more important from a developmental perspective, the relationships of early maternal behavior to the child's behavior at age six were greater than the contemporaneous relationships between both types of variables in the first year of life. As the foregoing findings illustrate, Tulkin's implicit paradigm not only specifies some proximal processes through which developmental change is brought about, but also makes it possible to investigate whether and how these processes vary as a function of the broader context in which the process takes place (in this instance, the family's social class position). Hence the name process-context model. Although Tulkin's research illustrates the critical features of a process-context paradigm, he did not exploit its scientific potentialities to the full; additional steps could have been taken that would have shed more light on the influence of context on developmental processes. Thus Tulkin showed that the effect of the mother's reciprocal interaction on school outcomes was stronger in middle-class than in working-class families. He also demonstrated that the correlation between maternal attitudes and behavior was stronger at higher socioeconomic levels. But he did not go on to determine whether the impact of reciprocal mother-child activity on school outcomes was greater for mothers who, when their children were only 10 months old, had viewed both the infants and themselves as more capable, and whether this relationship, in turn, was stronger for middle-class than for working-class families.

Interacting systems in human development research paradigms

35

Tulkin's implicit paradigm is noteworthy in yet another respect. His conceptions of process include not only objective behaviors (e.g., patterns of motherinfant interaction) but also subjective psychological states (e.g., maternal belief systems about child rearing), and, what turns out to be even more critical, the relationship between these two process domains. I suggest that such simultaneous consideration of subjective and objective factors, both in developmental processes and outcomes, constitutes a dual component for designing more effective research models. Although Tulkin's study reflects a significant advance in scientific strategy, the addition of yet another element can produce an even more revealing analytic system. Process-person-context models The nature and scope of these paradigms is perhaps best conveyed by a concrete example drawn from an analysis of data from a study of the developmental impact on children of the transition from home to school (Bronfenbrenner & Cochran, 1979; Cochran & Woolever, 1982). This research provided an index of the frequency of joint activities engaged in by the mother with her infant when the latter was 3 years old. This measure was positively related to teacher evaluations, obtained several years later, of the child's performance and behavior in school. Thus far, we have an example of a simple process paradigm. The next analysis revealed that, consistent with a pattern of findings reported earlier in this chapter, the relationship between joint mother-child activity and subsequent school outcomes was significantly stronger for boys than for girls. Our paradigm has now been expanded to what might be called a "person-process" model. In the next phase, a contextual variable was added to the regression equation, specifically, the number of years the mother had gone to school. Although maternal education and the mother's joint activity with the child were correlated with each other, each was found to exert an independent effect on school outcomes, and both of these effects were stronger for boys than for girls. Finally, to assess the joint influence of the mother's schooling and her joint activity, the product of these two variables was also added to the regression equation. Solutions for points of maximum and minimum value yielded an intriguing result. It turned out that the mother's joint activity facilitated the child's performance in elementary school only if the mother herself had some education beyond high school, and the more higher education she had, the greater the effect of her joint activity on the child's achievement at school. By contrast, for mothers with a high school education or less, the relationship was nonsignificant, and negative in sign. Here we see the process-person-context model in its entirety. The power and,

36

Urie Bronfenbrenner

in this instance, even the direction of the process is shown to vary as a joint function of the context (the mother's educational background) and of characteristics of the person (the gender of the child). Moreover, the effects of person and context are not merely additive, but interactive, since the joint impact of education and activity was significantly greater for boys than for girls. With this example before us, we are now in a position to specify the defining properties of a process-person-context model. The first requirement is that the design provide for systematic information in at least three separable domains: (1) the context in which development is taking place; (2) the personal characteristics (biological or psychological) of the persons present in that context; and (3) the process through which their development is brought about. But the mere presence of information in these three spheres is not enough. A second sine qua non of the paradigm is analysis of the possibility that the potency and even the direction of the process may vary as a joint function both of the properties of the context and of the characteristics of the developing person. In short, the process is subject to the interactive moderating effects of both person and context. As I have indicated elsewhere (Bronfenbrenner, 1984a), the nature and range of such moderating effects can be depicted in formal terms as segments of a geometric surface known as a hyperbolic paraboloid. The surface defines moderating effects of two general types - positive and negative. If we consider the minimal possible care of two moderating factors (for example, one varying attribute of the person, the other of the context), a positive moderating effect is one in which each factor enhances any positive moderating influence of the other, but buffers any negative influence the latter may have. Thus, in the example cited from the school transition study, each additional year of schooling tended to increase the positive influence of the mother's joint activity on the child in the case of mothers with some education beyond high school, but to reduce the negative impact of such activity from mothers with no education beyond high school. A negative moderating effect is one in which each factor exacerbates any negative influence of the other, or undermines any positive influence the latter may exert. A prime example of such a negative moderating factor is poverty. Its destructive impact is especially salient in a society like the United States, where money is a key to meeting essential family needs (such as health care). The example of poverty is also useful for highlighting the distinction between moderating factors and those that represent links in a direct or indirect causal chain. I refer to the latter as mediating factors. The same environmental condition can often function in both modes simultaneously, poverty being a case in point. For example, Elder (1974; Elder, Caspi, & Downey, 1986; Elder, Caspi, & van Nguyen, 1986) showed that loss of income through the father's unemployment increased the likelihood of family conflict, with resulting ill effects on the

Interacting systems in human development research paradigms

37

development of the children. Moreover, this disorganizing process was most severe in those families in which the father had been described as a difficult and irascible person. Finally, the destructive effects of this overall pattern were most pronounced in those families that had a low income to start with. In other words, poverty not only set the vicious circle in motion (a mediating effect), but also accelerated its downward course (a moderating effect). It is also possible, under certain ecological conditions, for moderating effects to change direction, that is, to transform negative forces into positive ones, or vice versa. An example of such a reversal is documented in Crockenberg's research on the effects of support networks on the mother-infant relationship. In her initial study (1981), she had demonstrated that such networks foster the development of maternal responsiveness, particularly if the infant is irritable. In a subsequent investigation, however, Crockenberg (1987) showed that, for mothers living under highly stressful environmental conditions, social networks not only cease to exert a positive influence, but became a source of stress. A similar result is reported in a recent paper by Riley and Eckenrode (1986). In a study of stresses and supports in mothers' lives, the investigators found that the influence of social networks on psychological well-being shifted direction from positive to negative as a function of three kinds of factors: (1) reduced socioeconomic status; (2) the occurrence of misfortune in the lives of significant others (e.g., a close relative suffers an accident); or (3) low levels of belief either in one's capacity to influence one's own life (i.e., locus of control) or in the probable success versus failure of one's own help-seeking efforts. Processes and outcomes of social support are set within a still broader social context in Crockenberg's study (1985) of English teenage mothers and a matched sample of their American counterparts. She found that "English mothers engaged in more smiling and eye contact, less frequent routine contact, and responded more quickly to their babies' crying than did American mothers" (pp. 413-14). Control of possibly confounding variables through regression analysis pointed to the amount and type of social support as a key factor accounting for the difference. Crockenberg elaborates: In the United States most mothers rely on private doctors to serve their own and their children's health needs. . . . Public health nurses or social workers may be assigned to families in need of special assistance, but there is no comprehensive system designed to provide health-related and child care advice to parents. In contrast, England incorporates community-based social support for parents in a comprehensive program of health care. This care begins before the child's birth and continues through the school years. . . . Midwives provide postnatal care for mothers and babies after they leave the hospital following delivery, and home health visitors see new mothers on a regular basis. . . . In England, mothers had only to be home and open their doors, (pp. 414-15) In a passage from an earlier version of this research report that was unfortunately excluded from the somewhat condensed published version, Crockenberg

38

Urie Bronfenbrenner

states that her data raise questions about a common view of health and social services in the United States summarized by de Lone as follows: "Large-scale efforts will be implemented with variable quality, and the modal performance will be, by qualitative standards, mediocre" (de Lone, 1982, p. 489). The foregoing examples illustrate the double-edged scientific power of the process-person-context model, for it is at once both more rigorous and more revealing. On the one hand, it operates conservatively to guard against unwarranted conclusions regarding the generalizability, across different types of people or environments, of particular processes affecting the course of human development. On the other hand, the paradigm can illuminate the ways in which particular characteristics of the person or the context can influence such processes, either as mediating factors, moderating factors, or as both. This last feature bears further examination since, in effect, it extends the analysis of process beyond the immediate setting to more remote regions both of the person and the environment. Thus, in the former, personal domain, it highlights the conceptual structures and strategies typically employed by the individual in interpreting and manipulating the outside world. This process of interpretation and manipulation in turn influences the ways in which the environment can affect the course of subsequent development. In short, the use of a process-person-context model permits assessment of the individual's contribution to his or her own development. On the environmental side, the paradigm's primary focus on process forces the investigator to confront the issue not only of the proximal mechanisms through which features of the immediate setting bring about developmental change (or stability) in the individual, but also of the distal mechanisms through which features of the environment beyond the immediate setting can influence the power and direction of the proximal processes that affect development directly. To be more specific, in previous publications (Bronfenbrenner, 1976, 1977, 1979; Bronfenbrenner & Crouter, 1983), I have proposed a conceptualization of contexts of development in terms of a hierarchy of systems at four progressively more comprehensive levels: 1. The microsystem, which involves the structures and processes taking place in an immediate setting containing the developing person (e.g., home, classroom, playground, etc.). 2. The mesosystem, which comprises the linkages and processes taking place between two or more settings containing the developing person (e.g., the relations between home and school, school and workplace, etc.). In other words, a mesosystem is a system of microsystems. 3. The exosystem, which encompasses the linkages and processes taking place between two or more settings, at least one of which does not ordinarily contain the developing person, but in which events occur that influence processes within the immediate setting that does contain that person, (e.g., for a child, the

Interacting systems in human development research paradigms

39

relation between the home and the parent's workplace; for a parent, the relation between the school and the neighborhood peer group). 4. The macro system, which is defined as an overarching pattern of ideology and organization of the social institutions common to a particular culture or subculture. In other words, the macrosystem comprises the pattern of micro-, meso-, and exosystems characteristic of a given society or segment thereof. It may be thought of as a societal blueprint for a particular culture or subculture. The relevance of this more differentiated conceptualization of the environment for the present discussion lies in the fact that the process-person-context design provides for its operationalization; that is, it makes possible the analysis of the mediating and moderating processes that constitute the linkages between and within the environmental systems shaping the course of human development. In addition, the paradigm incorporates as a key domain, having equal status with the environment, a set of factors all too frequently omitted in studies of development in context - the contribution to development of the biological and psychological characteristics of the persons involved in the process. These scientific virtues notwithstanding, full-fledged process-person-context models are still comparatively rare in developmental research. Somewhat more common are truncated designs in which one of the principal moderating elements, either the person or the context, is missing. Paradoxically, given the fact that most developmental research is conducted by psychologists, it is the differentiating characteristics of the individual that are most often left unspecified. There are more process-context models to be found in the literature than processperson paradigms. As a result, we are typically left, by default, with the usually unacknowledged assumption that the processes operating in the given context affect the development of all persons in that context in the same way. Such recognition of missing elements and unstated but challengeable assumptions can also be viewed in a more positive light; for, also by default, it defines a promising, and as yet little explored, terrain for future research. The prospect is scientifically promising because, to judge from the limited evidence generated thus far through the still infrequent use of more complex models, it would appear that prevailing research paradigms considerably underestimate the role of the person's own psychological characteristics in affecting the course of his or her future development, be it for better or for worse. If we now broaden the perspective further to add uncharted domains in the environment, the prospects for extending scientific knowledge appear even brighter. At the present time, studies of process in developmental research are limited almost entirely to the family. The extension of process-person-context models beyond this core microsystem should enable us to look beyond the superficial labels of place and person; to identify the human realities that lie behind such uninformative categories as class, gender, race, or residence; and to

40

Urie Bronfenbrenner

discover the processes through which these realities shape the kinds of human beings that we continuously become. Having sought to persuade the reader of the superiority of the process-personcontext model over most of its contemporaries, I shall now, perversely, point to a major lacuna in this powerful design, the omission of an entire domain that is critical for the study of development. The dimension is missing not only from all of the paradigms we have examined thus far; it is absent even from the revised Lewinian equation that has served as the conceptual framework for our analysis. The missing parameter is the very dimension along which development occurs - the dimension of time. Lewin's failure to include this factor in his formula was not accidental, but deliberate. In his view (1931), the Galileian mode of thought was by its very nature ahistorical. In psychology as in physics, he argued, present events can be influenced only by forces operative in the present situation. It was perhaps Lewin's predilection for the paradigms of physics, and their ahistorical orientation, that led him, and many other psychologists as well, to be far more interested in the study of behavior than of development. The fact remains, however, that, as mentioned above, the process of human development cannot be defined except in relation to time, since the central concern of developmental study is the nature of continuity and change in the biological and psychological structures of individual human beings throughout their life course. Moreover, it is along this same temporal dimension that another major advance has recently taken place in the design of research paradigms for investigating psychological growth. Chronosystem paradigms Traditionally in developmental science, the passage of time has been treated as being synonymous with chronological age, that is, as a scale for ordering individuals in terms of how long they have lived. Moreover, age was viewed as a purely personological construct, that is, one presumed to reflect developmental changes within the individual essentially unrelated to external conditions or events. Especially during the past decade, however, research on human development has projected the factor of time along a new axis. Since the mid-1970s, an increasing number of investigators have formulated research designs in which time is employed not only for ordering individuals according to age, but also for ordering events in their historical sequence and context. I have referred to designs of this kind as chronosystem models (Bronfenbrenner, 1984b). True to their nature and name, such models have a history that antedates their present popularity. Thus the first formal, systematic critique of chronological age as a purely personological construct appeared in the work of Baltes and Schaie in

Interacting systems in human development research paradigms

41

the late 1960s and early 70s (Baltes, 1968; Baltes & Schaie, 1973; Schaie, 1970). The sophisticated analyses conducted by these investigators demonstrated the confounding effects of age and cohort. At a substantive level, Baltes and Schaie argued that cohort differences reflect the influence of diverse historical contexts. Not only do persons in the same age group share a life history of common experience, but those of a given age in different generations could have quite diverse experiences, depending on the period in which they live. The research paradigms that these investigators employed to unravel the confounding were entirely statistical in nature. No effort was made to identify the particular experiences that influenced the person's development, or the specific effects that these experiences produced. Studies of the latter kind, however, were already present in the research literature, although their scientific significance and power generally remained unrecognized. For instance, in the late 1940s, Baldwin (1947) reported dramatic changes in the mother's behavior toward the first child before, during, and after the mother's pregnancy with the second child. A decade earlier Stone and Barker (1937, 1939) described the marked shifts in interests and preferences exhibited by adolescent girls before and after menarche, even after control for chronological age. Both of these early examples illustrate the defining property of a chronosystem model; its design permits one to identify the impact of prior life events and experiences, singly or sequentially, on subsequent development. These experiences may have their origins either in the external environment (e.g., the birth of a sibling), or within the organism (e.g., the first menstruation). A further distinction, introduced by Baltes and his colleagues (Baltes, 1979; Baltes, Reese, & Lipsett, 1980) differentiates between normative experiences (school entry, puberty, entering the labor force, marriage, retirement) and nonnormative events (a death or severe illness in the family, divorce, moving, winning the sweepstakes) . Experiences of both types occur throughout the life span and may often serve as the impetus for developmental change. At least three types of research strategies have been used in chronosystem designs. 1. The most common is a cross-sectional design in which developmental outcomes are compared between different groups of subjects who have and have not been exposed to a particular life experience, with appropriate controls for possibly confounding factors. Stone and Barker's (1937) study of the impact of menarche in altering patterns of motivation and interest among adolescent girls provides an early example. The groups were matched to rule out possible effects both of chronological age and socioeconomic status. While profiting from the advantages of a chronosystem design, this pioneering study lacked a number of key elements found in other paradigms that we have examined, features that considerably enhance their scientific power. For example, although Stone and Barker speculated that the psychological changes provoked by menarche may involve an interaction between biological and social factors, their model did not

42

Urie Bronfenbrenner

2.

3.

provide for any investigation of such joint effects (for example, the possible moderating influence of social class or personal belief systems on the psychological impact of first menstruation). In short, Stone and Barker did not incorporate a process-person-context model in their chronosystem design. As we shall see, such incorporation offers special promise for future research on human development. A second strategy for implementing a chronosystem model involves a shortterm longitudinal design in which data are obtained for the same group of subjects both before and after a particular life experience or life transition. This strategy is particularly applicable to normative events, since their occurrence can be anticipated. Baldwin's study of changes in parent-child interaction induced by the impending arrival of a sibling is a case in point. As has been argued elsewhere (Bronfenbrenner & Crouter, 1983), a design of this kind has special advantages from the viewpoint of scientific method, since it represents a ready-made experiment of nature in which the usual sources of confounding are eliminated, since each subject serves as his or her own control. Moreover, the fact that life transitions occur throughout the life course provides opportunities for studying development during the stage that is both the longest and the one about which we have the least knowledge - adulthood between the ages of 25 and 60. This same consideration applies with even greater force to the most powerful form of chronosystem design, a long-term longitudinal investigation that examines the often cumulative effects of particular sequences of transitions, what Elder (1974) has referred to as the life course. Elder's studies of the systematically different developmental paths set in motion by the Great Depression of the 1930s represent both a prototype and, in a number of respects, a paragon for this type of research design.

I use the term "paragon" advisedly, because, without being explicit on either issue, Elder manages to avoid two major pitfalls that characterize most of the research designs currently being employed in the study of life-course development. Both pitfalls involve a typically unrecognized theoretical oversimplification. The first danger derives from the uncritical use of an especially powerful methodological tool - causal modeling - especially in the highly sophisticated form in which it has been developed and adapted for computer use by the Swedish statisticians Joreskog and Sorbom (1976). Although path analysis as a statistical tool for analyzing causal processes was conceived and applied in the field of genetics by Sewell Wright half a century ago (1934), its application in developmental research has been relatively recent. Structural models, being purely abstract and mathematical in nature, of course do not in themselves constitute theoretical conceptions of development in context. They have nevertheless made a significant conceptual contribution by highlighting the importance of, and providing a method for analyzing, two types of phenomena that have turned out to be frequent and often critical features of developmental processes: bidirectional influences on the one hand, and direct versus indirect

Interacting systems in human development research paradigms

43

effects on the other. However, the increasing use of structural equations for exploring these important phenomena has often led investigators to overlook other process possibilities not allowed for in the orthodox path-analytic paradigm. The omission derives from the fact that, with all their mathematical and statistical complexity, structural equations, as typically employed, are essentially main-effect models, with no allowance made for possible interactions. To state the issue in substantive and theoretical terms, if an investigator relies exclusively on a conventional causal model, developmental sequences are implicitly conceptualized and explicitly analyzed as if they were invariant across both person and situation. No matter how many pathways through life are found, or how convoluted each course may be, once travelers embark on a particular route, they are seen as proceeding at the same pace to the same place, irrespective of who they are, whence they came, or the nature of the terrain they may be traversing. To move from metaphor to concrete cases, it is a common practice in applying path-analytic models in developmental research to introduce either as exogenous or control variables such factors as socioeconomic status, age, family size, or even sex of child. Such a procedure assumes that any developmental processes or effects under investigation are invariant across these parameters. As has been demonstrated in the preceding discussion, such invariance can hardly be taken for granted. The preceding discussion also points the way to a more powerful paradigm that preserves the advantages of a path-analytic model while avoiding its questionable universalistic assumptions. The preferred alternative is to incorporate structural equations within the larger framework of a process-person-context model. In effect, this means calculating separate path diagrams for each of the potential moderator factors included in the design; for example, whether and how path diagrams may differ for males versus females, lower-class versus middleclass families, single-parent versus two-parent households, and so on. 3 This was the strategy employed by Elder. It can yield new knowledge of considerable theoretical and practical significance. For instance, more than a decade ago, Scarr-Salapatek (1971) showed that middle-class children were more likely to realize their genetic potential than the offspring of lower-class families. In a different domain, research teams working independently both in West Germany 3 To be sure, no method exists as yet for testing the significance of differences between two formally analogous structural models. It is possible, however, to test differences between corresponding unstandardized regression coefficients. Moreover, purely technical considerations should not deter investigators from looking at relevant data from any perspective that can reduce the danger of unwarranted generalization by calling attention to possible variations in process as a function of context, personal characteristics, or some combination of the two.

44

Urie Bronfenbrenner

(Schneewind, Beckmann, & Engfer, 1983) and in the United States (Elder, Caspi, & Downey, 1986) have found that economic hardship exerts its impact on family processes and the child's subsequent development primarily as a function of personality characteristics of the parent of the same sex. What is a developmental outcome? The second pitfall besetting contemporary studies of the life course forces us to confront, at long last, the first and most definitive term in our revised Lewinian equation - the D that stands for Development. If development is the product of interaction between person and environment, how are we to distinguish it from its prototypic predecessor - the B that stands for Behavior? And are there other outcomes of the conjoint process that further complicate the task of identifying the true object of study - namely, continuity and change over time in the psychological characteristics of the person? The answers to these questions are by no means obvious. This fact becomes readily apparent when we examine some of the variables that have been taken as measures of developmental outcome in studies of life-course development. They have included, among other things, the number of years of schooling ultimately completed, achieved occupational status, and whether or not the person is on welfare. To be sure, these are clearly way stations on the life course, and often also ends of the line, but are they also appropriate measures of human development? Implicit in this question is a basic theoretical issue: What is the conceptual definition of a developmental outcome? Somewhat to my surprise, I have not been able to find a formal definition of this construct in the research literature. Operational definitions of course abound in great variety. They include everything from indices of life-course status of the type mentioned above to highly specific responses in laboratory experiments, results of psychological tests, clinical assessments, descriptions and ratings by personal acquaintances, and systematic appraisals based on coded interviews or observations conducted in real-life situations. Clearly, such a diverse array precludes a quick and easy calculation of a theoretically lowest common denominator. Under these circumstances, to pursue the issue to this critical point, I am faced with two alternatives, neither of which is pleasant to contemplate. The first is to resort to the hackneyed device of declaring the problem a priority for further study, clearly a case of passing the buck. The second option is to go out on a limb by proposing a presumably more adequate conceptual formulation of my own. If the proposal is to have any theoretical coherence, it must of necessity rule out one or more operational definitions that are currently regarded as quite acceptable by one or another group of established researchers in the field. Since the first

Interacting

systems in human development

research paradigms

45

alternative risks criticism from all sides, whereas the second may be somewhat more selective in provoking objections, I shall opt for the latter. The proposed conceptual definition of a developmental outcome takes the form of a series of propositions as follows: 1.

2.

3.

4.

5.

Psychological development is a process that, in the last analysis, takes place within the mind; it involves the evolution, through the life course, of established patterns of mental organization and content that are characteristic of the particular person. Such established patterns are referred to as developmental outcomes. Patterns of mental organization and content cannot be observed directly in the brain (at least not yet); they can only be inferred from the characteristic ways in which the person subjectively experiences and objectively deals with the world in which he or she lives (including perceptions of the behaviors toward the self). The term "evolution" refers to change over time; yet, the expression "established pattern" similarly implies some degree of stability in the emergent product of change. Hence the demonstration of a developmental outcome requires evidence of patterns of subjective experience and objective behavior that exhibit some degree of continuity across space and time but have their origins in conditions, events, and processes taking place at an earlier period in the life of the person. Moreover, since psychological development has been defined as a process, a full demonstration of its occurrence must also identify one or more mechanisms through which particular prior conditions or events affect subsequent patterns of subjective experience and behavior. The expression "patterns of subjective experience and objective behavior" also implies a functional relationship between these two domains. Therefore, a full description of developmental processes should encompass both subjective and objective aspects of human activity and document the functional relationship between them.

Operational definitions of developmental outcomes typically address one or the other of the several elements outlined in the foregoing propositions, but rarely, as they were in Tulkin's work, have more than one or two of them been integrated in the same research design. For this reason, it is useful to distinguish between a narrow, or minimal, definition of a developmental outcome and one that is more comprehensive. A.

B.

Under a narrow operational definition, a developmental outcome is considered established if it can be shown that observed patterns of subjective experience or of objective behavior can be predicted from conditions or events taking place at an earlier period in the life of the person. Implicit in this minimal formulation is the dual assumption that, if past circumstances predict present patterns, then the former must have had some role in producing the latter, and the latter, in turn, must have exhibited some kind of functional continuity over time. Two important elements are missing from the above definition and are therefore added in the broader formulation. 1. The first is the requirement to specify one or more processes through which past conditions or events bring about the evolution of established patterns of psychological functioning. As the reader will have observed, the narrow operational definition does not call for such specification. Hence a social

46

Urie Bronfenbrenner

2.

address model, a personal attributes model, or a combination of the two in a person-context model can be adequate to the task. By contrast, the broader operational definition requires the addition of process terms to the equation. The second additional requirement calls for the specification of developmental patterns in both the subjective and the objective domains, as well as an analysis of the functional relationship between them. In the narrow definition, the identification of patterns in only one or the other of these domains is deemed sufficient. When a dual domain model is employed, it is also essential, again as illustrated in Tulkin's research, that the subjective and objective elements included be selected on a theoretical rather than an ad hoc basis, so that there must be some possibility of demonstrating a functional relationship between the particular beliefs and behaviors under investigation.

Up to the present time, only a few studies have followed the broader operational definition, but these few have been especially valuable scientifically. In addition to the work by Tulkin and his colleagues referred to earlier in the chapter, another outstanding example is Elder's demonstration in a series of studies extending over a decade (Elder, 1974; Elder, Caspi, & Downey, 1986; Elder, Caspi, & van Nguyen, 1986) of the impact of the Great Depression in shaping both the behavior systems and belief systems of his subjects in later life. A third investigation that approximates the broader criteria is Mortimer's longitudinal investigation of life-course patterns leading to professional and managerial success (Mortimer & Kumka, 1982; Mortimer & Lorence, 1979; Mortimer, Lorence, & Kumka, 1982). Mortimer and her colleagues have shown that occupational attainment in American men is the product, in part, of value orientations fostered by processes beginning in the family of origin but ultimately encompassing experiences and linkages in educational and work settings, as well as in the men's marital relationships. The research falls short of the broader operational definition only in the fact that measures of occupational achievement are limited to indices of career progress and income and do not extend to actual performance on the job. Retrospect and prospect Despite the diversity of models and materials reviewed in this chapter, it is not difficult at a purely abstract level to suggest combinations of paradigm properties, and their constituent interacting systems, that could appear especially likely to lead to scientific progress and effectiveness of future research on human development in context. It may be no easy matter, however, to integrate and implement these properties and systems within a single research design. To begin with the easier task, in the light of the ideas and evidence we have examined, a productive design for research on development in context might well incorporate the following system elements:

Interacting 1. 2. 3.

systems in human development research paradigms

47

The use of a process-person-context model involving a contrast between at least two settings, and between two groups of subjects distinguished by differing personal characteristics. Observations in all three domains of the above model, as well as data on developmental status, should be obtained at two or more points in time. The measures of developmental status should include assessments of behavioral and experiential functioning, as well as assessments of the interrelations between these two domains.

It will be observed that the above elements encompass key features not only of a ' 'person-process-context'' model but also properties of a number of other system paradigms examined in this chapter, in particular, the chronosystem, meso- or exo-system, and a developmental outcomes model. It is even more obvious that few researchers will have the time, resources, or, for that matter, the scientific "chutzpah," to undertake so ambitious an endeavor. To be sure, as we have seen, a number of investigators have already come fairly close to bringing off such an awesome enterprise quite successfully; only one or two of the specified elements were missing. Invaluable as such comprehensive efforts are, it would be a mistake to conclude that the far more modest undertakings that most of us engage in are not, in the long run, equally if not more important for scientific advance. The fact that developmental processes do vary as a joint function of process, person, and context does not mean that these variations are infinite in number and variety, for they are subject to the constraints and imperatives of a particular biological species. A basic tenet of the ecological perspective implies a fit between the characteristics of a living organism and its surroundings. Like other species, we are limited in the kinds of ecologies in which we can survive and develop. It becomes the task of today's developmental science to identify the nature of this finite set of ecological conditions and to assess their as yet unprobed potential for sustaining and fostering human growth. The ideas and models presented here are offered as an aid to that endeavor. Their purpose is not to establish a set of criteria that every researcher should strive to meet, but rather to provide an array of promising systems paradigms from which the investigator can choose the most suitable theoretical and practical alternatives, and, at the same time, be able to recognize and to specify the ambiguities of interpretation created by the omission of important elements in the selected design, so that others can fill in the gaps. If this discussion contributes to the achievement of this modest goal, it will have served its highest hope.

References Baldwin, A. L. (1947). Changes in parent behavior during pregnancy. Child Development, 18, 2939.

48

Urie Bronfenbrenner

Baltes, P. B. (1968). Longitudinal and cross-sectional sequences in the study of age and generation effects. Human Development, 11, 145-71. Baltes, P. B. (1979). Life-span developmental psychology: Some converging observations on history and theory. In P. B. Baltes & O. G. Brim (Eds.), Life-span development and behavior (Vol. 2). New York: Academic Press. Baltes, P. B., Reese, H. W., & Lipsett, L. P. (1980). Life-span developmental psychology. Annual Review of Psychology, 31, 65-110. Baltes, P. B., & Schaie, K. W. (1973). Life-span developmental psychology: Personality and socialization. New York: Academic Press. Bell, R. Q. (1968). A reinterpretation of the direction of effects in studies of socialization. Psychological Review, 75, 81-95. Bronfenbrenner, U. (1976). The experimental ecology of education. Teachers College Record, 78, 157-204. Bronfenbrenner, U. (1977). Toward an experimental ecology of human development. American Psychologist, 32, 513-31. Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press. Bronfenbrenner, U. (1984a). The graphic analysis of moderating effects. Ithaca, NY: Cornell University, Department of Human Development and Family Studies. Bronfenbrenner, U. (1984b). The ecology of the family as a context for human development: Research perspectives. Position paper prepared at the request of the Human Learning and Behavior Branch of the National Institute of Child Health and Human Development as a contribution to the preparation of its Five-Year Plan. Bronfenbrenner, U., & Cochran, M. (1979). The comparative ecology of human development. Research methods and designs. Ithaca, NY: Cornell University, Department of Human Development and Family Studies. Bronfenbrenner, U., & Crouter, A. C. (1983). The evolution of environmental models in developmental research. In W. Kessen (Ed.), Handbook of child psychology: History, theories, and methods (Vol. 1). New York: Wiley. Cochran, M., & Woolever, F. (1982). Beyond the deficit model: The empowerment offamilies with information and informal supports. Ithaca, NY: Cornell University, Department of Human Development and Family Studies. Crockenberg, S. B. (1981). Infant irritability, mother responsiveness, and social support influences on the security of infant-mother attachment. Child Development, 52, 857-65. Crockenberg, S. B. (1985). Professional support and care of infants by adolescent mothers in England and the United States. Journal of Pediatric Psychology, 10, 413-28. Crockenberg, S. B. (1987). Support for adolescent mothers during the postnatal period: Theory and research. In C. F. Z. Boukydis (Ed.), Research on support for parents and infants in the postnatal period. New York: Ablex. Crockenberg, S. B. (in press). English teenage mothers: Attitudes, behavior, and social support. In E. J. Anthony (Ed.), International yearbook series of the International Association for Child Psychiatry and Allied Professions. de Lone, R. H. (1982). Early childhood development as a policy goal: An overview of choices. In L. A. Bond & J. M. Joffee (Eds.), Facilitating infant and early childhood development. Hanover, NH: University Press of New England. Elder, G. H., Jr. (1974). Children of the Great Depression. Chicago, IL: University of Chicago Press. Elder, G. H., Jr., Caspi, A., & Downey, G. (1986). Problem behavior and family relations: Life course and intergenerational themes. In A. Sorensen, F. Weinert, & L. Sherrod (Eds.), Human development and the life course: Multidisciplinary perspectives. Hillsdale, NJ: Erlbaum.

Interacting systems in human development research paradigms

49

Elder, G. H., Jr., Caspi, A., & van Nguyen, T. (1986). Resourceful and vulnerable children: Family influences in hard times. In R. K. Silbereisen, K. Eyferth & G. Rudinger (Eds.), Development as action in context: Problem behavior and normal youth development. New York: Springer. Fisher, R. A. (1925). Statistical methods for research workers. Edinburgh: Oliver & Boyd. Gal ton, F. (1876). The history of twins as a criterion of the relative power of nature. Anthropological Institute Journal, 5, 391-406. Joreskog, K. G., & Sorbom, D. (1976). Statistical models and methods for analysis of longitudinal data. In D. J. Aigner & A. S. Goldberg (Eds.), Latent variables in socioeconomic models. Amsterdam: North Holland Press. Kagan, J., & Moss, H. A. (1962). Birth to maturity: A study in psychological development. New York: Wiley. Lewin, K. (1931). The conflict between Aristotetlian and Galileian modes of thought in contemporary psychology. Journal of Genetic Psychology, 5, 141-77. Lewin, K. (1935). A dynamic theory of personality. New York: McGraw-Hill. Mortimer, J. T., & Kumka, D. (1982). A further examination of the "occupational linkage hypothesis." Sociological Quarterly, 23, 3-16. Mortimer, J. T., & Lorence, J. (1979). Work experience and occupational value socialization: A longitudinal study. American Journal of Sociology, 84, 1361-85. Mortimer, J. T., Lorence, J., & Kumka, D. (1982). Work and family linkages in the transition to adulthood: A panel study of highly educated men. Western Sociological Review, 13, 50-68. Riley, D., & Eckenrode, J. (1986). Social ties: Subgroup differences in costs and benefits. Journal of Personality and Social Psychology, 22, 770-8. Scarr, S., & McAvay, G. (1984). Predicting the occupational status of young adults: A longitudinal study of brothers and sisters in adoptive and biologically-related families. Unpublished manuscript, Department of Psychology, University of Virginia. Scarr-Salapatek, S. (1971). Race, social class, and IQ. Science, 174, 1285-95. Schaefer, W. S., & Bayley, N. (1963). Maternal behavior, child behavior and their intercorrelations from infancy through adolescence. Monographs of the Society for Research in Child Development, 28, 1-27. Schaie, K. W. (1970). A reinterpretation of age-related changes in cognitive structure and functioning. In L. R. Goulet & P. H. Baltes (Eds.), Life span developmental psychology: Research and theory. New York: Academic Press. Schneewind, K. A., Beckmann, M., & Engfer, A. (1983). Eltern und Kinder. Stuttgart, West Germany: Kohlhammer. Schwabe, H., & Bartholomai, F. (1870). Der Vorstellungskreis der Berliner Kinder beim Eintritt in die Schule. In Berlin und seine Emwickelung: Stadtisches Jahrbuchfur Volkswirthschaft und Statistik: Vierter Jahrgang. Berlin: Guttentag. Stone, C. P., & Barker, R. G. (1937). Aspects of personality and intelligence in postmenarcheal and premenarcheal girls of the same chronological ages. Journal of Comparative Psychology, 23, 439-55. Stone, C. P., & Barker, R. G. (1939). The attitudes and interests of premenarcheal and postmenarcheal girls. Journal of Genetic Psychology, 54, 27-71. Tulkin, S. R. (1970). Mother-infant interaction in the first year of life: An inquiry into the influences of social class. Unpublished doctoral dissertation, Harvard University. Tulkin, S. R. (1977). Social class differences in maternal and infant behavior. In P. H. Leiderman, A. Rosenfeld, & S. R. Tulkin (Eds.), Culture and infancy. New York: Academic Press. Wright, S. (1934). The method of path coefficients. Annals of Mathematical Statistics, 5, 161-215.

Children, families, and communities: ways of viewing their relationships to each other

Jacqueline J. Goodnow

It is standard practice in the social sciences to say that people are influenced by their environments as well as by internal forces and that the influence occurs in a two-way fashion. It is no easy task, however, to specify how effects occur or how particular environmental features are linked to particular qualities of people or particular developmental outcomes. The task is difficult enough when we limit descriptions of environmental contacts to face-to-face interactions. It becomes even more difficult when we begin to heed Bronfenbrenner's (1979) call to pay attention to such settings as school, workplace, or neighborhood and at the same time to treat these in a more dynamic fashion than by referring to the usual "social address labels" of socioeconomic status, ethnicity, or culture. How should we proceed if we wish to take account of children, other family members, and people or conditions outside the family? I shall do so by singling out several themes in discussions of family-community relationships, all related to identifying the processes that may be involved. The first two sections deal with some orientations that get in the way of cleareyed accounts: the treatment of families or settings as units, and the presence of value judgments. As these orientations are rethought, new possibilities emerge for the descriptions of settings, development, and process. In the remaining sections I focus on particular processes that might underlie relationships among children, parents, and communities. Broadly speaking, accounts of relationships contain three kinds of proposals for processes. I label these as "enhancement," "transmission," and "access" proposals. All three will be This chapter has some large intellectual debts: to the demand for a new way of thinking about children and families once issues of social policy began to be of interest, to time and colleagues at the Stanford Center for Advanced Study in the Behavioral Sciences (1984-5), and to the SRCD study group at Cornell. The financial help of the Spencer Foundation is gratefully acknowledged, together with the special contribution of Robert Goodnow's fascination with word processors - may it continue.

50

Children, families, and communities

51

considered. I give least attention to the first type, however, since the "enhancement" model is already well known (largely through the writings of Elder: e.g., Elder, 1974, 1981; Elder, Caspi, & van Nguyen, 1986) and is well represented in this volume in Elder and Caspi's discussion (chapter 4). Moreover, I have only a few new points to add to it. Basically this model may be summarized as one in which effects ' 'flow on'' from one sector to another in interactive fashion, with an emphasis on one particular interaction, namely, the way that the qualities of community settings accentuate some qualities of individuals (in particular, their vulnerability to stress, their resilience, or their tendency toward explosive losses of control). This approach is most common in work that deals with the effects of economic conditions ("hard times" especially) on family functioning and child development. I give more space to transmission models than enhancement models largely because with the former some reconceptualization is occurring and might be taken further. The basic concept is again one of "flow on," but in this case parents are seen as acting like conduits, as agents of society who pass on the local mores. The background concern is often cultural continuity rather than economic conditions. I also consider positions that emhasize the organization and perception of access between settings or people. This kind of emphasis appears in Barker and Wright's (1955) account of the ways in which children move from home settings to community settings, and in discussions about the effects of welfare and legal policies on families. What, for instance, are the rights of access between children and divorced parents? Who has the right to remove children from families? Under what conditions should children have their own direct access to legal or medical advice? As in the case of transmission models, I give special attention to proposals phrased in access terms because I believe they can be extended in some interesting ways. There is a further reason as well: I wish to break from a metaphor often used in discussions of the relationships among children, parents, and communities. The metaphor refers to "nested" sectors. An alternate offered in the discussion of access proposals is the metaphor of eternal triangles, which prompts one to think in terms of direct access between children and representatives of the wider community or state and in terms of relationships that may be marked by competition, varying degrees of tolerance for "interference," and concern for control over the access of any one party to the other. In each section of the chapter I wish to point out the way in which particular orientations to the question of relationships (orientations based on theoretical assumptions or on what one is trying to explain) give rise to some particular descriptions of settings, development, and the processes by which effects take place. In each section, I wish also to ask about limitations, to seek alternatives, and to identify questions that still need to be asked. We are clearly in no position

52

Jacqueline J. Goodnow

to choose one account of process over another, and we may well need a variety of models. What we can do, however, is to clear away some limiting ideas and to see how far we can take some of the promising accounts that are emerging. Families and settings as units: taking a more differential view It is tempting to refer to "the community," "the culture," or "the family" as if each of these were a single unit. A little thought tells us that this is not the case and that this approach is likely to be limiting. What are the alternatives? Bronfenbrenner (1979) offers one alternative for the description of community or culture. Briefly, he suggests that any social environment or ecological setting consists of a number of settings. Houses may contain public space and private space. Schools contain classrooms and playgrounds. Play areas cover backyard and street. The critical questions to ask about settings, in his view, have to do with the way they are related to one another. To what extent, for example, are the same people found in each? To what extent do the people in one setting know about what happens in another? To what extent is there spillover from one to another? The critical questions to ask about development, then, have to do with the way each setting is perceived, the extent to which it is found satisfying, and whether settings are seen as linked or as discontinuous, with different people and different rules in each. These perceptions and feelings are said to be influenced by a variety of conditions, one important condition being the way in which we make a transition or are introduced into any new setting. Children's feelings about school and adjustment to school, to give a concrete example, are affected by the way in which children are first introduced to school: by an accompanying parent, by advice from an older sibling, or by no prior knowledge and a parting at the gate. Bronfenbrenner's views about settings are relatively well known. For psychologists, they are also a recognizable extension of ideas about behavior settings made familiar by the work of Barker, Wright, and their colleagues (e.g., Barker & Wright, 1955; Schoggen, 1983). To summarize, that work distinguished among settings in terms of place, activity, and the degree of undermanning. Undermanned settings - small schools or small towns, for example were seen as providing more opportunity and more pressure to be a participant or performer rather than an onlooker (e.g., Schoggen, 1983). Participation is then seen as leading to further developmental outcomes, such as the opportunity to play a variety of roles, and a greater knowledge of community settings and of adults outside the family. Less familiar to developmental psychologists are parallel attempts to go beyond references to the family as a solid unit. The general argument is that any family is a collection of people whose interests are not always in agreement; who

Children, families, and communities

53

negotiate among themselves various alliances, divisions of labor, and divisions of rewards; who hold expectations about exchange and reciprocity; and who maintain ''balance sheets of merit and indebtedness." This argument is prominent in therapists' analyses of family systems (e.g., Baszormenyi-Nagy & Spark, 1973). It is also to be found in a number of economists' current views of families. Hartmann (1981) and Edwards (1985) argue for closer attention to divisions of labor, struggles for equality, and the flow of goods - for example, money or family income - within families. Finally, an emphasis on individual family members is to be found in analyses of social policy (e.g., Strober, 1982). When people speak of the "good of the family," for instance, the rejoinder is increasingly likely to be, Whose good? When people speak of the elderly or the handicapped being returned to "the care of the family," the question is likely to be asked, Whose care? Where does this image of families lead us? As a start, we may begin to put more research time into describing relationships among family members, drawing upon the descriptions of alliances, exchanges, and implicit contracts found in descriptions of family systems, or upon the burgeoning literature on describing relationships in general. Hinde (1979) has summarized and expanded a number of positions contained within this literature, noting attention to such qualities as the presence of symmetry or balance (e.g., Heider, 1958), of mutuality in commitment or sharing (e.g., Levinger, 1974), and of underlying beliefs about equity and the nature of legitimate competition (e.g., Lerner, 1974). Such descriptions take us part of the way toward describing relationships among family members. We need, however, to link family relationships to settings outside the family. If we now have to consider different family members and take a differentiated view of the community, how shall we proceed? One way of doing so is to examine divisions among family members of the labor involved in establishing and maintaining contact with various parts of the community. Hareven (1982) notes that in many families one member specializes in becoming "the kinkeeper." Why, we might ask, does that particular person take on this task? For one answer, we might turn to analyses of family networks, starting with Bott's (1957) questions about parallels between divisions of labor in the household and divisions of labor in maintaining family and friendship networks. Do women in families with "traditional" divisions of household labor, she asks, also take the lion's share in maintaining social networks while their spouses specialize in workplace and worktime friendships? Such questions are still in need of exploration. To direct more attention to children, we might also ask about the work of family members in teaching about the nature of the outside world. Do parents divide the labor so that one of them, perhaps, informs children about feelings and friendships while the other teaches about money, competition, and the need to be

54

Jacqueline J. Goodnow

tough? That type of division is suggested by Graeme Russell's (1983) work on differences between mothers and fathers. The question could readily be added to with a question prompted by Cashmore and Goodnow's (1985) work on agreement among family members: Do parents subvert one another or do they agree in the messages they send about what is needed to be a good family member or to succeed in the world? Finally, a nonbloc view of families and of settings may prompt us to look at development in some different ways. We know that agreement between parents can affect the accuracy with which children report the position of parents and the extent to which they agree with their parents' views (Cashmore & Goodnow, 1985). That study, however, dealt mainly with the general importance of certain values (e.g., being obedient, understanding easily, being interested in how and why things happen) and with occupational aspirations. The design might well be extended to include children's views about how to approach or cope with various parts of one's community, or to analyze how children develop skills in detecting disagreement, forming alliances, and locating the family member who can provide the information or the introductions needed in order to act effectively outside the family. Similarly, we might also ask whether particular patterns or divisions of labor in teaching about the world alter children's views of the outside world. If most of this teaching is carried out by fathers, even when mothers are in the waged work force, do children emerge with the view that the outside world belongs more to men than to women? In short, there is no lack of interesting questions to ask or research directions to take when we adopt a nonbloc view of both families and their larger settings. The presence of values: taking alternate views Social science often contains value judgments. I single out three, each of which affects the way we view relationships between families and settings. The first is that "wider is better"; the second that "continuity is good"; the third - the assumption I give most space to - that "the world is open." "Wider is better." I use this phrase to refer to the assumption that people will regularly benefit from being part of larger settings than the family provides. Travel broadens the mind, it is said. The outcome is not, in fact, always so positive. Consider, for example, going to school at the age of 5 or 6 years. This is usually regarded as having many benefits. Essential academic and social skills are learned, children enter a larger world, acquire friends, become less dependent on their parents, and so on. School is not, however, always a positive experience. It may, in fact, give rise to a lesser independence of thought than, say, being taught at home until the age of 8 years (a suggestion from

Children, families, and communities

55

Margaret Mead's autobiography). And it may diminish one's view of oneself. As Entwisle and Hayduk (1978) have pointed out, schools provide comparisons that may make it difficult for parents to believe that their child is above average or for children to maintain a high self-regard. This type of outcome, they suggested, might occur especially in areas that have been elective until one goes to school, so that school offers the first encounter with social comparisons. Parents and children, for example, may come to realize for the first time that a child sings flat or has two left feet when following a rhythm or drill. Cairns and Cairns (this volume) make a similar point more strongly in discussing peer groups. They may indeed be a positive part of development. They can also be highly restrictive in their demands for conformity and can painfully whittle away at one's selfesteem. "Continuity is good." It is easy to believe that continuity between settings is a good thing and that it is an advantage when people in one setting know about your life in another. Such assumptions surface especially in discussions about the "two worlds" of immigrant children or about gaps between teachers of one socioeconomic background and children of another (e.g., Burns & Goodnow, 1985, chap. 4; Heath, 1983). Consider, however, the benefits of some lack of similarity, if only by virtue of novelty. Consider also the benefits of people in one setting not knowing what you are like or supposed to be like in another. Such a pattern provides opportunities to try out new styles, to operate free of labels that have come into use in an old setting, to be a different person. The critical conditions, it would appear, are not simple ones of continuity or similarity. What counts is the extent to which the new setting offers or closes out opportunities for various forms of development. That kind of approach to the description of relationships between people and settings is more complex and will be more difficult to develop. I suggest, however, that it is a necessary approach if we are to move to a more satisfying - and a more useful - view of how connections between settings affect developmental outcomes. "The world is open." One way to view the world is to consider it open for exploration, use, and enjoyment. Advancement is limited only by the individual's capacity to take advantage of what is readily available or can be gained with the willingness to put forth effort. This "open frontier" view is perhaps less popular than it has been. At least there is now an alternate view, captured in the recognition that opportunities are not equal and that it may take major social changes to make them so. When do such views of the world affect analyses of connections between family members and community settings? As a start, an open view of the world can cloud analyses of social policy and the probable effects of policy changes

56

Jacqueline J. Goodnow

upon family members. A specific example comes from discussions of how to improve the position of immigrants in a country such as Australia, especially those who do not speak English on arrival (see Burns & Goodnow, 1985, chap. 4; Poole, 1985). Their problems are seen as readily solved once they overcome the difficulties within them that limit their ability to benefit from a society otherwise open to them. In particular, all is expected to go well once the immigrant becomes linguistically competent, and most programs of support are directed exclusively toward that purpose. It is only occasionally pointed out that the immigrant faces difficulties beyond language. In some cases, the information needed may not be available at all. In others, what is faced is a "maze of misinformation" (Bullivant, 1976) that reflects a lack of concern or is designed to foster the interests of others (employers or entrenched labor groups that now move into an "aristocracy of labor"). These sources of difficulty do not spring readily to mind when one starts from an open view of the world, a view more likely to prompt one to wonder what features of individuals lead them not to make use of the available opportunities. The assumption of openness can also lead one away from a closer look at the extent to which others share one's open view or hold a more wary assessment of life's possibilities. It is rare to find, for instance, research such as that by Margaret Spencer (1985) of Emory University on the way black parents in the United States prepare their children for encounters with discrimination and prejudice in schools. She sees a historical shift from active preparation for a harsh world to the expectation that since "things are better," preparation can be less and action may be best deferred until a negative experience occurs. Spencer herself wonders whether the shift is to the real benefit of children. Her concern should prompt us to ask, Under what conditions do people present their children with various views of the outside world? When is it more effective for children to be well defended in advance or to believe that the world is kind unless proved otherwise? How are we to account for the fact that a high proportion of adolescents in a study by Lewis (1981) were not alert to the possibility of vested interests behind the advice of adults? Is this attributable to a general human tendency to prefer the view that the world is generally kind and people in authority are to be trusted? Or is it the result of this sample being white and Californian, with no direct experience of active guile or oppression and a surrounding ideology of a just world? Finally, asking about the assumption of openness prompts us to think of other ways to describe settings and the process of development. I shall work out some of these possibilities in terms of comments on a particular approach to cognitive development: namely, Piagetian theory. I have chosen this theory because it contains some basic assumptions about open worlds (in the form of free informa-

Children, families, and communities

57

tion), and because of my own relatively recent recognition of some limitations and alternatives. Piagetian accounts of cognitive development generally assume a world that is open, benign, and full of redundant information. The main limitations to the development of knowledge lie in the individual's capacity to pick up and process the information so freely available. There are, it is true, some limitations noted in the concern that an authoritarian world will dampen the intrinsic curiosity that children bring to the task of constructing reality (e.g., Lautrey, 1980), but the major barriers are the structures of one's mind. The Piagetian model, I have come to think, may work relatively well for part of what has been called "physical cognition": concepts of amount, number, weight, volume, space, and the like. These are areas in which adults are perfectly willing to have children explore freely, learn for themselves, make amusing errors, or hold odd "naive" theories. The model does not seem to hold as well, however, for some other areas in physical cognition: for example, knowledge about the sexual or reproductive aspects of the human body. Nor does it seem to hold particularly well for those areas of social cognition in which adults have some vested interest in seeing that children acquire particular views, rather than developing their own constructions, areas that have to do with ideas about privacy, aggression, cheating, obligation, respect, charity, or the nature of "rich" and "poor." In everyday social life, adults use a variety of techniques to block access to information. Some areas of information are tagged "to be learned about when you are older." Some forms of knowledge are not appropriate to one's gender, social group, or ability group. Some questions are "not nice" to ask. Some activities, as Valsiner (1984) points out, are encouraged, whereas others are indicated as tolerable at best or not to be engaged in under any circumstances. In fact, the more one comes to recognize the nonopen nature of many settings, the more one begins to consider the possibility that the benign Piagetian view may apply only to areas of knowledge that parents regard as either relatively unimportant (so that errors are of no great consequence) or as requiring no active intervention (the world being so full of information of the right kind that acquiring the right ideas is inevitable) (Goodnow, Knight, & Cashmore, 1985). What happens in other areas? In these - and they are probably the larger part of knowledge about the world - information is managed by others. It comes in forms that others feel we can manage or need to know. Such management or "prepackaging" (Shweder, 1982) of information about oneself and about the social order is probably more frequently the case for children (or adults) than not. My remarks are not intended simply as a comment on Piagetian theory. What I propose is that an altered view about the freedom of information can be helpful

58

Jacqueline J.

Goodnow

in opening up ideas about connections among children, parents, and settings outside the family. It can lead first of all to an expanded description of settings and of the way they may differ from one another, as the following examples show: 1.

2.

3.

4.

Settings may vary in the extent to which there is open or restricted access to information about options, pathways, and prerequisites. Some societies are secret. All societies, in the view of some sociologists, contain some equation between knowledge and power, a relationship that encourages people to protect a privileged state by restricting information or by maintaining control over "symbolic capital" and procedural knowledge (Bourdieu, 1977; Bourdieu & Passeron, 1977). Age groups may also work toward "insider" status. Teenagers, for instance, seem to promote a closed group by changing "in" words and styles, relegating people who are older, or who do not belong, to an outer world of last year's clothes and vocabulary. Settings may vary in the domains for which information is open or restricted, the domains to which they attach "red" or "green" lights. This tagging of content areas may apply to all or most people. In societies with a traditional religious base, for example, questioning the existence of God is not likely to be encouraged. More often, some particular areas of knowledge become tagged as "owned" in monopolistic fashion by one gender, one age group, one social group, or one clique. The most obvious examples are settings in which particular occupations are restricted to particular castes, or to the children of people already in that occupation. More subtle (and for that reason perhaps more difficult to deal with) are cases in which an area of knowledge is defined as being theoretically open to all - in the United States and in Australia, mathematics for males and females is an example - but as nonetheless more relevant, more appropriate, and easier for one group than another (in this case for boys than for girls). It is not to be wondered at that a recurrent question in analyses of culture has to do with the conditions that give rise to particular domains of knowledge and action being tagged as "owned" rather than as open (e.g., Berger, 1977; Foucault, in Sheridan, 1980; Sharrock, 1974). Settings may vary in the way they encourage particular routes to the gaining of information. Stodolsky (1988) argues that school settings are arranged in such a way that children come to see knowledge in social studies as achievable by thinking about it themselves and by putting together the results of projects and experiments. In contrast, knowledge in mathematics is seen as achievable only by going to an expert and being given prescriptive advice. Finally, settings may vary in the legitimations offered for availability arrangements in various domains, legitimations ranging from dangers and difficulties that require special guides or mentors to the presence of a natural or ordained order, imposed by biology or sanctioned by tradition.

Do such differences among settings affect development? Suppose we were to define development as the achievement of a coherent and differentiated view of the surrounding world. Such development must be slowed down by information withheld until one is older and wiser, by inconsistent feedback, or by the presence of prepackaged information (Shweder, 1982), especially if the packaging requires special effort and skill to untangle the pieces and decode the message.

Children, families, and communities

59

For a concrete example of difficulty, I draw from children's answers to some questions used in a large-scale survey (Goodnow & Burns, 1985). Children were asked, What would you change if you could change one thing - in the morning, at school, in the world for other children? Answers for the first two categories were quite different from those given for the third. Whereas children had been inventive and fluent in their suggestions for changes at home and school, they were less so for the third. "Food for the Kampucheans" (the question had been asked in the Year of the Child) was a frequent answer, along with "clothes or money for the poor," "housing for the people in Darwin" (a region hit by a typhoon), and "parents for children orphaned by the Khmer Rouge." Answers were marked by cliches, by recourse to images from television, and by vagueness about how many of the shifts from the haves to the have-nots would or could occur. Why the difficulty? I have come to feel (Goodnow, 1986) that it does not lie in the children and their general intellectual capacity to work with hypothetical material, but in other factors: in the poverty and the packaging of information that most adults provide about the haves and have-nots, the rich and poor, us and them, in the overriding vividness of television's images and in the complexity of the question. Like the question, "What are you going to be when you grow up?" a question about improving the world is not easy to answer because answering it well, on adult terms, depends on meeting several criteria. The respondent needs to display the conceptual ability to come up with a reasonable and feasible solution. "Food for the blind" - to use one second grader's solution - is not an acceptable answer to children or adults, and many might not accept as feasible an answer such as "all the countries have to get together and agree to have no more wars" (a sixth grader's suggestion). In addition, a good answer also needs to display some compassion and a lack of smugness: to show that one is kind as well as sensible. Few children, we suspect, would regard as a proper answer statements such as "They should take all those union people who' won't work and send them over to build hospitals for the Kampucheans," "They should give all the poor people $100 and tell them that's it," or "They should send back all the immigrants who are taking our jobs." Children certainly hear such negative statements, and a few offer them as solutions. Their rare occurrence in this survey, however, seemed a sign that children knew these were not the kinds of answers required. Where does this example fit in a general picture of developmental outcomes? Imagine a setting that is full of questions with double-barreled criteria for success and that at the same time provides little of the information needed or packages it in unhelpful ways. That type of setting is not conducive to developing coherent or differentiated views of oneself and the world. Mystification on the one side is likely to lead to bewilderment or stereotyping on the other.

60

Jacqueline J. Goodnow

Analyzing process: questions about enhancement Up to this point I have argued that it is important to take a differentiated view of families and communities and to look closely at some assumptions that may lead us away from a clear-eyed view of relationships between various family members and their various settings. Unfortunately, however, there has been relatively little direct analysis of the processes involved in the mutual influence of people and settings. Ideally, the processes we need to consider are those that take into account the qualities of people and the qualities of settings. There are at least three candidates, all requiring some in-depth analysis. In this section I comment on the first of these very briefly. The process of enhancement or accentuation has been studied by Elder (e.g. Elder, 1974, 1981, this volume), who points out that the stresses imposed by a harsh economic climate add to the stresses already present by virtue of one's personality or family life. The individual feels an increasing gap between what is hoped for and what seems possible, and a "loss of control" over his or her situation. (As Elder, 1981, notes, the concept of "control" in the sense of matches and discrepancies between "claims" and "resources" is adapted from W. I. Thomas.) Any tendency to explosive outbursts will then be accentuated, and the adult's actions - by way of reinforcement or modeling - produce an impact on others that may then continue through generations. The model clearly raises some interesting questions. Are outbursts or other negative behaviors directed more to some family members than others? The differential direction toward more attractive and less attractive daughters (Elder, Caspi, & Van Nguyen, 1986) is an interesting example. When do individuals respond to the gap by scaling down hopes or working to increase resources, rather than by becoming angry? When and how do people decide to break the flow-on effect: stopping on the way home from work to shift moods, or reminding themselves that a spillover of mood - kicking the family cat - is unjust? Who stands in the way of flow-on effects? We may expect to see more such questions, and a number of answers, growing out of this view of process. A variation on an enhancement model is the contextual theory of parental competence outlined by Belsky, Robins, and Gamble (1982). Briefly, three qualities are seen as important determinants of parental sensitivity and involvement: patience, endurance, and commitment. The first of these three is thought to be particularly enhanced by the presence of emotional support in family or community settings, the second by instrumental assistance (e.g., by goods and services that free up physical energy), the third by social expectations. Social expectations have a further interplay with the qualities of individuals in the sense

Children, families, and communities

61

that emotional and instrumental support are more likely to be provided to those who meet rather than violate social expectations. The proposal is of particular value because it makes an explicit attempt to match specific qualities of parents with specific qualities of settings, and because it proposes that social or economic conditions can provide sources of support as well as of stress. Analyzing process: exploring the nature of transmission In this section I review some of the recent findings about transmission as a process and indicate some of the questions that remain to be studied. The emphasis is on what Furstenberg (1971) calls the most common form of transmission models: namely, parents seen as agents of the wider society, passing the prevailing cultural values from one generation to another. The classical model contains two transmission steps: from culture to parents, and from parents to children. To analyze the pattern, one might consider both steps and all parties in one frame. One might compare the total process to the passing on of rumors and use that literature as a base. One might also ask about second-order effects and the way in which the relationship between parties A and B affects the development of C in ways over and above the independent effects of A or B on C. Or, in more concrete fashion, one might ask who speaks to whom in any three-party situation. Margaret Steward (personal communication) has drawn my attention to the way pediatricians, for instance, often use a language of explanation that can only be decoded by adults, even though the explanation is officially directed to the child as patient. The underlying assumption appears to be that the parent(s) present will serve as memory banks and as translators even when not directly invited to do so. Most of the analysis of transmission, however, has concentrated on two parties (parents and a younger generation) and on the degree of agreement between them on a number of values (usually political, religious, or occupational). The assumption is that the analysis of this form of agreement will yield a test of transmission effects or point to new aspects of process. I shall make the same assumption and concentrate on agreements between generations, with the understanding that the conditions and processes emerging in this relationship probably apply as well to the way children receive messages from peers, or to the way parents are sent messages by nurses, physicians, welfare agencies, or parent educators. What findings have emerged from the work on agreement between generations? One problem that has surfaced has to do with discrepant data. What do you do if transmission is not successful, if the younger generation does not display values that parents presumably try to transmit? For these cases of nontransmis-

62

Jacqueline J. Goodnow

sion, either we need a new model, or we need to keep the usual one and see if we can develop a more specific and flexible form that will help explain the discrepant cases. Suppose we take studies of agreement between generations on values. These studies are often considered tests of the model of parents as transmitters of social values and sometimes contain the further proposition that identification is the main underlying process leading to agreement. The difficulty is that the data on agreement are not especially convincing, a point made with particular clarity by Furstenberg (1971). The actual level of agreement between generations on most measures of values is often low (Acock & Bengston, 1980; Furstenberg, 1971). In addition, the relationship is often weak between degree of agreement and a condition that seems likely to reflect identification, namely the warmth of the parent-child relationship (Clausen, 1966; Furstenberg, 1971; Jennings & Niemi, 1968; Rollins & Thomas, 1979). Even when agreement is demonstrated, it may be difficult to determine how much is due to intrafamily transmission and how much is due to a common response to a culture shared by both generations (Troll & Bengston, 1979). Yet it is difficult to believe that parents have no success in transmitting their values, that the nature of the parent-child relationship is irrelevant, and that transmission is not part of the processes involved. How do we resolve the discrepancy between these expectations and the data? One solution has been to argue that most measures have dealt with very specific values (e.g., political allegiance), when agreement may be more likely to occur on general rather than specific values. A study of student activists, for instance, led to the proposal that generations tend to agree on the importance of commitment, but not on the specific cause or goal to which commitment is made (Troll, Neugarten, & Kraines, 1969). Another suggestion put forth is that agreement is more likely when the values involved are important to parents and this importance has led parents to provide clear cues about their positions (Jennings & Niemi, 1968). These salient areas - these "family themes" (Hess & Handel, 1959) - may vary from one family to another and may not be tapped by questionnaires or interviews. A third argument is that we may need to wait, in our measurements of agreement, until the second generation is at the same life-cycle stage as that of the first, noting then whether a pattern or position is repeated or altered (e.g., Huesmann, Eron, & Lefkowitz, 1984). A fourth - the one I wish to emphasize - is that the apparent lack of supporting data reflects the need for a closer analysis of process and a different statement of theory. An important step in this direction would be to separate two forms of agreement: actual and perceived. The former may be low, but the latter high. Parents may comfortably assume that their children share parents' ideas when they do not; and children may perceive parents as agreeing with them when in fact they

Children, families, and communities

63

do not. The distinction between actual and perceived agreement is made with particular force by Acock and Bengston (1980), and by Furstenberg (1971), all of whom point to evidence that perceived agreement is higher than actual agreement. A second step is to propose that the perception of the others' position becomes a mediating variable for actual agreement. The positions taken by the younger generation, for example, may be better predicted by the positions they perceive parents to take than by parents' actual positions. There is some evidence to support that proposition (Acock & Bengston, 1980; Furstenberg, 1971). There are also some data indicating that as accuracy of perception increases, so also does actual agreement (Furstenberg, 1971; Smith, 1982; Tedin, 1974), although we do not yet know whether this effect varies with the age of the younger generation or what content areas it is more or less likely to occur in (Cashmore & Goodnow, 1985). A further step is to ask what gives rise to either actual agreement or perceived agreement. Since actual agreement is what cultural transmission is supposed to aim for, let us concentrate on that. As noted elsewhere (Cashmore & Goodnow, 1985; Furstenberg, 1971), it appears to involve two processes: accurate perception of the other's message, and acceptance/rejection of that message. The final theoretical step is to propose that a number of conditions may be differentially related to these two processes. "Some . . . factors would improve the child's chances of correctly perceiving their parents' views; others may increase his willingness to adopt these views once he knows what they are" (Furstenberg, 1971, p. 599). Articulate statements by parents, for example, or the wish to perceive similarity between generations, might have particular effects on the accuracy of perception. Warmth in the parent-child relationship, or the presence of competing messages from peers, might have particular effects on the acceptance of the message, once accurately perceived. Would these expansions to a bald model of transmission meet all one's needs and also fit situations in which parents or children are the targets of messages from outside the family? They go some distance toward accounting for cases in which the two parties do not end with the same views, that is, for cases of apparently unsuccessful transmission. They do not, however, yet capture an aspect of transmission that I consider critical: namely, the active management of information. Suppose we take parents as a starting point. Parents mark information as having varying degrees of importance. They provide it, withhold it, or frame it on the basis of judgments about value and about the other party's need, age, or capacity to cope. They are selective in the social messages they pass on. Some they endorse, others they subvert or try to exclude from the child's awareness. Such active management is not restricted to parents. We have a great deal yet to

64

Jacqueline J. Goodnow

learn about the criteria that lead children to decide what information they will transmit, which members of an older generation will be the recipients, and under what circumstances. "It will only upset them" is one criterion familiar to us all. Research of this kind is more likely to occur readily if we depart from a relatively mechanical notion of information transmission and reception and add to it the more dynamic notion of management by parties who have vested interests in the messages they send, receive, or choose not to hear. Analyzing process: exploring the nature of access In this section, I ask four questions: (1) What observations prompt the need for attention to access arrangements among children, parents, and communities? (2) What does the term "access" cover? (3) How can one use access notions to describe the objective and perceived qualities of settings? (4) How can one combine ideas about access with attention to the qualities of individuals? Observations prompting questions about access The notions of enhancement and of transmission do not account for the full range of connections that exist between children, parents, and communities. What are the further interactions or relationships we need to account for? The following are some that call attention to the need to explore ways in which children, families, and communities have access to each other and use that access: Using the outside community to manage or to alter parent-child interactions. Parents may use references to the wider society as a way of enforcing or strengthening the position they adopt toward children. They may do so by reference to policemen, truant officers, reform schools, juvenile courts. Or they may take the line that children should act in particular ways because otherwise parents will be judged harshly by others. To quote a mother commenting on how she responds when children leave their rooms untidy: "I say to them, what would people think if they came in and saw that mess? They would think I was a pretty poor mother." These appeals to the wider society are not unique to parents. Appeals to what "other kids" and "other families" do are familiar arguments from children. Despite their frequency of occurrence, there seems to be little information about the occasions when these outside references are used, which outside people are mentioned as significant, and the degree of persuasiveness the references to others have. We also need information on the consequences when outside courts of appeal become involved. This subtle issue was brought to my attention by Margaret Steward (University of California personal communication), in the

Children, families, and communities

65

form of a comment on children's changed perceptions of the power of parents when a family becomes involved in divorce and goes to court. The power of the court may come to be recognized as being above that of parents, weakening for the first time the child's view that parents are powerful and can offer protection as well as love. Such effects do not seem easily accounted for by processes of enhancement or of transmission. They are closer to the alliances and power plays often noted in descriptions of family systems and of intergeneration relationships (e.g., Baszormenyi-Nagy & Spark, 1973). Nor do they seem to fit well with a metaphor of nested sectors, since one feature of these relationships is the possibility that children have direct access to the community without the mediation of parents. Children may use these contacts as a buffer against family effects or they may act as agents of society, bringing home to the family concepts, skills, and values ihat are more readily available to their generation than to that of their parents. Competition and concern about "interference." It was to capture these aspects of relationships that we (Goodnow & Burns, 1980) began to think in terms of eternal triangles rather than in terms of nested sectors. The traditional triangle - two spouses and a lover - immediately calls to mind competing parties, with conflicts over rights and obligations, agreements not to interfere as long as certain standards are met, and so forth. These are precisely the relationships that seem best to describe the interweaving of children, parents, and the state. In this interweaving there are both agreements and conflicts about responsibilities. Who has the right, for instance, to teach a child about sexuality, to administer corporal punishment, or to provide contraceptive help? There are divisions of labor that most people accept. Most parents now accept, for example, that the state will provide schooling, and one sees fewer objections to compulsory schooling and fewer private arrangements than existed, at least in Australia, at an earlier time in history (Burns & Goodnow, 1985). There are implicit contracts about responsibility and interference. The state can step in and remove children when parents are considered not to be fulfilling their obligations. Protective and gatekeeping strategies have been developed on all sides to maintain the degrees and forms of autonomy, privacy, control, and constraint that each sees as proper. This expanded range of relationships has to be considered in analyses of social policy. Historical shifts in policies for children and families, for instance, often take the form of altering the acceptable patterns of collaboration and competition - altering conditions such as those under which the state may remove children from families or under which children may have direct access to legal or medical services without the mediation of parents. Comments about "the breakdown of the family" are also often comments on perceived or claimed shifts in patterns of

66

Jacqueline J. Goodnow

responsibility (Bronfenbrenner, Moen, & Garbarino, 1984; Goodnow, 1981; Goodnow & Burns, 1984). Schools see themselves as needing to take on the task of sex education because parents have abdicated their responsibility. Politicians argue for mothers to stay at home rather than press for day-care services at cost to the state, for fathers to pay continued maintenance after divorce rather than have the state pay supporting benefits to mothers, and for parents to accept the responsibility of supporting their children economically rather than tolerating their 16year-olds collecting unemployment benefits. Once again, the relationships do not seem easily covered by processes related to enhancement or to transmission, so that we must look further. Moments of entry or exit from one setting to another. Developmental psychologists have become increasingly aware that individuals to some extent socialize themselves by selecting some settings or niches rather than others (e.g., Whiting, 1980). They have also become aware of the fact that parents work at exercising some control over children's encounters with the outside world. Selecting the right neighborhood, the right school, the right summer camp, the right babysitters; keeping a watchful eye on the child's friends, monitoring television; instilling in children a particular view of the outside world (which they may call a "dangerous place," for instance, full of "heathens" or "parents who do not really care what their children do"). All these are steps toward managing the child's access to and interpretation of segments of the wider society. How this control is exercised is far from clear, and several commentators have pointed to the need for more information about the ways in which parents remain informed or influential when a child is away from home and face-to-face interactions (Maccoby, 1984; Rubin & Sloman, 1984). We clearly need to add to the two processes discussed so far (enhancement and transmission). Some ways of doing so are organized around the concept of access. Two meanings to "access" Access may refer to physical movement from one setting to another, physical participation in each, and the way in which both of these are managed. This is the meaning applied by Barker and Wright (1955). They distinguish between home settings and community settings, and, within the latter, between community settings that are "adult" and those that represent transitional settings, designed predominantly for particular age groups and sometimes attached to adult settings (e.g., Sunday schools attached to churches as opposed to the usual preschools). Such distinctions among settings prompt questions about the proportion of time that children of various ages spend in various settings and descriptions of how

Children, families, and communities

67

children enter new settings. They progress, it is argued, by a process of anchoring, establishing secure places or "home bases" and then fanning out from these. They also progress through several forms of participation or involvement: as onlooker, audience or invited guest, member or customer, active functionary, joint leader, and single leader. Involvement in the last three levels - wherein one becomes a performer in a setting - rises steadily in frequency with the child's age. Access may also denote entry or "penetrance" (Barker & Wright, 1955), a concept that can clearly be applied and expanded to provide further descriptions of settings and of development. I propose as well a psychological expansion, one represented by a shift in terms from "entry" to "access." The expansion is prompted by the need for ways to talk about access not only to places but also to activities, opportunities, people, or information (information about oneself, about others, or about the nature of the social order). This need emerges on several occasions. It emerges, for instance, when one discovers that children have different ideas about how to learn about social sciences versus mathematics (Stodolsky, 1988), or when one wishes to talk about whether the world is truly open or whether information is managed and often restricted. Finally, it arises when we consider what children have in mind when they talk about what makes a good parent. One of the features that emerges is the way parents can tell you about yourself. In the children's words, parents "can tell you what you were like as a baby" and "can tell you about your grandparents" or "about the old country" (Goodnow & Burns, 1985). The term "entry" does not fit these relationships as well as "access." Using access notions to describe settings and the perceptions of settings How can we use the concept of access to describe relationships between children, other family members, and communities? As a start, one may describe settings in terms of objective features of access. Settings may vary, for instance, in the pattern of steps or prerequisites for access. In some cases, one step may allow easy access to subsequent steps. In others, each new step may bring a new hurdle. In some settings, going to the right high school or university allows all other good things to follow. In others, this step may help, but the competition for moving to each new rung of the ladder calls for continued effort. Settings may also vary in the arrangements for exit: from a phased series of steps to a single bound, from careful preparation by others to a more ruthless pushing out. Arrangements for retirement or unemployment are classic examples. The several ways in which people may be dropped from social groups provide other examples. Settings may also vary in regard to whether they allow for departures from the standard path. Some settings, for instance, allow shortcuts (e.g., accelerated

68

Jacqueline J. Goodnow

promotion, jumping school grades), whereas others insist on step-by-step progression. Some settings allow time-out for acceptable reasons (one may take legitimate time-out for military service, for example, or for maternity or paternity reasons). In addition, some settings allow for easier recovery from an earlier step than others do. Until this year in Australia, if you wished to be a plumber, you needed to become an apprentice around the age of 15 or 16. There was no other route into the occupation, and no way to make up for failing to take the required step at this particular age. The situation in a country such as the United States was quite different. The features noted above may be regarded as objective features. One may also describe settings in terms of the way they are perceived. This is the form of description I wish to emphasize. The objective features of settings may alter one's degrees of freedom in action, the trajectory established by an action taken at a particular time, and the chances of recovery from an earlier action. If we wish to concentrate on development, however, perceptions of access and of options provide the content area in which change is likely to occur. In addition, there are reasonable grounds for arguing that the actions people take are based as much on what is seen as feasible as on what is actually the case. In any description of perceptions, which might turn out to be important? Four forms of perception strike me as interesting candidates. They are sufficiently general to apply to a variety of concrete settings, are sufficiently specific to suggest some research possibilities, and can supply some indications of effects on particular aspects of development. 1. Perceptions about ways to gain access. Let me provide two examples that go beyond some perceptions noted earlier about settings linking open or restricted access to particular features of people (their age, gender, social class, natural talent, etc.). One is the surprise Australians often feel when responding to the questions or approaches of Southeast Asians. The latter see access to most, if not all, situations as critically dependent on introductions. The former see introductions as critical for a smaller set of situations. For the second example, the data are more accessible. In Jackson and Marsden's (1966) study of education and the working class in England, working-class children and their parents who aimed at university education often did not know until the moment of applying that some particular high school subjects were prerequisites. Poole (1985) makes a number of similar observations about factors that affect the choice of occupation when Australian adolescents first leave school. In both cases, socioeconomic background creates a particular view of how one gets from X to Y. Furthermore, particular views are created about who is a reasonable person to ask for information or help. Perceptions about "mentors" or "introductions" are by no means an aspect only of cultural variations.

Children, families, and communities

69

2. The perception of arrangements in a setting as based on nature as opposed to being a social option and therefore changeable. It is not necessarily true that patterns based on nature are difficult to alter, but the equation of the two terms seems to be part of much everyday Western thought. This aspect of perceptions appears to be linked especially to views of oneself, whereas the first aspect (ways to gain access) would be expected to alter mainly one's choice of actions or decisions. "Perceived self-efficacy" (Bandura, 1977) is one area in which one might look for an effect. Concepts of self provide another. It is difficult to imagine one's self as changing over the life course if one believes that development is strongly determined by gender or genetics. It is also easy to imagine that the self is deviant - unnatural or sinful - if one resents a current social order but still regards it as part of nature's way or God's. For any developing individual, it takes effort to struggle out from under images that depict the group one belongs to as occupying a particular social place and as being restricted to it because members of the group are inherently flighty in mind, have a good sense of rhythm but little of logic, are field dependent, are unattractive, have poor spatial sense, and so forth. Many people, in fact, do not develop the awareness that these are cultural images justifying a social order rather than features of one's self. 3. The perception that arrangements in a setting have a rational rather than an irrational base. An example comes from the work of Lautrey (1980), on French household rules and the extent to which parents offer an explanation or rationale for these, and of Dasen (1980), on the presence of clear sequence, preplanning, or offered rationale for tasks assigned to Kikuyu children. In both cases, the aspect of development likely to be affected is cognitive: A link is reported between performance on Piagetian tasks such as conservation and the extent to which tasks contain some perceivable degree of order. The intervening step implied is the child's perception that there is some value (some likelihood of success) in attempting to make sense of the phenomena encountered. The sense of there being no structure present inevitably dampens the alertness to discrepancies and the questioning of earlier schemas that, for Piagetians, are essential conditions for cognitive development. The argument makes intuitive sense. Any effort to make sense of the world must be based on the assumption that there is meaning to be found. The argument also prompts the research question: What are the features of settings that encourage or discourage the perception that meaning can be found and that it is likely to be found in some forms of explanation rather than others? The perception that the settings one will encounter are reasonably safe rather than dangerous. This perception has been noted as important for parents'

70

Jacqueline J. Goodnow

views about their own health and that of children. Is good health viewed as a given that is not easily lost, or as a precarious state that will readily disappear with every period of not eating well, every occasion of being caught in the rain or not wearing warm clothes? I owe my awareness of that issue again to Margaret Steward (personal communication). For examples outside the area of health, Douglas's (1966) classic work on purity, pollution, and contamination is a prime source. Social groups vary, Douglas argues, in their perception of the outside world as containing various kinds and degrees of pollution and contamination, with sources ranging from bad air, bad company, and people with colds, to menstruating women. They vary also in the rituals and strategies developed to avoid contamination or to purify oneself after contact (from hand-washing to religious ceremonies for the spiritual cleansing of women who have recently given birth). Such perceptions of danger, and of risk (Douglas, 1982), must inevitably alter one's concept of self and of others, as well as the actions one chooses and the options in life that one takes up. Why an emphasis on perceptions of access? I have clearly given more space to perceptions than to the objective features of access, even though the latter affect the options available and the chances of recovery or change from a trajectory once established. The reasons for emphasizing perceptions are fourfold. Two have already been noted. One is that the actions people take are as likely to reflect their views of settings as the settings themselves. The other is that we can regard shifts in the perceptions of settings as a form of development. What seemed a closed door may come to be recognized as an option; what seemed incomprehensible may come to be seen as containing some order; what seemed overwhelmingly dangerous may come to seem manageable. Two further reasons also apply. One is that by exploring perceptions one is able to ask about relationships between people in terms of consensus or disagreement about the nature of a setting or its function. Silbereisen and Noack (this volume) provide an example, asking both adolescents and security guards about the reasons for going to a disco and about the rules that apply there. Finally (the most important point for this discussion), an emphasis on perceptions makes it possible to come to grips with a constant difficulty: namely, paying attention to interactions between the particular features of individuals or families and the particular features of settings. Combining concepts of access with qualities of people One of the particular virtues of enhancement models is that they do pay attention to the qualities of individuals, rather than treat people solely according to their family position as father, mother, firstborn, son, daughter, or the like. Similarly,

Children, families, and communities

71

in transmission models, individuals are seen as varying in the vested interests they have in agreeing with each other or in hearing the other's message. Any model would, in fact, be seriously deficient, in the view of psychologists, if it did not take into account the fact that the effects of social settings may combine or interact with the qualities of individuals. How do concepts of access do this? For one thing, they describe individuals in terms of their perceptions of access, and - one step further back - in terms of further qualities that prompt particular perceptions. This will go some distance toward resolving the problem. We do know, at the start, that there are individual differences in perceptions about the nature of the outside world. If we concentrate on parents, for instance, we know - at least from our everyday experience - that parents differ in the extent to which they see the world as a dangerous place that undermines their own efforts at socialization. Some seem especially afraid of contamination. The child's virtues or socialization seem always precarious. Others seem more assured that exposure to other values will "not take" or that its effects will soon "rub off" in the face of home training. We might well ask what accounts for these differences and how they are translated into the way parents regulate the access of children to settings outside the home or help the child internalize the parental picture of the world. Similarly parents appear to differ in the extent to which they see themselves as able to influence their children's development either in general or in specific domains. Such differences in perceptions of influence, and the actions these give rise to, are matters about which we have begun to learn a little (see Bugenthal & Shennum, 1984; Goodnow, 1984, Goodnow, Knight, & Cashmore, 1985). However, a great deal more could be added to round out descriptions of parental qualities that interact with their perceptions of access or with effects these have on relationships among children, parents, and settings outside the family. These particular qualities, however, do not include a source of individual differences often emphasized by developmentalists, namely, age. I am sympathetic to the argument that analyzing community effects on families and children calls for "a developmental perspective - the recognition that community has different meanings for children of different ages and for their families" (Bronfenbrenner et al., 1984, p. 321). I also agree that the processes involved in any community effects may shift, depending on the age of the child (Bronfenbrenner, chapter 2). Given this preferential stance, why do I not proceed to develop a picture of individual differences with age as a primary quality? The reasons are twofold: (1) some data on effects suggest that age does not make a difference, and (b) the real individual differences (perhaps underlying age) may have something to do with one's novice or expert understanding of a setting and ability to influence it. Those reasons call for some elaboration. Evidence suggesting that age may not

72

Jacqueline J. Goodnow

make a difference concerns perceptions of school and of the morning at home before going to school. These were expressed in the form of answers to the following questions: If you could change one thing in the morning, what would you change? If you could change one thing at school, what would you change? These questions were asked of children in over 200 schools throughout Australia, and the details of their answers are given in Goodnow and Burns (1985). The striking result is the similarity of answers between children in the younger and the older sets of grades. The complexity of the language varies, but the issues remain the same. In effect, the nature of the setting seems to override the expected effects of age change. Consider, for instance, the wishes for change in the morning, with reference to changes at home. At both age levels, the major issues mentioned are problems of time and the lack of choice as to what one does or when one does it. Grade level does make a difference to the extent to which children mention problems of time and choice as being interrelated. The older children more often make statements along the lines, "I wish school didn't start until later and then I wouldn't have to . . . , " but at both age levels the theme of choice ("I wish I could . . . I wish I didn't have to. . . .") is the predominant refrain. The results for changes in school present a similar picture. Most of the wishes of juniors (Grades 1-3) and seniors (Grades 4-6) have to do with changes in what is taught, as they want to learn 4'different" things, generally more useful or more interesting things (33% and 35%, respectively). For both, the next largest category of wishes is to reduce the amount of time spent at school (17% and 22%) or not to go at all (20% and 7%). In addition, both are particularly concerned about changes in teachers. This concern increases among seniors (9% vs. 16%), either because teachers have changed or because children in Grades 4-6 are especially concerned with issues of ''fairness." The remaining wishes have to do with physical improvements to the school as a place (6%, 8%); to the nature of the work required, especially in terms of effort (7%, 6%); and peers (4%, 5%). Overall, a quality of children that we generally regard as being of prime importance - their age - contributes little to these perceptions of setting. Once in harness, one might say, young and old horses have much the same regrets. Therefore, I am prompted to ask what qualities other than age could make a difference to the perceptions of a setting such as school and to their impact. The qualities that appeal are those that have to do with being able to alter a setting, with the power one has to bring about change or to have one's wishes taken into account. When we consider the impact of school on children, for instance, the quality of individuals to which I consistently look for some interaction with the quality of settings - namely age and the difference it brings in intellectual understanding - may be less significant in many settings than such qualities as the ability to cajole the powers that be, to bend a teacher's agenda without

Children, families, and communities

73

attracting punishment (Mehan, 1979), to leave the field in deed or fantasy, or in other ways to soften the extent to which a setting such as school organizes the lives of all who enter. These qualities may have less to do with intellectual capacity than they do with know-how or, as Bern (this volume) suggests, with differences in personality and style.

Some final comments I have clearly not exhausted the possible relationships that may exist among children, other family members, and particular segments or settings in a community. Nor have I even touched on critical questions of method, such as the question of how one establishes that a community setting or condition has effects over and above those brought about by face-to-face interactions. Questions of method have been treated fully elsewhere (Bronfenbrenner et al., 1984; Parke, 1984; Lewis, 1984). My goal has been to isolate some issues that apply to a number of community settings (school, court, playgrounds) and community conditions (economic, legal, or social), to look for some general processes, to analyze these in order to bring out difficulties, expansions, alternatives, and research directions, and to borrow where feasible from studies of two-party relationships. Fitting community conditions and settings into our more traditional concern with face-to-face interactions is a challenging but not an easy task, but there are some positive ways to proceed. My main concern has been to outline these.

References Acock, A. D., & Bengston, V. L. (1980). Socialization and attribution processes: Actual versus perceived similarity among parents and youth. Journal of Marriage and the Family, 42, 501 — 15. Bandura, A. (1977). Self-efficacy: Toward a unified theory of behavioral change. Psychological Review, 84, 191-215. Barker, R. G., & Wright, H. F. (1955). Midwest and its children: The psychological ecology of an American town. New York: Harper & Row. Baszormenyi-Nagy, I., & Spark, G. (1973). Invisible loyalties: Reciprocity in intergenerational family therapy. New York: Harper & Row. Belsky, J., Robins, E., & Gamble, W. (1982). Characteristics, consequences and deteminants of parental competence: Towards a contextual theory. In M. Lewis & L. Rosenblum (Eds.), Beyond the dyad. New York: Plenum. Berger, P. L. (1977). Facing up to modernity. Harmondsworth: Penguin. Bott, E. (1957). Family and social network: Roles, norms and external relationships in ordinary urban families (2nd ed.). London: Tavistock. Bourdieu, P. (1977). Outline of a theory of practice. Cambridge: Cambridge University Press. Bourdieu, P., & Passeron, J. (1977). Reproduction in education, society and culture. Sage studies in social and educational change (Vol. 5). Beverly Hills, CA: Sage.

74

Jacqueline J. Goodnow

Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press. Bronfenbrenner, U., Moen, P., & Garbarino, J. (1984). Children, family, and community. In R. Parke (Ed.), Review of child development and research: The family (Vol. 7). Chicago: University of Chicago Press. Bugenthal, D. B., & Shennum, W. A. (1984). "Difficult" children as elicitors and targets of adult communication patterns: An attribution-behavioral transactional analysis. Monographs of the Society for Research in Child Development, 49 (No. 1, Serial No. 205). Bullivant, B. (1976). Social control and migrant education. Australian and New Zealand Journal of Sociology, 12, 174-83. Burns, A., & Goodnow, J. J. (1985). Children and families in Australia (rev. ed.). Sydney: Allen & Unwin. Cashmore, J., & Goodnow, J. J. (1985). Agreement between generations: A two-process approach. Child Development, 56, 493-501. Clausen, J. A. (1966). Family structure, socialization and personality. In L. W. Hoffman & M. L. Hoffman (Eds.), Review of child development research (Vol. 2). New York: Russell Sage Foundation. Dasen, P. (1980). Psychological differentiation and operational development. Newsletter of Laboratory of Comparative Human Development, 2, 82-86. Douglas, M. (1966). Purity and danger: An analysis of the concepts of pollution and taboo. London: Routledge & Kegan Paul. Douglas, M. (1982). Risk and culture: An essay on the reflection of technical and environmental dangers. Berkeley, CA: University of California Press. Edwards, M. (1985). Individual equity and social policy. In J. J. Goodnow & C. Pateman (Eds.), Women, social science, and public policy. Sydney: Allen & Unwin. Elder, G. H., Jr. (1974). Children of the Great Depression. Chicago: University of Chicago Press. Elder, G. H., Jr. (1981). History and the family: The discovery of complexity. Journal of Marriage and the Family, 43, 489-519. Elder, G. H., Jr., Caspi, A., & Van Nguyen, T. (1986). In R. K. Silbereisen, K. Eyferth, & G. Rudinger (Eds.), Development as action in context: Problem behavior and normal youth development. New York: Springer. Entwisle, D. R., & Hayduk, L. A. (1978). Too great expectations: The academic outlook of young children. Baltimore, MD: Johns Hopkins University Press. Furstenburg, F. F., Jr. (1971). The transmission of mobility orientation in the family. Social Forces, 49, 595-603. Goodnow, J. J. (1981). Some aspects of social policy in Australia and beyond. (Annual Konopka Lecture). Minneapolis: University of Minnesota Press. Goodnow, J. J. (1984). Parents' ideas about development and parenting: A review of issues and recent research. In M. Lamb, A. Brown, & B. Rogoff (Eds.), Advances in developmental psychology (Vol. 4). Hillsdale, NJ: Erlbaum. Goodnow, J. J. (1986). Organizing and reorganizing: Some lifelong everyday forms of intelligent behavior. In R. Sternberg & R. Wagner (Eds.), The origins of intelligence in everyday life. Hillsdale, NJ: Erlbaum. Goodnow, J. J., & Burns, A. (1980). Children and society. In R. Brown (Ed.), Children in Australia. Sydney: Allen & Unwin. Goodnow, J. J., & Burns, A. (1984). Factors affecting policies in early childhood education: An Australian case. In L. Katz (Ed.), Current topics in early childhood education (Vol. 5). Norwood, NJ: Ablex. Goodnow, J. J., & Burns, A. (1985). Home and school: Child's eye view. Sydney: Allen & Unwin. Goodnow, J. J., Knight, R., & Cashmore, J. (1985). Adult social cognition: Implications of parents'

Children, families, and communities

75

ideas for approaches to development. In M. Perlmutter (Ed.), Minnesota symposia on child development: Social cognition (Vol. 18). Hillsdale, NJ: Erlbaum. Hareven, T. (1982). Family time and industrial time. Cambridge: Cambridge University Press. Hartmann, H. (1981). The family as the locus of gender, class, and political struggle: The example of housework. Signs, 6, 366-94. Heath, S. B. (1983). Ways with words. Cambridge: Cambridge University Press. Heider, F. (1958). The psychology of interpersonal relationships. New York: Wiley. Hess, R. D., & Handel, G. (1959). Family worlds. Chicago: University of Chicago Press. Hinde, R. A. (1979). Towards understanding relationships. London: Academic Press. Huesmann, L. R., Eron, L. D., & Lefkowitz, F. (1984). Stability of aggression over time and generations. Developmental Psychology, 20, 1120-34. Jackson, B., & Marsden, D. (1966). Education and the working class. Harmondsworth: Penguin. Jennings, M. K., & Niemi, R. G. (1968). The transmission of political values from parent to child. American Political Science Review, 62, 169-184. Lautrey, J. (1980). Classe sociale, milieu familial, intelligence. Paris: Presses Universitaires de France. Lerner, M. (1974). Social psychology of justice and interpersonal attraction. In T. L. Huston (Ed.), Foundations of interpersonal attraction. New York: Academic Press. Levinger, G. (1974). A knee-level approach to attraction: Toward an understanding of pair relatedness. In T. L. Huston (Ed.), Foundations of interpersonal attraction. New York: Academic Press. Lewis, C. (1981). How adolescents approach decisions: Changes over grade seven to twelve and policy implications. Child Development, 52, 538-44. Lewis, M. (Ed.). (1984). Beyond the dyad: Genesis of behavior. New York: Plenum. Maccoby, E. (1984). Middle childhood in the context of the family. In W. A. Collins (Ed.), Development during middle childhood. Washington, DC: National Academy Press. Mehan, H. (1979). Learning lessons. Cambridge, MA: Harvard University Press. Parke, R. (Ed.). (1984). The family: Review of child development and research (Vol. 7). Chicago: University of Chicago Press. Poole, M. (1985). Pathways and options for ethnic adolescents. In M. Poole, P. deLacey, & B. Randahwa (Eds.), Australia in transition: Culture and life possibilities. Cambridge: Cambridge University Press. Rollins, B. C , & Thomas, D. L. (1979). Parental support, power and control techniques in the socialization of children. In W. Burr, R. Hill, K. Reiss, & F. I. Nye (Eds.), Contemporary theories about the family (Vol. 1). Glencoe, IL: Free Press. Rubin, Z., & Sloman, J. (1984). How parents influence their children's friendships. In M. Lewis (Ed.), Beyond the dyad. New York: Plenum. Russell, G. (1983). The changing role of fathers? Brisbane: Queensland University Press. Schoggen, P. (1983). Behavior settings and the quality of life. Journal of Community Psychology, 11, 144-57. Sharrock, W. W. (1974). On owning knowledge. In R. Turner (Ed.), Ethnomethodology, Harmondsworth: Penguin. Sheridan, A. (1980). Michel Foucault: The will to truth. London: Tavistock. Shweder, R. A. (1982). Beyond self-constructed knowledge: The study of culture and morality. Merrill-Palmer Quarterly, 28, 41-69. Smith, T. E. (1982). The case for parental transmission of educational goals: The importance of accurate offspring perceptions. Journal of Marriage and the Family, 44, 661-74. Spencer, M. (1985). Black families in the U.S.A. Paper presented at conference on Families, Cultural Transmission and Cultural Change. Stanford Center for Advanced Study in the Behavioral Sciences, March 1985.

76

Jacqueline J. Goodnow

Stodolsky, S. (1988). The subject matters: Classroom activity in math and social studies. Chicago: University of Chicago Press. Strober, M. H. (1982). Market work, housework and child care: Burying archaic tenets, building new arrangements. In P. W. Berman (Ed.), Women: A developmental perspective. Washington, DC: National Institutes of Health Publication No. 82-2298. Tedin, K. L. (1974). The influence of parents on the political attitudes of adolescents. American Political Science Review, 68, 1579-92. Troll, L., & Bengston, V. (1979). Generations and the family. In W. Burr, R. Hill, K. Reiss, & F. I. Nye (Eds.), Contemporary theories about the family (Vol. 1). Glencoe, IL: Free Press. Troll, L. E., Neugarten, B. L., & Kraines, R. J. (1969). Similarities in college students and their parents. Merrill-Palmer Quarterly, 15, 323-36. Valsiner, J. (1984). Construction of the zone of proximal development in adult-child joint action. In B. Rogoff & J. Wertsch (Eds.), Children's learning in the zone ofproximal development. San Francisco: Jossey Bass. Whiting, B. B. (1980). Culture and social behavior: A model for the development of social behavior.

Ethos, 8, 95-116.

Human development and social change: an emerging perspective on the life course

Glen H. Elder, Jr., and Avshalom Caspi

Epochs of rapid and drastic change repeatedly draw attention to important issues in the study of human development. These are the times that generate problems of human dislocation and deprivation, as illustrated most recently by the mass immigration of the 20th century and the turmoil of the 1960s. They also force us to recognize that any serious study of human development must consider time, process, and context. Periods of rapid change call for temporal perspectives on individuals, families, and environments. These concerns are at the base of our longitudinal research on human development in a changing society. The central hypothesis relates social changes in the Great Depression and Second World War to individual and family development. We therefore ask the following questions: What are the connections between social change and life patterns? What are the processes by which these effects are expressed and persist across time and place? And how do individuals surmount the limitations of a deprived childhood and manage to turn around their fortunes and lives? The life-course perspective (Elder, 1975, 1978a, 1978b, 1984, 1985) has become one way of bringing all of these elements together in a programmatic study. The life-course perspective is temporal and contextual in the sense that it locates people and family members in history through birth years and in the life course through the social meaning of age (i.e., age-graded events and social roles). From this perspective, multiple factors contribute to the shape of human growth and development, linkages between social change and lives provide microtheories or explanations regarding the influence of social change, and people represent both the agent and consequence of their changing life paths. This chapter is based on a program of research on social change in the family and life course (Social Change Project), currently supported by NIMH Grants MH-34172 and MH-41827. Glen H. Elder, Jr., was supported by a NIMH Senior Scientist Fellowship (MH-00567). We are indebted to the Institute of Human Development, University of California, Berkeley, for permission to use archival data from the Oakland Growth Study and the Berkeley Guidance Study.

77

78

Glen H. Elder, Jr., and Avshalom Caspi

These ideas are clearly illustrated by our research on the stressful antecedents of children's problem behaviors in the 1930s and on their life-course implications (the Social Change Project). In this chapter, we present the main outlines of this work, completed and proposed, and explain the logic of our inquiry. We begin with preliminary studies based on the Oakland Growth Study at the Institute of Human Development, University of California, Berkeley, and then turn to the Berkeley Guidance sample on which the current work is based. We conclude by offering general, integrative observations about linkages between social change and human development. Before we describe the project research, we should explain the intellectual history and underpinnings of life-course analysis. As a field of study, the life course has known two periods of theoretical and research vitality: the prewar (prior to 1940) phase, which is most closely identified with the early Chicago School of Sociology; and the decades including and following the 1960s. In both periods, social change acquired problematic significance for people and society. The history of life-course analysis The study of lives in a changing society The first era of life-course thinking began with the accelerating growth of cities through migration and socioeconomic development from about 1900 to 1940. On the eve of World War I, Walter Lippmann (1914, p. 152) offered a cogent description of this age: 'There isn't a human relation, whether of parent or child, husband or wife, worker and employer, that doesn't move in a strange situation." Some years later in their classic study of Middletown, the Lynds (1937) observed that "we today are probably living in one of the eras of greatest rapidity of change in the history of human institutions." At the time, the Chicago School of Sociology was known for its research on the multiple problems of a rapidly changing society - waves of immigrants to cities, high rates of delinquency and crime, family disorganization, and mental illness. W. I. Thomas played a leading role in this school and in the study of lives amidst changing times. Inspired by Thomas and Znaniecki's The Polish Peasant in Europe and America (1918-20), others began to use life records for the analysis of historical change and individual lives. Immigrant lives embodied the discontinuities and strains of the age. They were socialized for a world that soon became only a memory. The pioneering figure in this movement, W. I. Thomas (Volkart, 1951, p. 593), recommended that priority be given to "the longitudinal approach to life history." Studies, he claimed, should explore many "types of individuals with regard to their experiences and various past periods of life in different situations"

Human development and social change

79

and follow "groups of individuals into the future, getting a continuous record of experiences as they occur." Though promising and compelling, Thomas's prescriptive agenda saw few empirical applications between the Great Depression and the War on Poverty of the 1960s. The use of subjective life records advocated by Thomas was not in keeping with the new positivism of social science in postwar America. Crosssectional surveys became the dominant mode of data collection, ensuring a measure of contextual blindness with respect to lives and times. The second wave of research on the life course did not emerge until the 1960s. The life course as a new configuration The decade of the 1960s gave birth to three developments that shaped a new configuration of life-course study. First, social strife and discontinuities of the decade prompted new questions regarding ties between the generations, historical influences, and life-span development. Out of this intellectual context came preliminary concepts of a life course embedded in social institutions and ever subject to historical forces. Second, data resources emerged as if by design to match the needs of a temporal and contextual approach. New longitudinal, panel, and follow-up samples were established, yet their purposes often had little to do with a life-course study. A prime example is the Michigan Panel Study of Income Dynamics (Elder, 1985). Initiated at the end of the 1960s to address policy issues regarding poverty, the still active panel represents an unparalleled source for life-course research. Techniques of analysis appropriate to an evolving life course represent the third development. These include causal models based on systems of equations and event history analyses. The intellectual milieu of the 1960s offers a distinctive view of life-course dynamics as a field of inquiry. First, we see a renewed consciousness of and sensitivity to the bond between social change and the life course of individuals. Second, a number of theoretical developments, based on an understanding of age in society and history, helped to illuminate the relation between social and individual change. Third, research and theoretical efforts were mutually influential. Longitudinal studies in the 1960s were typically launched without theoretical knowledge of the life course, and yet their findings had important implications for this perspective. Consistent with Robert Merton's (1968) observations on theory and research, the story of life-course dynamics includes the interplay of concepts and findings. Several lines of inquiry gave birth to a concept of life patterns in particular historical times: an interactional perspective on life and social history. Norman Ryder's (1965) thoughtful essay, "The Cohort as a Concept in the Study of

80

Glen H. Elder, Jr., and Avshalom Caspi

Social Change," proposed the cohort as a concept for studying the life course. With its "life-stage principle," the essay provided a useful point of departure toward greater understanding of the interplay of social change and cohort life patterns: The impact of a historical event on the life course of a cohort reflects the stage at which the change was experienced. The implications of mass unemployment or military conscription obviously vary among individuals of different ages, from children to the middle aged and elderly. Differences in life stage tell us something about the adaptive resources, options, and meanings that become potential elements in linking social change to life outcomes. Ryder's essay on the bond between age and time brought more sophisticated awareness to the connections between historical and life time, and, in doing so, provided fresh insight into the temporal aspects of lives. Once individuals are placed in historical context, it is not easy to avoid issues regarding this context the impact of historical events and circumstances. The central message of Ryder's analysis posed worrisome implications for the study of age differences in psychological functioning. If social change differentiates the life trajectories of adjacent birth cohorts, then any age difference in behavior could be attributed to both historical and life change. The second wave of inquiry on life-course dynamics is uniquely distinguished by a convergence of research and theory on the separate dimensions of both social and historical age. The exemplar of this integration appears in Aging and Society (Riley, Johnson, & Foner, 1972). Age represents a basis of historical differentiation through cohorts and of social differentiation according to agegraded statuses and role sequences. Age strata thus order both people and social roles. Socialization and role allocation link people and social roles in the process of aging and cohort succession. This connection between historical age and cohort, on the one hand, and social age in the life course, on the other, has alerted investigators to the variable meaning of life events in history. Divorce is a case in point. With the sharply rising divorce rate from 1960 to the 1980s, one might expect some change in the meaning and psychological effects of divorced parents for children. From its rudimentary stage in the 1960s to the present, perspectives on the life course and its dynamics represent a theoretical orientation that defines a context for empirical inquiry. They identify relevant problem foci and variables, and structure the generation of evidence and hypotheses. In this orienting function, the perspective suggests questions, rationales for why they are important, and actual specifications of the questions for research purposes. The bearing of research on theory is equally important - the active role of longitudinal studies in shaping perspectives on the life course. Concerning this influence in general, Merton (1968, pp. 162-8) refers to unexpected findings and the discovery of new data that exert pressures for new theory.

Human development and social change

81

MACRO SOCIAL CHANGE

Economic Collapse of 1930s RESOURCES AND OPTIONS Education, Savings, Problem-solving Skills, Supports, Personal Efficacy, etc.

SITUATIONS.

CHANGE Economic Loss, Unemployment DEFINITION OF SITUATION Causal Attribution, Cognitive Appraisal ADAPTIVE RESPONSE Job seeking, Woman enters Labor Force, takes in Boarders, etc.

/OUTCOMESx

Sense of Efficacy ^ Assertiveness Resilience Economic Recovery

\ A Sense of Victimization Withdrawal Somatic Symptoms Continued Hardship

Figure 4.1: Theoretical model relating macrosocial change to individual adaptations and outcomes

Children under stress Children under stress in the Great Depression can be viewed as a group at risk of problem behavior. The task is to examine the persistence, consequences, and change of this behavior into the adult years. The development of competence in stressful situations represents another side of children's experience under stress (Garmezy, 1981). From studies of family history, Bleuler (1974, p. 106) concludes that one t4is left with the impression that pain and suffering can have a steeling - a hardening - effect on some children, rendering them capable of mastering life with all of its disadvantages.'' This theme emerged from our initial research on children of the Great Depression, using the birth cohort of 1920-2 in the Oakland Growth Study (Elder, 1974, 1979), and is now coupled with a perspective on the vulnerable child (Elder, Caspi, & Van Nguyen, 1986). The same stressor can have both pathogenic and salutary effects. The challenge is to understand the development of such outcomes. A schematic outline of our analytic model linking social change and individual development is presented in Figure 4.1 (Moen, Kain, & Elder, 1983). Two features of the framework deserve special attention. First, the model draws attention to the options and resources available to the family before the occur-

82

Glen H. Elder, Jr., and Avshalom Caspi

rence of economic loss. A second feature of the analytic framework is its stress on the critical role played by adaptive responses as mediating linkages between economic realities and their consequences for children. The Oakland Study, "Depression children" Two general lines of inquiry structured the Oakland study: (1) the social and psychological consequences of family income loss on children through the adolescent years; and (2) the long-term implications of these early outcomes of Depression hardship for the adult life course and personality. These questions led to the formulation of micromodels that specified plausible connections between a family's loss of income in the 1930s and the developmental course of the child. The Oakland Growth Study was launched around 1930 with fifth- and sixthgrade children from elementary schools in the northeastern sector of Oakland, California (Eichorn, 1981; Elder, 1974). The study was designed to investigate normal development - biological, psychological, and social. Data were collected annually from multiple sources - teachers, study participants, peers, classmates, and staff observers. Mothers were interviewed on three occasions, 1932, 1934, and 1936. None of the fathers were interviewed or otherwise contacted for information. One of the first challenges was to come up with a way of thinking about families over time. A related task involved devising ways of characterizing children's lives from the early 1930s into the adult years. Both of these tasks led to concepts of the life course. Adult follow-ups at the time were carried out in 1953, 1958-60, and 1964. Each follow-up included interviews, health assessments, personality inventories, and face-sheet questionnaires. The point of departure for the study was variation in economic change among the Oakland families during the worst years of the Depression. Some families lost heavily and others fared well. Indeed, specific families experienced even better times in the 1930s, owing to the sharp decline in cost of living. Keeping all of this variation in mind, along with a 25% drop in cost of living, we classified families with a loss of more than 34% in family income (1929-33) as economically deprived. All other families were defined as nondeprived. The basic design included deprived and nondeprived families within the middle and working class as of 1929. Initial class position provided a way of determining what families brought to the 1930s, their expectations, values, and resources. Class position in 1929 thus defined a cultural and economic context in which families worked out lines of adaptation to deprivation. Three micromodels were proposed for connections between family hardship and the Oakland children (Elder, 1974; Elder, Van Nguyen, & Caspi, 1985). These are shown in Figure 4.2. First, we hypothesized that income loss altered the household economy in ways likely to make a difference in the lives of the

Human development and social change

83

FAMILY PATTERNS AS LINKAGES

DEPRIVATION

HOUSEHOLD ECONOMY

CTDJIIIC

RELATIONS

STRAINS

CHILD-ADOLESCENT OUTCOMES Figure 4.2: Family patterns as linkages

Oakland children. Drastic loss of income shifted the household economy toward more labor-intensive operations. These new modes of economic maintenance included entry of mother and children into productive roles as earners; involvement of children in household operations, from food preparation to laundry, cleaning, etc.; doubling up of family units; and a reduction in expenditures. Such changes altered domestic and economic roles, shifting responsibilities to mother and the older children. As we moved to the adult years of the Oakland children, this household change acquired significance as an explanation for why women from hard-pressed families were likely to value homemaking, family activity, and responsibility in the role of parent. Men from deprived families who played an economic role in their household were more likely than other men to establish a stable career line at an early point and to value industry and responsibility in their children. The second micromodel involved change in family relations. Father's loss of earnings and resulting adaptations in family support increased the relative power of mother, reduced the level and effectiveness of parental control, and diminished the attractiveness of father as a model. Mother became a more central figure in the family, both concerning affection, authority, and the completion of basic tasks. This change in relative status appears in the attitudes of the Oakland boys and girls. However, prominent mothers frequently lost status with their children over the long run to middle age, especially if they played the "martyr role." More important, such family changes in response to loss of income accelerated the social independence of boys and enhanced the family dependence of girls. While boys from deprived families often lost a significant male figure, family change reinforced the subordination of girls to the household and to the mother. As a result, girls from deprived families acquired strong preferences for a domestic way of life, an orientation that they expressed as adults on matters of family life, conjugal relations, and children.

84

Glen H. Elder, Jr., and Avshalom Caspi

The third micromodel centered on social strains in the family. Heavy income loss magnified the likelihood of family discord, disorganization, and demoralization. Discord refers to tension, conflict, and even violence in both marital and parent-child relations. Discord contributed to family disorganization, to the loss of control over member behavior, and to erratic, unpredictable actions, such as arbitrary discipline. Demoralization refers to the depressed state of family climate and member outlook. These conditions were especially prevalent in working-class families, and it is among their offspring that we find the only long-term adverse effects of hard times in the 1930s. Their background entailed sacrifices in education and health at midlife. By comparison, family deprivation in the middle class is associated with psychological health among men and women. As a study of historical variations within a single cohort, the Oakland research inevitably raised questions concerning the boundaries of generalization. Indeed, the life-course framework (Elder, 1975; Riley, Johnson, & Foner, 1972) makes a case for expecting differing effects of Depression hardship between successive cohorts (the life-stage principle). Each cohort encounters a historical event, such as depression and war, at a different point in the life course. Such differences entail variations in the meaning of the event, in adaptive potential and options, and thus in linkages between the event and the life course. A comparison of deprivational effects in successive cohorts during the 1930s would most likely require explanatory models unique to each cohort in some aspects. The timetable of the Oakland adolescents enabled them to avoid the developmental risks of a young child in the social upheaval of the 1930s as well as the trauma of unemployment among adults. They were beyond the critical years of family dependency during the worst years of the 1930s and left school during a period of rising prosperity. This timing (according to a life-stage hypothesis) would have minimized long-term disabilities resulting from a deprived childhood, but we could not verify this account without data on adjacent cohorts. Temporal considerations of this sort led to a second phase of research using another longitudinal study at the Institute of Human Development. The Berkeley cohort, 1928-9 The comparison group is the Berkeley Guidance sample with birth dates of 1928-9 (Macfarlane, 1938). All members of the sample were assigned to matched groups: an intensively studied group (known as the Guidance subgroup) and a less intensively studied group. The two groups were identical on socioeconomic characteristics in 1929. From the original sample, 214 children were followed through the 1930s and early 1940s; 182 were members of the study through the age of 40 or 1970. Data on the parents' origin and their early life course to 1930 were obtained from the parents in 1929-30. Annual data

Human development and social change

85

(1929-45) on the parents, children, and the family unit as a whole came from observations by Institute field workers and teachers, and from self-reports (interviews, questionnaires). Postwar data from the parents were collected in 1969 and 1973, and from their children, the Berkeley subjects, in 1960, 1969-71, and most recently in the 1980s. Family and life records thus extend from the preDepression era to the postwar years and beyond. For purposes of comparison during the adult years, we focused on the age 40 follow-up. The Oakland men and women were contacted in 1958-60 at this life stage; the follow-up in 196971 is the comparable time for the Berkeley sample. The Berkeley archive extended research options in three major respects: (1) the availability of pre-Depression information on the family and individual members; (2) the presence of data on parents and children at three points relative to the hard times of the 1930s - before, during, and after; and (3) the availability of data on children's competent and problem behavior. Using data collected before the Great Depression, we could determine whether and how economic misfortune interacted with life experience and history. Was behavior in the Depression merely an extension of prior behavior problems? Did economic pressure accentuate the initial problem dispositions of parents and children and also enhance competent behavior in other families? Pre-Depression data thus enabled us to determine the protective and risk factors in children's lives, a task that could not be addressed in the Oakland project. The wealth of family information across the Depression decade also made it possible to investigate how the same family stress or income loss could have very different implications for children, as mediated by variable parent-child relationships. The overall design provides an approach to the life course that is comparative, contextual, longitudinal (temporal and process-oriented), and explanatory through an understanding of variations in historical circumstances. It is comparative in terms of subgroups that were differentiated by exposure to the economic collapse (nondeprived and deprived). Even more than the Oakland archive, the Berkeley data permit a contextual analysis of economic deprivation and its effects relative to what families brought to the Depression (such as class standing, marital quality, parent-child relations, values), and an explanatory probe of the course by which deprivation influenced the lives of children. Comparing the Oakland and Berkeley cohorts: the life-stage principle Comparative research on these cohorts has addressed the life-stage hypothesis, that environmental change, such as the Great Depression, affects people of different ages in different ways. This variability is reflected, in part, by the options and coping resources people of different ages bring to the adaptation process. The Oakland adolescents in Children of the Great Depression (Elder,

86

Glen H. Elder, Jr., and Avshalom Caspi

Table 4.1. Age of Oakland and Berkeley cohort members by historical events Age of cohort members Date

Event

1880-1900 1890-1910 1921-2

Birth years of OGS parents Birth years of GS parents Depression

1923 1923-9

Great Berkeley fire General economic boom; growth of "debt pattern" way of life; cultural change in sexual mores Onset of Great Depression Depth of Great Depression Partial recovery; increasing cost of living; labor strikes Economic slump Incipient stage of wartime mobilization Major growth of war industries (shipyards, munitions plants, etc.) and military forces End of World War II Postwar economic growth Korean War and the McCarthy era Civil Rights Era begins Civil rights mobilization: urban civil strife; Vietnam War End of affluent age in postwar America: energy crisis; rising inflation

1929-30 1932-3 1933-6 1937-8 1939-40 1941-3 1945 1946-9 1950-3 1954-9 1960-73 1974-

Oakland

Birth (1920-1) 2-3 1-9

Berkeley

9-10 11-13 12-16 16-18 18-20 20-3

Birth (1928-9) 1-2 3-5 4-8 8-10 10-12 12-15

24-5 25-9 30-3 34-9 40-53

16-17 17-21 22-5 26-31 32-45

1974) encountered hard times when they were beyond the years of family dependency and they entered adulthood after the economy had begun to revive. By contrast, children in the Berkeley cohort, seven to eight years younger than their Oakland counterparts, were less than 2 years old when the economy collapsed and they remained exclusively within the family through the worst years of the decade. Table 4.1 represents the key contrast in this comparison of the two cohorts. Cohort differences between Oakland and Berkeley appear in this age contrast and in the variable sequence of hard times and prosperity from early childhood to the adult years. Comparison of the Oakland Growth and Berkeley Guidance samples thus offered a rare opportunity to examine age and sex variations in vulnerability to drastic environmental change. Both studies were launched by the same research institute and they relied upon similar measures of social, psychological, and family behavior. The two cohorts also shared the same ecological setting within the eastern region of the San Francisco Bay area.

Human development and social change

87

Children's age and sex during the Depression crisis had implications for their roles and vulnerability within the family. In the Oakland families, adolescent boys and girls in deprived circumstances were often called upon to assume major responsibilities when their mothers worked and a good many held paid jobs, as did boys from similar circumstances. Family change of this sort enhanced the social independence of boys. But for girls, greater family involvement also meant greater exposure to family discord and tension. From theory and research, we expected the effects of Depression hardship to be even more severe in the lives of young children (Berkeley cohort), when compared to adolescents in the Oakland cohort. Moreover, this difference should be especially pronounced among young boys in view of their known vulnerability to family stress and marital tension (Rutter, 1979). To assess the effect of children's age in their family experience, we used similar measures from the Oakland and Berkeley cohorts to construct a simple causal model. The effects of the family deprivation (1929-33) and social class (1929) were estimated on scales of adolescent and adult (age 40) functioning. For both cohorts, income change was measured by comparing the 1929 family income and the figure for 1933, or the worst year. Taking into account the sharp decline in the cost of living in the San Francisco Bay area (about 25% in 1933) and the correlation between income and asset loss, all families that lost at least 34% of their income were classified as economically deprived. Smaller losses placed families in the nondeprived category. The two-factor Hollingshead index of family standing before the Depression (in 1929) was used in both cohorts. One factor is father's education and the other is his occupational status. California Qsort items were available on both cohorts for the adolescent and adult years (Block, 1971), and we constructed identical sets of measures of psychosocial functioning for the two cohorts in both time periods (see Elder, 1979; Elder, Caspi, & Van Nguyen, 1986). In adolescence and at midlife, the effects of family deprivation were negative primarily among the Berkeley males. The sons of deprived parents were more likely to be judged relatively low on self-adequacy, goal-orientation, and social competence in adolescence when compared with their nondeprived counterparts. A series of other analyses identified this group as more disadvantaged even up to the middle years (Elder, 1979). Heavy drinking, lack of energy, low psychological health, and dispirited attitudes toward work were more characteristic of the Berkeley men from deprived family environments. By comparison, no such disadvantage was observed among the Oakland males in adolescence. Indeed, the costs of family deprivation for young boys in the Berkeley cohort were matched in some respects by the developmental gains of older boys in the Oakland cohort. In the latter group, boys from deprived homes were judged even more resilient and resourceful than the nondeprived.

88

Glen H. Elder, Jr., and Avshalom Caspi

Among females, the short-term effects of family deprivation were negligible in the Berkeley (childhood) cohort and moderately negative in the Oakland (adolescent) cohort. The same pattern emerged in adulthood although it was not as pronounced. At least on psychological well-being, the Berkeley girls from deprived families had more in common with the Oakland boys than with males of similar age. Consistent with other research (e.g., Rutter & Madge, 1976), these findings suggest that family stress is more pathogenic for boys than for girls in early childhood, though not in adolescence. In adolescence, girls appear to be at greater risk than boys (see Werner & Smith, 1982). Acute social pressures and developmental change in early adolescence may have contributed to the psychological risk observed among adolescent girls. Other longitudinal studies (e.g., Simmons, Blyth, Van Cleave, & Bush, 1979) are beginning to identify early adolescence as a time of distinctive vulnerability for girls. In combination, our results suggest that risk factors vary in type and relative influence along the life course and between the sexes. Though separated by only 8 years, the two cohorts followed strikingly different historical trajectories. The timing of lives and historical events combined to produce a contrasting sequence of material well-being and scarcity. For the Oakland men and women, good times coincided with childhood and scarcity with adolescence. By comparison, the Berkeley adults experienced childhood during the worst years of the Great Depression and left this decade for the pressures and opportunities of wartime adolescence and an economic boom. As very young children when the economy collapsed, they were more vulnerable than their Oakland counterparts to family strains and they were exposed to a longer phase of economic hardship and its persistence up to the adult years. The developmental significance of this difference in timing is clearly suggested by the different life trajectories of children from different cohorts. The results of this cohort comparison support a perspective on families and children that locates them within the life course and its age-graded tasks and experiences. The ever-changing environment of developing individuals implies that risk factors should vary in type and relative influence along the lifeline. Sensitivity to this complexity of interactions among social, psychological, and biological factors is consistent with Kagan's (1979, p. 886) criticism of the search for ''absolute principles which declare that a particular set of external conditions is inevitably associated with a fixed set of consequences for all children." A number of important research questions were not open to investigation through a comparison of the Oakland and Berkeley cohorts, especially questions about risk and protective factors that affect the health consequences of economic stress on children's lives (Rutter, 1979; Rutter & Garmezy, 1983). Only the Berkeley Guidance archive provided measurement of family and individual be-

Human development and social change

89

havior before, during, and after the Great Depression. In addition, only the Berkeley Guidance study included measures of actual problem behavior. We now turn to these issues in the Berkeley research. Linking hardship and human development Any study of human development in a rapidly changing environment must attend at some point to the principal ways by which antecedents and outcomes are linked. Linkages may be thought of in terms of overlapping regions of different conceptual systems (Elder, 1973), such as the individual and the interactional context, the small group and the larger social organization, and the latter in relation to its surrounding environments. As an illustration, consider the work experience of father as an influence on the intellectual functioning of sons through his cognitive skills. The complexity and self-directed nature of father's work increases his level of cognitive skill and through this the intellectual development of sons (see Kohn & Schooler, 1983). Other linkages could be added, such as the connection between father's level of cognitive skill and choice of reading or leisure activity. Another version of this process might connect father's work with the achievement motivation of children through a specific type of family environment, one characterized by the achievement of successful role models. To bring out the motivational relevance of these family patterns, we must identify relatively concrete manifestations of an achievement model (such as father's personal standards, aspirations, and social prestige), and follow with the same procedure on achievement training - the encouragement of self-direction, emotional support, the presentation and reinforcement of standards of excellence. Without losing sight of the structural variable (father's work), we specify aspects of the family environment that provide explicit linkages to the ambition of children. The relative potency of these linkages can be tested empirically. Much dissatisfaction with studies of social structure and social change, however, is due to inadequacies of interpretation, to a failure in the conceptualization and appraisal of linkages. Literally hundreds of studies have correlated measures of socioeconomic status and change with indicators of child behavior, leaving the analysis without the barest feature of an interpretation or empirical test of linkages (Bronfenbrenner & Crouter, 1983). Zigler and Child (1968) make the general observation that a psychologist is not likely to be content with evidence that shows the effect of social class on cognitive functioning; psychologists do not feel they understand the effect of social class on cognitive functioning until they have "reduced the sociological variable by conceptualizing it as a set of psychological events that cause the behavior being 'explained.' " This position is expressed also in Block's (1971, pp. 271-4) argument on "the necessity of

90

Glen H. Elder, Jr., and Avshalom Caspi

psychologizing the notion of social class." If the term 'reduce' has troubling connotations, the intended message is clear enough. We must ask what it is about class that has relevance to thinking. To answer this question, we would specify concrete manifestations of life experience in differing social strata, both social and psychological. How does class position structure life situations, social options, and subjective reality? The concept of linkages helps to remind us of the connections between people and the social order. Of course, social class remains a vital part of the analytic model, no matter how successful we are in accounting for the paths through which it affects thought processes. Life-course studies in a changing environment combine the challenge of crossing levels of analysis with the complex task of crossing temporal phases. A case in point is the challenge of linking social conditions in childhood and early child behavior to the middle years. What is the process by which early experience persists across the life span? What are the mechanisms that allow problem behaviors to persist or be extinguished over time? Why is it, for example, that only one third of the problem children (in terms of delinquency, acting out) eventually become problem adults (Robins, 1978)? Consistent with a life-course model, Macfarlane, Allen, and Honzik (1954) argue that developmentalists must study situation changes to understand the problem behavior in question. They conclude their longitudinal study of problem behavior in the Berkeley cohort by noting that most behaviors in their classification are problem-solving adaptations to new situations. Whether externalizing (such as overactivity, quarrelsomeness, negativism) or internalizing (fearfulness, oversensitivity, withdrawal), children's problems express ways of dealing with a changing environment. Economic hardship and income loss represent such change. Between 1929 and 1933, the number of Berkeley Guidance families at the bottom of the economic ladder (below $1,500) more than tripled. Deprived families in the middle and working class (losses greater than 34%) experienced substantial increases in financial and psychological strain, household routines became disorganized, and family roles shifted to a pattern in which father represented a more peripheral figure in authority and emotional significance. Within this context we set out to study the antecedents and consequences of children's problem behavior. Research to date on the Berkeley cohort is characterized by five themes. The first centers on the resources and variations that were brought to the Depression crisis and examines the interaction of pre-crisis histories and hardship. The second focuses on the variable responses of families to changing situations and examines linkages between hardship and children's behavior. The third concerns the link between childhood behavior and the adult years, and its variation across

Human development and social change

91

situations or contexts. The fourth focuses on the intergenerational transmission of behavior patterns and attitudes. The final theme on turning points brings us to circumstances that enable people to change their lives around. The interaction of' pre-crisis histories and hardship In theory and in the research literature, the effects of drastic change and stress are generally contingent on what the individual brings to the situation (Elder, Liker, & Jaworski, 1984). Individual and relational attributes, such as parent coping, the marital bond, and parent-child relations, can be viewed as pre-crisis factors that impinge on the implications of family hardship. The Berkeley archive has yielded fresh insights about individual differences in coping with stress, and more specifically, about interactions between pre-Depression histories and economic stress. The first study (Liker & Elder, 1983) addressing this theme examined the "accentuation" hypothesis that personal dispositions interact with income change to influence behavior. The accentuation concept is drawn from the work of Feldman and Newcomb, who found that the distinguishing attributes of entering college students were apt to be "reinforced and extended by the experience incurred in those selected settings" (1969, p. 333). Along similar lines, we expected extreme hardship to produce more inadequacy among the unstable, and possibly even greater competence among the stable. Overall, the results support the view that unstable behavior in the hardpressed 1930s depended on the joint influence of economic hardship and prior psychological resources. Men who entered the 1930s calm and stable did not as a rule become more explosive under intense economic pressure. Indeed, our results suggest that men who were successful copers moderated any inclination toward erratic behavior when they lost jobs and income. Only one group became increasingly unstable under economic deprivation: men who were irritable, tense, and moody before hard times. Among these men we see an accentuation of characteristics already evident when entering the 1930s. Unstable men prior to the Depression were likely to become more unstable if they lost income, whereas calm, even-tempered men remained relatively unaffected (e.g., Elder, Liker, & Jaworski, 1984). From theoretical accounts of social change and stress, there is also good reason to expect adaptive variation by marital support (e.g., Emery, 1982). This conditional hypothesis has long been part of family stress theories in the prediction that relationships are differentially vulnerable to external pressures (e.g., McCubbin et al., 1980). Pioneering studies of families under stress in the Depression suggest that highly integrated families drew even closer under economic pressures, while loosely integrated families became even more disorganized

92

Glen H. Elder, Jr., and Avshalom Caspi

(Cavan & Ranck, 1939). One possible implication is that couples without chronic disputes and friction before hardship were better able to adapt to the economic crisis than couples who brought their conflicts into the Depression. Our results provide some support for the concept of marital integration as an adaptive resource (Elder, 1979; Liker & Elder, 1983). Conflicts in hard-pressed families generally reflected marital tensions before hard times, but marital quality was more likely to be diminished by economic pressures when marital relations were weak before misfortune. We made a more specific attempt to test the integration effect of economic crisis by focusing on the notion that marital bonds were actually strengthened in families that were highly integrated as of 1930. We divided the sample into subgroups of marital tension, 1930, by using the median as the dividing line. The effects of income loss turned out to be more negative in the high- than in the low-tension group, as expected, but hard times also made life more conflicted for couples who ranked high on marital quality before the Depression. This difference is merely one of degree, though it is clear that initially strong marriages were resilient in the sense that mounting financial conflicts did not alter the quality of the relationship to any appreciable degree. A final source of pre-Depression variation is found in parent-child relations prior to drastic income change. The Berkeley children entered the Depression crisis with different relationships to their mothers. Some were recipients of maternal warmth and affection, whereas others lacked such nurturance. The implications of this pre-Depression variation are suggested by the dependency of young preschool children on their family environment and especially on mothers. Maternal responsiveness to child cues affects both immediate and long-term adaptations by children (Schaffer, 1977) and, in times of stress, the emotional support and assurance of mothers can protect children from conflicts and overwhelming demands (Caplan, 1976). Using data from the Berkeley archive we examined the effects of family hardship and fathers' behavior on children whose mothers differed sharply on expressed affection prior to stressful environmental change (Elder, Caspi, & Van Nguyen, 1986). To investigate this question we set up a general causal model linking economic hardship to children's undercontrolled behavior and proceeded to test it under conditions defined by mother's relations to the child before the Depression. The model defined children's undercontrolled behavior as an outcome of a process whereby fathers' instability and Depression income loss increased the chances of inconsistent behavior by fathers. From prior research, we know that such parents, and especially fathers, increased the undercontrolled behavior of children (e.g., Elder, Liker, & Cross, 1984). We compared this model in groups of families where mothers ranked above and below average on nurturant, accepting attitudes toward the study child. Two central findings emerged from this analysis. First, fathers' emotional

Human development and social change

93

instability most sharply increased their inconsistent behavior and its influence on children's undercontrolled behavior when mothers were not supportive toward the child. Second, the influence of economic deprivation on fathers' inconsistent behavior was also greater in the nonsupportive group. Subsequent analyses have provided evidence that suggests that the anger-out behavior of children is also more likely to persist when mothers are not supportive. The parenting behavior of mothers prior to stressful times clearly played an important role in shaping the nature and course of father-child relations during stressful times. Indeed, mothers' relation to the study children before the onset of stressful times emerged as an important contextual determinant of how children fared in Depression families. The aversive dynamics of family stress were relatively limited when mothers were affectionate toward the study child. Family units with strong affective mother-child bonds restrained the arbitrary treatment of children by Depression fathers, and children with warm, affectionate mothers were under some protection from economic hardship. In summary, drastic change and life stress do not have uniform effects on all individuals. Whether an individual is adversely affected by income loss depends on other variables that moderate the impact of that change. What is different about people that rise above disadvantage? What are the protective or ameliorating circumstances? Conversely, what factors accentuate the adverse influence of stressful times? We have addressed here only some factors; others are no doubt relevant. We do believe, however, that the implications of family hardship can be understood only in relation to the resources and attributes - psychological and social, individual and relational - that are brought to change situations. In addition, the influence of environmental stressors on individual functioning must be considered in terms of the family context in which these operate, and personal (dyadic) relationships must be examined in terms of the network of interpersonal interactions in which they take place (Hinde, 1979). These elements are central to the ecological study of human development (Bronfenbrenner, this volume). Linking hardship to children's lives Studies of families under stress draw repeated attention to the problem of linkages. Economic setbacks clearly make a difference in psychological functioning (Barrett, 1979; Pearlin, Lieberman, Menaghan, & Mullen, 1981). But how is this difference manifested in children's lives? To understand the impact of economic hardship on children's lives requires knowledge of the adaptations chosen and played out by their parents. The adverse consequences of stressful economic times are not necessarily exercised directly. They are more often produced indirectly, through their disorganizing effects on family relations. The schematic model shown in Figure 4.3, which is based on findings from a

94

Glen H. Elder, Jr., and Avshalom Caspi FATHER IRRITABLE V, \ ARBITRARY DISCIPLINE

u

"A

CHILD EXPLOSIVENESS

MARITAL CONFLICTS

1_

MOTHER IRRITABLE Figure 4.3: Linking hardship to children's lives

series of studies, provides some insight into the dynamics of family behavior under stress and to the increased vulnerability of children in hard times (Elder, Caspi, & Downey, 1986; Elder, Liker, & Cross, 1984). Our model linking stressful events to family dynamics includes personality characteristics, qualities of relationships, and reciprocal influences between parents and their children. Personality is defined here as a characteristic way of responding to environmental stimuli, a relatively enduring quality of individuals. Like personality, social relationships have their own developmental history. Although partly dependent on the unique dispositions of individuals engaged in interaction, relationships evolve over time as emergent interaction patterns that cannot be explained solely in terms of individual personalities (Gottman, 1979). More specifically, we have focused on three qualities of disorganized families in stressful times: irritability, marital tension, and punitive parenting. These variables are assumed to mediate the effects of extrafamilial stressors on the child. To the extent that economic stress increases levels of irritability, marital discord, as well as rates of unpredictable, hostile exchanges, children are at greater risk. On the other hand, as long as irritable parents are kept in check, as long as discipline remains consistently applied, and as long as marital conflicts are peacefully resolved, children should be protected from the psychological costs of environmental stressors. On a general level, our analytic model bears some resemblance to Patterson's schematic outline of factors contributing to antisocial child behavior (1982, this volume): A major crisis, such as unemployment, increases the likelihood of a disruption of family management practices, which in turn increases the risk of antisocial child behavior. The key, of course, is to disentangle the various elements in multiproblem families. Because aversive interactions, whether between spouses or with children, so frequently occur in the context of a disordered

Human development and social change

95

family, their effects cannot be easily separated from the total pattern. We shall make an effort here to summarize what we do know about the process of influence. Two major conclusions are warranted on the basis of our work. First, the influence of family income loss on the undercontrolled behavior of children is indirect. Drastic income loss increases the prospects of children's problem behavior, but only by increasing aversive interactions within the family. Second, the master link between environmental stress and child behavior is inconsistent and arbitrary parenting. Inconsistent and extreme parental punishment have been repeatedly implicated in the development of aggression and undercontrolled behavior (e.g., Glueck & Glueck, 1950; McCord, 1979; Parke & Deur, 1972), and our findings suggest that economic change may well increase the probability of punithe behaviors by adults. Indeed, as long as discipline is consistently applied, children seem to be highly resilient under a wide array of parental styles. It is the inconsistent or arbitrary pattern that appears most problematic for healthful development in families under stress (Rutter, 1980). Two additional factors must be considered in accounting for the effects of environmental stress on children: marital discord and irritability. Marital discord. The recent appreciation of marital relations and their role in parent-child interaction represents one of the more important developments in the field of family studies and mental health (Belsky, 1984; Emery, 1982). "Every investigation, whether of questionable or sound methodology, has found marital turmoil to be related to some form of undercontrolled behavior" (Emery, 1982, p. 325). The critical question, however, is the process by which marital conflict or tension increases the risk of externalizing problems. Two potential dynamics are suggested by our findings (see Rutter & Garmezy, 1983). First, marital discord may undermine effective, consistent discipline. Under conditions of stress, parents become inconsistent themselves and relative to each other. Indeed, consistency between parents may be just as crucial as the consistency of each parent across situations (Block, Block, & Morrison, 1981). Second, marital discord may establish a setting for coercive parent-child exchange. Under conditions of stress, aggressive behavior in marital disputes is adopted as one means of achieving one's way. Indeed, reactions to sudden income loss conform to a theory of force in regaining control over life circumstances (Goode, 1971). Force is one means to this end and its consequences are reflected across family relationships. Irritability. Both parenting practices and marital tensions have profound consequences for children's behavior. But these characteristics appear to be specific manifestations of problematic behavioral histories that adults bring to

96

Glen H. Elder, Jr., and Avshalom Caspi

their family roles. Irritability in particular appears to set in motion aversive interactions among family members that eventuate in child disturbance. From a series of nonrecursive models, we know that irritability plays a major role in marital discord (discord has little effect on irritability in the short term), and that both factors increase the chances of punitive and inconsistent parental behaviors. The disposition to react irritably reflects a generalized coping response which has short-term purposes and long-term disadvantages. Indeed, what may be functional in the short run for the individual may inadvertently have negative consequences for the family unit. In Patterson's studies (1982, this volume), irritability has also been identified as the key variable for the understanding of children's social aggression. Parental irritability undermines family management practices, the rule structure of the household, the supervision of child behavior, and problem solving. Child influences on family relationships. Children also have an impact upon family activities as indicated by the feedback loop in Figure 4.3. It is highly probable that children residing in an aversive system make a significant contribution to the continued disruption of the family, and children's influence on parents should be especially prominent when economic pressures and discord are overwhelming. A quarrelsome, explosive child could easily exacerbate the irritability and arbitrariness of parents. Our findings support the premise of a mutually reinforcing process involving the problem behavior of parents and children in the Depression decade. For example, the arbitrariness of fathers increased the likelihood of children's temper outbursts, and the latter increased father's arbitrariness. A problem child is clearly an arousing and aversive stimulus. Moreover, undercontrolled behavior on the part of the child increases significantly the likelihood that aversive interactions will occur between other family members and in other family relationships. An especially noteworthy process of influence emerged with regard to child effects on the marital relationship. A problem child had little direct bearing on marital discord in hard-pressed families. However, problem children indirectly contributed to marital tension by increasing their parents' level of irritability. Thus, we can see the importance of viewing the reciprocal influences of family change on children and of children's behavior on the responses of families to economic change. Much more work is required to clarify the links between family responses to economic loss and the socialization of children, but it is obvious that such links have profound consequences. Unfortunately, the connections between family hardship and deleterious child outcomes are still drawn more often by conjecture than by empirical evidence. Whether family change results in negative effects on children's behavior clearly hinges on the adaptations chosen and played out by other family members.

Human development and social change

97

Childhood problem behavior and the adult life course The third emphasis of our research centers on the long-term implications of problem behavior in childhood. How does problem behavior shape individual lives? Through what mechanisms or processes are continuities in personality functioning established or maintained? Prospective life histories collected by longitudinal studies can help answer such questions as, How do the lives of undercontrolled children differ from the lives of inhibited, timid children? What aspects of the life structure are influenced by the inability to modulate impulses? Does withdrawn behavior interfere with an individual's ability to initiate action and make critical life decisions? The challenge is clear. As Robins (1984, p. 44) notes, "The potential contribution of longitudinal research in discovering the determinants of the duration of childhood problems is so enormous that it should inspire the next generation of research efforts." We address issues of continuity and change by viewing personality in the childhood years as an organizing force in life patterns. Our perspective on problem behavior as a precursor of the adult life course focuses attention on the age-related life schedules of individuals and their patterned movements into, along, and out of multiple role paths such as education, marriage, work, and parenthood. According to this framework, successful transitions and adjustments to age-graded roles are the core developmental tasks faced by the individual across the life course. The corresponding research agenda is to examine how various decisions and actions are shaped by childhood problem behaviors (Caspi & Elder, 1988). In our research, life-course trajectories are organized in terms of military, education, work, marriage, and parent roles. Our strategy is to obtain measures of behavior in specific situations relative to specific time periods (or developmental phases). This strategy enables us to study what people do in specific situations at specific points in the developmental process. Three sets of adult variables enter the analysis at this point: (1) the transition to adulthood and career beginnings, expressed especially in the differential timing and sequencing of life events such as education, marriage, parenthood, and work; (2) the adult trajectories of marriage, parenthood, and work, their stability, quality and coherence, and (3) midlife attainment, in the form of socioeconomic achievement, relations with kin, children, and friends, as well as in psychological health. We assume also that the life-course implications of children's problem behavior vary according to whether the behavior is a form of undercontrol or overcontrol. Undercontrol or externalizing problems disrupt relationships and thus bear directly on the impairment of achievement prospects, whether in education or work life, in marriage, or in parenthood. Achievement refers to the attainment of status or goals and to the maintenance of relationships. By comparison,

98

Glen H. Elder, Jr., and Avshalom Caspi

internalizing problems represents a passive behavioral style, suggesting impaired initiative and timing that may, in turn, lead to further difficulties in the adult years. Our initial empirical efforts have centered on childhood temper tantrums and social withdrawal (Caspi, Elder, & Bern, 1987, 1988). Consider, first, children who in late childhood exhibited a stable pattern of temper tantrums. We have shown that when these children, as adults, faced military authority and its demands, they adjusted poorly. An explosive, undercontrolled relational style in childhood also led men to occupy jobs that were not in keeping with their social backgrounds or intelligence, in part because they failed to conform to the demands of the educational system. On the job, men with a history of temper tantrums were more prone to conflicts and explosive relationships, especially when they were placed in highly structured circumstances that may have evoked defiant reactions. Among women with a tantrum history, these achievement differentials were expressed in downward mobility through marriage. Finally, in family interactions, adults characterized by explosive behaviors in childhood were at greater risk for marital discord, divorce, and ill-tempered, inadequate parenting. The life-course patterns of withdrawn children are more benign, though no less problematic. We have shown that men with a childhood history of social withdrawal adjusted poorly when they were called upon to make critical choices and decisions in their transition to adulthood. A withdrawn and reserved relational style in childhood took the form of relatively late and off-time transition behavior in entering marriage, the parental role, and work. Over the course of adulthood, off-time transitions gained added significance. Men with a childhood history of social withdrawal who delayed entry into adult roles frequently encountered a structurally incompatible nexus of roles. And such incompatibility, expressed by ongoing efforts to settle into a career and sustain a marriage, increased their risk of disordered work careers and marital instability. With a broader sample of childhood behaviors in each problem category - externalizing and internalizing - we plan to follow up these issues more fully. The most challenging question we currently face is why problem children persist in approaching and responding to the world in maladaptive ways. What are the processes that sustain maladaptive behaviors across time, in diverse situations, and in new relationships? Elsewhere we have proposed two processes: interactional continuity and cumulative continuity (Caspi & Elder, 1988; Caspi, Elder, & Bern, 1987, 1988; see also Bern, this volume). Interactional continuity refers to the reciprocal, dynamic transaction between the person and the environment: The person acts, the environment reacts, and the person reacts back. For example, Patterson's work with aggressive boys (this volume) has shown in elegant detail how family

Human development and social change

99

interactions can create and sustain destructive patterns of behavior. By extension, we suggest that children who coerce others into providing short-term payoffs in the immediate situation may thereby learn a behavioral style that continues to "work" in similar ways in later years. The immediate situation short-circuits the learning of more appropriate interactional styles that might have greater adaptability in the long term. Cumulative continuity refers to dispositional continuity that is maintained through the progressive accumulation of the consequences of the behavioral style. We refer here to the case in which the individual's disposition systematically selects him or her into particular environments, environments that, in turn, might reinforce and sustain those dispositions (e.g., Scarr, this volume). For example, when extroverts preferentially seek out social situations, they thereby select themselves into environments that further nourish and sustain their sociability. Maladaptive behaviors can similarly select individuals into environments. Thus, the ill-tempered boy who flunks out of school may thereby limit his future career opportunities and select himself into frustrating life circumstances that evoke a pattern of striking out explosively against the world. His maladaptive behaviors increasingly channel him into environments that perpetuate those behaviors; they are sustained by the progressive accumulation of their own consequences. Of course, interactional and cumulative continuity may, and frequently does, operate in concert to maintain both externalizing and internalizing problems. Our research task is to assess these sustaining processes across the life course. Intergenerational transmission of behavior and attitudes A fourth concern of research on the Berkeley cohort has been to trace the influence of Depression hardship through the family and into the subsequent lives of parents and children. What are the legacies, valued and unwanted, that are passed on to generations who have no direct contact with the Depression years? The theoretical framework for this work brings together two lines of research within a life-course perspective (Elder, 1984; Elder, Caspi, & Downey, 1986): behavior across the life span and the intergenerational transmission of behavior. The first line of inquiry (defined above) focuses on the course or history of behavior patterns, personalities, and social relationships within the life span of an individual. The second line of inquiry focuses on the dynamics of intergenerational transmission, a process that bears directly on the puzzle of how society is possible. Our joint emphasis on individual and family development is based on an approach to the life course that adds conceptual distinctions from the field of

100

Glen H. Elder, Jr., and Avshalom Caspi

kinship to an age-based model of the individual life course. Age and kinship represent complementary sources of temporality and context for life-course analysis. Chronological age brings a concept of age-related trajectories and transitions to studies of problem behavior and aversive family interactions. When individual development is depicted as a series of behavioral reorganizations around a series of developmental tasks, it becomes possible to define maladjustment in terms of deviation from expected and normative patterns. Concepts from the kinship domain (generation, lineage) broaden the lifecourse framework beyond the single life span. Following the assumption of intergenerational dependence, events in one generation often have consequences for events in the older or younger generation. By attending to individual lives across time and their interweave through emergent family patterns, life-course analysis highlights the interactive relation between individual and family functioning. This interactive view of individual and family implies that people are changed by changing family environments, and families are changed by changing people. Using archival data on four generations in the Berkeley Guidance Study, three orienting hypotheses were identified regarding the relation between problem behavior and unstable family relations (Caspi & Elder, 1988; Elder, Caspi, & Downey, 1986). Problem behavior refers to a cluster of behaviors that identify high-strung, irritable, and explosive personalities who display an inability to cope with external pressures and difficulties. Unstable family relations refer to aversive systems of family interaction, as in high levels of marital discord, parent-child hostility, and punitive parenting. The first hypothesis concerns the reciprocal dynamic linking problem behavior and unstable family relations in one generation. As depicted by the A link in Figure 4.4, the primary flow of influence runs from problematic behavior to problematic relationships in the family. The second hypothesis views the family environment as a link between the problem behavior of parents and the problem behavior of their children in the next generation. This causal pathway combines the A and B linkages in Figure 4.4. The third hypothesis views the problem behavior of children as one route by which unstable family relations in one generation are replicated in the next - the B and C linkages in Figure 4.4. Thus, children acquire problem dispositions and they carry them into marital and parent-child relations many years later as adults. These pathways represent the primary links in the hypothesized causal chain over time. The ideal design for investigating this interlocking pathology would cover the entire life span of multiple generations. Behavioral transmission across the generations should be linked to processes of life-span stability and change from early childhood to the middle years of adulthood to old age. To capture this dynamic

Human development and social change Gl Generation

101

Problem f^ Unstable Behavior ^ ^ > F a m i l y (explosive, Relations etc.)

IB

G2 Generation

Problem ^C^ ^ Unstable Relations

Figure 4.4: Intra- and intergenerational processes linking problem behavior and unstable family relations

we would need to follow members of each lineage unit over their life span as they age and interact with one another. Intergenerational relations at a point in time could thus be viewed in relation to both individual aging and a relational history of self-other interaction. Our work in this area represents a compromise between the ideal model noted above and the realities imposed by the Berkeley data archive. The Berkeley Guidance Study archive includes members of four generations: The cohort of children (G3), born in 1928-9, which has been studied prospectively up to the middle years, along with their grandparents (Gl), parents (G2), and children (G4). Because not every generation was studied prospectively, we utilized ordered dyads as the primary domain for our intergenerational analyses and not the full sequence of generational units. In the first phase of our study, we examined grandparents (Gl) and parents (G2). This pattern was replaced by parents (G2) and children (G3) in the second phase. The last part of the picture centered on the grown-up children (G3) and their offspring (G4). Across the four Berkeley generations, we find substantial documentation of a reinforcing dynamic between problem behavior and unstable weak ties in the family. Aversive family patterns mediate the influence of unstable parents on offspring, and they are reproduced in the next generation through the development of offspring who are least able to sustain and nurture enduring relationships. From Gl to G2 generations, unstable personalities are reproduced in part through marital tension and parent hostility. This socialization environment linked unstable grandparents to a similar style of behavior in their middle-aged sons and daughters. Likewise, unstable parents in the G2 generation markedly increased the likelihood for ill-tempered difficult children by increasing marital tensions and arbitrary tension. In this cohort, the pattern was also fueled by stressful events associated with the Great Depression. The pattern finally repeats itself in the G3 generation during their active years of parenting. Indeed, weak

102

Glen H. Elder, Jr., and Avshalom Caspi

family ties were characteristic of adolescents in the fourth generation (G4) who ranked highest on lack of self-control. The intergenerational cycle of problem behavior and problem relationships resembles a general dynamic across the four Berkeley generations. The evidence reviewed here - on the enduring structure of early personality and on the intergenerational replication of aversive dynamics - provides insights about the origins, transmission, and sequelae of problem behaviors. The course of individual differences, however, is probabilistic; disadvantages - personal, familial, or environmental - are clearly not immutable, and for this reason "biographies cannot be written in advance" (James, 1890). Equally deserving of analysis, then, are the factors that enable individuals to surmount the limitations of a problematic childhood. What are the mechanisms that enable some men and women to turn around their fortunes and lives?

Turning points in the life course Change may be explained by the fact that at times in people's lives a fortuitous event is encountered that intervenes in the developmental process rather than simply confirms an established mode of relating to the world. Bandura (1982) has referred to this process in an essay on chance encounters. Some experiences channel behavior into predictable and unprofitable directions, but other experiences may direct people toward new life paths. Little research has documented such processes, however. A central aim of our research is to identify key turning points in the life course of children from deprived backgrounds. According to project research, three life transitions represent potential change or turning points in which the life course is recast: the quality of the first marriage, entry into college, and entry into military service. The quality of the first marriage bears directly on the development of a primary support system that can provide direction and soften the blows of adversity. The stabilizing influence of an understanding spouse on adults with a deprivational history has been documented by Robins (1966) and Quinton and Rutter (Quinton, Rutter, & Liddle, 1984; Rutter & Quinton, 1984). The role of college and military service has not received as much systematic attention, but both events may open up a new opportunity system of significant figures, incentives, and goals.

Entrance into higher education. As we follow the Berkeley and Oakland children from deprived families to the middle years, one early transition stands out above all others as the critical line of demarcation between achieve-

Human development and social change

103

ment and continuing disadvantage - entrance into higher education. College entrance opened the door in both cohorts to career lines with advancement opportunities, and college men from deprived families generally made the best of them (Elder, 1974; Elder & Rockwell, 1979). Even when college amounted to little more than a year or so, it made a significant difference in the career entry and progress of men who grew up in deprived homes. In the case of women, college entry created social opportunities for advancement through marriage. From early adulthood to middle age, the men who succeeded in entering college also managed to advance well beyond worklife expectations based on their background, rising to the occupational level by midlife of men from more affluent homes. But when coupled with less schooling, Depression hardship forecast an adult life that had some parallels with the 1930s. These men from deprived homes fared poorly in work (more instability, less advancement) than the nondeprived. They also ended up with a lower occupational position by midlife (age 40).

Military experience in the life course. Life trajectories are formed through the influence of early childhood on subsequent life changes and by events or experiences that enable people to alter the course of their lives. Unpromising origins may set in motion a cumulative sequence of life disadvantages, but they can also be surmounted through opportunities and initiative. An understanding of such change through military service presents a major challenge, especially in relation to the mechanisms that enable men to turn their lives around. As the Oakland and Berkeley men entered the service, they experienced a disjuncture in their lives that resembled the disjunctures of the Depression decade. For the most part, the war years and military service broke with the past, a past defined by hardship, frustration, and limited opportunity (Elder, 1986). Placed within this context, military service (from 1940 through the mid-1950s) acquired special historical significance in relation to life-course development and aging. Indeed, service in the military has emerged as a plausible source of life change. The more we probed the evidence on military service, the more certain we became of the need to consider this event along with that of Depression hardship (Elder, 1986; Elder & Bailey, 1986). In the transition to adulthood, military entry closes options, creates others, and postpones still others. It establishes a moratorium from decision pressures in a structured environment, and rearranges the traditional pathway to adulthood, delaying career progress and commitments. Early entry into the service, in particular, should extend the transition to adulthood, postpone family events, and

104

Glen H. Elder, Jr., and Avshalom Caspi

their responsibilities, and provide a route to future opportunities through separation from home, exposure to new places and people, and access to educational benefits. As a whole, military service may well have placed children from deprived circumstances on a more rewarding trajectory. Consider the notion that there is an optimal time for life events. In the case of military service, the optimal time seems to be early entry, as soon after high school as possible. On the "early" track, the service comes before family and serious work careers. In many cases, men who enter later do so after marriage, children and launching a career. Service time would be disruptive, but it could also build upon accomplishments and enhance continuity. From one perspective, among others, early military service represents a moratorium or holding operation relative to the age-graded career, a legitimated "time out" from commitment pressures. This concept of service time appears in the account of a young Berkeley man at age 30. Uncertainty about future directions made the army especially attractive. "I didn't know what I wanted, though I probably wanted to get away from home and community. I didn't know what to take in college. I recall arguing with myself about maybe I'd grow up and settle down a bit if I got into the service." Coupled with this moratorium experience is the developmental payoff of a physical break from home and greater independence, as well as exposure to male models in a structured program. At first thought, the timing of military service in wartime seems unlikely to be a matter of choice. However, military service does combine elements of choice and constraint. Within the limits of the draft and local draft board policy, young men could decide whether to enter immediately or to stay out as long as possible, perhaps aided by a student or work deferment. Assuming that life choices are made within the context of a life history, we hypothesized that early entry among the Oakland and Berkeley men is most strongly linked to military ideologies and felt deprivations that made joining up an attractive option. In both cohorts, the early entrants were characterized by attitudes favoring the military role and by disadvantages of one kind or another. Early entrants in the Oakland cohort were more likely than the late entrants to come from large families and from the less successful ranks in school. They were also more positive toward the U.S. military. In the Berkeley cohort, early entry was most common among those who held positive attitudes toward the armed forces, and among those who experienced at least one of the following disadvantages - an economically deprived background in the 30s, academic trouble or difficulty, and low self-esteem. These factors were less predictive of military service as a whole. Military service entailed a delay in the timing of family events (marriage, first birth, etc.) among veterans in both cohorts and especially among the early

Human development and social change

105

entrants. The late entrants were more likely to have entered the service with a wife and child. Military service also led to a delay in the completion of education. As a whole, the early entrants tended to follow a less conventional path of adult achievement than the late entrants, and this observation applies to both cohorts. The early joiners had less education and they did not fare as well on occupational achievement by midlife. However, they frequently acquired additional education after military service and managed to narrow the gap in life achievement relative to the late entrants. On the basis of background, the early entrants should have ranked higher on marital discord and divorce than the late entrants, given their more disadvantaged background. But we find no such difference in either cohort. Marital stability consistently favors the early veterans. An important explanation for this change in outcome involves the consequences of a large delay in marriage. The early veterans married at a much later time than did the nonveterans, a timing that increases the prospect of marital permanence. In both cohorts the developmental influence of military service appears among veterans who entered the service at a relatively early age. According to social and psychological evidence, these men show the greatest development and career achievement when measured from baseline. At midlife, a larger percentage of the early veterans linked military service to personal growth and improved life chances, when compared to the late veterans. Military service was clearly a turning point for a good many of these men, though a precise understanding of the mechanisms awaits more detailed studies. Control cycles and the evolving life course: a concluding note Human development in a changing society poses many demanding questions about the life course, such as how it evolves or changes. One approach to the study of change focuses on the societal or institutional level. A changing society alters the course of people's lives by changing social institutions and complex organizations. Major school reform is one example of such change (Featherman & Sorenson, 1984). The extension of mandatory public education to older age groups adds an element of standardization to the pathway from childhood to the adult years. Another approach focuses on individual maturation and aging, emphasizing the behaviors and choices of individuals and how these shape life pathways or trajectories (Scarr, and Silbereisen & Noack, this volume). Particular choices reflect individual socialization and development, as expressed through family, school, and neighborhood. Both of these perspectives are linked in a third approach. The evolving life course of individuals and families is a product of the interplay between individual

106

Glen H. Elder, Jr., and Avshalom Caspi

Resources Figure 4.5: Discontinuity between soaring claims and available resources (historical examples: initial stage of rapid economic growth, World War II)

and environmental changes. Age-graded schools establish a range of options for moving through the educational system. The option followed leads back to the individual. Some students qualify for the honors track, whereas others remain in general education. Studies in the Social Change Project have relied on this interactive approach to study families and individuals across time in rapidly changing environments. The connecting thread across levels of analysis involves the core idea of control, as initially developed by W. I. Thomas in his writings on crisis situations. Thomas (1910) viewed control of desired outcomes as a function of the changing relation between claims relative to resources. Control is achieved when resources make it possible to fulfill claims. In Thomas's formulation, resources include the potential means to claim satisfaction, both psychological and social. A widening gap between claims and resources creates a disturbance of habit; habitual ways of behaving no longer meet expectations. Figure 4.5 shows how this gap may result from soaring claims during periods of rapid economic growth. Even in the late 1930s, aspirations rose quickly and spurred feelings of frustration. Durkheim (1951, p. 248) refers to this dynamic as the "malady of unlimited aspiration." The more one has, the more one wants. Figure 4.6 shows that a gap may also arise when resources decline sharply, for example, through income loss during the Great Depression. Both aspiration gain and resource loss entail a loss of control over desired outcomes, but the solution to each problem differs. Resource loss requires efforts to create more income or resources; it also calls upon reductions in level of expectation and expenditure. Soaring aspirations call for limits on demand, as well as efforts to increase resources or income. The same dynamic applies to individuals. Both of these new conditions, from soaring aspirations to loss of resources, represent a transition to a new situation and a loss of control potential. They also represent well-known disjunctures between family or individual change and the

Human development and social change

Ti

107

T2 Claims-,

§Lack of control! -"Crisis"!

1 Resources^ Figure 4.6: Loss of control: Discontinuity between claims and resource loss (historical examples: economic depression, severe inflation)

surrounding environment. In our studies of Depression families, adaptations to drastic income loss were efforts to regain control potential, to regain a balance between claims and resources in the achievement of desired outcomes. The adaptive options available to families were determined by the times (for example, the public aid option changed markedly over the 1930s). The option selected had much to do with what families and individuals brought to the Depression decade, both psychological and social resources and vulnerabilities. The choice also had much to do with the nature of the economic crisis or loss. Some men lost income through unemployment, while others lost similar amounts through reductions in work week or through loss of customers. The jobless sought jobs as many of their wives and children earned small wages on odds and ends; the shortweek condition led men to carry more than one job, if possible; and customer loss motivated efforts to retain customers, for example, in providing larger charge accounts. What we see in family and individual adaptations is a process by which the social life course is constructed. Loss of control sets in motion a life-course dynamic involving efforts to regain control. Cases of soaring aspirations and loss of resources can be found across the programmed and idiosyncratic transitions of the life course. One of the better examples comes from the literature on marriage and the pitfall of idealization. Rising expectations of what marriage will offer place the new relationship under greater strain and increase prospects for disillusionment. The profound and often devastating economic loss following divorce illustrates the implications of resource loss. Retirement also provides examples of this disparity. But perhaps even more relevant to life transitions is the loss of control potential that, as shown in Figure 4.7, arises from discontinuity between resources before and after a transition, or between claims before and after an event. In Thomas and Znaniecki's landmark study, The Polish Peasant, discontinuity is vividly illustrated by the process of migration to America from the

108

Glen H. Elder, Jr., and Avshalom Caspi

Claims-^

WLack of control! -"Crisis"!

-Resources*^ 'Some inappropriate for adaptive requirements of new situation.

Figure 4.7: Loss of control: discontinuity between resources before and after a transition (historical examples: rural European immigrants in urban America)

feudal countryside of 19th-century Poland. Immigrants brought skills to the new land that were not useful in an urban, industrial environment. The same could be said for claims or aspirations. Unplanned transitions in the life course often place individuals in predicaments where they lack the essential skills to cope effectively. Even with education for prospective parents, the birth of a first-born places them in a situation where they feel ill-equipped to meet expectations, theirs and others. Within the age-graded life course, transitions to new roles entail some loss of control potential and prompt efforts to regain control through adaptations of one kind or another. A predictable transition does not ensure control over desired outcomes. The control cycle dynamic in constructing the life course rests on the link between losing control and efforts to restore control over life outcomes. This connection has been documented by studies of personal control and reactance behavior. Reactance feelings occur whenever one or more freedoms or expectations are threatened or eliminated. Such feelings motivate efforts to regain control. The Brehms (1982, p. 375) refer to the overwhelming evidence for such motivation in their review of the field. "A problem threatens control, whether it is a matter of understanding, attitudinal disagreement or explicit loss of control. It is the threat to control (which one already had) that motivates an attempt to deal with the environment. And the attempts to deal with the environment can be characterized as attempts to regain control." On the larger screen of the life span, defining the immediate situation (see Figure 4.1) has much in common with Thomas's ideas regarding life planning and life organization. Life plans are based on concepts or definitions of the future life course. Revised thinking about the future prompts new life-course planning the situation has changed. As a means of controlling social reality for personal needs, the individual in Thomas's analysis devises general schemes of situations with goals, a life organization that provides a plan for subsequent action. This

Human development and social change

109

organization represents a continuing "project" of construction and reconstruction as a new experience is encountered that cannot be assimilated. The more drastic the change in life experience, the more one's life organization must be recast and the more the transition marks a probable turning point in trajectory. Major transitions, then, are times that increase the likelihood of a reorientation in life structure, direction, and purpose. They are times that make life planning a conscious act. Four general points from our studies summarize the relation between control cycles and the evolving life course: 1. The disparity between claims and resources may occur through rising claims, declining resources, and a discontinuity between acquired and needed resources. 2. The experience of losing control over one's life situation evolves from the preceding disparity. 3. Efforts to restore control or alternative modes of resignation involve adjusting claims, resources, or both in terms of their relation. 4. The life course may be altered through new lines of adaptation and their consequences. The precipitating event for this process is an event that substantially changes the balance between claims and resources, the actor's control potential. Options depend on conditions - the structured situation; responses to life change and the loss of personal control entail choices among given options, and this constraint is one way that a social institution shapes the life course. Action alternatives also depend on the resources people bring to the new situation. Control emerges as a central theme across many examples of linkages in the Social Change Project. We have seen how the connection between "losing control and life disadvantage" derives from a cumulative dynamic in which selfdefeating efforts to regain control produce even greater life disadvantages. Force within the family initiates a dynamic that can eventually make it ineffectual; force generates counterforce. Another example comes from the decision to enter military service, which can be viewed as one element of the dynamic in regaining a sense of control - an effort to gain the mastery and recognition so long denied the children of misfortune. Conclusion This chapter describes a program of research that began 25 years ago in the archives of the Institute of Human Development on the campus of the University of California, Berkeley. Although few analysts at the time even used the term "life course" in their research, the data archives of the Institute called for an approach that is now labeled the life-course perspective, an approach that views people and families over time as a dynamic process in context. The unparalleled

110

Glen H. Elder, Jr., and Avshalom Caspi

archives of the Institute served as a mentor in life-course thinking, as did the intellectual currents of the time. The approach we have followed in this program of work reflects the times as much as it matches the requirements of the problems at hand. Sociological insights are fashioning a distinctive approach to human development by relating it to social structure and history. Twenty years ago we saw this imagination at work in probing differences between middle- and working-class parents on values and behavior (Kohn & Schooler, 1983). More recently, we see its guiding hand in shaping a view of development across the life span and in specific historical settings. Growing awareness of an ever-changing environment is giving overdue visibility to matters of time, context, and social dynamics in the study of human development. Problems of social change and human development have gained significance as research foci, along with the concept that people make social history even as they are influenced by it. Views of this dialectic have emerged in the form of a life-course perspective that is based on the premise that aging entails interacting processes (social, psychological, and biological) from birth to death, and that life-course variations among individuals and cohorts are shaped by historical conditions. The dialectical perspective of life-course analysis, relating social history and life history, brings a number of issues to the fore that call for fresh thinking. Unfortunately, in most studies, structural and psychological variables still have a timeless quality. Social structures are seldom rooted in historical contexts that give them particular meaning, nor are they linked to psychological characterizations of the individual. Yet we know that the psychological effects of a social structure depend on historical context. Three emphases distinguish our theoretical orientation to the study of lives from the usual study of life-span development and aging. The most distinctive feature is our emphasis on social change and its effects on individual behavior and family patterns. Related to this theme is attention to linkages between social change and life patterns, and to the effects of differential life paths on developmental processes. A third feature is the focus on intracohort processes, as opposed to comparison of whole cohorts. At present, life-span research and developmental psychology are largely descriptive, relying on cohort comparisons instead of intracohort comparisons and offering little understanding of social change, the life course, and their relationship. Greater social science imagination (Mills, 1959, p. 12) is needed to "grasp history and biography and the relations between the two within society."

References Bandura, A. (1982). The psychology of chance encounters and life paths. American Psychologist, 37, 747-55.

Human development and social change

111

Barrett, J. E. (Ed.). (1979). Stress and mental disorders. New York: Raven. Belsky, J. (1984). The determinants of parenting: A process model. Child Development, 55, 83-96. Bleuler, M. (1974). The offspring of schizophrenics. Schizophrenia Bulletin, 8, 93-107. Block, J. (1971). Lives through time. Berkeley, CA: Bancroft Books. Block, J. H., Block, J., & Morrison, A. (1981). Parental agreement-disagreement on child-rearing orientations and gender-related personality correlates in children. Child Development, 52, 965-74. Brehm, S. S., & Brehm, J. W. (1982). Psychological reactance: A theory offreedom and control. New York: Academic Press. Bronfenbrenner, U., & Crouter, A. C. (1983). The evolution of environmental models in developmental research. In P. Mussen (Ed.), Handbook of child psychology: History, theory, and methods (Vol. 1). New York: Wiley. Caplan, G. (1976). The family as support system. In G. Caplan & M. Killiken (Eds.), Support systems and mutual help: MultidiscipUnary explorations. New York: Grune & Stratton. Caspi, A., Elder, G. H., Jr., & Bern, D. J. (1987). Moving against the world: Life course patterns of explosive children. Developmental Psychology, 22, 303-8. Caspi, A., Elder, G. H., Jr., & Bern, D. J. (1988). Moving away from the world: Life course patterns of shy children. Developmental Psychology. Caspi, A., & Elder, G. H., Jr. (1988). Childhood precursors of the life course: Early personality and life disorganization. In E. M. Hetherington, R. M. Lerner, & M. Perlmutter (Eds.), Child development in life-span perspective. New York: Cambridge University Press. Caspi, A., & Elder, G. H., Jr. (1988). Emergent family patterns. The intergenerational construction of problem behavior and problem relationships. In R. A. Hinde & J. Stevenson-Hinde (Eds.), Relationships within families. New York: Oxford University Press. Cavan, R., & Ranck, K. (1939). The family and the Depression. Chicago, IL: University of Chicago Press. Durkheim, E. (1951). Suicide. New York: Free Press. Eichorn, D. H. (1981). Samples and procedures. In D. H. Eichorn, J. A. Clausen, N. Haan, M. P. Honzik, and P. A. Mussen (Eds.), Present and past in middle life. New York: Academic Press. Elder, G. H., Jr. (1973). On linking social structure and personality. American Behavioral Scientist, 16, 785-800. Elder, G. H., Jr. (1974). Children of the Great Depression. Chicago, IL: University of Chicago Press. Elder, G. H., Jr. (1975). Age differentiation and the life course. In Annual Review of Sociology (Vol. 1). Palo Alto, CA: Annual Reviews. Elder, G. H., Jr. (1978a). Approaches to social change and the family. In S. Boocock & J. Demos (Eds.), Turning points (special supplement to the American Journal of Sociology, 84). Chicago, IL: University of Chicago Press. Elder, G. H., Jr. (1978b). Family history and the life course. In T. Harevan (Ed.), Transitions. New York: Academic Press. Elder, G. H., Jr. (1979). Historical change in life patterns and personality. In P. B. Bakes & O. G. Brim, Jr. (Eds.), Life-span development and behavior (Vol. 2). New York: Academic Press. Elder, G. H., Jr. (1984). Families, kin, and the life course: A sociological perspective. In R. D. Parke (Ed.), Review of child development research: The family (Vol. 7). Chicago: University of Chicago Press. Elder, G. H., Jr. (Ed.). (1985). Life course dynamics: Trajectories and transitions, 1968-1980. Ithaca, NY: Cornell University Press. Elder, G. H., Jr. (1986). Military timing and turning points in men's lives. Developmental Psychology, 22, 233-45. Elder, G. H., Jr., & Bailey, S. (1986). The timing of military service in men's lives. In J. Aldous & B. Klein (Eds.), Families and lives. New York: Guilford.

112

Glen H. Elder, Jr., and Avshalom Caspi

Elder, G. H., Jr., Caspi, A., & Downey, G. (1986). Problem behavior and family relations: Life course and intergenerational themes. In A. Sorensen, F. Weinert, & L. Sherrod (Eds.), Human development and the life course: Multidisciplinary perspectives. Hillsdale, NJ: Erlbaum. Elder, G. H., Jr., Caspi, A., & Van Nguyen, T. (1986). Resourceful and vulnerable children: Family influences in hard times. In R. K. Silbereisen, K. Eysferth, & G. Rodinger (Eds.), Development as action in context: Problem behavior and normal youth development. New York: Springer. Elder, G. H., Jr., Liker, J. K., & Cross, C. (1984). Parent-child behavior in the Great Depression: Life course and intergenerational influences. In P. B. Baltes & O. G. Brim, Jr. (Eds.), Lifespan development and behavior (Vol. 6). New York: Academic Press. Elder, G. H., Jr., Liker, J. K., & Jaworski, B. (1984). Hardship in lives: Historical influences from the 1930s to old age in postwar America. In K. McCluskey & H. Reese (Eds.), Life-span developmental psychology: History and cohort effects. New York: Academic Press. Elder, G. H., Jr., & Rockwell, R. (1979). Economic depression and postwar opportunity in men's lives: A study of life patterns and health. In R. G. Simmons (Ed.), Research in community and mental health. Greenwich, CT: JAI Press. Elder, G. H., Jr., Van Nguyen, T., & Caspi, A. (1985). Linking family hardship to children's lives. Child Development, 92, 310-30. Emery, R. E. (1982). Interparental conflict and the children of discord and divorce. Psychological Bulletin, 92, 310-30. Featherman, D., & Sorensen, A. (1984). Social transformation in Norway and change in the life course transition into adulthood. Acta Sociologica, 26, 105-26. Feldman, K. A., & Newcomb, T. M. (1969). The impact of college on students. San Francisco, CA: Jossey-Bass. Garmezy, N. (1981). Children under stress: Perspectives on antecedents and correlates of vulnerability and resistance to psychopathology. In A. I. Rabin, J. Aronoff, A. M. Barclay, & R. A. Zucker (Eds.), Further explorations in personality. New York: Wiley. Glueck, S., & Glueck, E. (1950). Unraveling juvenile delinquency. New York: The Commonwealth Fund. Goode, W. J. (1971). Force and violence in the family. Journal of Marriage and the Family, 33, 624-36. Gottman, J. M. (1979). Marital interaction. New York: Academic Press. Hinde, R. A. (1979). Toward understanding relationships. Ontario/London: Academic Press. James, W. (1890). The principles of psychology. New York: Macmillan. Kagan, J. (1979). Family experience and the child's development. American Psychologist, 34, 88691. Kohn, M., & Schooler, C. (1983). Work and personality: An inquiry into the impact of social stratification. Norwood, NJ: Ablex. Liker, J. K., & Elder, G. H., Jr. (1983). Economic hardship and marital relations in the 1930s. American Sociological Review, 48, 343-59. Lippmann, W. (1914). Drift and mastery. New York: Mitchell Kennedy. Lynd, R. S., & Lynd, H. M. (1937). Middletown in transition: A study on cultural conflicts. New York: Harcourt & Brace. Macfarlane, J. W. (1938). Studies in child guidance. I. Methodology of data collection and organization. Monographs of the Society for Research in Child Development, 3, (Serial No. 6). Macfarlane, J. W., Allen, L., & Honzik, M. P. (1954). A developmental study of the behavioral problems of children between twenty-one months and fourteen years. Berkeley, CA: University of California Press. McCord, J. (1979). Some child-rearing antecedents of criminal behavior in adult men. Journal of Personality and Social Psychology, 37, 1477-86.

Human development and social change

113

McCubbin, H. I., Joy, C. B., Cauble, A. E., Patterson, J. M., & Needle, R. H. (1980). Family stress and coping: A decade review. Journal of Marriage and the Family, 42, 855-971. Merton, R. K. (1968). Social theory and social structure. New York: Oxford University Press. Mills, C. W. (1959). The sociological imagination. New York: Oxford University Press. Moen, P., Kain, E. L., & Elder, G. H., Jr. (1983). Economic conditions and family life: Contemporary and historical perspectives. In R. Nelson & F. Skidmore (Eds.), American families and the economy. Washington, DC: National Academy Press. Parke, R. D., & Deur, J. L. (1972). Schedules of punishment and inhibition of aggression in children. Developmental Psychology, 7, 266-9. Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castallia. Pearlin, L. I., Lieberman, M. A., Menaghan, E. G., & Mullen, J. T. (1981). The stress process. Journal of Health and Social Behavior, 22, 337-56. Quinton, D., Rutter, M., & Liddle, C. (1984). Institutional rearing, parenting difficulties, and marital support. Psychological Medicine, 14, 107-24. Riley, M. W., Johnson, M., & Foner, A. (1972). Aging and society. New York: Russell Sage Foundation. Robins, L. N. (1966). Deviant children grown up. Baltimore, MD: Williams & Wilkins. Robins, L. N. (1978). Sturdy childhood predictors of adult antisocial behavior: Replications from longitudinal studies. Psychological Medicine, 8, 611-22. Robins, L. N. (1984). Longitudinal methods in the study of development. In S. A. Mednick, M. Harway, & K. M. Finello (Eds.), Handbook of longitudinal research: Birth and childhood cohorts (Vol. 1). New York: Praeger. Rutter, M. (1979). Protective factors in children's responses to stress and disadvantage. In M. W. Kent & J. E. Rolf (Eds.), Primary prevention of psychopathology: Social competence in children (Vol. 3). Hanover, NH: University Press of New England. Rutter, M. (1980). Changing youth in a changing society. Cambridge, MA: Harvard University Press. Rutter, M., & Garmezy, N. (1983). Developmental psychopathology. In P. Mussen (Ed.), Handbook of child psychology: Personality and social development (Vol. 4): New York: Wiley. Rutter, M., & Madge, N. (1976). Cycles of disadvantage: A review of research. London: Heinemann. Rutter, M., & Quinton, D. (1984). Long-term follow-up of women institutionalized in childhood: Factors promoting good functioning in adult life. British Journal of Developmental Psychology, 18, 225-34. Ryder, N. (1965). The cohort in the study of social change. American Sociological Review, 30, 84361. Schaffer, R. (1977). Mothering. Cambridge, MA: Harvard University Press. Simmons, R. G., Blythe, D. A., Van Cleave, E. F., & Bush, D. M. (1979). Entry into early adolescence: The impact of school structure, puberty, and early dating on self-esteem. American Sociological Review, 44, 948-67. Thomas, W. I. (1910). Sourcebook for social origins. Boston: Badger. Thomas, W. I., & Znaniecki, F. (1918-20). The Polish peasant in Europe and America (Vols. 1 & 2). Chicago, IL: University of Chicago Press. Volkart, E. H. (1951). Social behavior and personality: Contributions ofW. I. Thomas to theory and social research. New York: Social Science Research Council. Werner, E. E., & Smith, R. S. (1982). Vulnerable but invincible: A study of resilient children. New York: McGraw Hill. Zigler, E., & Child, I. L. (1969). Socialization. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology (Vol. 3). Reading, MA: Addison-Wesley.

Family process: loops, levels, and linkages

Gerald R. Patterson

The relation of family to the culture contains many contradictions. On the one hand, the family's function is to make the next generation ready for socialization by agents outside the family. On the other hand, it is the function of a small number of families to prepare individuals who will function as agents of change. The family must somehow retain its sense of identity while one half of its members prepare to enter and the other half prepare to leave society. This tiny, transitory social system floats in a societal sea that is constantly making contradictory demands for changes and, simultaneously, for constancy and equilibrium. Clinicians and developmentalists alike would agree on the necessity for the study of change. The official imprimatur for such studies has been issued and reissued for decades. It is not a lack of interest or commitment that prevents our working in this area; it is simply that we do not know how to proceed. This chapter outlines some metaphors that seem appropriate to the study of process. Data will be presented at several points to illustrate the utility of thinking about it in this way. As employed here, the term process refers to behavior changing over time along a discernible trajectory. A plot of the changes over time in children's height and weight would be one example of a process. Changes over time in musical skill, or in the skills that maintain acquaintances, friends, or building intimate relationships would be other examples. As illustrated in Figure 5.1, each individual is moving along in multiple processes at any given point in time. We assume that children find themselves in many different processes, but make continuous movement in only a restricted number, and by middle age very few individuals make a commitment to initiate a new process. Commitments to initiating activities such as music, sports, getting an education, or starting a profession would all be examples of socially acceptable processes. Many young people are initiated to such processes, but for most, they do not move very far. For example, graduation from high school completes the education process for many, and few graduates read a book after that time. 114

Family process: loops, levels, and linkages

115

Figure 5.1: That's life

Because of a long-term commitment to clinical interests, I have focused my investigations on what might be called pathologic processes in families. This preoccupation is reflected in the present report by its emphasis on the coercion process; the details of these preoccupations are summarized in Patterson (1982) and Patterson, Reid, and Dishion (in press). Levels Each process implies the presence of some underlying mechanism(s) that determine the continuous changes in behaviors over time. In the maturation and aging process, a number of biological and environmental factors come to mind. However, if we consider processes such as learning a musical skill, getting an education, becoming neurotic, or developing conduct disorders, then the mechanisms are not so self-evident.

116

Gerald R. Patterson

Social processes such as those represented in the development of conduct or character disorders are complex affairs, in which changes take place simultaneously at several levels. There are at least two primary levels that must be examined in order to understand how the process is initiated and why the changes continue over time. Each process must differ in terms of what the relevant levels, or mechanisms, might be, but for the coercion process, it is hypothesized that the initial changes in behavior are molecular (or microsocial) in nature. This would mean that the process starts at the level of dyadic exchanges. The second, accompanying hypothesis is that many of these initial changes occur when one or both members are subjected to high levels of stress. According to the model (Patterson, 1982, 1983), increases in the daily level of stress are associated with commensurate increases in the irritability for social interaction exchanges with other family members. Persons with dispositions to antisocial and depressive behaviors are presumably more vulnerable to these increases. In a sense, dyadic exchanges become a kind of transformer whose function is to transform forces impinging from outside the family into disrupted patterns of social exchange. The third hypothesis is that for the process to move to a second stage, it is necessary for the social patterns altered by the dyadic exchange to generalize to other social settings. For example, the coercive patterns acquired by the child in his or her family interactions generalize to include interactions with his or her peer group. Data from the first cohort of a longitudinal study on juvenile delinquency being conducted at the Oregon Social Learning Center in Eugene were used to test the hypothesis that the effects of dyadic training in the home generalize to the peer group. The findings were consistent with the hypothesis. The correlation between the measure of peer (report) aggression and the construct measuring observed child coercion (start up 4- continuance + synchronicity) in the home was .31 (p < .001). The fourth hypothesis identifies the mechanism that keeps this stage of the process moving. It requires that the newly generalized behaviors elicit some predictable pattern of reactions from a significant constituency in the social environment. For example, the reaction of peers to musical or athletic skills taught at home might have a great deal to do with the child's continuing in these processes. The noncompliance and coerciveness of the child with a conduct disorder produce the same short-term reinforcement at school as at home. In effect, these immediate reactions sustain and perhaps extend the home-based training program. However, the members of the peer group also tend to reject the coercive child in the long run. To test this hypothesis, the child coercive behaviors observed in the home were correlated with the data from the peers describing their relation with the child. The findings were consistent with the hypothesis, with a correlation of — .35 (p < .001). It is the long-run reaction that makes the

Family process: loops, levels, and linkages

117

significant contribution to the maintenance of the problem child in the conduct disorder process. It may also contribute to speeding the process up, and the child becomes increasingly deviant relative to his or her peer group. It seems that data focused on two very different levels are required to understand the coercive process. Data at the molecular level would require observation data collected in sequential form in natural settings. The analyses would identify action-reaction patterns that occur repeatedly enough to be identifiable. The data describing stressors and the reactions of the peer group would be based in large part on more macrosocial measures such as self-reported stress, peer nomination data, and teacher reports. Short-term and long-term effects As a general case, we believe that two kinds of mechanisms are required for a process to go into action. At the molecular level, there needs to be immediate short-term effects that significantly alter the behavior of one or both members. As noted above, a second mechanism could be the reinforcing reactions to coercive behavior at school that would keep the child in the process. It does seem that there are some second mechanisms for the coercion process that actually make the individual more vulnerable to reinforcement provided at home and by the peer group. As noted above, not one but two reactions of the social environment relate to the process. In the new setting in which the short-term coercive and avoidant behaviors pay off, this maintains the process. But the fact that the peer group also rejects the child serves a long-term function. It is hypothesized that long-term reactions such as this function as positive feedback loops. The function of these mechanisms is to maintain and to speed up the process. Over a period of months and years, the effect of peer rejection or isolation would be to create a situation of limited opportunity. According to Youniss (1980), the function of the peer group is to teach the child cooperation with equals, reciprocity in social exchange, and empathy. The adolescent who has failed to acquire these skills will be perceived as increasingly childlike by peers and employers alike. In a culture that demands increasing social skills (work and relational) of its young adults, this individual will become increasingly salient as unskilled. He or she will find him- or herself using coercive and avoidant behaviors at increasing rates. Relatively speaking, he or she is identifiable as being increasingly deviant. In this sense, one might say that the process speeded up. This is also the sense in which one might think of the long-term effects of peer rejection as functioning as a positive feedback loop. The remainder of this report is divided into a discussion of the microsocial and the macrosocial levels.

118

Gerald R. Patterson

Microsocial mechanisms for change It is the social interactional perspective (Cairns, 1979; Gottman, 1981; Lamb, Suomi, & Stephenson, 1979; Patterson & Reid, 1984) that implicates dyadic exchanges as an important mechanism for change in families. In the dyadic exchange, there are literally hundreds of "trials" each day. If one observed the dyad interacting in the home on each of 16 days, it would be possible to calculate the mean daily values for the likelihood of person A reacting in a coercive fashion to person B. Figure 5.2 illustrates what such values might look like. Previous analyses have shown that individual differences are quite stable over a 1-year period. For example, in one study the stability correlation was .74 for a 12-month test/retest interval. The first estimate was based on three sessions and the second on two sessions. We believe that some of the unaccounted-for variance in such estimates is really noise or errors of measurement, but also that there is a significant component that reflects genuine shifts in interactional patterns over time. In our experience, daily plots of coercive exchanges show a good deal of fluctuation. In fact, we believe this ebb and flow of likelihoods is the daily bill of fare for most dyads. The assumption, again, would be that not all of the variance in daily fluctuations is due to errors of measurement, but that some of it is in fact determined. Most shifts in likelihood could not be detected by the participants, but only by trained observers and careful statistical analyses. Figure 5.2 shows that for the first 8 days there was no change in the slope. During that time, all the likelihoods fluctuated about a shared grand mean. What is important for our purposes is the slope described after Day 8. Notice that now the values are shown to fluctuate about a nonzero slope. Given more data points, it would be possible to use a time-series analysis (Gottman, 1981) to demonstrate that this is a significant shift in dyadic exchange. What is interesting about this is that, given observational data collected in sequential form, it is possible to specify precisely the initiation of change in dyadic process. The assumption is that for the target child in a longitudinal study, many of the basic changes leading to prosocial and deviant outcomes begin in the dyadic exchanges with the mother and the siblings. What begins in a simple shift in probability values at the level of dyadic exchange may generalize to a surprisingly wide spectrum of effects, what members of the dyad feel about each other, and the kinds of attributions they make about each other. If one has an empirical anchor point on the question of when the dyad process shifts, one can focus on the question of which variables determine that shift. This is a kind of bootstrapping operation, beginning with precise descriptions of action-reaction sequences occurring in the real world, then moving to a precise statement about when minute changes have occurred.

B.

o

Probability A coercive to B

120

Gerald R. Patterson

For convenience, the examination of recurring dyadic exchange patterns and their accompanying probability values are labeled microsocial analysis. This is also essentially what most writers, such as Cairns (1979) and Lamb et al. (1979), mean by a social interactional approach. It describes the likelihood of a certain reaction by person B given a certain action by person A; it may also include more complex three-step patterns (Patterson, 1982). Rather than a finite pattern such as A-B-C, some formats such as the sequential lag analysis described by Sackett (1978) examine the likelihood of event A occurring at each interval or event from tx through tl6. Given the microsocial anchor point illustrated in Figure 5.2, one is in a good position to examine several different questions. For example, how do these changes at the microsocial level relate to changes in more macrosocial (e.g., trait-like) behaviors? It would also be possible to use the reference point to search for the variables that produce the dysjunction in process. Given the contemporary developments in code systems that accurately describe interacting dyads in sequential form (e.g., Dishion et al., 1983), then it seems reasonable to suppose that we can specify when a process is changing. Given information about baseline levels, we could also specify the magnitude of change over time. However, there is another way of thinking about how much things have changed over time. It is possible that most processes move through recognizable stages, or sequences, in a progression. It is our strong impression that children in the coercion process move through definable stages, not only in the development of symptoms (Patterson & Dawes, 1975) but also in more complex ways as well. The assumption is that there are definable sequences, transitive in form, that characterize both the microsocial and macrosocial levels. This possibility will be examined in a following section. Before that can be explored, however, it is necessary to understand how the coercion process produces significant changes in dyadic exchanges. This material is examined in the next section. How do changes in microsocial exchanges come about? As suggested by Skinner (1969) and others, people can and probably do use positive reinforcers to alter each other's behavior all the time. However, if one looks only at immediate outcomes, the contingent use of aversive stimuli may be a more efficient method for changing behavior. But in the long run, these aversive contingencies have some negative outcomes that are extremely disruptive to family process. It is a sad fact that many adults in our society choose to maximize short-term gains by employing aversive contingencies at very high rates, to the detriment of those living with them. In the long run, the contingencies work against the person who employs them, but the fact that the payoffs are

Family process: loops, levels, and linkages

121

immediate, while the negative outcomes are lagged, creates a reinforcement trap for the unskilled and unwary. The details of the coercion mechanism emerged slowly, beginning with our first sequential analysis of family interaction data (Patterson & Cobb, 1971). The data showed that aversive consequences suppressed the child's ongoing prosocial behaviors, but increased the likelihood that the child would repeat prior coercive behavior! This was a surprising finding. We decided to begin a systematic search for the antecedents and the consequences that controlled children's ongoing coercive behaviors (Patterson, 1977; Patterson & Cobb, 1973). After a decade of effort, it finally became clear that in addition to rewards and punishments, there was a very important third contingent arrangement that also altered the performance of social behaviors. This arrangement was also the key to understanding coercion in families. In the operant literature, this third arrangement is called negative reinforcement (Hineline, 1977). Generally, it is a variant of escape-avoidant conditioning defined by the sequence: aversive stimulus - response - termination of aversive stimulus. Because many investigators are confused by the term negative reinforcement (it is not punishment), we use the rather awkward phrase escapeavoidant conditioning as the appropriate label. The experimental findings and field study data related to this mechanism are spelled out in detail in Patterson (1982, ch. 7). Escape-avoidant conditioning and punishment contingencies are at the heart of the coercion mechanism, but their long-range effects tend to slip by unnoticed, lost in the flood of ongoing social interaction. In keeping with the speculations by Langer (1984) and the experimental studies reviewed by Nisbett and Wilson (1977), we believe that most of the minutiae of changes in probability during social exchanges occur outside the awareness of the individual. Each of us has better things to do with our limited channel capacity than to try to focus on whether this reinforcer or punisher occurred and what the effect was on our behavior. As shown in the experimental tests of the reinforcement change without awareness hypothesis, it is a relatively simple affair to arrange for an individual to be attending to one set of stimuli while his or her ongoing behavior is altered by some other set of contingencies s/he is not tracking (Gewirtz & Boyd, 1977; Rosenfeld & Gunnel, 1985). One of the characteristics of dyadic interchanges that makes them so difficult to track and store lies in the multiplicity of effects, introduced by a single aversive event. One of the most fascinating features of social interaction is the possibility of simultaneous short- and long-term outcomes for a single event. After observing it in family after family being treated because of an antisocial child, we came to label the simultaneous outcomes as the reinforcement trap. From the situation illustrated in Figure 5.3, one may obtain five different

122

Gerald R. Patterson Mother

Mother

Mother

less

scolds

talks

t0

Child whines

Da

V

like| sc

If child whines, y

mother more likely

<> ld

Child stops whining

1

t0

Jalk

Child more likely to whine Day 2

Figure 5.3: The reinforcement trap (from Patterson, 1982)

outcomes from a single aversive exchange. The situation there tells a familiar story. There have been many previous trials in which the mother nagged the child about a messy room. The scene opens with mother scolding. As previously rehearsed, the child begins to whine. For the sake of illustration, assume that this time the child's whining effectively terminates mother's scolding. She now talks to the child in a soothing manner. Thus, the child's whining produced mother's talking. In the example, the mother and child both behave in such a way as to maximize short-term gains. The child behaves in such a way as to maximize the likelihood that the mother will terminate her scolding. The mother behaves in such a way as to maximize the likelihood the child will immediately stop whining. This second outcome is in keeping with findings from the laboratory analogue study by Dengerink, Schnedler, and Covey (1978). In a Buss-type shock situation, when the other member terminated attacks, subjects quickly reduced their level of aggression. In a family setting, does the coercer terminate the attack within seconds of victim compliance? In a study of three mother-child pairs, Woo (1978) found that during baseline the mean duration of the child's whining was 28 seconds. Given an aversive stimulus followed by the child whining and mother complying, the child stopped whining in an average of 4.6 seconds! What makes this interesting is that both participants failed to see that for both of them there were long-term effects that would, in turn, produce increases in future misery. The mother did something that stopped the child's whining, but in so doing, she increased the likelihood of the child whining in the future. This is an operational definition of what is meant by the reinforcement trap. The child has trained the mother to stop trying to teach social skills; in the long run, this will cost him dearly. This lack of agreement between short-term and long-term effects makes it extremely difficult for the participants to understand what has happened. It is, in other words, an ideal situation for behavior change without awareness.

Family process: loops, levels, and linkages

123

In the short term, both members reduced the number of average events impinging on them. It should be noted, however, that the child's room remained a mess. The mother lost because she opted for an immediate reduction in her "pain" schedule, not knowing, of course, that it was at the expense of an increased likelihood that messy rooms will occur in the future. The findings from the laboratory and field studies summarized in Patterson (1982) are all consistent with these ideas. More recent studies by Snyder and Patterson (1986) showed that when mother submitted to coercive child behaviors, it increased the probability that the child would behave coercively in future interactions. The coercion mechanism has recently been used to investigate dyadic process in a wide range of pathologies. For example, Biglan et al. (1984) showed that mothers' depressed behaviors may serve the useful function of suppressing spouse aggressive behaviors. Gottman and Levenson (1986) have begun the analyses of the function of spouse affect during conflict, which suggest that what works in reducing conflict may become strengthened over time. Microsocial structure A network of ideas is involved in this next problem. The first part concerns the possibility that there are changes in the structure of social interaction accompanying reinforcing contingencies. This implies that there are not one, but two, outcomes of programmatic reinforcement: (1) increases in the likelihood of occurrence for the reinforced responses, and (2) regular structural changes that accompany these programs, a kind of bonus as it were. These structures are transitive in nature. In a perfectly transitive progression, all individuals classed in an advanced stage would have passed through each of the prior stages; however, not all individuals in a given stage move on to the next one. The second assumption about these structures is that the longer the child has been involved in the process, the further he or she has moved into the progression. The third assumption is that the further into the progression, or the more advanced the stage, the more likely the training is to generalize to macrosocial levels (including traits and settings). The microsocial process starts after there has been an increase in the rate of coercive behaviors for one or both members of the dyad. When the rate of coercive responses is well above the mean for a period of several weeks, then certain structural changes in the dyadic exchanges begin to emerge. We assume that the startup variable is the first to show an increase. Startup means that even though the other member of the dyad is in a positive or neutral state, the child is likely to initiate an aversive behavior. Figure 5.4 summarizes the data from a sample of the first 44 families in Cohort 1. It can be seen that 72% of the boys did this one or more times with the siblings during the three observation sessions.

124

Gerald R. Patterson Transitional probabilities

Alternate path probability values Figure 5.4: Microsocial progression: target child and sibling exchanges

Given that the child engaged in one or more startup behaviors, the likelihood that he would move on to the next stage in the structure was .59. The comparable conditional for boys' reactions to mothers was .64. This next stage was defined as synchronistic negative exchange. Given the sibling reacted to him in an aversive fashion, then the target subject reacted in kind. Dyads who display frequent negative synchronistic reactions have made considerable progress into the coercion process. For example, Levenson and Gottman (1983) found that couples with high negative synchronicity in a laboratory task were extremely likely to be dissatisfied with their marriage. As might be expected, boys who had high negative synchronistic scores were also described as highly antisocial by parents (Patterson, 1982), and in another study by teachers, peers, and parents (Patterson et al., in press). Given that the child engages in one or more synchronistic reactions, the likelihood that he will move on to the next stage was .79; the comparable value for his reactions to mothers was .64. This next stage, continuance, was defined as the likelihood that the child would follow one coercive response with another (regardless of what the other person did). The significance of this stage lies in the

Family process: loops, levels, and linkages

125

fact that longer-duration coercive exchanges set the stage for increases in the amplitude of responses. It is during the extended exchanges that one or the other member is likely to try a higher-amplitude response, and when s/he does, the odds are very good that the other will withdraw. In keeping with this formulation, the analyses by Reid, Taplin, and Lorber (1981) showed that hitting was more likely to occur during coercive chains longer than 18 s. This held for both normal and clinical samples, but as might be expected, the family members in the clinical samples were more likely to engage in the longer chains. Given that the child engages in one or more continuance reactions with siblings, then the likelihood of engaging in physical aggression with some family member (most likely siblings) was .55. The finding provides support for the earlier analyses of data from clinical samples by Reid et al. (1981). The last bit of information contained in Figure 5.4 describes the likelihood that a child arrived at a given state via some path other than the progression described there. Notice that the odds were generally less than one in five that if a child engaged in that behavior, he did not also engage in the prior behavior in the progression. The data presented in Figure 5.4 are consistent with the notion of a microsocial structure that accompanies family-based training in coercion. What might the implications of this be? It is hypothesized that there is a parallel progression for antisocial traits such that the further one progresses in the microsocial progression, the further one advances in the trait progression. Presumably, it is simply the amount of time in the microsocial process that determines how far into the trait progression one moves. This possibility will be examined in the following section. Traits define progressions Antisocial process is characterized by simultaneous movement along both microsocial structural progressions and a corresponding movement along an antisocial trait progression. Both are transitive in nature. Best of all for our purposes is that both are eminently measurable. We believe traits to be relatively stable across time and across settings. We have expanded the meaning of the term slightly to have it serve as a referrent for the menu of settings likely to be selected by the child. For example, it would be expected that the aggressive child would more likely choose unsupervised activities rather than supervised ones, and excitement-provoking settings rather than intellectual ones. In the present context, a trait should include in its definition the general reactions the (trait) behavior elicits from the social environments in which they are practiced. For example, the careful studies of peer interactions by Coie and

126

Gerald R. Patterson

his colleagues attest to the consistency of peer reactions to coercive behavior (Coie & Kupersmidt, 1983). This is also in keeping with our own studies noted earlier, where child coercion in the home correlated .34 with peer rejection at school. The other characteristic to be included in defining a trait would be a specification of its position in the transitive sequence. The trait progression reflects the fact that there is an underlying shared process determining all the traits in the sequence. One of the problems in using progression is that, as yet, there is no effective means for deciding when a progression exists, or even for deciding where one should look. The technique proposed by Guttman (1944) initiated a great deal of research activity, and that activity attests to the fact that he posed the right question. But as Robinson (1973) and many others have pointed out, the statistical basis he used for deciding on the existence or nonexistence of a progression proved to be an absolute disaster. As we shall see later in the discussion, if one can live with the idea that there is no single index for deciding about an entire progression, then it might be said that a simple application of conditional probability analyses is a serviceable way of describing each juncture in the progression, as well as the likelihood of alternative routes to the same stage. The data used to illustrate the progression in Figure 5.5 were taken from the first 50 families in Cohort 1 from the longitudinal study. Both the mother and the father filled out the Oregon Child Aggression questionnaire when the boys were 10 years old. If one or both parents indicated that noncompliance, temper tantrums, or fighting occurred at least once a week, it was counted as evidence that the child had progressed that far. The child was also included in that stage if one parent claimed once a week but the other once a month. The child was placed in the stealer stage if both parents agreed that he stole at least once per month or more. Five of the children had been involved in one or more court offenses at the time of the study or during the ensuing 12 months following the study. The base rates for the sample of fourth-grade boys are shown in the lower line of the box for each symptom. The parents describe surprisingly frequent occurrences of noncompliance and temper tantrums for boys of this age (a temper tantrum per week!). We were also surprised to see that 33% of the boys stole at least once a month. The transitional probability values describe the likelihood that, given the child was in one stage, he had also been described in the next stage in that progression. For example, two thirds of the boys described as noncompliant had also been described as having temper tantrums. Notice that the majority of children with temper tantrums did not move on to fighting. A substantial number of children who were fighters were also stealers, and about one third of the children who were stealers had court offenses.

Family process: loops, levels, and linkages

127

Transitional probability values

Alternate path probability values

Figure 5.5: Microsocial progression

The third bit of information required to understand progression is the probability value defining the likelihood of an alternative path. For example the p value for fighting describes the fact that 82% of the boys who were fighters were also described as having temper tantrums. If one believes these particular data, then 18% arrived at the fighting stage from some other path. This probability value is of particular interest when it describes the alternative path for stealing. The general model stipulates that the majority of the cases referred for treatment would follow the path described in Figure 5.5 (noncomply - fight - steal) (Patterson et al., in press). In that model, about one quarter of the clinical referrals are thought to have passed along a sneaky aggression path that does not include fighting, and where the children have an acceptable level of social skills. The present findings reaffirm the need to study that alternative path because the majority of the stealers in this sample do not seem to fit the coercion model. The alternative model described in Patterson (1982) was based on the idea of a distancing parent who could cope with discipline confrontations but was

128

Gerald R. Patterson

Table 5.1. Relating microsocial to trait progression Trait sequence Fight {.22)a

Microsocial progression p(X/ Startup) p{XI Startup + Synch) p{XI Startup + Cont) p(X/Startup + Synch + Cont + Phys Agg)

Steal (.33)

Mom

Sibling

Mom

Sibling

.22 .32 .44

.34 .33 .38

.38 .63 .47

.45 .50 .36

.60

.56

.40

.33

Court offense 11) Mom

Sibling

.40

.40

° Figures in parentheses indicate baserate.

not involved enough to effectively monitor the child. According to the earlier study, the trait progression defining this path seemed to be noncomply - lie steal - set fires. Integrating microsocial and trait progressions It was implied earlier that in some sense the microsocial and the trait progressions were tied together. Other things being equal, the longer the time spent in the coercion process, the further info the microstructure the child would progress. Presumably, the further the child progresses at this level, the greater the risk for movement through the trait progression. The data from the first 44 families in Cohort 1 were analyzed to form a preliminary basis for examining the hypothesis. Prior studies suggest that the key microsocial exchanges probably take place between the target child and both siblings and mother. For this reason, the two dyadic exchanges were analyzed separately. Each stage in the microsocial progression was analyzed separately to determine its relation to each stage in the trait progression. The findings are summarized in Table 5.1. The baserate value for each trait is given at the head of each column. For the analysis to be of interest, the conditional probability value should exceed the baserate by a good deal. Because this analysis is just illustrative, no effort was made to make a statistical comparison of the two values. It became clear very quickly in this example that the microsocial progression provided no basis for predicting transition to the noncomply and temper tantrum stages in the trait progression. The baserate values for these two behaviors were

Family process: loops, levels, and linkages

129

simply set too high; they also exceeded the conditional probability values from the microsocial progression. Evidently some artistry is required to systematically test the relation between two progressions. Note, however, the steady increment in prediction to the fight stage for both mother and sibling exchanges. Given that one knows the child used one or more startups does not provide a solid basis for predicting fighters, but the fact that the child has moved from startup to synchronicity to continuance seems predictive. If one knows the child uses physical aggression during the observation sessions, it provides a conditional that more than doubles the baserate values. We simply do not know what to make of the next column describing the situation for stealing. Knowing the child used startup plus synchronistic reactions seems to be a useful basis for predicting stealing; moving further into the coercion process seems to function as a suppressor! It seems to be an iteration of the earlier theme that there is a second path leading to stealing, and perhaps to court offenses as well. Given the small sample size, no effort was made to even calculate the various conditionals for the court offense column. The last one in the column, however, was simply too suggestive to omit. It seems plausible to assume that there might well be an empirical linkage between the two levels. The pilot data are consistent with the idea that the variable that connects the two is really time in process. If the dyad has reached the high-amplitude stage in the microsocial progression, the child would be more likely to have moved far into the trait progression, for example, to fighting. Stages The idea of process can be approached in several ways. The least interesting would be to think of a single behavior increasing in rate over time. Most clinical processes consist of much more than an increase in rate of deviant behavior over time. The hundreds of hours of clinical histories presented by the families treated at Oregon Social Learning Center suggest a sequence of stages. The idea is that arrival at each new stage places the child at even greater risk to continue on toward a career of antisocial behavior. The sequence of stages that characterizes the clinical histories is summarized in Figure 5.6. As shown there, the first stage consists of the microsocial training of a target child in the home by family members. They teach him to use aversive behaviors, to avoid responsibility, and above all else to be noncompliant. These behaviors generalize to the school setting. The reactions of the persons in that setting define Stage 2. The coercive interpersonal style quickly leads to rejection by the normal peer group. In this sense, the trait of coercion might include as part of its definition the social environmental reaction, "peer rejec-

Gerald R. Patterson

130

Stage 5

Stage 4

Stage 3

Stage 2

Stage 1 Figure 5.6: Stages model (from Patterson, Reid, & Dishion, in press)

tion." As noted in Figure 5.6, the reaction of the social environment should also specify that the parents tend to reject the child, and for the same reasons that peers do. The correlation between parental and peer rejection was .43 (p < .001). The obdurate noncompliance includes noncompliance to classroom

Family process: loops, levels, and linkages

131

setting rules, homework assignments, and responses to teacher instructions and requests. This, in turn, places the child at grave risk of academic failure. As shown in Figure 5.6, the effects of rejection by parents is thought to lead to the development of poor self-esteem. Rejection by peers and academic failure are thought to lead to the development of depressed mood. These are outcomes or effects produced by what had previously happened at Stage 2. The assumption is that these outcomes, as well as those at Stage 2, would function as positive feedback loops that serve the function of keeping these children in the process. The boys have been in the process for some time. They are both antisocial and socially incompetent. They have failed at almost everything important that they have attempted. They feel they are not good people, and find themselves increasingly caught up in dysphoric mood swings. Should they continue in the process, they will become at risk for membership in the cadre of likefellows that comprise the deviant peer group. This support group will facilitate their entry into substance abuse and delinquency. The sequence model is designed to describe what the data should look like when generated in a well-designed longitudinal study. The data have been collected for the first two waves for two cohorts of 100 families each. The families are living in the high-crime areas of Eugene, Oregon; they began their third wave of data in the fall of 1985. In the following sections, the relations among variables that should characterize each of the first three stages will be examined by using data from the first wave. Each of the stages in Figure 5.6 defines which constructs should covary at any point in time. If the correlational structure does not fit, then the models are wrong and would not be suitable for the longitudinal analyses that will follow. The findings relevant to each of the first two stages are briefly examined as illustrations. Stage one: micro social and macro social together It would seem useful at this point to attempt to put the microsocial and the macrosocial levels together. The structural model presented here examines the relation between two microsocial variables defining the Parental Inept Discipline and Child Coercion constructs on the one hand and an across-setting measure of Antisocial Behavior on the other. It is assumed that outside stressors or a lack of prior training in family management skills can lead to a situation in which the microsocial exchange between parent and child becomes badly disrupted. The relation between outside stressors (employment, illness, daily hassles) and parenting skills has been explored in a series of modeling studies reported in Patterson (1986a), Forgatch, Patterson, and Skinner (in press), and Viken (1986). We are also interested in

132

Gerald R. Patterson

exploring the Elder hypothesis that the effects of stress on child discipline confrontations may be mediated by the father's general disposition to be irritable (Elder, Van Nguyen, & Caspi, 1985). The effort to replicate the Elder mediational model for stress was pursued in a study of single-parent families (Forgatch et al., in press). The analyses of these data showed clear support for the Elder hypothesis. For single mothers, the effect of stress on discipline practices was mediated by their general disposition to be irritable. The Stage 1 model, basic training for antisocial behavior, requires a demonstration that antisocial behavior occurring in the home and the school is significantly related to coercive exchanges between parent and child. In Figure 5.7, the parent contribution to the microsocial exchange is to be found in two of the four indicators that define the Inept Discipline construct (likelihood of nattering, likelihood of explosive discipline). The child's contribution is defined by the Child Coercion construct (likelihood of startup and duration of coercive chains). Note that the relation between the parent and child constructs is assumed to be bidirectional (e.g., as the child becomes more coercive, he is even more difficult to discipline). There are major problems associated with the estimation of bidirectional effects. In that we are still struggling with this problem, the precise values for these paths remain in doubt. In the model in Figure 5.7, multiple indicators were used to define each of the three latent constructs. The data from Cohort 1 served as the derivation sample, and Cohort 2 as the replication sample. The findings are summarized in Figure 5.7. The numbers on the diagonals connecting the constructs to the indicators refer to the factor loadings. Notice the substantial stability in loadings across cohorts for 10 of the 11 indicators (mother reported discipline will be dropped). According to the clinical model, inept discipline sets into motion the microsocial process whereby siblings and mothers, and to a lesser extent fathers, train the antisocial child in the basic antisocial skills. This, in turn, makes it more difficult to discipline him or her. All the path coefficients for the bidirectional effect were found to be significant for both cohorts. The findings are therefore consistent with the clinical model. In the theory-driven model, the malleable determinant for antisocial behavior is thought to be parental discipline. In the treatment procedures, training parents to use nonphysical forms of punishment effectively reduces the level of antisocial behavior. The model requires a strong path coefficient connecting the Antisocial Behavior and Inept Discipline constructs. As shown in Figure 5.7, the path with the largest coefficient in the entire model is the one connecting these two constructs. Two chi-square tests showed a satisfactory fit of the data to the overall model. For both cohorts, the model accounted for about 40% of the variance in the criterion measure of antisocial behavior. This more than meets our requirements

EXPLOSIVE DISCIPLINE

OBSERVED STARTUP TO PARENTS

OBSERVED PARENT NATTERING

OBSERVED STARTUP TO SIBLINGS OBSERVED DURATION OF CHILD COERCION

OBSERVER IMPRESSIONS MOTHER REPORT DISCIPLINE STRATEGIES

PARENT REPORT

ANTISOCIAL BEHAVIOR *p<.05

V.60

Figure 5.7: Basic training for antisocial behavior (from Patterson, Reid, & Dishion, in press)

CHILD TELEPHONE

134

Gerald R. Patterson

for an adequate performance model and suggests a nice fit between what we expected on the basis of our field and clinical studies. Stage two: reactions of social environment to the symptoms The clinical model at Stage 2 stipulated that the parents and peer group would react negatively to the increasing rates of coercive child behavior. Child noncompliance to adult requests and setting rules would result in lower academic achievement. For one thing, noncompliance would relate to lower completion rates for homework assignments: studies demonstrate a consistent relation between homework completion and academic achievement (Knorr, 1981). An additional set of findings demonstrates a relation between antisocial behavior and low homework completion (Dishion, Stouthamer-Loeber, & Patterson, 1984; Patterson et al., in press). In addition, the coercive child spends less time on task in the classroom itself, and this is in turn related to low academic achievement, as shown in the studies by Cobb (1972), Cobb and Hops (1973), and Hops and Cobb (1974). The analyses of data from a planning study for the present OSLC longitudinal study of juvenile delinquency strongly supported the relation between the daily round of discipline confrontations and conflicts with siblings on the one hand and maternal rejection on the other (Patterson, 1986b). The multiple correlation coefficient was .69 (p < .01). Additional analyses of the longitudinal samples show comparable correlations, but none of the variables have been employed, as yet, in multivariate analyses. The report by Patterson (1986a) presented the findings from Cohort 1 for a structural modeling that included parts of Stages 2 and 3. The findings are presented in Figure 5.8. Stage 2 in the clinical model stipulates that the coefficients should be significant for the paths from the Child Antisocial Behavior construct to the Peer Acceptance and Academic Performance constructs. This was shown to be the case, and the findings were replicated by the data in Cohort 2 presented in Patterson et al. (in press). For both Peer Acceptance and Academic Performance, the single Child Antisocial Behavior construct accounted for 20% to 50% of the variance. The findings are consistent with the Stage 2 hypothesis. Antisocial child behavior seems to be related to very negative reactions by parents and peers alike, and with academic failure. The studies of Stages 3 and 4 are still under way, but in all cases the findings are consistent with the stages hypotheses (Patterson & Capaldi, in press). The replicated modeling studies showed that it was the lack of support by the parents (low positive reinforcement, low involvement) that was associated with the child's development of low self-esteem. The findings are in direct support of the developmental hypotheses outlined by Harter (1983). The modeling studies also

PARENT QUESTIONNAIRE

PARENT CBC

TEACHER CBC

TEACHER CBC

PEER NOMINATION

PEER NOMINATION

CHILD TELEPHONE INTERVIEW

PARENT QUESTIONNAIRE

CHILD SKILL CHECKLIST

CHILD QUESTIONNAIRE

CHILD PERCEPTIONS QUESTIONNAIRE

CHILD TELEPHONE INTERVIEW

CHILD INTERVIEW

PARENT CBC

.83 ^

STANDARD TEST SCORE

<6p WRAT SCORE Figure 5.8: Stage 2: the reactions of others (from Patterson, 1986)

136

Gerald R. Patterson

showed consistent relations between peer relations, academic failure, and the development of depressed moods. For both sets of models, 40% to 70% of the variance was accounted for in the criterion measures. The findings from the static analyses are consistent with the notion of stages. If the modeling studies were not successful, then it would not be possible to proceed to the next step in the studies. Starting in the fall, we will use the stage models to study process. The following section identifies some of the problems involved in testing a stage model in longitudinal studies. Problems in testing stage models: longitudinal designs It was implied in the stage model that some effects were relatively immediate whereas others might have a delayed effect on process. The focus for the present section is to delineate these two contributions to process. An example of a determinant that has a relatively immediate impact on child behavior would be the Stage 1 model described earlier. According to clinical experience, changes in some parental discipline practices would be followed by significant changes in child behavior in a matter of a few days or weeks. In keeping with the excellent discussion of longitudinal study designs by Dwyer (1983), we have labeled these short-term effects synchronistic. The stage model implies the idea that some changes in the family might have a delayed effect on child behavior. For example, the Stage 2 effect of being rejected by members of the normal peer groups could serve to increase the child's risk for remaining in the coercion process over a long period of time. In this discussion, those effects that unfold slowly over a period of months or years are labeled lagged effects. Inasmuch as this latter set of variables are also produced by an earlier stage of the same process that they in turn affect, they can be thought of as lagged feedback loops. It seems reasonable that a study of change in families should begin with the simplest phenomena and then proceed to the more complex. In our mind, the study of effects that are relatively immediate would be the easiest. For the preadolescent, family management skills of the parent are hypothesized to be the prime determinant for most deviant child behavior, and many prosocial behaviors as well. Shortterm changes in parenting practices produce relatively immediate effects on child behavior. Specifically, it is hypothesized that disruptions in family management practices, particularly in discipline, can set microsocial processes in motion. As already noted for preadolescent boys, the particular parent skill central in producing deviant behavior is discipline practices. For the preadolescent boys, the model will be expanded to include parental monitoring (Patterson & Stouthamer-Loeber, 1984). Parental use of positive reinforcement, parent involve-

Family process: loops, levels, and linkages

137

ment, and skills in problem-solving situations are critical to the development of prosocial skills (Patterson, 1982). Two decades of clinical experience in training parents of antisocial children indicate that these skills are critical. Training the parents to use these common sense practices is accompanied by significant reductions in both observed and telephone-reported incidences of deviant behavior for both the problem child and siblings (Arnold, Levine, & Patterson, 1975; Patterson, 1985). Twelve-month follow-up observation data collected in the home show that the treatment effects persist (Patterson & Fleischman, 1979). A review of outcome studies involving random assignment comparisons and preand postobservation data showed significant changes for the parent training groups, but not for waiting-list controls, leaderless parent discussion, or community agency treatment groups. More recently, Marlowe, Reid, Patterson, and Weinrott (1986) reported a random assignment comparison design for families of adolescent chronic delinquents. The parent training group showed significantly fewer police offenses for the first year of the 3-year follow-up, and significantly less time in institutions for all 3 years. The hypothesis generated by clinical experience is that the disruptions in family management skills are the prime determinants for deviant and prosocial child behaviors. To test this, it was necessary to spend several years developing tailor-made assessment methods for each of the constructs involved in parent skills and prosocial and deviant child behaviors. The general model is defined by a dozen more key constructs described in Patterson et al. (in press). Multiple indicators Even with the general model solidly grounded in clinical experience, how one might set about measuring the key ideas was still unclear. Most investigators have used questionnaire or interview data obtained from a single agent as a means of evaluating both the parenting practices and the deviant status of the child. For example, Schuck (1974) pointed out (in his reanalysis) that the interview study by Sears, Maccoby, and Levin (1957) showed a significant correlation of .23 between parental permissiveness and child aggression. However, this finding did not replicate in the study by Yarrow, Campbell, and Burton (1968). An assumption in the present report is that each method of assessment carries its own disturbance, noise, or bias in measuring a phenomenon such as monitoring, discipline, or antisocial behavior. These disturbances, and our means for addressing them, are discussed in detail in Patterson and Bank (1985, in press). Generally speaking, the use of multiple indicators in measuring all variables in a model should provide more generalizable scores. This emphasis is in keeping with the logic of structural equation modeling (Bentler, 1980; Dwyer 1983). We assume that such scores should stand up better in replication studies.

138

Gerald R. Patterson

There are six interlocking models that define the general model. Thus far, three models have been replicated (Patterson et al., in press). Although this is consistent with the idea that multiple indicators provide more generalizable scores, of course, equally plausible reasons might be given for the preliminary success in replication. The assessment battery was tailored to fit the major constructs in the model and tested in a planning study based on 200 normal families. The first sample for the multilevel assessment approach included boys in 4th, 7th, and 10th grades. The battery included interviews, observations, telephone reports, questionnaires, laboratory tasks, and diaries. An effort was made to define each concept by data from two or more agents (mother, father, teacher, peer, self, observer, interviewer). The constructs built with data from this planning sample were then used to differentiate delinquents (Patterson & Dishion, 1986; Patterson & StouthamerLoeber, 1984), fighters (Patterson, Dishion, & Bank, 1984), and substance abusers (Dishion & Loeber, 1985). Following this first wave of construction and analysis, an effort was made to improve the assessment procedures for all key constructs in the model. Two cohorts, each consisting of 100 families of fourth-grade boys living in high-crime areas, were assessed with the revised battery. The structural equation-modeling format was used to examine the contribution of disruptions in parental discipline and monitoring to antisocial behavior of the child (Patterson et al., in press). Each construct was defined by three or more indicators. As already noted, the basic training (home) model accounted for substantial amounts of variance in the criterion measure of antisocial behavior (defined by peer, teacher, parent, and child reports) for both cohorts. No matter how elegantly modeled, correlations cannot tell us very much about the status of variables as determinants. One problem to be addressed here concerns the means by which we can ascribe causal status to key constructs in the model. With this in mind, we have designed and are currently carrying out two studies. One consists of a randomized longitudinal experiment and the other a passive longitudinal design. We will briefly review each design because we see these methodological considerations as being key issues in the study of change in families. The second problem to be considered concerns the means by which one evaluates the long-term impact of Stage 2 or Stage 3 status. How are we to get a grasp of concepts such as positive feedback loops and subject them to empirical test? We will consider some of the available techniques in this next section. Causal status As noted by Dwyer (1983), the traditional passive longitudinal design has little utility in testing for causal status of, for example, a Family Management con-

Family process: loops, levels, and linkages

139

struct. We could say that if the path from Family Management to Child Behavior were not significant at tx, then it is unlikely that Family Management has causal status. But given that the correlations were significant between Family Management at tx and Child Behavior at tx and again at t2, it would still be an unconvincing argument for causal status. This argument by Dwyer (1983) will be examined in detail in the following section. Assume that Child Behavior was tested each year on three different occasions, while Family Management was assessed only at Year 1 and Year 3. As shown in Figure 5.9a, the effect of Family Management is specified as a direct effect at tx and at t3. Reviews of empirical findings from longitudinal studies of antisocial children (Loeber & Dishion, 1983) indicate that the autocorrelations of Child Behavior at tx with t2 and t2 with t3 should be significant. Similarly, the autocorrelation of Family Management measured at tx with the same construct measured at t2 will be significant. We believe that the explanation for the apparent stability of Child Behavior is that behaviors are driven by an equally stable set of parenting practices. In effect, both the Child Behavior measures and the Family Management measures will be about equally stable. The effect of Family Management on Child Behavior is expressed as a synchronistic effect at tx and at t3. Within the structural modeling format, the structural relations among the constructs are expressed as path coefficients. The fact that they are expressed in standardized form makes them even more useful because it is possible to compare the magnitude of one path relative to another. In the present context, it would be required, then, that the path coefficient describing the synchronous effect of Family Management on Child Behavior at t3 will account for some variance normally accounted for by the path coefficient describing the stability of the Child Behavior. The reason for this assumption is that this is a causal variable that is, itself, stable over time. Its contribution at t3 reflects this prior contribution. A recent test of this model with longitudinal data showed that this effect was obtained (Patterson & Bank, in press b). Given that the synchronistic model was accepted, there is a second means for demonstrating that it could not have causal status. As shown in Figure 5.9(b), a third unmeasured variable, Z, is introduced as a determinant for both Child Behavior and Family Management, presumably present at Wave 1 and Wave 3. In Dwyer's (1983) language, this would be a synchronous common-factors model. If the model were accepted, it would mean that the investigators should stop to consider the possibility that there is some more general factor determining both the child's behavior and parenting practices (e.g., something like genetic similarities between parent and child in activity or intelligence status) than with describing the important developmental sequences unfolding for the family. Both Rogosa (1980) and Dwyer (1983) attribute the essential idea for the testing of spuriousness models to an article by Kenny (1975). The spuriousness model is viewed by Dwyer as a weak correlational analogue for an experimental

140

Gerald R. Patterson

(a) Synchronistic effect model

(b) Spurious correlation model

(c) Cross-lagged model

Figure 5.9: Passive longitudinal designs

manipulation. Incidentally, he would also propose to carry out this analysis prior to testing the substantive models, the reason being that if the spurious variable model were accepted, it would seriously weaken the conclusions to be drawn from any further modeling for that data set.

Family process: loops, levels, and linkages

141

Dwyer (1983) makes the important point that there may be more than one common spurious variable. Therefore, the fact that one is rejected does not mean there are not others. His conclusion, and our own as well, is that correlational analyses are simply preludes to the experimental manipulations that must inevitably be carried out. The present analysis is thought of as a necessary prelude. It is an exploratory correlational analysis, identifying those components that should be tested further in the manipulation studies that must follow. Randomized longitudinal designs. As noted earlier, surviving a modeling for synchronistic effects at tx and at t3 is a necessary but not sufficient basis for confidence in the causal status of the Family Management construct. This status can only be established by placing the construct at risk in a series of experimental manipulations. It is hypothesized that using random assignment for treatment conditions that have previously been demonstrated to be effective or ineffective could provide an appropriate test. As a research strategy, we envision a series of treatment manipulations beginning with the group design outlined below. The variables that survive this will undergo a series of single-subject designs (Sidman, 1960), where a single variable can be manipulated with great precision and tested with modern time-series analytic formats (Gottman, 1981). In a randomized longitudinal experiment design currently under way at this center, 70 families referred for treatment were randomly assigned to a community agency and to an experimental group in which they receive parent training. Families in both samples participate in the entire assessment battery prior to, at termination, and following treatment. Findings from an earlier comparison study using random assignment (Patterson, Chamberlain, & Reid, 1982) led me to assume that only the parent training approach will effectively alter antisocial behavior. The hypothesis to be tested is that the change is brought about because of increased effectiveness in parent discipline practices. In keeping with the suggestions by Dwyer (1983), the design is the familiar synchronistic plus autocorrelated effects model described in Figure 5.9a. Now there is one additional path: It describes the effect of participating in treatment. This can be thought of as a categorical variable (e.g., the family either participated in the social learning parent training procedures or it did not). In that all the therapy tapes are coded, the amount of social learning training can also be specified as a continuous variable. In either case, the assumption is that the path from the Treatment construct to Family Management at termination will be significant. In this model, the baseline to termination stability values will tend to be low and perhaps nonsignificant, and those from termination to follow-up will be stable. The key idea is that the effect of Treatment on Child Behavior will be completely mediated by its effect on Family Management.

142

Gerald R. Patterson

If the model survives the passive and experimental longitudinal studies, a new set of proposals will be submitted that precisely manipulates determining variables and tests for the spread of effects specified in the general model. The programmatic studies are designed to provide an empirical base for concluding that variables identified by clinical practices as determinants do, indeed, fulfill these functions.

Lagged effects: a case for feedback loops The hypothesis is that some variables may have an effect on child behavior that is lagged over a period of months or years. The kind of variables we have in mind do not seem to have an immediate or synchronous effect. As noted earlier, both clinical experience and a substantial body of empirical findings reviewed in Patterson et al. (in press) emphasize the presence of low self-esteem, low academic performance, and rejection by the normal peer group as significant correlates with antisocial behavior. Each of them has been presented at one time or another as the determinant for antisocial behavior. However, treatment outcome studies suggest that rather than determinants, they are really products of the same process that gives rise to the antisocial behavior (see studies reviewed in Patterson et al., in press a). As pointed out by Rogosa (1980) and others, there are undoubtedly causal factors in child development that exert their effects slowly and over long periods of time (e.g., physical maturation and its effect on social behavior). In this regard, one of the puzzles for the longitudinal investigator concerns the problem of estimating the ideal time intervals required to test for the presence of such delayed-incremental processes. The first step in the study of delayed effects is to conceptualize a mechanism that might underlie such long-term effects. The idea we have termed shopping is offered as such a possibility. The general idea is that there are certain experiences outside the family that, if omitted, bring the child under even greater control of the contingencies that govern the coercion process. In this sense, some of the Stage 2 and 3 experiences would place the child at risk for not shopping for certain activities, settings, and groups. Presumably, antisocial children are less free to select, or shop for, some of the major socializing experiences available to normal children. For example, if certain children are extremely coercive or aggressive, the likelihood is high that they will be rejected by the normal peer group. Peers, therefore, are not going to teach them to cooperate, nor teach them sports skills, nor teach them how to form good relationships with equals. Antisocial children's refusal to cooperate with classroom rules means they spend very little time learning and tend to be consistent underachievers. This, in

Family process: loops, levels, and linkages

143

turn, means a reduction in the options open to academically unsuccessful children (e.g., recognition by adults, plans for continuing education, self-confidence, etc.). In a similar vein, children with low self-esteem or antisocial attitudes may also avoid certain kinds of settings or groups. The fact that these alternative experiences do not occur enhances the impact of the basic training in coercion occurring in the home. Over time, this increases the likelihood that antisocial children will shop for and select a social support system that will react somewhat positively to their behavior: the deviant peer group. As noted by Elliott, Huizinga, and Ageton (1985), the commitment to a deviant peer group is one of the prime determinants for career offenders. The hypothesis is that products such as low self-esteem, rejection by peers, or academic failure function as positive feedback loops for the criterion, antisocial behavior. The feedback hypothesis imposes several interesting requirements in a structural modeling analysis. As depicted in Figure 5.9(c) the hypothesis states that at tx the path coefficient will move from antisocial to peer relations, but the path coefficient from peer relations at tx to antisocial at t2 and t3 will also be significant. These cross-lagged paths must contribute something significant relative to both the synchronistic effects of family management at t3 and the stability coefficient for antisocial from t2 to t3. These predictions were recently tested in the study by Patterson and Bank (in press b). All of the hypothesized relations were obtained. Notice that one interesting second implication of such an effect is that the presence of such positive feedback loops indicates a dynamic system. Such families tend not to stabilize, but are still in the process. Even if cross-lagged effects are found, Dwyer (1983) and others are quick to point out it is shaky ground on which to construct arguments for causal connections. Dwyer's (1983) second major common-factors model would be used as a test prior to testing for the cross-lagged effects model. He elaborates on the difficulties of applying even structural modeling formats to questions concerning cross-lagged effects. With the biasing effects of strong synchronous relations at tl9 t2, and t3, he stipulates that it is increasingly important to first test for an unmeasured variable having a lagged effect on both X and Y. The family in context The family is highly vulnerable to a multitude of variables that can be influential in starting up the irritable exchanges among its members. In addition to the antisocial outcome detailed in this report, the irritable exchanges probably have much to do with marital satisfaction, depressed mood, and a general sense of well-being. This section is intended as a brief overview of the variables thought

144

Gerald R. Patterson

Figure 5.10: The family in context (from Patterson, Reid, & Dishion, in press)

to relate to risk for irritable exchange processes getting out of hand. The variables are summarized in Figure 5.10. Temperament Some children are more difficult to raise than others. There is also a convincing (not yet compelling) set of empirical findings that buttress this statement. For example, the programmatic studies by Lee and Bates (1985) have demonstrated stable correlations for difficult infant temperament at 6 months against ratings and observation data at 12 and 24 months. Furthermore, there are low-level but significant correlations between infant temperament ratings at 6 months with observed parent-child conflict at 2 years. What is particularly interesting is that

Family process: loops, levels, and linkages

145

this conflict looks very much like the coercion patterns we have found for families of older antisocial children. It is not hard to imagine that difficult children would be a strong stimulus for the parent, in that they elicit more irritable reactions from the parent. They may also be differentially responsive to parental contingencies, as suggested by the studies reviewed in Patterson (1982). It also seems quite possible that these difficult infants and toddlers may be related to the more daring boys who also tend to be at risk for delinquency (West & Farrington, 1973, 1977). Several studies make a clear case for an interactional effect (see Bronfenbrenner, chapter 2) in delineating the impact of child temperament and parent skills on child behavior. The longitudinal studies by Werner and Smith (1977) and more recently Sameroff and Seifer (1983) both demonstrate an interaction effect between infant risk status at birth and parent status. It is the effect of the difficult child and the mother's level of skill that significantly determines the level of child adjustment. Data from Cohorts 1 and 2 were not collected until the children were well beyond the preschool years. However, Olweus (1980) demonstrated the utility of variables based on parents' retrospective account of child temperament during preschool years. To replicate his findings, parents in the longitudinal study were asked to recall difficulties in rearing the child during the first 4 years (Patterson et al., in press). The alpha analyses showed the resulting scale to be internally consistent. It was expected that the difficult child construct would covary with disrupted parenting skills (i.e., Discipline and Monitoring constructs). However, the data did not support this hypothesis. None of the correlations for the difficult child score with family management measures were significant. The primitive measure of temperament did, however, correlate significantly with the composite child antisocial score (.27, p < .01). Since parental reports served as a-primary basis for both dependent and independent variables, this is hardly a satisfactory effort. Fagot's longitudinal studies of toddlers and their families at this center should give better coverage for this problem. Beliefs, values, and the neighborhood One would expect that parental values and beliefs about appropriate roles for adults and children would affect one or more of the family management skills. For the present, we have made no effort to assess these attitudes and beliefs. It is generally accepted that some residential areas in the city place the child at greater risk for delinquent involvement than do others, but it is also true that the vast majority of children living in such areas do not have police records. Cultural

146

Gerald R. Patterson

values and beliefs about the family and child rearing, and the immediate neighborhood in which the family resides, must have an impact on family process. As shown in Figure 5.10, the assumption is that these effects on child behavior are mediated by their impact on family management practices. For example, parents living in ghetto areas who are unskilled in family management put their children at greater risk than do skilled parents, or unskilled parents living in middle-class neighborhoods. It is interesting to speculate about the implication of this emphasis on family management constructs as central mediators. If the incidence of inept parenting were as high as 30% in our society (an entirely arbitrary figure), the likelihood of a child having two inept parents would be 9%, and the likelihood of having two effective parents would be 49%. This means that the likelihood of having one adequate and one inept parent would be 42%. Let us assume, along with Rutter and Guiller (1983), that a good relationship with one parent may be sufficient for adequate adjustment. This would mean that, as long as families remain intact, about 91% of the children are in pretty good shape; they have either one or two adequate parents. Since roughly half of the families will be divorced, then about 9% of the children in single-parent families will have an inept parent. It is easy to imagine what might happen to a mother recently divorced who has previously depended on her husband to discipline and socialize their child(ren). Overall, then, slightly more than 10% of the families should have problem children. Discussion The perspective for the study of family process presented here involved a set of interlocking assumptions and accompanying procedures. The study of process means that we are essentially studying change. To study change, we must have some idea about the kind of variables in family life that really produce changes in child behavior. Each investigator approaches this task in different ways. In our own case, the list of determining variables emerged as a result of several decades of effort to help parents change the behavior of their own children. The clinical studies led to the emphasis on five family management skills that seemed to relate to various kinds of child behavior (Patterson, 1982). If our studies had stopped at this macrosocial level, we would certainly have obtained the correlational support for the clinical-based hypotheses (e.g., poor discipline and monitoring covary with antisocial child behavior). However, we would not have understood the mechanisms that explain why the correlations were obtained. To answer this second question, "Why does poor discipline produce aggressive behavior?" we needed a whole new level of assessment procedures.

Family process: loops, levels, and linkages

147

The development of procedures for coding family sequences of interaction required 10 years of methodological studies. These kinds of data led to the study of microsocial exchange patterns. It became clear as the studies progressed that there are really two kinds of changes occurring simultaneously, and they occur at microsocial and macrosocial levels. It eventually became clear that they were interconnected, and that the connection was bidirectional. We think of the two levels in terms of progressions or structures. If such structures can be given a sound empirical base, it would mean we have indicators that could tell us how far into the process a dyad or a family has moved. It has been the microsocial level that has given us the most difficulty. We feel there is still much to be learned about how such data should be collected, and even more to be learned about how it should be analyzed. For example, the new codes developed in the last 5 years describe events in real time. Most of the statistical forms cannot use the additional information (e.g., a Markov process that takes into account duration of events is being developed by Gardner in Virginia; Griffin & Gardner, 1987), but the difficulty with microsocial data lies not so much here as it does in the nature of the phenomena themselves. Coercive exchanges seem to run like beads off a string. The patterns that we have studied make up only a minuscule part of what is going on in a family. These relatively banal events seem to describe processes that keep getting out of hand in troubled families. When a process does go awry, it can produce profound consequences for all members of the family. One of the questions about this level of the process concerns the fact that for the dyad to change, one or both members must react in a systematic manner over a long series of trials. For example, the stressed newly single mother must not only initiate more irritable exchanges with the target child, but in addition, systematically reinforce the child when he or she reacts in kind. How can it be that a reaction to outside stressors is to become a systematic reactor in this pattern? It would be less surprising, perhaps, if their reactions became more uncertain. It is obvious that this is not the only way to study family process; but it is one way. With this method or another, it is time that we initiate studies that developmental psychologists and clinicians alike so badly need, time to study causal relations in complex processes where behavior changes over time. Acknowledgments I gratefully acknowledge the support provided by MH 38318 (Family Mental Health and Policy Research Branch) for the preparation of this chapter. This

148

Gerald R. Patterson

latest revision is in part a reaction to the careful critique by Glen Elder of an earlier draft and a lively seminar discussion with my colleagues at the center. Both the readers and I have been rescued from a failed attempt to employ the ubiquitous boxes and arrows to describe process by the artistic activities of Will Mayer.

References Arnold, J., Levine, A., & Patterson, G. R. (1975). Changes in sibling behavior following family intervention. Journal of Clinical and Consulting Psychology, 43, 683-8. Bentler, P. M. (1980). Multivariate analysis with latent variables: Causal modeling. Annual Review of Psychology, 31, 419-56. Biglan, A., Hops, H., Sherman, L., Thorsen, C , Friedman, L., Arthur, J., & Osteen, V. (1984). Problem solving interactions of depressed women and their spouses. Paper presented at the annual meeting of the American Psychological Association, Toronto, Ontario. Cairns, R. B. (1979). The analysis of social interactions: Methods, issues, and illustrations. Hillsdale, NJ: Lawrence Erlbaum Associates. Cobb, J. A. (1972). The relationship of discrete classroom behaviors to fourth-grade academic achievement. Journal of Educational Psychology, 63, 74-80. Cobb, J. A., & Hops, H. (1973). Effects of academic survival skill training on low-achieving first graders. Journal of Educational Research, 63, 74-80. Coie, J. D., & Kupersmidt, J. B. (1983). A behavioral analysis of emerging social status in boys' groups. Child Development, 54, 1400- 16. Dengerink, H. A., Schnedler, R. S., & Covey, M. K. (1978). The role of avoidance in aggressive responses to attack and no attack. Journal of Personality and Social Psychology, 36, 104453. Dishion, T. J., Gardner, K., Patterson, G. R., Reid, J. B., Spyrou, S., & Thibodeaux, S. (1983). The family process code: A multidimensional system for observing family interaction. OSLC Technical Report. (Available from OSLC, 207 E. 5th, Suite 202, Eugene, OR 97401.) Dishion, T. J., & Loeber, R. (1985). Male adolescent marijuana and alcohol use: The role of parents and peers revisited. Journal of Alcoholic and Substance Abuse. Dishion, T. J., Stouthamer-Loeber, M., & Patterson, G. R. (1984). The Monitoring construct. OSLC technical report. (Available from OSLC, 207 E. 5th, Suite 202, Eugene, OR 97401.) Dwyer, J. H. (1983). Statistical models for the social and behavioral sciences. New York: Oxford University Press. Elder, G. H., Jr., Van Nguyen, T., & Caspi, A. (1985). Linking family hardship to children's lives. Child Development, 56, 361-75. Elliott, D. S., Huizinga, D., & Ageton, S. S. (1985). Explaining delinquency and drug use. Beverly Hills, CA: Sage. Forgatch, M. S., Patterson, G. R., & Skinner, M. (in press). A mediational model for the effects of divorce on antisocial behavior in boys. In E. M. Hetherington & J. D. Arasteh (Eds.), Impact of divorce, single parenting, and step-parenting on children. Gewirtz, J., & Boyd, E. (1977). Experiments on mother-infant interaction underlying mutual attachment acquisition: The infant conditions the mother. In T. Alloway, P. Pliner, & L. Krames (Eds.), Attachment behavior. New York: Plenum Publishing Corp. Gottman, J. M., & Levenson, R. W. (1986). Assessing the role of emotion in marriage. Behavioral Assessment, 8, 31-48. Gottman, J. M. (1981). Time-series analysis: A comprehensive introduction for social scientists. London: Cambridge University Press.

Family process: loops, levels, and linkages

149

Griffin, W. A., & Gardner, W. (1987). Continuous time Markov analysis of social interaction. Manuscript submitted for publication. Guttman, L. A. (1944). A basis for scaling qualitative data. American Sociological Review, 9, 139— 50. Harter, S. (1983). Developmental perspectives on the self system. In E. M. Hetherington (Ed.), Handbook of child psychology: 4. Socialization, personality, and social development. New York: Wiley. Hineline, P. (1977). Negative reinforcement and avoidance. In W. Honig & J. E. Staddon (Eds.), Handbook of operant behavior. Englewood Cliffs, NJ: Prentice-Hall. Hops, H., & Cobb, J. A. (1974). Initial investigations into academic survival-skill training: Direct instruction and first-grade achievement. Journal of Educational Psychology, 66, 548-53. Kenny, D. A. (1975). Cross-lagged panel correlations: A test for spuriousness. Psychological Bulletin, 82, 887-903. Knorr, G. L. (1981). A synthesis of homework research and related literature. Paper presented to the Lehigh Chapter of the Phi Delta Kappa Society, Bethlehem, PA. Lamb, M. E., Suomi, S. J., & Stephenson, G. R. (Eds.). (1979). Social interactional analysis: Methodological issues. Madison, WI: University of Wisconsin Press. Langer, E. G. (1984). Rethinking the role of thought in social interaction. Invited address presented to the Annual Meeting of the Oregon Psychological Association, Portland, OR. Lee, C. L., & Bates, J. E. (1985). Mother-child interaction at two years and perceived difficult temperament. Child Development, 56, 1314-25. Levenson, R. W., & Gottman, J. M. (1983). Marital interaction: Physiological linkage and affective exchange. Journal of Personality and Social Psychology, 45, 587-97. Loeber, R., & Dishion, T. J. (1983). Early predictors of male adolescent delinquency: A review. Psychological Bulletin, 94, 68-99. Marlowe, J. H., Reid, J. B., Patterson, G. R., & Weinrott, M. (1986). Treating adolescent multiple offenders: A comparison offamily management and traditional community approaches. Unpublished manuscript. Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 231-59. Olweus, D. (1980). The consistency issue in personality psychology revisited - with special reference to aggression. British Journal of Social and Clinical Psychology, 19, 377-90. Patterson, G. R. (1977). A three-stage functional analysis for children's coercive behaviors: A tactic for developing a performance theory. In D. Baer, B.C. Etzel, & J. M. LeBlanc (Eds.), New developments in behavioral research: Theories, methods, and applications. In honor of Siganey W. Bijou. Hillsdale, NJ: Lawrence Erlbaum Associates. Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castalia. Patterson, G. R. (1983). Stress: A change agent for family process. In N. Garmezy & M. Rutter (Eds.), Stress, coping, and development in children. New York: McGraw Hill. Patterson, G. R. (1985). Beyond technology: The next stage in developing an empirical base for parent training. In L. L'Abate (Ed.), Handbook offamily psychology and therapy (Vol. 2). Homewood, IL: Dorsey Press. Patterson, G. R. (1986a). Performance models for antisocial boys. American Psychologist, 41, 43244. Patterson, G. R. (1986b). Maternal behavior: Determinant or product for deviant child behavior? In W. Hartup & Z. Rubin (Eds.), Relationships and development. Hillsdale, NJ: Erlbaum. Patterson, G. R., & Bank, L. (1985). Bootstrapping your way in the nomological thicket. In R. McFall (Ed.), Behavioral Assessment, 8, 49-73. Patterson, G. R., & Bank, L. (in press a). When is a nomological network a construct? In D. Peterson & D. Fishman (Eds.), Assessment for decision: Psychology in action. Patterson, G. R., & Bank, L. (in press b). Some amplifying and dampening mechanisms for coercive

150

Gerald R. Patterson

family processes. In M. Gunnar (Ed.), Minnesota symposia on child psychology. Hillsdale, NJ: Erlbaum. Patterson, G. R., & Capaldi, D. (in press). A two-stage model for children's depression. In A. Masten & J. Rolf (Eds.), Developmental psychology: Risk and protective factors. Patterson, G. R., Chamberlain, P., & Reid, J. B. (1982). A comprehensive evaluation of a parent training program. Behavior Therapy, 13, 638-50. Patterson, G. R., & Cobb, J. A. (1971). A dyadic analysis of "aggressive" behaviors. In J. P. Hill (Ed.), Minnesota symposia on child psychology (Vol. 5). Minneapolis, MN: University of Minnesota Press. Patterson, G. R., & Cobb, J. A. (1973). Stimulus control for classes of noxious behaviors. In J. F. Knutson (Ed.), The control of aggression: Implications from basic research. Chicago, IL: Aldine Press. Patterson, G. R., & Dawes, R. M. (1975). A Guttman scale of children's coercive behaviors. Journal of Clinical and Consulting Psychology, 43, 594. Patterson, G. R., & Dishion, T. J. (1986). The contribution of families and peers to delinquency,

Criminology, 23, 63-79. Patterson, G. R., Dishion, T. J., & Bank, L. (1984). Family interaction: A process model of deviancy training. Aggressive Behavior, 10, 253-67. Patterson, G. R., & Fleischman, M. J. (1979). Maintenance of treatment effects: Some considerations concerning family systems and follow-up data. Behavior Therapy, 10, 168-85. Patterson, G. R., & Reid, J. B. (1984). Social interactional processes within the family: The study of moment-by-moment family transactions in which human social development is imbedded. Journal of Applied Developmental Psychology, 5, 237-62. Patterson, G. R., Reid, J. B., & Dishion, T. J. (in press). Antisocial boys: Social learning series (Vol. 4). Eugene, OR: Castalia. Patterson, G. R., & Stouthamer-Loeber, M. (1984). The correlation of family management practices and delinquency. Child Development, 55, 1299-1306. Reid, J. B., Taplin, P. S., & Lorber, R. (1981). A social interactional approach to the treatment of abusive families. In R. Stuart (Ed.), Violent behavior: Social learning approaches to prediction, management, and treatment. New York: Brunner/Mazel. Robinson, J. P. (1973). Toward a more appropriate use of Guttman scaling. Public Opinion Quar-

terly, 37, 260-7. Rogosa. D. (1980). A critique of cross-lagged correlation. Psychological Bulletin, 88, 337-83. Rosenfeld, H. M., & Gunnel, P. K. (1985). Pragmatics vs. reinforcers: An experimental analysis of verbal accommodation. In M. Giles & R. St. Clair (Eds.), Recent advances in language communication and social psychology. Hillsdale, NJ: Lawrence Erlbaum Associates. Rutter, M., & Guiller, H. (1983). Juvenile delinquency: Trends and perspectives. New York: Penguin Books. Sackett, G. P. (1978). The lag sequential analysis of contingency and cyclicity in behavioral interaction research. In J. Osofsky (Ed.), Handbook of infant development. New York: Wiley. Sameroff, A. J., & Seifer, R. (1983). Sources of continuity in parent-child relations. Paper presented at the meeting of the Society for Research in Child Development, Detroit, MI. Schuck, J. R. (1974). The use of causal nonexperimental models in aggression research. In J. De Wit & W. W. Hartup (Eds.), Determinants and origins of aggressive behavior. Paris: Mouton Publishers. Sears, R. R., Maccoby, E. E., & Levin, H. (1957). Patterns of child rearing. Evanston, IL: Row, Peterson. Sidman, M. (1960). Tactics of scientific research. New York: Basic Books. Skinner, B. F. (1969). Contingencies of reinforcement. New York: Appleton-Century-Crofts. Snyder, J. J., & Patterson, G. R. (1986). The effects of consequences on patterns of social interac-

Family process: loops, levels, and linkages

151

tion: A quasi-experimental approach to reinforcement in natural interaction. Child Development, 57, 1257-68. Viken, R. (1986). Stress, family process, and child antisocial behavior. Manuscript in preparation. Werner, E., & Smith, R. (1977). Kauai's children come of age. Honolulu, HI: University of Hawaii Press. West, D., & Farrington, D. P. (1973). Who becomes delinquent? London: Heinemann Educational Books, Ltd. West, D., & Farrington, D. P. (1977). The delinquent way of life. London: Heinemann Educational Books, Ltd. Woo, D. (1978). Experimental studies of the reinforcement trap. Unpublished master's thesis, University of Oregon, Eugene, OR. Yarrow, M. J., Campbell J. D., and Burton, R. V. (1968). Child rearing: An inquiry into research and methods. San Francisco: Jossey Bass. Youniss, J. (1980). Parents and peers in social development: A Sullivan-Piaget perspective. Chicago: University of Chicago Press.

6

On the constructive role of problem behavior in adolescence

Rainer K. Silbereisen and Peter Noack

Introduction Is substance use good for adolescents? This chapter demonstrates that a simple t4 no" may obscure the intricate ways in which problem behaviors and normal, adaptive adolescent development are intertwined. In making the transition from childhood to adulthood, adolescents face a variety of demands and challenges and they must acquire appropriate skills to master these challenges. Resorting to problem behavior such as substance use may be viewed as one way of mastering the developmental difficulties of adolescence. This postulated connection between the typical problems adolescents face and problem behaviors leads us to expect a sharp, but temporary, increase in substance use during this period of the life span. Indeed, prospective longitudinal studies have documented that the increase in substance use in adolescence is transient (Huba & Bentler, 1983; Jessor, 1986; Kandel & Logan, 1984; Magnusson, Duner, & Zetterboom, 1975). Our perspective on adolescence highlights the constructive function of problem behavior in coping with age-typical demands and difficulties. We do not, of course, dismiss the health and psychological risks resulting from the use of legal and illegal substances. Rather, we propose that young people's problem behaviors can be understood adequately only in the broader context of normal adolescent development. In this chapter we first outline a conceptual framework linking adolescent development and problem behavior. We have referred to this framework as "development as action in context" (Silbereisen & Eyferth, 1986). Second, we describe a naturalistic study of adolescent leisure settings undertaken to illuminate the role that these settings play in the mastery of developmental tasks and the function of substance use in this process. Third, a model of dynamic relationships between self-esteem, leisure-time orientations, and substance use is formulated and tested with longitudinal data from a large, represen-

152

On the constructive role of problem behavior in adolescence

153

tative sample of adolescents. Finally, we suggest several directions for future research on the multiple roles of substance use and other problem behaviors in developmental processes during adolescence and early adulthood.

Development and problem behavior: a conceptual framework Initiative and self-regulation As is the nature of heuristic models, "development as action in context" does not constitute a specific theory. Rather, it serves to delineate objectives for future research and helps to identify empirical strategies to pursue them. Viewing development in terms of goal-directed action puts the adolescent's initiative and planning in focus. This perspective pays particular attention to the goals that adolescents set for their own development and the ways in which they try to attain these goals. Although it highlights adolescents' active role in shaping their development, it does not imply that they have complete control of developmental changes. For a full exposition of the action-theoretic approach to human development the reader is referred to recent overviews by Brandtstadter (1984) and Silbereisen (1986). One way of characterizing control over development is in terms of selfregulation (Carver & Scheier, 1982; Silbereisen & Kastner, 1986). Individuals gradually form developmental orientations (i.e., personal goals) concerning future steps in their development. These orientations can be seen as individual responses to normative social expectations or "developmental tasks" (Havighurst, 1952; Peterson & Spiga, 1982). They function as a frame of reference for further action. If the present state of development is perceived as deviating from these orientations or goals, individuals may respond by changing'their internal conditions (e.g., goals) or their outer contexts (e.g., peer group) in order to reduce the discrepancy. Such self-directed interventions may lead to a change in the state of development. The result of this intervention is compared, in turn, with the corresponding orientation. Of course, each loop of control is subject to other external and internal influences on the system. The self-regulation of development is organized in a hierarchical-sequential way. That is, long-term changes, such as establishing peer relationships or undergoing professional training, are composed of a series of shorter steps. Thus, the goal of an intermediate step may be the outcome of a previous developmental intervention that, in turn, was directed by more abstract long-term developmental orientations. These interlocked regulative processes may simultaneously address different goal domains.

154

Rainer K. Silbereisen and Peter Noack

Problem behaviors in developmental perspective Behaviors are considered problematic when they conflict with social norms (see Brandtstadter, 1977). Thus, substance use becomes problematic when a legal consumption age is violated and/or when such use leads to long-term health problems. Viewed developmentally, adolescent problem behavior may be seen as an effort to master age-typical difficulties in psychological growth. This does not imply that substance use or delinquency is always directed at gaining control over development. Moreover, even if problem behavior is an attempt to cope with developmental difficulties, the relationship remains complex. A particular problem behavior may reflect attempts to cope simultaneously with quite different developmental difficulties, and, likewise, any given developmental difficulty may result in various problem behaviors. Few theoretical approaches to date deal systematically with the constructive role of problem behavior in adolescent development. The theory of "problem behavior proneness" is a major exception. Jessor and Jessor (1977) view problem behavior, whether alcohol use or precocious sexual experience, as an attempt to consummate the transition to adulthood in a demonstrative manner in cases where other ways of adopting adult privileges are unavailable. Empirical support for the functional role of problem behavior in development has also been provided by Magnusson, Stattin, and Allen (1985), who found that early onset of menstruation is associated with girls' use of marijuana in middle adolescence. They suggest that the violation of norms by smoking marijuana reflects the girls' efforts to adjust their social behavior to concurrent biological changes (i.e., to make their behavior in social contexts more adultlike).

Leisure time as a context of adolescent development In this chapter, we apply our model of "development as action in context" to a particular developmental phenomenon: the interplay between person and context in adolescent substance use. A variety of contexts offer opportunities for the formation of developmental orientations as well as for performing behaviors that may contribute to the attainment of goals reflecting these orientations. Despite this truism, developmental psychology is overly preoccupied with the two traditional contexts of development - family and school (Bronfenbrenner & Crouter, 1983). Thus, research on other, neglected developmental settings deserves particular attention. Places of leisure are a particularly relevant, yet underresearched setting of adolescent development (see Silbereisen & Noack, 1986). This is surprising considering the tradition of qualitative sociological studies, dating back to the 1920s, on the urban life-space of youth. These studies emphasized the fringe groups of society and the psychologically impoverished contexts

On the constructive role of problem behavior in adolescence

155

in which these young people lived. Thrasher's (1927) detailed work on youth gangs is a case in point. He suggested that the unambiguous role structure of the gang compensated for the ''social disorganization" of deprived city districts. More recently, research at the Centre for Contemporary Cultural Studies in Birmingham (Brake, 1980; Willis, 1977) has identified strategies employed by youth in their particular subcultures to cope with collective problems that are associated with their age group as well as their position in the social structure. Shifting from deviant peer groups to the everyday environment of normal adolescents, Becker, Eigenbrodt, and May (1984) applied the Birmingham theoretical and methodological approach to their study of ice cream parlors and various street settings. This study focuses on developmental changes in the appropriation of public spaces (see Harms, Preissing, & Richtermeier, 1985). This interest in developmental phenomena, however, is not shared by the majority of researchers studying development in settings such as streets, parks, and playgrounds (e.g., Eick, 1979; Grabow & Salkind, 1976; Holahan, 1978; Medrich, Roizen, Rubin, & Buckley, 1982; Payne & Jones, 1977). Thus, the ecology of adolescents' developmental opportunities remains uncharted. Even when both individual and contextual influences on development are integrated in empirical research, the models employed are frequently flawed. They are, more often than not, social address models (Bronfenbrenner & Crouter, 1983). Although the context is taken into account, it figures as little more than a name, "with no attention to what the environment is like, what people are living there, what they are doing, or how the activities taking place could affect the child" (pp. 382-3). In our opinion, the preference for simple models of this type reflects a conceptual gap in thinking about contextual "affordances" for development (Gibson, 1977). Specifically, we lack theoretical concepts to describe situational demands and corresponding personal competencies (Fuhrer, 1985; Kaminski, 1986). One approach that holds promise for overcoming these limitations is a multilevel assessment of individuals and contexts. Martha Muchow's research, carried out half a century ago, comes closest to meeting these goals (Muchow & Muchow, 1935; Wohlwill, 1985). Interviews with students about their activities outside the home led her to select several significant settings (mainly street settings and one department store) in which to carry out extensive observations. She described the department store as a place where adolescents could develop orientations toward the world of consumption and as an exercise ground where adolescents could practice acting like adult customers. The department store also accommodated the first shy approaches to the other sex. Our Berlin Longitudinal Study on Adolescent Development (Silbereisen & Eyferth, 1985) adopts the multilevel approach that characterizes Muchow's work. Annual assessments of favorite and most frequented leisure places are

156

Rainer K. Silbereisen and Peter Noack

being carried out on a large representative sample of West Berlin youth. The resulting data have guided the selection of major settings for further analyses. Developmental issues in leisure places: a naturalistic Held study In this section we report a series of naturalistic field studies carried out in adolescent leisure settings, including discotheques and shopping centers. These studies combine both interviews and observations. The semistructured interviews conducted in these leisure settings focus on the goals motivating adolescents to visit such settings and on the strategies they use to attain their goals. Thus, insights into the meaning of leisure places as developmental contexts are gained from the adolescents' perspective. The frequency and target of peer contacts were assessed through systematic observations. Previous research has shown that such contacts play a central role in adolescent leisure settings (Silbereisen, Noack, & Eyferth, 1986). In addition, the prevalence of substance use at the settings was recorded. An interview study of adolescent perspectives A major purpose of the interviews was to identify the normative developmental tasks reflected in the personal goals motivating adolescents' visits to leisure settings. The developmental tasks considered crucial for this age period include contacts with the other sex and peer group integration (e.g., Dreher & Dreher, 1983; Havighurst, 1952; Newman & Newman, 1979). Peer group integration refers to an adolescent's efforts to become a member of a clique and to assume and maintain a certain role in such a group. A second purpose of the interview was to develop an understanding of how abstract, long-term developmental goals are manifested in the concrete goals motivating visits to leisure settings. Method Sample. The data are based on 190 interviews conducted with adolescents (age range: 11 through 18 years) at 10 leisure places, 2 each of discos, shopping centers, ice-skating rinks, and indoor and outdoor pools. These settings were chosen from a list of favored leisure places indicated by adolescents in a representative questionnaire assessment (this sample is described later in the chapter). Interviewees were randomly selected at these locations. The distribution of age and gender was relatively even across the different places. Procedure. The adolescents entering a place were first asked about the goals that prompted the visit. The interviewer had to judge from an adolescent's

On the constructive role of problem behavior in adolescence

157

response whether a developmental task had motivated the visit and, if so, which one. Was the goal a topic of conversation at the place? What problems were encountered in the pursuit of the goal? The interviewer then explored how goals identified as important by the adolescent were dealt with in that setting. The order of questions was flexible in order to maintain conversational flow. Interviews lasted between 15 and 45 minutes and were coded directly from tape recordings. The categories used in coding the goals were contacts with the other sex, peer group integration, formation of a personal style, physical exercise and development, adopting a consumer role, and substance use. These categories reflect the aggregation of the adolescents' goals in terms of personal orientations for development. These orientations refer to developmental tasks that many consider to be of central importance during the adolescent years (e.g., Havighurst, 1952; Newman & Newman, 1979). Descriptions of the categories are listed in appendix A. Our discussion of the means adolescents use to attain developmental goals is restricted to those used to establish contacts with the other sex and to achieve integration into the peer group. Information on adolescents' ideas about ways of attaining developmental goals, gathered in a previous questionnaire assessment (Noack, Muchowski, & Silbereisen, 1986), provided the basis for summarizing the interview responses using the following categories: verbal contact, offering substances, visual contact, mediation by friends, using equipment or program of the place, attracting attention, change to other places, and personal attributes. In establishing the categories, we aimed at a level of abstraction that captures what is going on in concrete situations and that is applicable to same-sex and other-sex contacts and to the place types analyzed. The resulting categories parallel the general distinction made between individual efforts, external help, and being subject to autonomous events in the service of developmental progress (see Dreher & Dreher, 1983). A short description of the content of these categories appears in appendix B. Further information on the specific interview technique and the coding procedure is contained in Schuhler and Kastner (1985). Results. The distribution of interview responses according to the goals that adolescents described as guiding their visit is given in Table 6.1 (data are aggregated over the two instances per place type; breakdowns according to individual places are not shown). Use of legal drugs is also presented wherever it is mentioned as a goal of the visit. On the one hand, the data reveal place-dependent goals. The layout and the program of certain places are specifically designed to provide opportunities to pursue specific goals such as physical exercise, formation of personal style, and shopping. On the other hand, there are place-independent goals related to friendship and peer group activities. At one of the shopping centers, for example,

158

Rainer K. Silbereisen and Peter Noack

Table 6.1. Number of interviews in which goals related to developmental tasks were indicated by adolescents as motivating their visit to the leisure place

Goal Contacts with the other sex Peer group integration Formation of a personal style Physical exercise and development Adopting a consumer role Substance use Column total* Number of interviews

Disco

Shopping center

Indoor swimming pool

Outdoor swimming pool

Iceskating rink

24 14

22 5

1 19

10 15

10 22

67 75

10

1

0

0

0

11

0

0

24

7

6

37

0 1 49 (111) 40

15 0 43 (95) 41

0 0 44 (69) 40

0 0 32 (69) 29

0 0 38 (91) 40

Row total

15 1 206

Note: Multiple responses were possible; default category was omitted because of empty cells. a Numbers in parentheses refer to total number of instances in which goals related to developmental tasks were addressed in general.

contact with the opposite sex was the most frequently mentioned goal. At the other shopping center, this goal was also important, but consumer-related activities ranked first among the goals mentioned as pursued there. Substance use was rarely mentioned by adolescents. In fact, it was mentioned only once as an important goal motivating visits. Thus, substance use is not, in itself, a primary goal motivating visits to leisure settings. Strategies serving the attainment of developmental goals are only reported for contacts with the other sex and peer group integration, the dominant goals at virtually all places analyzed. Table 6.2 shows the overall results. Although the preferred way of establishing contacts with the opposite sex is clearly direct verbal address, discussants reported that the principal means of making contacts with same-sex peers often involved personal attributes such as appearance, attitudes, or personality traits. Often same-sex relationships are conducted on the tacit assumption that things will move along by themselves without any extra effort "if you are the right type of person." Within this general pattern, there were some setting-specific strategies for initiating peer contact. At sports settings, for example, the use of facilities was frequently mentioned as a strategy for initiating interactions (typically this involved playing tricks on others). Discos were the only places where the program (i.e., dancing) played a direct role in fostering opposite-sex contacts. Discos

Table 6.2. Percentage of strategies at various settings indicated by adolescents as means of attaining contacts with the opposite sex and peer group integration Shopping center (N = 41)

Disco (N = 40)« Strategies Verbal contact Offering substances Visual contact Mediation by friends Using equipment or program Attracting attention Change to other places Other (e.g., personal attributes) Number of strategies

Indoor pool (N = 40)

Other sex

Peers

Other sex

33 9 6 15

18 6 — 15

54 12 3 6

12 12 3

3 3 15

3 — 6

36

9 75

39 54

15 61

42 63

Peers

Other sex

Peers

Outdoor pool (N = 29)

Ice rink (N = 40) Other sex

Other sex

Peers

30 6 15 24

36 3 3 24

27 6 12 21

15 6

18 3 6

18 3 3

9 20

9 39

12 59

27

Peers

Note: Total number of strategies per goal at each place type = 100%. Percentages are not given when number of strategies is less than 20. a N = number of interviews.

160

Rainer K. Silbereisen and Peter Noack

were also described as the best place to attract attention from the opposite sex. For example, the fashion-oriented disco context provides the opportunity to use eye-catching clothes instrumentally for social ends. Although general reference to the strategy of verbal contact may be presumed to include offering substances, quite a few adolescents also explicitly described this behavior as their way of establishing or extending relationships with other adolescents. In 30 interviews (21% of the verbal contacts) adolescents indicated that the first sentences directed at another person focused on cigarettes or alcohol. However, the frequency with which this strategy was mentioned was strongly contingent on place type and depended on whether the use of a substance was permitted and common in a given place, as well as on the availability of other strategies such as borrowing a towel at the swimming pool.

Systematic field observation The systematic observations provide information on several general features of the leisure places, such as number of visitors and patterns of space use. A major component of the observations was the assessment of the extent and target (other sex, peers) of contacts among adolescents and substance use at each setting. Method Sample. A total of 386 ten-minute observation periods were conducted at the 10 leisure settings sampled. Individual visitors and face-to-face groups were observed. The adolescents selected for observation were stratified according to situation type (e.g., hanging out and dancing; see Noack, 1985). Usually 10 randomly selected instances of each situation type were observed at a given place. Across the 10 places, the situations observed show a relatively even distribution with respect to age and gender of the adolescent participants. Procedure. For each 10-minute interval, the observers recorded all instances of contacts with other people. A contact was recorded if any participant in the observed situation started to talk to, or was addressed by, a nonparticipant. Contacts involving same-sex peers were distinguished from those involving opposite-sex peers, and contacts between previously acquainted adolescents were distinguished from those between previously unacquainted adolescents. The observers tried to proceed unobtrusively. Thus, the data were recorded after the 10minute observation interval, and the notes were transferred to protocol sheets later on, outside the leisure setting (Noack, 1985, provides additional details). Use of legal substances was scored if at least one participant in the situation smoked a cigarette or drank alcohol during the observation interval.

On the constructive role of problem behavior in adolescence

161

Table 6.3. Frequencies of situations with participants making contact with other adolescents (percent) Indoor swimming pool

Outdoor swimming pool

Iceskating rink

Type of contact

Disco

Shopping center

Same sex Acquainted Not acquainted

27 (34.5) 5 (6.4)

3 (5.0) 0(0)

9 (11.3) 4 (5.0)

20 (25.0) 0(0)

24 (27.3) 0(0)

Other sex Acquainted Not acquainted

25 (32.1) 10(12.8)

3 (5.0) 4 (6.7)

3 (3.8) 3 (3.8)

11 (13.8) 0(0)

25 (28.4) 0(0)

Number of situations observed

78

60

80

80

88

Note: Figures in parentheses indicate situations in which contacts were established as a percentage of all situations observed.

Results. Table 6.3 shows the frequencies (as percentages) of observed situations in which contacts were established between adolescents participating in the situation and people outside of the situation. (The figures were calculated across all settings.) The initiation of short- or long-term interactions with peers was the most commonly observed behavior at the leisure situations. However, the frequency of these behaviors depended on place types. The discos surpassed all other places in number of contacts made, followed by ice-skating rinks. The latter ranking is somewhat surprising because other sports-oriented places, such as the indoor pools, emerge at the bottom of the list. The vast majority of contacts in all places involved adolescents who were already acquainted. At several places, all observed contacts were of this type. Most contacts between previously unacquainted adolescents were found at the discos. The observations could not, however, distinguish whether the adolescents knew each other from prior face-to-face interaction or from mutual gazing during previous visits to the place. In any event, some common history seemed to be a necessary prerequisite for initiating conversations. Table 6.4 shows the number of situations in which cigarettes or alcohol were used during the observation interval across all leisure places. The highest proportion of situations (almost 33%) involving the use of cigarettes and alcohol occurred at the discos. Most of these situations involved cigarette smoking. Outside the discos, no situations involving alcohol were observed, a fact that is certainly due to its limited availability. Cigarette smoking could occur anywhere except for the indoor pools. The extent of smoking varied considerably across settings with the frequency at shopping centers approximating that observed at the discos.

162

Rainer K. Silbereisen and Peter Noack

Table 6.4. Frequencies of situations with participants smoking cigarettes or drinking alcohol (percent) Indoor swimming pool

Outdoor swimming pool

Iceskating rink

Substance

Disco

Shopping center

Cigarettes Alcohol

30 (38.5) 20 (25.6)

17 (28.3) 0 (0)

0(0) 0(0)

8 (9.1) 2 (2.3)

3 (3.8) 0(0)

Number of situations observed

78

60

80

80

88

Note: Figures in parentheses indicate smoking and drinking situations as a percentage of all situations observed.

Discussion It is reasonable to characterize the leisure contexts studied as instrumental for pursuing concrete developmental goals, especially integration into peer groups and establishing contacts with the opposite sex. In this regard, the adolescents' subjective assessment of the function of leisure settings obtained in the interviews concurs with the observational results (Silbereisen, Noack, & Eyferth, 1986). Even where a place has an officially designated purpose, such as sport or shopping, adolescents do not identify activities associated with this official function as more important than activities associated with the social function of the place. Silbereisen, Noack, and Eyferth (1986) describe how adolescents' behaviors during visits to such places consist mainly of looking around, chatting, and body contact. The social orientation of adolescent activities at leisure settings is paralleled by results from a more qualitatively oriented interview and observational study that Harms and his colleagues (1985) conducted in two West Berlin neighborhoods. They describe how several cliques of younger male adolescents use a department store as a ''playground." This gave rise to conflicts with the staff who tried to enforce the official and unofficial rules designed to protect and facilitate the operation of the department store for its intended purpose. Similarly, they report on the importance of an ice-skating rink as a meeting place for peer groups for whom establishing cross-sex contacts is a major goal. Adolescents rarely visited the observed settings to partake of substances, but substance use did facilitate the developmental goals. A 16-year-old boy tacitly exposes this function: "The best way to make contact is to ask somebody for a cigarette, and then tear the filter off, and then crumble the cigarette - she just must realize that I mean something else." Cigarette use serves as a prop in mastering the developmental goal of establishing contacts with the opposite sex.

On the constructive role of problem behavior in adolescence

163

Self-esteem, leisure-time orientations, and substance use Our field studies have identified what adolescents seek at leisure settings and what they actually find there. Although the use of legal substances per se does not motivate visits to certain leisure places, it is an integral part of activities involved in the pursuit of specific developmental goals. These results, however, provide only a snapshot. Whether certain kinds of leisure activities actually influence the course of development remains an open question. Our field studies can only support the plausibility of the notion that substance use is instrumental for adolescents in mastering their developmental tasks. An explicit test of the constructive role of problem behavior requires an investigation of the long-term interplay between developmental difficulties and substance use.

Toward a process model Our model assumes a bidirectional relationship of person and context. Specifically, developmental difficulties are assumed to impair self-esteem. This, in turn, promotes leisure-time orientations that lead to more extended contacts with peer groups, and these contacts lead to increased substance use. As a result of increased contact with peers, self-esteem increases. Our model presumes that selfesteem is a prerequisite for successful mastery of other kinds of developmental difficulties. We refer to these leisure-time orientations as ''cathartic activities." Adolescents perceive these cathartic activities as responses to frustrating experiences resulting from failure to meet demands at school or at home. Examples of cathartic activities include acting out on the stands of a soccer stadium, "getting into" a rock concert, and roaming about the neighborhood "letting off steam." Our assumptions about the relationship between self-esteem and substance use draw upon Kaplan's theory of deviance in defense of self (1980). Kaplan suggests that adolescents' repeated failure to meet norms and age-graded expectations held by parents and teachers results in a loss of self-esteem. This situation is assumed to lead adolescents to turn to groups and behaviors deviating from adult norms. Behaving in a way that meets the expectations of the new reference group results in renewed recognition and helps adolescents to regain positive self-esteem. At present, findings that support our model only address parts of the proposed sequence. In a longitudinal study focusing on problems of fit between adolescents' expected and actual contacts with the opposite sex, Caspi and Silbereisen (1986) showed that adolescents who failed to minimize this mismatch had substantially lower self-esteem than adolescents who managed to attain a fit. Kaplan, Martin, and Robbins (1984) found that lowered self-esteem in response to rejection by normative reference figures and groups facilitated the use of illegal

164

Rainer K. Silbereisen and Peter Noack

substances later in life. Finally, Huba and Bentler (1982) observed that the use of cannabis in early adolescence resulted in higher self-acceptance scores four years later. The question of whether these relationships are components of a larger sequence of interrelated changes awaits an empirical answer. Our model postulates that leisure-time activities involving peers link low selfesteem with subsequent increases in substance use. Studies on this mediating role of leisure activities in adolescent development are scarce. An exception is the research on the impact of social networks on health (Cohen & Syme, 1985). Wills and Vaughn (1985) studied social networks accessible to adolescents, focusing on adolescents' willingness to seek advice and help from members of their peer group when coping with problems. Those who relied on social support as their primary coping mechanism showed a higher use of cigarettes, particularly during middle adolescence. Ways that peer group activities influence the extent of substance use have been suggested by Jessor and Jessor (1977) and Kandel, Kessler, and Margulies (1978). Specifically, these studies suggest that behavior models at the settings that adolescents frequent are mainly responsible for this effect. The following study provides a partial test of the model outlined above. We propose that cathartic leisure-time orientations are the crucial mediator between self-esteem and substance use in adolescence.

Method Sample. The participants belonged to two groups: (a) The middle adolescence group comprised 504 students from Grades 7, 8, and 9; approximately 50% were female. This group was representative of West Berlin in terms of school tracks1 and city districts, (b) The early adolescence group comprised 326 students from Grades 5 and 6 in elementary school. Assessments for both groups were carried out at an interval of 12 months. 1 German school track names are retained in order to avoid the misleading use of English names that refer to a different school system. In the Berlin school system, Hauptschule includes Grades 7 to 10 and is nonacademic; Realschule includes Grades 7 through 10 and is somewhat more academically oriented than Hauptschule; Gymnasium includes Grades 7 through 13 and is the traditional German pre-academic institution; Gesamtschule includes Grades 7 through 13 and offers both academic and nonacademic tracks. Although the transition from primary to secondary school after Grade 6 is similar to many U.S. grade-level organizational patterns, the quality of the shift is not comparable. It implies not only a change in the ecology (child-centered neighborhood elementary school versus subject-centered, larger secondary school), but a placement according to prior school achievement. Thus, tracking usually leads to even more disruptive effects ("top-dog" phenomenon) than it does in the United States (see Blyth, Simmons, & Carlton-Ford, 1983).

On the constructive role of problem behavior in adolescence

165

Variables. Self-esteem (four items) was assessed employing items such as ' T m satisfied with myself" (Silbereisen, Reitzle, & Zank, 1986). Leisuretime orientations (six items) was measured by asking the adolescents to indicate how likely they would be to release the frustration that arose on a day marked by failure at school or at home by roaming about the neighborhood, playing pinball or video machines, attending a rock conceit, listening to music full blast at home, and the like. An example is as follows: "You've had a bad day. School, your parents - everything, gets on your nerves. Now assume you have a chance to go and really shake off your frustrations to the beat of fantastic live music. How likely would you be to do this?" Slightly different versions of this assessment were employed for each age group to take account of the age difference. During the first assessment, respondents reported on substance use in terms of lifetime prevalence. Twelve months later the respondents reported how frequently they consumed beer or wine, hard liquor, and cigarettes during the past 12 months. Analysis. Structural equation models (Figure 6.1) of the relationships between the three constructs at both measurement points were estimated separately for each age group and type of substance. These models were analyzed using LISREL V (Joreskog & Sorbom, 1981) as follows: (a) Measurement models for the constructs were tested using confirmatory factor analysis. All measurement models provided adequate fits to the data, (b) The goodness-of-fit between the structural model and data was tested using maximum-likelihood methods, (c) We hypothesized a particular pattern of cross-lagged paths among the different constructs. To test this proposition, we compared a model that included all time-lagged paths (cross-lagged model) to a model that included only paths between the same constructs over time (stability model). The former provided a better fit to the data, (d) Multiple-population maximum-likelihood methods were applied to assess whether the structural relationships were invariant across sex and school track. No significant subgroup differences were found. The tested models assume time-lagged as opposed to contemporaneous effects among the constructs. This is justified on the grounds of plausibility of the time lag in the effects of substance use on self-esteem: Regaining positive self-esteem through increased alcohol use, for example, involves several months rather than a couple of days. Further details about the statistical model and its analysis may be found in Silbereisen, Reitzle, and Dwyer (1986). Test of model for middle adolescents Our first analysis examined the relation between self-esteem and alcohol use (wine and beer) in middle adolescence (approximately age 14). The structural

n.s. Age

'Significant standardized structural coefficients among constructs and standardized regression coefficients for background variables are reported. Figure 6.1: Structural model for alcohol use in middle adolescence.

18

High Track

15

Middle Track

.12

High Track

n.s.

Gender

On the constructive role of problem behavior in adolescence

167

model for self-esteem, leisure-time orientations, and alcohol use is shown in Figure 6.1. Only significant paths are depicted.2 The pattern of time-lagged relationships among the constructs confirms our hypothesis. Adolescents with lower self-esteem were more oriented toward cathartic leisure activities (-.14). Moreover, adolescents oriented toward these activities were significantly more likely to drink in the future (.20). Finally, alcohol use had a positive effect on self-esteem (.13). Adolescents who drank more at time 1 thought more highly of themselves at time 2. Does this pattern also hold for cigarette use? The results of structural modeling with smoking substituted for alcohol use are depicted in Figure 6.2. Again, only significant paths are shown.3 Parallel to alcohol use, there was a significant path from prior leisure orientations to later cigarette smoking (.10). A positive effect of smoking on subsequent self-esteem, however, was not found. Although cathartic leisure orientations increased the use of both substances, only the consumption of alcohol affected future self-esteem. The absence of an effect of cigarette smoking on self-esteem may indicate that by middle adolescence smoking no longer evokes particular recognition in peer groups. If smoking does enhance self-esteem at all, this effect should show up in younger adolescents.

2 Using maximum likelihood analysis, a model for alcohol was estimated in which all lagged paths among the constructs (self-esteem, cathartic leisure-time orientations, alcohol use) were free parameters. Given the sample size, the goodness-of-fit of the model was acceptable (x 2 = 168.40, df = 95, p < .001). The hypothesis that there are no lagged relationships among the constructs was tested by comparing this model with an alternative model that included stability paths only. The difference in goodness-of-fit was significant A x 2 = 190.45 - 168.40 = 22.05, df = 101 95 = 6, p < .005), and thus the model including lagged paths was interpretable. However, only significant paths are depicted in Figure 6.1. The significance tests used are based on estimated standard errors. Disturbance covariances are not shown. Two of the disturbances at the first measurement were significant: those between self-esteem and leisure-time orientations (—.14), and those between leisure-time orientations and alcohol use (.24). Thus, the process linking these constructs predates the development represented by the first measurement. In order to rule out the possibility that different cross-lagged paths in the model are due to age differences, the sample was divided into three age groups. Next, a model with invariant crosslagged paths across groups was compared with a model involving no such constraints. The difference in fit between the model was not significant ( A \ 2 = 10.78, df = 12, p > .50). 3 The same logic used in the case of alcohol was applied to the model including cigarettes. The difference in goodness-of-fit between the stability model and a model that included all cross-lagged paths was significant (A x 2 = 135.40 - 119.59 = 15.81, df = 72 - 66 = 6, p < .025). Here, more disturbance covariances were significant: between cathartic leisure-time orientations and cigarette use at both waves of measurement (.29 and .07, respectively); for the first wave between self-esteem and leisure-time orientations (-.15); and between self-esteem and cigarette use (-.17). The model is age-invariant (A x 2 = 14.56, df = 12, p > .25). All reported tests refer to models that include the exogenous variables (age, gender, school tracks).

.12 Age

Age

Low Track

-.29

High Track

.25

15

High Track

15

Middle Track

n.s.

High Track

.08

Gender

Gender

Low Track

.25—1to'Cathartic

[ ^

High Track Low Track

Leisure 1982

.21

Middle Track High Track

-.34—• [

Gender

.12^"^

Age

.29*^

V

Use 1982

'Significant standardized structural coefficients among constructs and standardized regression coefficients for background variables are reported. Figure 6.2: Structural model for cigarette use in middle adolescence.

On the constructive role of problem behavior in adolescence

169

Test of model for early adolescents Figure 6.3 shows the structural equation model for alcohol use in the early adolescence group. 4 Only significant paths are depicted. A negative relationship between initial self-esteem and subsequent leisure-time orientations was also found among early adolescents. Adolescents with low self-esteem were more oriented toward cathartic activities (-.30). The stability of self-esteem is somewhat lower than in the older group (.51 as compared to .67). This makes sense when the division of students into different school tracks, a notorious source of strain in West Germany, is taken into account (Schwarzer, 1983; Fend & Schroer, 1985; Silbereisen, Reitzle, & Zank, 1986). Even though this division took place after the second measurement, students usually became aware between time 1 and time 2 of their probable placement - an event that has longterm consequences for social mobility. Leisure-time orientations did not influence future alcohol use. Moreover, current alcohol use did not predict self-esteem one year later. Bear in mind that fewer behavior models for alcohol use are available in settings preferred by early adolescents. Thus, alcohol was unlikely to be relevant to self-esteem for this age group (approximately age 11). Figure 6.4 shows the results for cigarette use. Only significant paths are depicted. Cigarette use was related to self-esteem, but the effect is not mediated by leisure performances. Moreover, the effect was opposite in sign ( - .20) to that predicted. It appears that the more heavily young adolescents smoked, the lower was their subsequent self-esteem. The results described thus far clearly indicate that drinking and smoking are part of different adolescent developmental processes. In middle adolescence, both behaviors may be induced by the context of leisure activities, but only alcohol has significant bearing on self-esteem. In early adolescence, neither drinking nor smoking appear to be promoted by cathartic leisure orientations. However, smoking may be a source of negative self-esteem. Although the structure of the model did not vary by sex or school track, there were some mean level differences by sex and school track. Particularly impressive is the contrast between the lowest and highest school track {Hauptschule

4 Again, cross-lagged models resulted in a significant increase in goodness-of-fit for both substances (alcohol: A x 2 = 88.59 - 73.40 = 15.19, df = 54 - 48 = 6, p < .025; cigarettes: A X 2 = 89.50 — 65.67, df = 54 — 48 = 6, p < .01). There were significant disturbance covariances: between leisure-time orientations and cigarette consumption at both measurement points (.24 and .14, respectively), and, in addition, at the first wave between cigarette use and self-esteem (—. 16), and between leisure-time orientations and alcohol consumption (.16). A breakdown into two age groups demonstrated the age invariance of the model (alcohol: A x 2 = 5.96, df = 6, p > .25; cigarettes: A x 2 = 10.80, df = 6, p > .05).

-.26

Gender -.16

'Significant standardized structural coefficients among constructs and standardized regression coefficients for background variables are reported. Figure 6.3: Structural model for alcohol use in early adolescence.

Age

Gender -.16

-.26

'Significant standardized structural coefficients among constructs and standardized regression coefficients for background variables are reported. Figure 6.4: Structural model for cigarette use in early adolescence.

Age

172

Rainer K. Silbereisen and Peter Noack

vs. Gymnasium).5 This finding clearly supports earlier research indicating that students put into the lowest track show lower self-esteem and higher preferences for cathartic leisure activities and drink alcohol more often (Dusek & Flaherty, 1981; Fischer, Fischer, Fuchs, & Zinnecker, 1981; O'Malley & Bachman, 1983; Schwarzer, 1983). In the course of the year covered by our study, subjects in the highest track caught up with students in the lower tracks with respect to drinking, but their self-esteem also increased. A characteristic of cigarette use suggested from previous literature was also confirmed (Johnson, O'Malley, & Bachman, 1984). The rates of use for girls are higher than those for boys. Moreover, this effect was accentuated over the study period. Discussion A negative relationship between self-esteem and cathartic leisure-time orientations was found for both boys and girls, in the lowest and highest school tracks, and in early and middle adolescence. The relationship between cathartic leisuretime orientations and substance use, however, differed significantly for the two age groups. A clear cross-lagged effect in the older cohort contrasts with no effect in the younger cohort. Our interpretation presumes differences in the demand characteristics of agetypical leisure settings. The finding that in the older cohort cathartic leisure-time orientations are related to subsequent alcohol and cigarette consumption can be understood as a facilitation effect, especially in light of our field studies. At adolescent leisure settings such as discos or soccer grounds, we observed that it is important for adolescents to fit the behavioral expectations at a given setting in order to feel comfortable. In attempts to conform to contextual pressures, the adoption of appropriate behaviors such as smoking and drinking is an essential aspect of the emerging "fit" between person and context (Lerner & Lerner, 1983). Presumably the pressure toward substance use is lower for younger adolescents in leisure settings, and this explains the absence of a significant crosslagged relationship between leisure and substance use in the younger cohort. Substance use and self-esteem. Our results suggest that drinking and cigarette smoking may be aimed predominantly at integrating the youngster into the peer context. This effect, in turn, would be responsible for the subsequent gain in self-esteem. Although the data on alcohol consumption fit this line of 5 For purposes of analysis, the four tracks were "effect-coded" (see Cohen & Cohen, 1975), i.e., coded into three trichotomous variables: on "low track" scores are Hauptschule = 1, Gesamtschule = -1, Realschule and Gymnasium — 0. "Middle track" {Realschule = 1, etc.) and "high track" {Gymnasium = 1, etc.) were coded similarly.

On the constructive role of problem behavior in adolescence

173

argument, the data on cigarettes do not, for no relationship with future selfesteem was observed. According to Flay and his colleagues (1983), one could try to solve this discrepancy by employing a model of "stages" of substance use. For example, the functional role of smoking for peer integration may be most prominent during the stage of initiation. Middle adolescence may be too late, then, to investigate this relationship, and we would expect it to hold only in our group of early adolescents. This was not the case, however. Although both cohorts show significant relationships between substance use and self-esteem, the effects are opposite in sign. Whereas for early adolescents smoking was negatively related to future self-esteem, for middle adolescents drinking alcohol showed a positive effect on later self-esteem. This difference in consequences is not surprising given the differences in the presumed developmental processes indicated by the structural models. Smoking in the younger cohort was not induced by cathartic leisure-time orientations and there was thus no indirect influence of earlier self-esteem operating through cathartic leisuretime orientations. Prospect In this chapter we have emphasized the constructive role of substance use in adolescence. Specifically, it is viewed as a goal-directed behavior aimed at mastering the developmental tasks of adolescence. However, we do not claim this role for all consumption patterns of all people. We proposed this function for the highly prevalent, quasi-normative use of legal substances, such as alcohol, in a normal population of adolescents, but not for substance abuse or addiction, which is restricted to small fractions of an age cohort. Even in such cases, however, a constructive perspective on problem behavior may shed light on overlooked aspects of heroin use, for example (Projektgruppe TUdrop, 1984; Uchtenhagen & Zimmer-Hofler, 1985). Moreover, we might ask whether the use of a given substance serves similar functions for different developmental difficulties. Our model was designed to account for developmental difficulties that represent minor discrepancies in the fit between goals and opportunities in domains such as peer friendship and intimate relationships. As the observations in leisure-time settings demonstrated, offering cigarettes may simply play a facilitating role in overcoming shyness when initiating contacts with the other sex. There are, however, more severe threats to self-regulation in development, such as what we would term perceptions of "meaninglessness of life" or alienation resulting from a general rejection of societal norms and values. This lack of orientation to normative goals may obstruct adolescents' systematic efforts to bring about developmental progress. In these cases, substance use and participa-

174

Rainer K. Silbereisen and Peter Noack

tion in related life-styles could substitute for developmental goals and become a major frame of reference. Substance use as part of the value system that regulates the principal aims and course of development can hardly be compared with substance use serving instrumental functions in approaching intermediate developmental goals in everyday contexts. Thus, we stress that it is the perceived function of substance use in the self-regulation of development rather than its prevalence or intensity that determines developmental outcomes. Agenda for future research In this section we identify two major topics for future research using a constructive approach to problem behavior. First, it is of considerable practical interest to identify functional equivalents for substance use in mastering developmental tasks. We use the term functional equivalents to refer to behaviors that are not problematic in terms of social norms or criteria of well-being, but that serve equivalent functions in the course of development. In our view, prevention can only be successful if it is based on concrete knowledge about the functional equivalents that may substitute for problem behaviors. Indeed, the issue of "alternatives" is receiving increased attention from the perspective of primary prevention (Gulotta & Adams, 1982; Botvin, 1984; Perry & Jessor, 1985). However, there is a striking lack of empirical evidence showing that problem behaviors and positive behaviors can play functionally equivalent roles in the mastery of developmental difficulties during adolescence. This gap represents a serious obstacle to more effective prevention of substance use. Although we do not claim to redress this gap, our data provide suggestive evidence that the relationship between leisure-time activities and goal attainment is more complex than assumed by advocates of positive alternatives. Consider, for example, the case of sport, which has been proposed as a positive alternative to discos. Our field observations showed that discos and ice-skating rinks may play a comparable role in helping adolescents achieve the goal of peer group integration. However, the contextual demands for smoking in these nonsporting and sporting leisure settings may be similar. Thus, the simple advocacy of sport (e.g., ice-skating) as a positive, alternative cathartic leisure-time activity may not lead to the desired objective of reducing substance use among youth. A second issue for future research is whether leisure-time orientations other than cathartic orientations play similar roles with regard to both self-esteem and substance use. "Substitutive" leisure activities may provide one alternative to cathartic leisure activities for coping with adolescent frustrations and strains. These leisure activities are characterized by goal achievements equivalent to those blocked in school (e.g., organization of a celebration at school, practicing

On the constructive role of problem behavior in adolescence

175

to play a musical instrument for a public performance). Using the existing sample, a measure of "substitutive" leisure-time orientations was constructed.6 Preliminary tests of structural models that replaced cathartic with substitutive leisure-time orientations showed that substitutive orientations have a distinctive association with self-esteem and substance use. In contrast to cathartic orientations, they are not influenced by self-esteem, but there is a slight protective effect on future smoking. Furthermore, girls show more pronounced substitutive orientations. Thus, what is seen in leisure research as alternative ways of compensating for job-related strains actually represents a quite different process in our sample. Thus far, few studies have examined the functions of problem behavior in normal adolescent development and the different ways that substance use is related to developmental changes in adolescence. Our results point strongly to the importance of a constructive perspective on problem behavior. This approach allows us to go beyond merely registering the coincidence of developmental transitions and peaks in the prevalence and intensity of substance use during adolescence; it highlights the importance of scrutinizing the instrumental value and psychological functions of problem behaviors in the course of adaptation during adolescence. Although our findings must be viewed as a first step, we hope they encourage further work in this direction and contribute to the design of prevention programs. Appendix A Categories for adolescents' goals related to developmental issues Contacts with the other sex. Initiating, maintaining, or extending heterosexual contacts, be they sporadic or steady (getting to know an adolescent of the other sex; getting to know each other better; establishing a steady relationship; dealing with the problems arising from such a relationship). Peer group integration. Establishing or improving contacts with adolescents of the same sex or adolescent cliques (making friends; hanging out with a good friend or a group of peers and gaining, consolidating, or changing one's position among peers). 6 An example of the items for substitutive leisure-time orientations is as follows: "You had big plans for this year at school, but you've been having trouble keeping up in some subjects. Now imagine that you play a musical instrument and a bunch of kids ask you to practice regularly with their group. You like playing with them, and you practice music at home, too, to be sure you'll be able to keep on playing with this group." The reported results refer to the older cohort.

176

Rainer K. Silbereisen and Peter Noack

Formation of a personal style. Establishing or consolidating a personally distinctive pattern of appearance, habits, and interests (finding an orientation toward music and dress, and identifying coherent patterns of leisure behavior, in particular; interest or participation in subcultural groups, and attempts at making a corresponding impression). Physical exercise and development. Dealing with the physical changes taking place during puberty, and changing attitudes and behaviors related to one's body and health (exercise aimed at an improved condition of one's body). Adopting a consumer role. Developing attitudes and behavior patterns of a conscious consumer (gathering information on prices; comparing quality and prices of goods; budgeting). Substance use. Smoking cigarettes or drinking alcohol. Other. Default category; all goals related to other issues (escaping boredom and looking for a lively atmosphere; relaxing and recovering from school; general preference for a place that cannot be elaborated on).

Appendix B Categories for means to reach the goals of contacts with the other sex and peer group integration Verbal contact. Speaking to somebody, for example, exchanging names, talking about the place. Offering substances. Offering or asking for a cigarette or a sip of beer/wine. Visual contact. Eye contact; winking at somebody. Mediation by friends. Being introduced to somebody by a mutual acquaintance; asking a friend to establish the contact; seeking peer situations with a lot of outside contacts. Using equipment or program of the place. Dancing together; splashing water at somebody while swimming; bumping into somebody while skating; going to a dark room to make out.

On the constructive role of problem behavior in adolescence

111

Attracting attention. Dressing up for effect; engaging in strange or loud behavior; having conversations so that outsiders will eavesdrop. Change to other place. Making dates for other places; leaving the place to continue the interaction somewhere else. Other (e.g., personal attributes). Default category; especially personality traits, outer appearance, attitudes that are essential in the pursuit of the goal and will eventually lead to success.

Acknowledgment Our studies were supported in part by German Research Council Grant Si 296/1-1 (principal investigators: R. K. Silbereisen and K. Eyferth). The research was carried out in collaboration with Matthias Reitzle and Peter Kastner. It benefited from the painstaking assistance of Mechthild Niemann. This chapter was prepared during a sabbatical leave made possible by German Research Council Grant Si 296/2-2 to R. K. Silbereisen. We are grateful to Urie Bronfenbrenner for helpful discussion of this chapter.

References Becker, H., Eigenbrodt, J., & May, M. (1984). Pfadfinder heim, Teestube, Strafienleben. Frankfurt: Extrabuch-Verlag. Blyth, D. A., Simmons, R. G., & Carlton-Ford, S. (1983). The adjustment of early adolescents to school transitions. Journal of Early Adolescence, 3, 15-30. Botvin, G. J. (1984). Prevention research. In Secretary, Department of Health and Human Services (Ed.), Drug abuse and drug abuse research. Rockville, MD: National Institute for Drug Abuse. Brake, M. (1980). Soziologie der jugendlichen Subkulturen. Frankfurt: Campus. Brandtstadter, J. (1977). Normen. In T. Herrmann, P. R. Hofstatter, H. P. Huber, & F. E. Weinert (Eds.), Handbuch psychologischer Grundbegriffe. Miinchen: Kosel. Brandtstadter, J. (1984). Personal and social control over development: Some implications of an action perspective in life-span developmental psychology. In P. B. Baltes & O. G. Brim, Jr. (Eds.), Life-span development and behavior (Vol. 6). New York: Academic Press. Bronfenbrenner, U., & Crouter, A. C. (1983). The evolution of environmental models in developmental research. In W. Kessen (Ed.), Handbook of child psychology: History, theories, and method (Vol. 2). New York: Wiley. Carver, C. S., & Scheier, M. F. (1982). Control theory: A useful conceptual framework for personality, social, clinical, and health psychology. Psychological Bulletin, 92, 111-35. Caspi, A., & Silbereisen, R. K. (1986). Transitions into dating: Costs and benefits to adolescent self-esteem. Paper presented at the 2nd European Conference on Developmental Psychology, Rome.

178

Rainer K. Silbereisen and Peter Noack

Cohen, J., & Cohen, P. (1975). Applied multiple regression! correlation analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum. Cohen, S., & Syme, S. L. (1985). Issues in the study of social support. In S. Cohen & S. L. Syme (Eds.), Social support and health. New York: Academic Press. Dreher, E., & Dreher, M. (1983). Wahrnehmung und Bewdltigung von Entwicklungsaufgaben im Jugendalter. Miinchen: Unpublished manuscript. Dusek, J. B., & Flaherty, J. F. (1981). The development of the self-concept during the adolescent years. Monographs of the Society for Research in Child Development, 46 (Serial No. 191). Eick, K. (1979). Untersuchung der Freiraumnutzung durch Kinder. Unpublished dissertation, Technical University Berlin. Fend, H., & Schroer, S. (1985). The formation of self-concept in the context of educational systems. International Journal of Behavioral Development, 8, 423-44. Fischer, A., Fischer, R. C , Fuchs, W., & Zinnecker, J. (1981). Jugend '81, Lebensentwurfe, Alltagskulturen, Zukunftsbilder. Hamburg: Jugendwerk der Deutschen Shell. Flay, B. R., Ryan, K. B., Best, J. A., Brown, K. S., Kersell, M. W., D'Avernas, J. R., & Zanna, M. P. (1983). Cigarette smoking: Why young people do it and ways of preventing it. In P. J. McGrath & P. Firestone (Eds.), Pediatric and adolescent behavioral medicine. New York: Springer. Fuhrer, U. (1985). Das Konzept "Behavior Setting": Uberleungen zu seiner fur Psychologie relevanten Aufbereitung. In P. Day, U. Fuhrer & U. Laucken (Eds.), Umwelt und Handeln: Okologische Anforderungen und Handeln im alltag. Tubingen: Attempto. Gibson, J. J. (1977). The theory of affordances. In R. Shaw & F. Bransford (Eds.), Perceiving, acting, and knowing. Hillsdale, NJ: Erlbaum. Grabow, S., & Salkind, N. J. (1976). The hidden structure of children's play in an urban environment. In P. Suedfeld et al. (Eds.), The behavioral basis of design (Vol. 1). Stroudsburg: Dowden, Hutchinson, & Ross. Gulotta, T., & Adams, G. R. (1982). Substance abuse minimization: Conceptualizing prevention in adolescent and youth programs. Journal of Youth and Adolescence, 11, 409-24. Harms, G., Preissing, C , & Richtermeier, A. (1985). Kinder und Jugendliche in der Grofistadt. Berlin: Fortbildungsinstitut fur die padagogische Praxis. Havighurst, R. J. (1952). Developmental tasks and education (2nd ed.). New York: Plenum Press. Holahan, C. J. (1978). Environment and behavior. New York: Plenum Press. Huba, G. J., & Bentler, P. M. (1982). A developmental theory of drug use: Derivation and assessment of a causal modeling approach. In P. B. Baltes & O. G. Brim, Jr. (Eds.), Life-span development and behavior (Vol. 4). New York: Academic Press. Huba, G. J., & Bentler, P. M. (1983). Causal models of personality, peer culture characteristics, drug use, and criminal behaviors over a five-year span. In D. Goodwin, K. Van Dusen, & S. Mednick (Eds.), Longitudinal studies of alcoholism. Boston, MA: Nijhof. Jessor, R. (1986). Adolescent problem drinking: Psychosocial aspects and developmental outcomes. In R. K. Silbereisen, K. Eyferth, & G. Rudinger (Eds.), Development as action in context: Problem behavior and normal youth development. New York: Springer. Jessor, R., & Jessor, S. L. (1977). Problem behavior and psychosocial development: A longitudinal study. NY: Academic Press. Joreskog, K. G., & Sorbom, D. (1981). LISREL V: Analysis of linear structural relationships by maximum likelihood and least squares methods. Uppsala: University of Uppsala. Johnson, L. D., O'Malley, P. N., & Bachman, J. G. (1984). Highlights from drugs and American high school students 1975-1983. Rockville, MD: National Institute for Drug Abuse. Kaminski, G. (1986). Ordnung und Variabiltdt im Alltagsgeschehen. Das Behavior Setting-Konzept in den Sozial- und Verhaltenswissenschaften. Gottingen: Hogrefe. Kandel, D. B., Kessler, R. C , & Margulies, R. (1978). Antecedents of adolescents' initiation into stages of drug use: A developmental analysis. Journal of Youth and Adolescence, 7, 13-40.

On the constructive role of problem behavior in adolescence

179

Kandel, D. B., & Logan, J. A. (1984). Patterns of drug use from adolescence to young adulthood: I. Patterns of risk for initiation, continued use, and discontinuation. American Journal of Public Health, 74, 660-6. Kaplan, H. B. (1980). Deviant behavior in defense of self. New York: Academic Press. Kaplan, H. B., Martin, S. S., & Robbins, C. (1984). Pathways to adolescent drug use: Selfderogation, peer influence, weakening of social controls, and early substance use. Journal of Health and Social Behavior, 25, 270-89. Lerner, J. V., & Lerner, R. M. (1983). Temperament in adaptation across life: Theoretical and empirical issues. In P. B. Baltes & O. G. Brim, Jr. (Eds.), Life-span development and behavior (Vol. 5). New York: Academic Press. Magnusson, D., Duner, A., & Zetterboom, G. (1975). Adjustment: A longitudinal study. New York: Wiley. Magnusson, D., Stattin, H., & Allen, V. L. (1985). Differential maturation among girls and its relation to social adjustment: A longitudinal perspective. In D. L. Featherman & R. M. Lerner (Eds.), Life-span development and behavior (Vol. 7). New York: Academic Press. Medrich, E. A., Roizen, J., Rubin, V., & Buckley, S. (1982). The serious business of growing up. Berkeley, CA: University of California Press. Muchow, M., & Muchow, H. (1935). Der Lebensraum der Grofistadtkindes. Hamburg: Martin Riegel Verlag. Newman, B. M., & Newman, P. R. (1979). Development through life. Homewood, IL: Dorsey Press. Noack, P. (1985). Analyse von Freizeitorten im Berliner Jugendlangsschnitt: Manual Beobachtung. In R. K. Silbereisen & K. Eysferth (Eds.), Berichte aus der Arbeitsgruppe TVdrop Jugendforschung, 54/85. Berlin: Technische Universitat. Noack, P., Muchowski, L., & Silbereisen, R. K. (1986). Zur Strukturierung von Entwicklungsaufgaben aus der Perspektive von Jugendlichen. Unpublished manuscript, Technical University, Berlin. O'Malley, P. M., & Bachman, J. G. (1983). Self-esteem: Change and stability between 13 and 23. Developmental Psychology, 19, 257-68. Payne, R. T., & Jones, D. R. W. (1977). Children's urban landscapes in Huntington Hills, Calgary. In P. Suedfeld et al. (Eds.), The behavioral basis of design (Vol. 2). Stroudsburg: Dowden, Hutchinson & Ross. Perry, C. L., & Jessor, R. (1985). The concept of health promotion and the prevention of adolescent drug abuse. Health Education Quarterly, 12, 169-84. Peterson, A. C , & Spiga, R. (1982). Adolescence and stress. In L. Goldberger & S. Breznitz (Eds.), Handbook of stress: Theoretical and clinical aspects. New York: The Free Press. Projektgruppe TUdrop (1984). Heroinabhdngigkeit unbetreuter Jugendlicher. Weinheim: Beltz. Schuhler, P., & Kastner, P. (1985). Analyse von Freizeitorten im Berliner Jugendlangsschnitt: Manual interview. In R. K. Silbereisen & K. Eyferth (Eds.), Berichte aus der Arbeitsgruppe TUdrop Jugendforschung, 55/85. Berlin: Technische Universitat. Schwarzer, R. (1983). Unterichtsklima als Sozialisationsbedingung fiir Selbstkonzeptentwicklung. Unterrichtswissenschaft, 2, 129-48. Silbereisen, R. K. (1986). Entwicklung als Handlung im Kontext: Entwicklungsaufgaben und Problemverhalten im Jugendalter. Zeitschrift fiir Sozialisationsforschung, 6, 29-46. Silbereisen, R. K., & Eyferth, K. (1985). Der Berliner Jugendlangsschnitt: Projekt "Jugendentwicklung und Drogen." Dritter Fortsetzungsantrag an die Deutsche Forschungsgemeinschaft. In R. K. Silbereisen & K. Eysferth (Eds.), Berichte aus der Arbeitsgruppe TUdrop Jugendforschung, 50/85. Berlin: Technische Universitat. Silbereisen, R. K., Eyferth, K. (1986). Development as action in context. In R. K. Silbereisen, K. Eysferth, & G. Rudinger (Eds.), Development as action in context: Problem behavior and normal youth development. New York: Springer.

180

Rainer K. Silbereisen and Peter Noack

Silbereisen, R. K., & Kastner, P. (1986). Jugend und Problemverhalten. Entwicklungspsychologische Perspektiven. In R. Oerter & L. Montada (Eds.), Entwicklungspsychologie. Munchen: Urban & Schwarzenberg. Silbereisen, R. K., & Noack, P. (1986). Adolescence and the environment. In D. Canter & M. Krampen (Eds.), Ethnoscapes. Aldershot: Gower. Silbereisen, R. K., Noack, P., & Eyferth, K. (1986). Place for development. In R. K. Silbereisen, K. Eyferth, & G. Rudinger (Eds.), Development as action in context: Problem behavior and normal youth development. New York: Springer. Silbereisen, R. K., Reitzle, M., & Dwyer, J. (1986). Self-esteem, leisure motives, and adolescents' substance use. Unpublished manuscript, Technische Universitat, Berlin. Silbereisen, R. K., Reitzle, M., & Zank, S. (1986). Stability and change of self-concept in adolescence: Self-knowledge and self-strategies. Paper presented at the symposium "In Memoriam Herrmann Ebbinghaus," Berlin, German Democratic Republic. Thrasher, F. M. (1927). The gang. Chicago: University of Chicago Press. Uchtenhagen, A., & Zimmer-Hofler, D. (1985). Heroinabhdngige und ihre "normalen" Altersgenossen. Bern: Haupt. Wills, T. A., & Vaughn, R. (1985). Relationship of peer and adult support to smoking and alcohol use in middle adolescence. Unpublished manuscript. Willis, P. (1977). Learning to labour. Westmead: Saxon House. Wohlwill, J. F. (1985). Martha Muchow and the life space of the urban child. Human Development, 28, 200-9.

The sociogenesis of self concepts

Robert B. Cairns and Beverly D. Cairns

Children's self ratings of such characteristics as popularity, affiliation, and aggression often bear modest relations to ratings obtained from peers and teachers, or to direct observations (e.g., Cairns & Cairns, 1984; Ledingham, Younger, Schwartzman, & Bergeron, 1982). Nonconvergence between measures from the self and those from other sources has been found as well in the assessment of 4 'masculinity-femininity" (Sears, 1965), ''conscience" (Hartshorne & May, 1928, 1929; Yarrow, Campbell, & Burton, 1968), and "dependency" (Gewirtz, 1972). The question is, Which forms of measurement should be given priority and which should be made subsidiary or discarded: assessments generated by the self, evaluations assigned by other persons, or observations made by objective observers? Perhaps the most widely accepted answer is that all of the above procedures can yield useful information on personality development. Although such methodological eclecticism permits one to get on with data collection, it inevitably generates new problems. Once the data are in, the different methods must be combined to provide a coherent view of personality (see Binet & Henri, 1895, for an astute account of the combinational problem in the context of cognitive development). Without an explicit theory to guide integration, multiple measures may compound the issue, not solve it. We propose in this chapter that (1) the solutions to the problems of measurement nonconvergence and integration are as much theoretical as they are statistical, and (2) the properties of self cognitions cannot be understood without an analysis of their developmental and evolutionary functions. On the first point, we follow the lead of investigators who have addressed issues of causal attribution (e.g., Kelley, 1973; Jones & Nisbett, 1972) and the generality of personality dispositions (e.g., Bern & Allen, 1974; Bern & Funder, 1978; Bowers, 1973; Epstein, 1980; Mischel, 1968). In both areas, progress toward solution of the issues has required simultaneous consideration of measurement issues and those of substantive theory. On the second point, we adopt a functional analysis of the properties of self concepts (following the lead of Bern, 1972; Epstein, 1973; 181

182

Robert B. Cairns and Beverly D. Cairns

Greenwald, 1980; Kendrick & Stringfield, 1980; Linville, 1987; and Markus & Nurius, 1986). Our purpose is to explore the implications of a developmental and evolutionary perspective on the social functions of self cognitions. Hence the root term in sociogenesis refers jointly to the development of the individual (ontogenesis) and development of the species (phylogenesis). Measurement nonconvergence is hardly a new problem. Hartshorne and May (1928, 1929) clearly outlined the issues at the outset of modern studies of personality development. They argued that nothing could substitute for assessments of actual conduct in controlled settings. Verbal reports were not sufficient. Yarrow, Campbell, and Burton (1968) later provided an insightful analysis of the pitfalls of child-rearing research, giving special attention to how nonconvergence among measures could obscure efforts to replicate. As Yarrow et al. (1968) observed, discrepant measures provide special problems in studies of children. Lytton (1980) has found, for example, that there is little carryover between measures of compliance and interaction obtained in the laboratory and those obtained in the home or from parental reports. Given the original focus by Hartshorne and May (1928), it is ironic that Lytton concluded that controlled laboratory observations yielded data that stood apart from nonlaboratory measures and contributed little to their interpretation. Although the problem is not new, the issues of measurement nonconvergence have grown in scope over the past 20 years. It is now recognized that significant social patterns in development are rarely determined by single events. Hence multimethod research has now become the norm in developmental studies of social behavior. But when multiple methods are employed, they often yield clusters of correlations that invite projective interpretations (Scarr, 1986; Sears, 1965). It has been assumed that multivariate statistical solutions, such as LISREL, are required to bring order into multitrait, multimethod analyses. But there are limits to purely statistical solutions. Multivariate analyses by themselves cannot correct for unclarity in original conceptualization and measurement. Indeed, one of the attractive features of the LISREL methodology is that it requires researchers to specify the theoretical relations among measures prior to the analysis, not afterward. But modern theories of social development provide few decisive guidelines on when to expect convergence and when not, and why (Cairns & Green, 1979; Harter, 1983). Two issues of theory: measurement and development The empirical literature poses two related sets of questions for current views of personality and social development. The first concerns measurement theory. For instance, in what domains should one expect self-ratings to correspond with the social consensus? Precisely what sources of variance are assessed by multi-

The sociogenesis of self concepts

183

method measures of the self, ratings of the individual by other persons, and direct observations of the individual? Are these sources the same regardless of the developmental and cognitive status of the child being rated/observed? These matters of measurement theory are not merely concerns of psychometric theory and statistical analysis; rather, they direct attention to the substantive issues of theory and the linkages that exist between different levels of assessment (Magnusson, 1985). The second set of questions involves substantive concerns of developmental theory and construction of the self system. Major theories of cognition provide few decisive guidelines on when to expect convergence and when not. On the one hand, it has been broadly assumed that hierarchic cognitive integration across successive stages in development implies increasingly higher levels in development in convergence between the concepts of the self, concepts of others about the self, and reality. This assumption has been given explicit statement by Kohlberg (1969) in primacy of cognition in self-organization. On the other hand, the correspondence assumption leaves unresolved key issues about possible autonomous functions in the development of thoughts of oneself and thoughts of others. Why should self-other correspondences arise in the course of development, and must these components be equally balanced at all points in ontogeny? Are the functions performed by self-evaluations and social cognitions qualitatively the same for children, adolescents, and adults, or do these cognitions perform different services at different developmental stages? Then there is the issue of individual differences, and how they arise at any stage. Why is it, for instance, that depressed persons are likely to have more realistic self appraisals than nondepressed persons (e.g., Greenberg & Pysczczynski, 1985)? The two sets of theoretical issues - (1) the relations between different levels of interactional measurement and (2) the development and functions of self concepts and social cognitions - seem inextricably linked. The functions of self cognitions Implicit in most recent studies of social-cognitive development is the assumption that mature and well-adjusted persons should have a coherent sense of themselves and other persons. According to this view, evaluations by persons from without should converge with the individual's own assessments from within. The correspondence assumption is in line with a common sense view of the relations between social reality and the constructions that it imposes upon persons. This view is also consistent with the neobehavioristic explanation for the emergence of self ideas, in that primacy is assigned to the environmental control of self concepts (Cooley, 1922; Mead, 1934; Bandura, 1982). But the empirical findings available indicate that self-other correspondences

184

Robert B. Cairns and Beverly D. Cairns

are not inevitable in the course of development. In this regard, the social realism embodied in the correspondence assumption is only one of the epistemological alternatives available. Another solution to this version of the body-mind problems would start with the analysis of functions during development rather than with its structure at maturity. In a different context, we have proposed that self cognitions do not always have to be veridical in order to be functional (Cairns & Cairns, 1981; 1986; see also Epstein, 1973; Greenwald, 1980; Wallwork, 1982). Even if one is old and anxious and poor, there should be reason to get up in the morning. The essential idea is that one's view of one's self may serve different functions from those served by one's view of others. In this regard, social cognitions provide at least four separate adaptive services for the individual and the society in development; namely, (1) the promotion of the individual's personal integration and well-being; (2) the facilitation of societal organization by enhancing communications with other persons; (3) the projection of the person's actions into the future through the formulation of plans, goals, and dreams; and (4) the facilitation of adaptation to the physical world. Although these functions are often mutually supportive, they serve different masters and thus are not always compatible with the views that others hold or with objective observations of behavior.

Developmental integration Of the four functions of self cognitions, the primary one is the preservation of self integration and direction. It is primary both in ontogeny and personal adaptation. The "holistic" assumption is embedded within the nuclear theories of cognitive development (Baldwin, 1897; Piaget, 1936/1952) and developmental psychobiology (Kuo, 1967; Schneirla, 1966). Internal synchrony - whether in physiology or cognition - is not something that must be socialized or learned. It is a starting point for ontogenetic accommodation and evolutionary adaptation. According to these modern theories of development, internal synchrony is present at the outset of development, though it may be challenged by changes from within and from without. Moreover, despite age-related modifications in role and capabilities, the bias toward within-system integrity and balance should remain dominant. There is one central point of unclarity in "holistic" models of development. Precisely what stuff should be integrated in development? On the one hand, Baldwin, Piaget, and others who have followed their lead have concerned themselves primarily with the organization of mental phenomena, including the organization of self ideas. On the other hand, Kuo, Schneirla, and the developmental psychobiologists have dealt primarily with the integration of physiology and

The sociogenesis of self concepts

185

social behavior in development. To the extent that mental and biological domains are kept separate, this version of the body-mind problem may be skirted by parallelism or epiphenomenalism. But if the domains are permitted to interact, inevitable problems of causation and primacy arise. For instance, if self conceptions are not in accord with the social consensus, what should be brought into line, one's perceptions or one's social reality? Furthermore, the fact that some self-conceptions resist conformity to the social consensus suggests that there must be other, internal criteria for determining their establishment beyond social consensus. That is, the standards extend beyond those of the "looking glass" of Cooley (1922). Cognitions that the person holds about the self and others may be sufficient in themselves, to the extent that these cognitions promote internal integration and direction. Social accommodation A second function of social cognitions is to facilitate adjustments to the external world of social and nonsocial events. These adjustments presuppose, among other things, the ability to rapidly and correctly predict the actions, motivations, and intentions of other persons, and reliably identify one's own social skills and deficiencies in relations with others. In addition, commonality in social perceptions is a prerequisite for communications with other persons. The natural history of human beings suggests that there should be substantial veridicality among persons in their interpersonal evaluations of other persons. The ability to distinguish between one's friends and enemies - persons one should trust and persons one should fear - is a primitive function of social cognition in everyday life. Social attributions take advantage of an evolutionarily stable strategy that appears relatively early in social-cognitive development. One's personal survival depends in large measure upon accurate recognition of persons and things that are likely to produce threat, harm, or injury. Whatever services social concepts perform, they should be reasonably helpful in correctly diagnosing situations and persons that one should avoid, and situations and persons that one can trust. In addition, commonality in social perceptions on these basic matters is a prerequisite for adequate communication and continuing affiliations with other persons (Heider, 1958). To the extent that there is a basic confirmation and balance in these basic perceptions, one might further share in the information network. Conversely, to the extent that there are discrepancies on these matters, there may be corresponding gaps in communication skills and discrepancies in information transmission. Accordingly, children and adolescents normally develop the ability to observe, communicate about, and weigh information about other persons so as to predict behaviors directed toward themselves.

186

Robert B. Cairns and Beverly D. Cairns

Plans, goals, and dreams A third function of self concepts and social cognitions can scarcely be separated from the first two, for it arises upon the disputed territory existing between them. One's formulation of plans for tomorrow and dreams for the future - what one might do in the next minute, the next day, the next month, the next year, or the next decade - presupposes concepts of one's self and concepts of one's world, including the options and constraints that exist. But plans need not be dominated by either source. Indeed, many of the dreams and goals that make one's life worth living may win only slender support in objective evaluations. Unreal concepts about the self and one's potential - including dreams, religions, and values - may figure importantly in making life worthwhile. But the admission of goals and values into self-concepts does not require that there be a direct connection between one's thoughts and actions. To the contrary, some goals could lead to disastrous consequences if put into action. Objectively unattainable dreams may promote individual consistency and coherence primarily by serving to organize other beliefs, and thereby produce stable affective states. Accordingly, slippage between values, beliefs, plans, actions, and social consensus may not be a bad thing. The buffering between ideas and action can diminish internal stress while, simultaneously, promoting plans and dreams for living. In order for values and dreams to be functional, the persons who hold them can hardly accept that they are convenient myths. There is a certain unease associated with any suggestion that some of one's own deeply felt beliefs may be adaptive fictions.

The physical world and practical reason Compared to the freedom in private beliefs about the self, there is little tolerance for idiosyncratic views in the perception of the external world, whether physical or social. Indeed, much of the content of elementary education in most societies is concerned with the establishment and enforcement of consensus rules about the "right" solutions to problems of arithmetic combination, language usage, and the formation of a standard frame of reference for conceptualizing events in the external world. Similarly, progress in intellectual discourse and scientific analysis depends on the acceptance of certain public and absolute constraints on logical rules.

Developmental realignment and challenges to convergence Within the present perspective, internal synchrony and adaptive self concepts do not automatically presuppose high levels of accuracy. But should there not be increasingly higher levels of self-other correspondence as the child grows older,

The sociogenesis of self concepts

187

more experienced, and more perceptive? Not necessarily. Although cognitive development clearly implies greater complexity, it does not imply better social adjustment, higher degrees of ethical conduct, or increased levels of self-other agreement. Developmental change is not limited to cognitive modifications. At puberty, there are age-related effects on biophysical status - including physical and hormonal growth - that are more dramatic for social adaptation than modifications in cognitive competence. Indeed, adolescence refers primarily to changes in physical and sexual capabilities, not to changes in cognition. Similarly, the major transitions in midlife are associated with biophysical changes in function and social roles, not sharp changes in cognition. In the context of the organismic perspective, it would be reasonable to expect cognitive conceptions of the self and others to be integrated with basic biosocial and sexual-related changes at each developmental stage (Cairns, 1986). Four types of ubiquitous change in childhood and adolescence may be identified; namely, changes in ability to think abstractly, changes in sexual dimorphism and motivation, changes in communicative capabilities, and changes in social role. Each of these developmental shifts has implications for one's concept of self, and its relation to the concepts that others share about him/her. Abstractness and rationalization There is an age-related increase in the cognitive ability of the individuals to extend beyond the self, to perceive and to emphathize with the motives and actions of other persons. More generally, there is an increase in the capability of adolescents and adults to employ abstract concepts in describing the functioning of other persons, their own feelings and concerns, and the social ecology. A byproduct of this cognitive advance is not only an enhanced complexity in understanding, but an age-related "advance" in the skills of persons to invent compelling explanations for the acts of others and convincing self-deceptions. Rationalization works by reconstructing the present and reorganizing the past; cognitive advances provide distance between the here-and-now concrete features of existence and the person's conceptions of himself/herself and the social world. Whatever advances for veridicality that are won by a more precise discrimination of the external world may be lost by the creation of abstractions in which the details are submerged or redefined. Biophysical changes and motivation Biophysical, emotional, and social growth occurs from infancy to adulthood. Certain changes are universal to members of the culture of which the person is a part, but variations in the rate and content of the changes are distinctive to the

188

Robert B. Cairns and Beverly D. Cairns

individual. They lead to inevitable modifications in social needs and social status. For instance, there are age-related changes in the actual structure of the social systems in which children, adolescents, and adults participate, and in the social roles they fill. Children's peer clusters are typically unisexual; by midadolescence they are heterosexual. There are changes as well in biophysical status and motives associated with these changes. One of the more obvious is the onset of sexual and romantic concerns in adolescence, and the correlated changes in the heterosexual social structure and personal relationships. Similarly, new levels of physical maturation lead to changes in social status and responsibility at several critical points in life. These stages recur from early childhood and adolescence to maturity, middle age, and senescence. Communication networks and the Pollyanna bias There are age-related increases in the extent to which one may learn what other persons think about oneself directly from others. Communication about other persons is enhanced in large measure by changes in one's cognitive capabilities, on the one hand, and freedom of movement across social groups and social relationships, on the other. To the extent that one's social cognitions are shaped by communication with others in a social network, attention must be given to one's associates and how these are related to one's developmental and social circumstances. Furthermore, "Pollyanna biases" operate at virtually all developmental stages in communications about the self. Social protocol (i.e., etiquette) ensures that most interchanges are not "rude" or provocative; hence it is not polite to mention information that may be disturbing to the other person, however true it may be. The same sanctions do not apply to information given to others about the person when they are not aware (i.e., gossip). Skills in social concealment and skills in social detection increase with age, at seemingly the same rate. Social ecology and social roles Conjoined to changes in the social communication network are certain inevitable changes in one's social role and the expectations associated with that role (Bronfenbrenner, 1979). For example, one's view of oneself as a rebellious child or incorrigible student may be incompatible with one's view of oneself as a responsible parent or instructor. Much of the content of these roles at the various stages in living is defined by the social culture, which, in turn, becomes incorporated into the individual's values and perspectives. Even the "gender constancy" of the self is open to modification (Lewis & Brooks-Gunn, 1979). Although the person's gender classification is unlikely to change - in the eyes of the self as

The sociogenesis of self concepts

189

well as the eyes of others - the attitudes and behaviors associated with those roles are likely to undergo change over the life course. One reason that a developmental increase in cognitive functioning does not lead immediately to greater veridicality across the self-other reality domains is that the same abstract capabilities that increase skills in the perception of others may be employed to counterbalance those perceptions, through the construction of rationalizations and self-enhancing fictions. Given the balance between increments in abstractness, on the one hand, and increases in rationalization, on the other, across-the-board increments in self-other agreement seem unlikely. Hence it appears that these discrepancies will not be erased with age or advances in cognitive skills. Some of the reasons for self-other discrepancies in selfevaluations in adulthood and adolescence are the same as those in earlier developmental stages: namely, my thoughts of myself serve different-purposes than my thoughts of others. This principle is not age relative, and it should hold across the developmental course. What changes over time is the ability of the self to share in aspects of the broader communication network and to empathize, generalize, and rationalize. Another reason that few age-related changes may be expected in self-other correspondences is methodological. Developmental changes demand changes in method and in operational definition. In this regard, the research methods employed in developmental studies are typically age adjusted. This means that different operations and instruments are employed with younger subjects than older ones to assess what is purportedly the same concept, whether "aggression" or "popularity." (This strategy parallels that of cognitive assessment, where sensorimotor tests for infants and verbal tests for children have traditionally been labeled measures of "intelligence," despite the fact that quite different abilities are assessed by the procedures.) Further, standard scores or other adjustments to eliminate the effects of age-related changes have been introduced to equate the findings obtained at different developmental stages. Such statistical modifications ensure that developmental changes are held constant or are otherwise eliminated before the introduction of descriptive or inferential tests. Because these procedures have now become standard in experimental design and analysis, the differences between age-developmental stages become inevitably blurred. Multiple masters and the compromising ego In the present view, there may be an ongoing tension in development between beliefs that serve to enhance personal well-being and integration, on the one hand, and beliefs that capture and precisely reflect external circumstances and social evaluations, on the other. Moreover, in the course of development, cogni-

190

Robert B. Cairns and Beverly D. Cairns

tions about the self that promote organismic well-being and integration at one stage may not be the same as those that can serve the same function at a later stage. Certain thoughts become less tenable in the course of development. For instance, beliefs by 3-year-old children in their own omnipotence or benign others (e.g., Santa Claus) could enhance feelings of control and thereby reduce fear or create hope. Those beliefs for the very young child may be socially tolerated, in that their adoption and maintenance by 3-year-olds have few negative consequences for their personal relationships. But if the same ideas are retained in adolescents, they become the stuff from which ideas of paranoid schizophrenia are diagnosed. The persistence of modestly inflated views of one's competence through adolescence and beyond can often serve to enhance both well-being and interpersonal adaptations. In this regard, thoughts of the self that are too congruent with the dominant ideas that others hold about oneself can be - in various measures disappointing, depressing, and inhibiting. All this is to say that the concepts of the self that are most functional for the person may not be the ones that are most veridical with social consensus. And the criteria for functional concepts themselves will shift across age and circumstances. There are obvious limits on self-enhancement, however. Inflated views of one's own self-competence can lead to problems of social adaptation if they (a) are in serious conflict with public beliefs on the matter, and (b) are acted upon. But other private illusions about the self can be benign or productive. These ideas could involve beliefs about one's satisfaction with one's self and one's life, one's essential goodness and friendliness, and one's prospects for immortality, either spiritual or intellectual. Even illusionary concepts about personal freedom and potency should not diminish social adaptation, provided they are not acted upon and submitted to public rebuff (Bandura, 1982). Nor is it necessarily disadvantageous to underestimate one's own potential, especially in public statements. Such modesty can serve to buffer the individual from even greater disappointment and distress, particularly in those areas that are both public and central to one's self-organization. The young Swedish tennis star, Mats Wilander (then ranked third in the world) illustrates the point. Interviewed just before the 1985 Davis Cup match between Sweden and West Germany, Wilander said, "I never expect to win. The American attitude helps you to be more disappointed if you lose. You can't think that way because then you don't enjoy your tennis; you only enjoy winning. We want to be the best, too, but we don't tell the public that we expect to win things. It is the Swedish way" (News and Observer, 20 December 1985). The cognitive constructions of the self must simultaneously serve multiple masters. The relative weight assigned to these forces necessarily changes during ontogeny because of changes in personal abilities and changes in external con-

The sociogenesis of self concepts

191

straints, roles, and information. Nonetheless, priority should be given at all developmental stages to the attainment and maintenance of personal integration and self-coherence.

Relevant longitudinal findings In 1981, we initiated a multimethod, longitudinal study over childhood and adolescence (Cairns & Cairns, 1984; Cairns, Cairns, Neckerman, Ferguson, & Gariepy, 1988). This research involved longitudinally tracking two cohorts of subjects, one selected in the 4th grade and another in the 7th grade. In the most recent assessment, the younger cohort was in the 9th grade (i.e., 6th annual wave of testing) and the older cohort was in the 11th grade (i.e., 5th annual wave of testing). The combined sample was N = 695 at the beginning of the study (N = 364 girls and 331 boys). The subjects were tested annually. Over 96% of the original subjects were individually tested and interviewed in the last test wave (668/695). The first problem was to determine when specific measures would be useful, and when they would not. Yarrow et al. (1968) had cogently observed that efforts to replicate would be stymied if the measures employed in the initial and replication studies were themselves discrepant. This problem was controlled in the present work by using objectively identical measures to obtain year-to-year information from the subjects themselves and from others. We were guided by the assumption that some dimensions of self-evaluation private, abstract - should be relatively available to achieve internal balance and stability. On the other hand, public, concrete dimensions should be relatively more constrained by objective events and social consensus. The broader assumption is that all self-evaluations - public or private; concrete or abstract; younger or older - are primarily constructive and adaptive, not simply a mirror of an objective reality or social consensus. In this view, deviations from social consensus often reflect the operations of the compromising ego. The individual's self evaluations are seen as dynamic and change worthy, constantly adapting to developmental changes in internal and external frames of reference. Normal development should involve multiple deviations from the social consensus and reality that are constructive and adaptive. Information on the adequacy of this view of self construction comes from four sources in our data: (1) divergence between self ratings and other, nonself ratings, (2) developmental consistencies of self ratings, (3) idiographic balance and compensation within self ratings, and (4) convergence between the self and others on constrained phenomena (i.e., highly public events, concrete acts, and salient dimensions).

192

Robert B. Cairns and Beverly D. Cairns

Divergence in self ratings Perhaps the most curious feature of self ratings has been their lack of stable individual-difference convergence with information from other methods. Our results repeatedly illustrate the point. Even when the measurement instruments are objectively identical and when children and adolescents rate themselves on the same scales employed by others, low levels of self-other convergence are observed on most dimensions (Table 7.1). The dimensions were not equally divergent: self-assessments of academic skills usually show the highest levels of convergence with teacher ratings, self-assessments of aggressive behavior are next in rank, and self-assessments of popularity, affiliation, and "All-American" factors (i.e., appearance, sports, winning) are the lowest. A scan down the rows of the table indicates reasonable consistency across the years and across the cohorts studied. The summary mean correlations at the bottom of the table appear to be representative of the several samples considered separately. One explanation for the relatively poor agreement between individual differences in self ratings and those of other persons outside the self is that both types of measure are psychometrically inadequate (i.e., have low reliability and/or validity). Although each year's factor scores are based on multiple items, the children assessed themselves on only one occasion each year. Similarly, teacher ratings and peer nominations were obtained on only one occasion. If the ratings of the self are summarized over all years of testing, more stable "self" ratings should be obtained than if a single year is employed. The higher reliability may be achieved even though there are few changes in the level of the child's adaptation and performance from late childhood through adolescence. Similarly, aggregated teacher ratings should capture some shared sources of variance, even though actual changes have occurred. Such aggregated correlations indicated consistent increases on all factors measured. For example, in the younger cohort, the median single-year self-other correlation for aggression was r — .32 and the aggregated self-other correlation was r = .47. The aggregation spanned six years (from the fourth to ninth grades). In the older cohort, the gains through aggregation were less impressive. The median single-year self-other correlation for aggression was r = .26, and the aggregated correlation was r = .28. The outcome is consistent with the expectations of Epstein (1980) and Block (1977). Another proposal for the generally low levels of convergence of self-other ratings is that the children had not yet achieved an adequate level of abstract reasoning to permit them to share in the adult evaluations of their dispositions. On this argument, children in the operational stage (i.e., preadolescence) would function according to relatively nonabstract dimensions, and their self-judgments would necessarily be flawed. By adolescence, with the onset of formal operations, more adequate levels of agreement should be achieved. As shown in Table

The sociogenesis of self concepts

193

Table 7.1. Self-other congruence correlations between ICS-S (self) and ICS-T (other) factor scores over six annual tests (Cohort I) Grades 4

5

6

7

8

9

Aggregated (4-9)

Aggression Girls Boys Mean

.33* .34 .34

.22 .37 .29

.32 .39 .35

.23 .34 .28

.30 .24 .27

.37 .44 .41

.41 .53 .47

Popularity Girls Boys Mean

.17 .02 .10

.36 .13 .25

.10 .37 .23

.21 .34 .27

.40 .23 .33

-.02 .34 .16

.24 .39 .31

Academic Girls Boys Mean

.33 .17 .25

.34 .24 .29

.46 .50 .48

.35 .47 .41

.46 .33 .41

.20 .52 .28

.44 .59 .51

Ail-American Girls Boys Mean

.03 .04 .03

.21 .17 .19

.18 .31 .24

.07 .19 .13

.12 .10 .11

.13 .14 .14

.23 .28 .25

Social competence Girls Boys Mean

.21 .09 .16

.27 .32 .29

.09 .44 .25

.25 .36 .30

.27 .21 .24

.15 .46 .31

.20 .48 .33

a

Product-moment correlations from subject and teacher ratings of Cohort I (N = 220; 116 girls and 104 boys at fourth grade). Owing to school dropout and other factors, teacher ratings were available on 197 subjects at the ninth grade. Italicized correlations are not statistically significant; all other correlations are reliable at p<.05 or greater. The mean correlations were obtained by Fisher Z transformations; the aggregated self-other correlations reflect aggregated raw scores over six years.

7.1, there was little support for the proposal. There was no obvious age-related convergence on any of the individual-difference dimensions of self-other assessment. There is scant support in these data for the idea that there is a general developmental progression toward convergence of self-other ratings. Developmental stability and enhancement of the self Despite their independence from other information sources, global self ratings are relatively stable within individuals over time. When subjects were assessed after one, two, or three years, regardless of their age on the first assessment,

194

Robert B. Cairns and Beverly D. Cairns

predictable individual differences were identified in the self ratings. The median predictive correlations after one year range from r = .47 (aggressiveness) to r = .63 (popularity). When the same assessments are made six years later, the predictive correlations diminish to r = .18 (aggressiveness) and r = .30 (popularity). Self ratings predict future self ratings, but little else (Cairns & Cairns, 1984). Such predictive validity implies, among other things, significant selfmeasure reliability and internal constraints on future assessments. Beyond individual differences, the general developmental picture is that adolescents see themselves as better than average. They also describe themselves as becoming more popular, less aggressive, and unchanged in academic performance over the years. Their teachers, for the most part, agree on the age trends but not on age levels. The mean discrepancies in self ratings vs. other ratings show a consistent discrepancy toward self-enhancement for both males and females, for all ages and grades, and for both longitudinal cohorts. Children and adolescents consistently see themselves as being more academically talented and more popular than their teachers see them. One apparent departure from self-enhancement may be noted. Children at most age levels see themselves as being more aggressive than do their teachers. (The exception is preadolescent boys, who give themselves high aggressive ratings, and earn them from their teachers.) This discrepancy with self-enhancement may simply reflect a faulty view of the value of aggressiveness, in that adolescents may have a more positive view of assertiveness than do their teachers (on the "brutalization norm" of adolescence, see Ferguson & Rule, 1980; Cairns & Cairns, 1986). Hence the interpretation of what is self-enhancement and what is not can be confounded with different values held by the self and others on what level of aggression is optimal. Ironically, adolescents do not see themselves as improving in those academic areas where significant changes doubtless occur (i.e., math, spelling). Nor do their teachers. It seems likely that ninth-grade students - even dull ones - can spell better and multiply more accurately than they did in the fourth grade. The apparent "stability" in achievement ratings may arise because the raters - both child and adult - judge the child's performance relative to same-age, same-grade peers. The powerful effects of educational advance and developmental change are submerged by the shifting standards employed by the evaluators, both the self and the teacher. Beyond the "looking glass": idiographic balance There is evidence in the present work to support the proposition that self-evaluations can be compensatory and self-balancing. One relevant finding is counterintuitive in that adolescents see their own basic academic skills to be independent

The sociogenesis of self concepts

195

of each other. Self-reported performance in math is uncorrelated with selfreported performance in spelling. For example, in the older cohort at the ninth grade, the correlation between these items was null for both boys and girls. But the teachers see things differently. The same items were highly correlated for teachers who rated the same students. The items were correlated for both males and females. These findings replicated in each of the other years: In all grades, students themselves did not see their own areas of academic competence to be correlated, whereas their teachers did. Self-perceived deficiencies in one academic domain quantitative or linguistic - were unrelated to strengths in another. But would not true compensation require a negative relationship rather than a negligible one? One possible explanation is that the self-correlations were not negatively correlated because constructive compensation could occur in domains other than academics (e.g., sports, appearance, popularity). For example, academics may be of little significance for some adolescents, whereas doing well in sports (baseball, soccer, or cross-country) or being attractive could be considered critical. What would be general across persons is the tendency to balance and equate self-perceived strengths and weaknesses, with an egocentric compromise biased toward strengths. Social consensus and reality So far we have addressed only the relations between self ratings and those of adults who presumably know them well. Which information sources - self ratings or teacher ratings - agree with other, independent measures, such as the peer social consensus and/or with direct observations? To answer this question, we examined (a) year-to-year agreement among independent persons outside the self and (b) contemporaneous agreement among the several sources of information. Independent adult consensus Despite the several sources of error variance in teacher ratings, empirical findings support the idea that interpersonal judgments can be made reliably and with tolerable accuracy by teachers. The convergences appear among different teachers who offer independent ratings at the same time and among teachers who evaluate children at widely separated intervals of time and settings. The closer the relations between the two assessments in time, measurement technique, and setting, the greater the relation between the two (Cairns et al., 1988). Accordingly, two or three teachers who offer ratings on the same dimension in the same classes in the same year show high correspondence. If the ratings are separated

196

Robert B. Cairns and Beverly D. Cairns

both in time and place - as by intervals of one, two, four, or five years and in different schools - the relations are diminished but remain strong for the most salient public dimensions (i.e., academic performance and aggressive behavior). Peer nominations and direct observations Less robust but reliable correlations are usually obtained across measurement instruments, such as between teacher ratings and peer nominations or direct observations. More erosion occurs when there are differences both in measurement domains and in time/place of assessment (Cairns & Cairns, 1984). As in the case of the teacher ratings, peer nominations and behavioral observations are poorly correlated with the child's global self ratings. To the extent that teachers and peers are members of the same social system, they typically share access to a general social communication network. Via this network, information about the individual's ''reputation" is gained, and this information may further bias the evaluation. But ratings obtained from outside the social-communication system - such as from parents or other individuals in different microsocial networks - may be expected to show diminished convergence when compared with those obtained by members within a common communication system. To what extent do the social consensus descriptions of the individual go beyond reputation to correspond to the individual's observable interchanges? Rather close, if the occurrence of instances of interchanges - especially positive interchanges - are considered (e.g., Cairns & Cairns, 1984; Green, Beck, Forehand, & Vosk, 1980; Kazdin, Esveldt-Dawson, & Loar, 1983). Correspondence between social consensus and social reality is consistent with the proposition that certain public classes of ' 'modal'' social perceptions are based, in part, on actual behavior. Conditions of convergence It would be a mistake to dismiss the social evaluations obtained from persons about themselves as meaningless, despite the inconsistent relations between self ratings and those from other sources. The issue is not whether self-reports would converge with social reality and the reports of others, but when they would converge. Private selves: public persons One of the limits on constructions about the self would be the nature of the available information to individuals about themselves. If information is public,

The sociogenesis of self concepts

197

salient, and concrete, and the person shares in the communication network, there should be greater agreement than when the information is private, hidden, and abstruse. The agreement arises in public and concrete matters, in part, because there is less freedom for the personal information to be employed by the self in creative construction. On the other hand, the more distant the personal information is from public scrutiny and correction, the greater should be the variability in within-person compensation and reconstruction. Using informal criteria for judging the publicness, concreteness, and salience of personally relevant information, we found that the proposition was useful in ordering much of our data. For example, self-other correlations in the evaluation of scholastic skills and aggressive behavior were always reliable and typically higher than correlations in other domains (Table 7.1). This outcome was in line with our expectation that these areas would be more accessible and public through scholastic grades and behavioral reprimands - than other areas of social and emotional functioning. Even within the "aggressive" factor, certain items show higher self-other agreement than others. Accordingly, the item that persistently carries the weight of the self-other agreement on the aggression factor is one that involves public reprimands; namely, "gets in trouble at school." Presumably the authority (i.e., teacher/administrator) who makes the "other" rating and the adolescent (i.e., the "troublemaker") who offers the self-evaluation are both constrained in their ratings by the public, concrete nature of the referent events. A step-forward multiple regression analysis confirms that the "trouble" item accounts for most of the shared variance in self-other agreement on the aggressive factor. Social cognition and social maps From the foregoing, a special case of consensus among persons should appear in their concrete descriptions of the microsocial structure in which they participate. This means, among other things, that persons within a given microsocial network should be able to describe most of its salient properties, including clusters of persons within that network and relative roles of those persons. Children and adolescents who are embedded in relatively small microsystems are usually quite accurate in describing the social structures in that microsystem. Moreover, high levels of agreement have been obtained in the descriptions of the social structure by individuals within a given classroom or setting, even when the individual is not part of a given clique or group. By way of example, seventhgrade students were asked to describe the social groups that existed within their classroom (i.e., "Do any people in your class hang around together?"). High levels of agreement have been obtained in reports by adolescents on which persons are included in particular groups, and which persons are excluded

198

Robert B. Cairns and Beverly D. Cairns

by groups altogether (Caims, Perrin, & Cairns, 1985). The individual's placement of himself/herself in the social structure itself seemed to be reasonably consistent with the judgment of others, except for an egocentric bias to name themselves and their own groups first. Beyond this bias, the degree of consensus agreement was great, regardless of the identity of the other person. The task describing the "social map" of the classroom proved to be an easy one for the subjects, even in the absence of recall supports (such as photographs of the class members or lists of names). More broadly, the ability of children and adolescents to construct social maps illustrates a remarkable skill for perceiving, remembering, accumulating, and organizing concrete social information. These phenomenalistic social maps correspond, to a significant extent, to the frequency and types of interchanges that may be observed (Cairns et al., 1985). Our results on the effects of publicness-concreteness-salience on self-other agreement are consistent with the work of Herjanic and Reich (1982) in their attempts to develop a useful psychiatric interview for children aged 6 to 16. In work with 307 subjects and their mothers, Herjanic and Reich (1982) found that the highest agreement in mother-child pairs occurred on questions concerning symptoms that were concrete, observable, severe, and unambiguous. There was least agreement when mothers and children made judgments on the severity of the problem, and on general evaluations of adaptation (see also Kazdin et al., 1983). Beyond the clinic, there is a similar basic relationship between selfevaluations of medical student competence and judgments by preceptors in the years of medical school (Arnold, Willoughby, & Calkins, 1983). The more concrete the task and the more public the performance, the greater the agreement between reports of the self and the evaluations of other persons and observers. Concluding comment Recent multimethod investigations in developmental psychology have tended to blur distinctions between self conceptions, social attributions, and observed behaviors. The essential methodological message of this essay is that investigators should report and analyze separately the outcomes of different measurement operations of the "same" characteristic. It is the "same" characteristic if and only if the measurement operations yield convergent results. Often they do not especially when self-attributions are involved. To the extent that the measures do not participate in the same nomological net, the assumed "sameness" may only perpetuate the illusions created by self-evident theories of personality or common sense. The present proposal also raises questions about the utility of therapeutic strategies that are designed to help children and adolescents achieve more realistic global self concepts. If one is really intelligent, good-looking, influential,

The sociogenesis of self concepts

199

respected, wise, and sensitive, it may be all to the better to share in the social consensus in the design of the self-concept. But what if persons fall beneath the normative standard on desirable characteristics - as do 50% of the population? To accept automatically consensus social judgments for self-evaluations could be tantamount to judging one's self unworthy. For those who are autistic or garrulous, shy or aggressive, schizoid or melancholy, or below the average of one's reference group on any dimension of competence, a "looking glass" self could be counteradaptive. The self concept need not be veridical in order to be serviceable. Beyond epistemological issues on the self and social cognition, the perspective suggests an alternative approach to the pragmatics of multimethod design and analysis. Among other things, concerns about the clustering and coherence of measures must be raised at each stage of the investigation - design, operation, analysis, interpretation - not merely at its end. Reliance on statistical techniques, such as multivariate analyses or structural modeling, cannot substitute for a priori theoretical analyses of the measures and their structure. By clearly specifying the domains and limitations of measurement beforehand, greater precision may be achieved in understanding replications and failures to replicate. Moreover, as Magnusson (1987) has insightfully argued, each measurement device should be considered in terms of its own level, and not automatically assumed to apply to other levels of analysis. Magnusson presents a persuasive case for the need for integrative approaches where the person is the focus of integration, not the variable or measure. The beguiling finding is that self reports can be closely related to the reports of others or direct observations - or self reports can be virtually independent of them. These outcomes become less puzzling when explicit attention is given to the concreteness and publicness of the characteristics that are assessed, on the one hand, and the likely role of self-integration and adaptation, on the other. On this count, the "fictions" on life satisfaction of the poor, widowed, and aged black women become more understandable (Mills & Cairns, 1987). Their subjective dimensions for evaluation would seem to be different from those employed by the "social consensus" or "objective" observers. And such selfevaluations are not costly to the person in terms of social sanctions. This model of the compromising self incorporates the fact that there are often constraints on perceptions of other persons and in conceptions of the self. These constraints may be embodied in one type of measurement device (e.g., concrete self reports), but absent in others (e.g., global ratings of the self). Finally, one's concept of one's self is at once a cognitive abstraction and a subjective, palpable reality at the core of one's sense of being. Counterintuitive claims about its functions or its roles are likely to be resisted. This includes our contention that self concepts need not be veridical or regulatory in order to be

200

Robert B. Cairns and Beverly D. Cairns

serviceable. Moreover, the self is removed by time and by abstraction from the concrete features of experience that gave rise to it and to which it may be addressed. This distance permits, over the short run, the myth of in variance and the stability of the self. Over the long run, the contents of self-evaluations are clearly dynamic and susceptible to change. It is probably beyond the ken of the self to know fully its own determinants - otherwise its illusions of primacy may be compromised. Accordingly, empirical questions concerning self functions and determinants might productively be divorced from existential questions on one's own sense of being. But this does not mean that the self is beyond scientific scrutiny. Once the operations and goals of different measurements are delineated, self-attributions may prove to be no more elusive than are observed behaviors or social attributions. References Arnold, L., Willoughby, T. L., & Calkins, V. (1983). Self evaluation in undergraduate medical education: A longitudinal approach. Research in Medical Education: 1983, Proceedings of the 22nd Annual Conference, 172-7. Baldwin, J. M. (1897). Social and ethical interpretations in mental development: A study in social psychology. New York: Macmillan. Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37, 122— 47. Bern, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 6). New York: Academic Press. Bern, D. J., & Allen, A. (1974). On predicting some of the people some of the time: The search for cross-situational consistencies in behavior. Psychological Review, 81, 506-20. Bern, D. J., & Funder, D. C. (1978). Predicting more of the people more of the time: Assessing the personality of situations. Psychological Review, 85, 485-501. Binet, A., & Henri, V. (1895). La psychologie individuelle. L'Annee Psychologie, 2, 411-65. Block, J. (1977). Recognizing the coherencce of personality. In D. Manusson & N. D. Endler (Eds.), Personality at the crossroads: Current issues in interactional psychology. Hillsdale, NJ: Erlbaum. Bowers, K. S. (1973). Situationism in psychology: An analysis and a critique. Psychological Review, 80, 307-36. Bronfenbrenner, U. (1979). The ecology of human development: Experiments by natural design. Cambridge, MA: Harvard University Press. Cairns, R. B. (1986). A contemporary perspective on social development. In P. S. Strain, M. J. Guralnick, & H. M. Walker (Eds.), Children's social behavior: Development, assessment, and modification. New York: Academic Press. Cairns, R. B., & Cairns, B. D. (1981). Self-reflections: An essay and commentary on "Social cognition and the acquisition of self." Developmental Review, 1, 171-80. Cairns, R. B., & Cairns, B. D. (1984). Predicting aggressive patterns in girls and boys: A developmental study. Aggressive Behavior, 11, 227-42. Cairns, R. B., & Cairns, B. D. (1986). The developmental-interactional view of social behavior: Four issues of adolescent aggression. In D. Olweus, J. Block, & M. Radke-Yarrow (Eds.), Development of antisocial and prosocial behavior: Theories, research, and issues. New York: Academic Press.

The sociogenesis of self concepts

201

Cairns, R. B., Cairns, B. D., Neckerman, H. J., Ferguson, L. L., & Gaviepy, J.-L. (1988). Growth and aggression. I. Childhood to early adolescence. Developmental Psychology. Cairns, R. B., & Green, J. A. (1979). How to assess personality and social patterns: Ratings or observations? In R. B. Cairns (Ed.), The analysis of social interaction: Methods, issues, and illustrations. Hillsdale, NJ: Erlbaum. Cairns, R. B., Perrin, J. E., & Cairns, B. D. (1985). Social structure and social cognition in early adolescence: Affiliative patterns. Journal of Early Adolescence, 5, 339-55. Cooley, C. H. (1922). Human nature and the social order (rev. ed.). Glencoe, IL: Free Press. Epstein, S. (1973). The self-concept revisited: Or a theory of a theory. American Psychologist, 28, 404-16. Epstein, S. (1980). The stability of behavior: II. Implications for psychological research. American Psychologist, 35, 790-806. Ferguson, T. J., & Rule, B. G. (1980). Effects of inferential set, outcome severity, and basis of responsibility on children's evaluation of aggressive acts. Developmental Psychology, 16, 141-6. Gewirtz, J. L. (1972). Attachment and dependency. Washington: V. H. Winston. Green, K. D., Beck, S. J., Forehand, R., & Vosk, B. (1980). Validity of teacher nominations of child behavior problems. Journal of Abnormal Child Psychology, 8, 397-404. Greenberg, J., & Pyszczynski, T. (1985). Compensatory self-inflation: A response to the threat to self-regard of public failure. Journal of Personality and Social Psychology, 49, 273-80. Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision of personal history. American Psychologist, 35, 603-18. Harter, S. (1983). Developmental perspective on the self-system. In P. H. Mussen (Gen. Ed.) and M. Hetherington (Ed.), Handbook of child psychology (Vol. 4, 4th ed.). New York: Wiley. Hartshorne, H., & May, M. A. (1928). Studies in the nature of character. Vol. 1: Studies in deceit. New York: Macmillan. Hartshorne, H., & May, M. A. (1929). Studies in the nature of character. Vol. 2: Studies in service and self control. New York: Macmillan. Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley. Herjanic, B., & Reich, W. (1982). Development of a structured psychiatric interview for children: Agreement between child and parent on individual symptoms. Journal of Abnormal Child Psychology, 10, 307-24. Jones, E. E., & Nisbett, R. E. (1972). The actor and the observer: Divergent perceptions of the causes of behavior. In J. Thibaut, J. Spence, & R. Carson (Eds.), Contemporary topics in social psychology. Morristown, NJ: General Learning Press. Kazdin, A. E., Esveldt-Daswon, K., & Loar, L. L. (1983). Correspondence of teacher ratings and direct observations of classroom behavior of psychiatric inpatient children. Journal of Abnormal Child Psychology, 11, 549-64. Kelley, H. (1973). The processes of causal attribution. American Psychologist, 28, 107-28. Kendrick, D. T., & Stringfield, D. O. (1980). Personality traits and the eye of the beholder: Crossing some traditional philosophical boundaries in the search for consistency in all the people. Psychological Review, 87, 88-104. Kohlberg, L. (1969). Stage and sequence: The cognitive-developmental approach to socialization. In D. A. Goslin (Ed.), Handbook of socialization theory and research. Chicago: Rand McNally. Kuo, Z.-Y. (1967). The dynamics of behavioral development. New York: Random House. Ledingham, J. E., Younger, A., Schwartzman, A., & Bergeron, G. (1982). Agreement among teacher, peer, and self-ratings of children's aggression, withdrawal, and likeability. Journal of Abnormal Child Psychology, 10, 363-72. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. New York: Plenum.

202

Robert B. Cairns and Beverly D. Cairns

Linville, P. W. (1987). Self-complexity as a cognitive buffer against stress-related illness and depression. Journal of Personality and Social Psychology, 52, 663-76. Lytton, H. (1980). Parent-child interaction: The socialization process observed in twin and singleton families. New York: Plenum Press. Magnusson, D. (1985). Implications of an interactional paradigm for research on human development. International Journal of Behavioral Development, 8, 115-37. Magnusson, D. (1987). Individual development from an interactional perspective. Hillsdale, NJ: Lawrence Erlbaum Associates. Markus, H., & Nurius, H. (1986). Possible selves. American Psychologist, 41, 954-69. Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago Press. Mills-Byrd, L., & Cairns, R. B. (1987). Life satisfaction in elderly, poor, widowed, black women. Unpublished manuscript. Mischel, W. (1968). Personality and assessment. New York: Wiley, 1968. Piaget, J. (1952). The origins of intelligence in children (M. Cook, Trans.). New York: International Universities Press. (Originally published, 1936). Scarr, S. (1986). Cultural lenses on mothers and children. In L. Friedrich-Cofer (Ed.), Human nature and public policy: Scientific views of women, children, and families. New York: Praeger. Schneirla, T. C. (1966). Behavioral development and comparative psychology. Quarterly Review of Biology, 41, 283-302. Sears, R. R. (1965). Development of gender role. In F. A. Beach (Ed.), Sex and behavior. New York: Wiley. Wallwork, E. (1982). Religious development. In J. M. Broughton & D. J. Freeman-Moir (Eds.), The cognitive developmental psychology of James Mark Baldwin: Current theory and research in genetic epistemology. Norwood, NJ: Ablex. Yarrow, M. R., Campbell, J. D., & Burton, R. V. (1968). Child rearing: An inquiry in research and methods. San Francisco: Jossey-Bass.

8

Putting persons back into the context

Daryl J. Bern

The assertion that behavior is a function of the person and the environment must surely now rate as our most popular homily - outranking even such favorites as "it will never heal if you pick at it." But it is clear that the contributors to this volume have managed to raise this injunction above the level of cliche by attending seriously to the detailed specification of the environmental factors that have been neglected for so long. As a veteran of social psychology - the field that claims the copyright for "Behavior is a Function of . . . " - and as a newcomer to the field of human development, I am impressed. As an aspiring personologist, however, I was struck by Urie Bronfenbrenner's observation (this volume) that it is "the differentiating characteristics of the individual that are most often left unspecified" in contemporary developmental psychology and that "prevailing research paradigms considerably underestimate the role of the person's own psychological characteristics in affecting the course of his or her future development." Apparently interactional approaches to development - like contemporary "interactional" psychology in general - are purging persons in their pursuit of context. I have decided to heal this lacuna by picking at it. In particular, I present here a commentary on research paradigms in personality that complements Bronfenbrenner's commentary on research paradigms in developmental psychology. I end by pontificating on the "correct" strategies for putting persons back into the context. Research paradigms in personality In its most general terms, our shared scientific task is to abstract the regularities of human conduct from its instances and to explicate the principles that underlie those regularities. In the domain of personality, we seek to convert observations of particular persons behaving in particular ways in particular contexts into assertions that certain kinds of persons will behave in certain kinds of ways in 203

204

Daryl J. Bern

certain kinds of contexts. In the domain of human development, we seek to assert that certain kinds of persons will develop in certain kinds of ways in interaction with certain kinds of environments. In both domains, we strive to construct triple typologies or equivalence classes - of persons, of behaviors, and of contexts and to fashion theories that relate these equivalence classes to one another (D. J. Bern, 1983a). Nomothetic individual-differences approaches to personality: certain persons I certain behaviors! all contexts The traditional individual-differences approaches to personality - psychodynamic, personological, and trait/type - have generally focused on the first two equivalence classes of the triple typology to the virtual exclusion of the third. For each equivalence class of persons (e.g., anal personalities, conscientious persons), there is a corresponding equivalence class of behaviors (e.g., thrift, punctuality, fastidiousness) that are postulated to be phenotypic expressions of the genotype identified by the dispositional descriptor. But there is no explicitly defined corresponding equivalence class of contexts that specifies, delimits, or qualifies the genotype's domain of applicability. At the risk of caricature, these approaches might be described as asserting that certain kinds of persons will behave in certain kinds of ways in all contexts, and it is this feature that provokes recurring criticism (e.g., Lehmann & Witty, 1934; Mischel, 1968; Peterson, 1968; Vernon, 1964). Specifically, the critics charge that no pair of person/behavior equivalence classes has ever been shown to instantiate such situationally unmodified assertions; no cross-situational consistency of a nontrivial kind has ever been demonstrated in the personality domain. One response to this challenge has been to categorize persons into more finely differentiated equivalence classes. If conscientious persons do not behave conscientiously across all contexts, then perhaps certain subsets of conscientious persons will. Perhaps low self-monitors (Snyder, 1983), those with dispositional or induced self-focus (Gibbons, 1983), or those who tell us they are consistent or observable (Bern & Allen, 1974; Cheek, 1982; Kenrick & Stringfield, 1980) will behave in certain consistent ways across all contexts. This is the so-called moderating variable strategy (D. J. Bern, 1972; Kenrick & Dantchik, 1983). Idiographiclmorphogenic approaches to personality: this person!certain behaviors!all contexts When taken to its limit, the moderating variable strategy becomes completely idiographic, one person per equivalence class, and propositions about the triple typology reduce to assertions that this person will behave in certain ways in all

Putting persons back into the context

205

contexts. Personality psychology reduces to biography. But an idiographic orientation to personality represents a more radical departure from the traditional approaches than just the reductio ad absurdum of the moderating variable strategy. In particular, advocates of an idiographic approach to personality reject the traditional nomothetic assumption that a common set of equivalence classes of behaviors can be used to characterize all persons (Allport, 1937; Bern & Allen, 1974). Thus Allport questioned Hartshorne and May's (1928) conclusion that the children they observed in their study of character were not consistently honest or dishonest across situations by challenging the assumption that an honesty-dishonesty dimension could be used to characterize all of the children in the sample and that differences among them could be specified solely in terms of their degree of honesty. The low correlations ''prove only that children are not consistent in the same way, not that they are inconsistent with themselves" (1937, p. 250). For example, lying and stealing might have been uncorrelated in the study because one child may lie in order to avoid hurting the feelings of the teacher, whereas another may steal pennies in order to buy social acceptance from his or her peers. The first child might be consistently empathic and sensitive across situations; the second child consistently insecure about acceptance in many contexts. But for neither of these two children do the behaviors of lying and stealing belong in a common equivalence class representing an honesty-dishonesty genotype, an equivalence class that exists only in the head of the investigator, not in the behavior of the children. Thus the traditional nomothetic approach to personality will prompt the conclusion that a sample of individuals is inconsistent to the degree that their behaviors do not sort into the equivalence class that the investigator necessarily imposes when he or she selects the behaviors and situations to sample. The traditional inference of inconsistency is therefore not an inference about individuals, but a statement about a disagreement between a group of individuals and an investigator over which behaviors and which situations may properly be classified into common equivalence classes (Bern & Allen, 1974). There are actually two distinctions between traditional approaches and Allport's approach that are often confounded. The first contrasts nomothetic and idiographic assumptions about personality: Nomothetic approaches assume that a common set of descriptors, dispositions, or trait dimensions can be used to characterize all persons and that individual differences are to be identified with different locations on those dimensions (or different points in an ^-dimensional factor space); in contrast, the pure idiographic approach assumes that a different set of descriptors is needed for each person. The second distinction contrasts a variable-centered orientation with a personcentered orientation. The first is concerned with the relative standing of persons

206

Daryl J. Bern

across variables; the second is concerned with the salience and configuration of variables within the person. Allport (1962) clarified the difference between these two distinctions when he suggested that his earlier term idiographic be replaced by the term morphogenic - to signify that he was dealing with the intraindividual patterning of variables. Note that one can be morphogenic or person centered while still assuming, nomothetically, that there is a common set of descriptors for all persons. This is the approach commonly used in clinical assessment. For example, the MMPI defines a common set of scales to be applied to everyone, but the interpretive codebooks classify an individual only in terms of the particular subset of scales on which he or she is elevated and, within that subset, according to their configuration. Note that this approach explicitly produces a typology of persons; it constitutes a type rather than a trait approach to personality. Process approaches to personality: all persons! certain behaviors I certain contexts The moderating-variable, idiographic, and morphogenic approaches to personality provide one kind of response to the criticism that there has never been a nontrivial instantiation of the assertion that certain persons behave in certain ways in all contexts. But they do not address the more radical version of the critique, namely, that no person ever behaves in certain ways across all contexts, that cross-situational consistency of a nontrivial kind will never be demonstrated because cross-situational variability reflects our natural state of grace (Mischel, 1968). If one holds this more radical position, then one cannot finesse the problem by simply devoting more effort to the equivalence classes of persons and behaviors, but must seriously address the equivalence classes of contexts which brings us to the orientation that is at once the most nomothetic and the most idiographic, situationism. In terms of this discussion, situationism can actually be considered a special case of the more general orientation to human behavior adopted by most of experimental psychology, social psychology and, according to Bronfenbrenner, contemporary developmental psychology, the orientation that eschews individual differences altogether - or buries them in the error term - and collapses the triple typology to a double one: All persons behave in certain ways in certain contexts. It is in this sense that the orientation can be considered the most nomothetic position. Behaviorism, of course, provides the prototype. All persons - even all organisms - are placed into a single equivalence class; "stimuli" constitute the equivalence classes of situations; and "responses" constitute the equivalence classes of behaviors. The nomothetic "principles of behavior" relate these two

Putting persons back into the context

207

classes to one another (e.g., reinforcing stimuli increase the probability of operant responses that precede them). Theoretical debates within behaviorism have often focused explicitly on the characteristic properties of the equivalence classes (e.g., Are reinforcing stimuli a subset or a superset of tension reducers? Are operant responses in the same equivalence class as classically conditioned responses?), and metatheoretical debates have often focused on the conceptual nature of these equivalence classes (e.g., Skinner's 1935 article, 'The Generic Nature of the Concepts of Stimulus and Response"). But precisely because situationism makes no attempt to categorize persons into equivalence classes, it is also the most idiographic when it actually comes to understanding or predicting the particular case. Because there are no classes or kinds of persons, there are only persons in general or this person in particular. Thus, in speaking about the modern incarnation of behaviorism, Mischel notes that "social behavior theory does not label the individual with generalized trait terms. . . . Behavioral assessment involves an exploration of the unique or idiographic aspects of the single case, perhaps to a greater extent than any other approach. Social behavior theory recognizes the individuality of each person and of each unique situation" (1968, p. 190). Process and content. This quotation from Mischel reveals not only that the behaviorist's approach is both nomothetic and idiographic with respect to persons but, in a special sense, with respect to behaviors and situations as well. In particular, if one believes that an individual's behavior reflects only the idiosyncratic vagaries of his or her past reinforcement history and the arbitrary similarities of the current situation to that history, then there is no rational nomothetic basis for classifying behaviors and situations into substantive - as opposed to functional or process - categories. Formally speaking, there are operant behaviors, but not anal or conscientious behaviors. There are reinforcing, aversive, or anxiety-producing situations, but not Oedipal conflicts. In other words, processes of behavior are formalized in nomothetic ways, but content is treated idiographically; it is extratheoretical, and hence, requires ad hoc treatment in each particular case. Consider, for example, the phenomenon of sex typing. Social behavior theory invokes principles of reinforcement and observational learning to explain how young boys and girls become sex-typed (Mischel, 1970). The category of sex is simply a special case, a particular area of content, on which general principles or processes of learning happen to operate in the culture. A social behavior theorist may thus choose to focus on a particular area of content - like sex typing, antisocial behavior, or even Oedipal conflicts - and may learn many new and even startling things that were not simply predictable from the general theory. But the epistemology of behaviorism does not regard these "new" things as new

208

Daryl J. Bern

nomothetic principles of behavior but as content-specific empirical generalizations and substantive instantiations of the more general processes embraced by the larger nomothetic theory. Only rarely in the history of this tradition have new nomothetic principles been added to the general theory (e.g., observational learning, self-reinforcement). But the nomothetic treatment of process combined with the idiographic or ad hoc treatment of content is also found in areas of psychology quite remote from behaviorism or situationism. For example, a second theory of sex typing, Sandra Bern's (1981) gender schema theory, also regards sex as an arbitrary cultural choice on which general principles of schematic cognitive processing happen to operate. As with social behavior theory, this theory presumes that if a particular culture happened to select eye color rather than gender for differential socialization, eye-color typing would be the result. In contrast, psychoanalytic theory regards sex typing as a distinctive phenomenon, inextricably linked to sex-linked anatomical features and specific sex-relevant developmental events; it does not regard sex as a culturally arbitrary choice, and hence has a content theory as well as a process theory of sex typing. [Kohlberg's (1966) cognitive-developmental account of sex typing is an interesting mixture. See S. L. Bern, 1985, for a discussion of these points.] In the area of personality, George Kelly's (1955) theory of personal constructs has several nomothetic principles governing the characteristics of constructs and the processes of construing, but it treats the content of each person's construals idiographically. In fact, Kelly's Role Repertory Test is explicitly designed as an instrument for idiographic measurement. In social psychology, theories of attitude structure, formation, and change propose processes presumed to operate universally on attitudes irrespective of their content. Again, psychoanalytic theory provides the contrast. The study of the authoritarian personality was guided by a content as well as a process theory of attitudes. As we shall see, the process/content distinction emerges again when we consider the kinds of personality variables that ought to be included in a truly interactional account of behavior. The interactional approach to personality: certain persons/certain behaviors/certain contexts A reader of the current literature in personality would probably conclude that everything I have said up to this point is parody. There are no trait psychologists who believe that persons are consistent across all contexts; there are no personologists who believe that personality theory must be idiographically reduced to biography; there are no situationists who believe that personality differences are just error variance. There's nobody here but us interactionists. We believe that behavior is a function of both the person and the situation!

Putting persons back into the context

209

But, as many writers have pointed out, this "function" is a many-splendored thing and comprises many diverse meanings of the term interaction (e.g., Magnusson, 1981; Olweus, 1977). Herewith a selected catalog. Persons and situations both contribute to behavior. This simplest sense of interaction has long been a part of both traditional personality psychology and social psychology. For example, Cattell (1965) explicitly formalizes this meaning of interaction in a linear "specification equation" in which situationspecific coefficients are used to weight a general set of personality variables. Social psychologists have typically explored this kind of interaction by selecting persons who are high and low on some selected personality dimension (e.g., need for approval, need for achievement, self-monitoring, Machiavellianism) and then observing their differential behavior in theory-relevant situations. The more specialized statistical meaning of interaction also falls under this general category and has been around for some time. It is implicit in Eysenck's (1970) early observation that extroverts perform better when in groups, introverts perform better when alone; and it is explicit in studies that actually assess the relative percentages of variance accounted for by person variables, situation variables, and their statistical interactions (e.g., Endler & Hunt, 1966). These approaches typically characterize persons in context-free dispositional terms first and then combine this information with independently obtained information about the situation. A stronger version of this kind of interactionism maintains that persons and situations are inseparable from the outset and that the enterprise ought to begin with persons-in-situations as the fundamental unit of analysis: ' T h e organism and its milieu must be considered together, a single creatureenvironment interaction being a convenient short unit for psychology" (Murray, 1938, p. 40). Murray himself attempted to follow this injunction by matching his set of "needs" to a complementary set of situational "presses." The situation is a function of the person. This meaning of interaction has several subcategories. First, persons perceive the same situation differently from one another. This is, of course, the guiding principle behind the phenomenological approach to interactionism favored by social psychologists and embodied in the famous dictum that if persons "define situations as real, they are real in their consequences" (Thomas & Thomas, 1928). The person who perceives a hurtful act as the product of hostility reacts differently from the person who interprets the act as the product of insensitivity. Second, persons are differentially influenced by the same situation. As Scarr notes (this volume), a child's genotype influences how much impact the environment will have on him or her. Children not only differ in their ability to process information in their environments, but their personalities (and maturity) also

210

DarylJ. Bern

influence the degree to which they attend to that information. Some individuals can decode nonverbal information in an interaction better than others. Thinskinned individuals will be more affected by an interpersonal slight than thickskinned individuals. And, of course, the differential perceptions of situations, described in the preceding paragraph, also produce differential impacts. Third, persons differentially select themselves into situations. Scarr argues that this selection process becomes increasingly important in development as the individual leaves the constraints of early childhood and that it is, in fact, "the most powerful connection between people and their environments and the most direct expression of the genotype in experience" (also see Scarr & McCartney, 1983). Recently Caspi, Elder, and I proposed that continuities in personality across the life course may rest in part upon this dispositionally guided selection process. In particular, we noted that individuals can systematically select themselves into environments that reinforce and sustain the very dispositions that led them to that selection (Caspi, Elder, & Bern, 1987, in press; also see Wachtel, 1977b). For example, when extroverts preferentially seek out social situations they thereby select themselves into environments that further nourish and sustain their sociability. We further proposed that maladaptive behaviors can also select individuals into personality-sustaining environments, albeit more coercively. For example, children whose ill tempers lead them to drop out of school thereby limit their future career opportunities and select themselves into frustrating life circumstances that further evoke a pattern of striking out explosively against the world. Similarly, socially withdrawn children may fail to experience many of the role and rule negotiations important to the growth of social skills, thereby decreasing the likelihood that they will initiate or respond appropriately to social overtures and selecting themselves into further isolation. In both these cases, the maladaptive behaviors are sustained across time by the progressive accumulation of their own consequences. We called continuity of this kind cumulative continuity and were able to show how it operated across the life course of both individuals with histories of childhood temper tantrums (Caspi et al., 1987) and individuals with histories of childhood social withdrawal (Caspi et al., in press). Patterson's new stage model of antisocial behavior - wherein a boy's antisocial behavior creates and selects him into contexts that promote further antisocial behavior - is also, in part, a model of cumulative continuity (Patterson, this volume; Patterson, Reid, & Dishion, in press). Persons interact with situations in reciprocal transaction. The most dynamic meaning of interaction refers to the reciprocal transaction between the person and the environment: The person acts, the environment reacts, and the person reacts back. Caspi, Elder, and I have proposed that this process can serve

Putting persons back into the context

211

as a second mechanism for producing continuities in personality across the life course. For example, Patterson has shown in elegant detail how family interactions can create and sustain destructive and aversive patterns of behavior. By extension, we suggested that a child whose temper tantrums coerce others into providing short-term payoffs in the immediate situation may thereby learn a behavioral style that continues to "work" in similar ways in later years. The immediate reinforcement short-circuits the learning of more controlled interactional styles that might have greater adaptability in the long run. We found evidence for such continuity in our study of individuals with childhood histories of temper tantrums (Caspi et al., 1987). In our study of individuals with childhood histories of social withdrawal, we suggested that such children's lack of assertiveness in ongoing interactions may cause them to be ignored or overlooked, thereby reinforcing and sustaining their socially withdrawn style. Again, we found evidence for such continuity (Caspi et al., in press). In this analysis, we thus concur with Wachtel (1977a) and others before him (e.g., Cottrell, 1969; Sullivan, 1953) who have argued that it is not so much a personality trait or a psychoanalytic-like residue of early childhood that is maintained across time, but an interactional style that evokes reciprocal, maintaining responses from others. Accordingly, we called continuity of this kind interactional continuity. These interactional analyses also suggested to us that continuities in personality are most likely to appear in later life when circumstances re-create contexts with similar interactional properties. For example, the children who reacted to frustration and authority with temper tantrums became undercontrolled, irritable adults when they again faced frustration and controlling authority - as in school, the armed services, or low-level jobs - or were immersed in life situations requiring the frequent negotiation of interpersonal conflict - as in marriage or child rearing (Caspi et al., 1987). The socially withdrawn children, children who were reluctant to enter social settings, became adults who were repeatedly tardy in entering the new social settings required by life-course transitions into marriage, parenthood, and career (Caspi et al., in press).

Putting persons back into the context I noted at the outset that, as a veteran of social psychology and a newcomer to the field of human development, I am impressed by the progress reported in this volume in studying the person in context. Also as a newcomer to the field of human development, I have been pleased and honored to join Caspi and Elder in exploring the interactional antecedents of life-course continuities in personality.

212

DarylJ. Bern

As an aspiring personologist, however, I don't believe we've quite got it right yet. Response-style variables: process and content revisited In our studies of personality continuity, Caspi, Elder, and I have used the terminology of Horney (1945) to characterize our ill-tempered subjects as children who "move against the world," our socially withdrawn subjects as children who "move away from the world," and our dependent subjects as children who "move toward the world." This terminology makes it clear that such variables are not just a random collection of behavior problems, but refer systematically to a set of problematic interactional styles. As such they are a subset of a larger class of person variables I have elsewhere called response-style variables (D. J. Bern, 1983b). This larger class also includes variables that characterize ways in which persons encode and process different kinds of information [e.g., the psychoanalytic defense mechanisms, cognitive styles like Field Dependence/Independence (Witkin & Goodenough, 1981), and the Jungian typology of preferences for sensing vs. intuition, thinking vs. feeling, and so forth (Myers, 1976)]. Temperament variables and variables relating to ego processes (e.g., impulse control) also fall into this larger class. I like response-style variables because they refer to quite general processes and are not tied to particular cultural content. In other words, I subscribe to the practice of the process theorists, discussed earlier, in that I believe a successful personality theory will have to treat processes nomothetically and - with some exceptions to be discussed shortly - treat content idiographically and, hence, extratheoretically. In contrast, many traditional personality variables - including many versions of the "big two" of extroversion and adjustment - are arbitrary cultural hybrids of process and content, and it is precisely such hybrids that most frequently fail to show consistency and coherence across contexts. For example, the variable of "honesty" in the Hartshorne-May studies is a creation of our culture, existing as an equivalence class of behaviors in the minds of the adult investigators but not in the conduct of children who had yet to "learn" the trait. The traits of masculinity and femininity similarly refer to quite diverse behaviors and attributes that our culture has chosen to lump together into two mutually exclusive equivalence classes. Accordingly, the terms do not apply to everyone, and "androgynous" individuals will be judged to be inconsistent across situations (S. L. Bern, 1981). And finally, when I imposed my own anal conception of conscientiousness on a set of subjects, they responded with a correlation between their conscientiousness in schoolwork and their personal hygiene of —.61 (Bern & Allen, 1974).

Putting persons back into the context

213

It is sometimes possible to use hybrid variables successfully, but only if they tap universal situational factors. Many of Freud's concepts, for example, are arguably successful hybrids of process and content because he could assume that the situational experiences encoded in the concepts were universal (e.g., the Oedipal situation or the incest taboo). And in constructing the California Psychological Inventory, Gough tried to select variables that would tap culturally universal concerns such as achievement and sense of well-being (Megargee, 1972). It may be that variables involving the self - an inevitable and universal situational presence - will have the requisite universality. The mechanisms of ego defense would fall into this category, and they seem to have enjoyed greater utility and longevity than those parts of psychoanalytic theory that seem more culture-bound. It is also pertinent to note that many theorists who come out of the behaviorist tradition - the prototype of the nomothetic process/idiographic content tradition - turn to self variables. Examples include Rotter's (1954) internal/external locus of control variable, Bandura's self-system (1978), Patterson's recent introduction of self-esteem (this volume), and several of Mischel's (1973) person variables. But even after allowing for these exceptions, I believe we are safest if we try to avoid hybrid variables and keep our person and our context variables "pure" and explicitly distinct from one another. Let us choose to combine person variables and context variables systematically in our theories rather than having them confounded at the outset. And again, I believe that response-style variables best meet this criterion. If response-style variables constitute the equivalence classes of persons, then the corresponding equivalence classes of contexts would be organized as parallel sets of environmental features that reward, encourage, or demand the corresponding personal response styles. For example, a triple typology might contain an equivalence class of Field-Dependent persons; the corresponding equivalence class of behaviors might include a global, gestalt-like way of processing information, a low threshold for confusion, and a preference for nurturant interpersonal environments; and the corresponding equivalence class of situations might include settings in which interpersonal skills are required for effective functioning, settings in which structure is provided, and so forth. Intrapsychic interaction: the missing ingredient A list of person variables - even response-style variables - is not a theory of personality. Nor do I believe that we will ever be able to construct the person by assembling individual variables after the fact. In other words, I believe that a successful theory must be morphogenic or person centered from the outset, focusing not on the relative standing of persons across variables, but on the

214

DarylJ. Bern

salience and configuration of an interacting set of variables within the person. It is this model of the person as an intrapsychic system of interacting components that is still absent from most contemporary interactional psychology. [The most notable exception is Block and Block's (1979) longitudinal research on ego control and ego resilience. These variables form an intrapsychic system of conceptually interrelated variables; they do not confound personality processes with arbitrary cultural content; and there is a preliminary set of situational variables that coordinates with them and the related equivalence class of behaviors (Block & Block, 1981).] A person-centered theory of response-style variables would also permit us to talk about human development. First, the theory itself could explicate how the person's intrapersonal patterning of styles changes and develops in interaction with particular kinds of environments. Second, the theory could suggest the most likely environments that the person will choose or avoid at transition points in the life course. Third, the person variables could themselves encode trajectories over time. For example, in Lives through Time, Block (1971) concatenated two Qsorts on each subject, one representing the adolescent years and one the adult years. The subsequent inverse or Q-factor analysis then treated two subjects as similar if and only if they were similar at both time periods, that is, were similar in their developmental trajectories over the years. And finally, the equivalence class of contexts or environments could directly encode transitions from one agegraded role to another across the life course (e.g., entering the first job). This, then, is the person - the interacting, intrapsychic ensemble - we must invite into our triple typology. For it is then that we can begin to talk sensibly about personality. And about development.

References Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt. Allport, G. W. (1962). The general and the unique in psychological science. Journal of Personality, 30, 405-22. Bandura, A. (1978). The self-system in reciprocal determinism. American Psychologist, 33, 34458. Bern, D. J. (1972). Constructing cross-situational consistencies in behavior: Some thoughts on Alker's critique of Mischel. Journal of Personality, 40, 17-26. Bern, D. J. (1983a). Constructing a theory of the triple typology: Some (second) thoughts on nomothetic and idiographic approaches to personality. Journal of Personality, 51, 566-77'. Bern', D. J. (1983b). Toward a response style theory of persons in situations. In R. A. Dienstbier & M. M. Page (Eds.), Nebraska symposium on motivation 1982: Personality - current theory and research (Vol. 30). Lincoln: University of Nebraska Press. Bern, D. J., & Allen, A. (1974). On predicting some of the people some of the time: The search for cross-situational consistencies in behavior. Psychological Review, 81, 506-20. Bern, S. L. (1981). Gender schema theory: A cognitive account of sex-typing. Psychological Re-

view, 88, 354-64.

Putting persons back into the context

215

Bern, S. L. (1985). Androgyny and gender schema theory: A conceptual and empirical integration. In T. B. Sonderegger (Ed.), Nebraska symposium on motivation 1984: Psychology and gender. Lincoln: University of Nebraska Press. Block, J. (1971). Lives through time. Berkeley, CA: Bancroft Books. Block, J. H., & Block, J. (1979). The role of ego-control and ego-resiliency in the organization of behavior. In W. A. Collins (Ed.), Minnesota symposia on child psychology (Vol. 13). Hillsdale, NJ: Erlbaum. Block, J., & Block, J. H. (1981). Studying situational dimensions: A grand perspective and some limited empiricism. In D. Magnusson (Ed.), Toward a psychology of situations: An interactional perspective. Hillsdale, NJ: Erlbaum. Caspi, A., Elder, G. H., Jr., & Bern, D. J. (1987). Moving against the world: Life-course patterns of explosive children. Developmental Psychology, 22, 303-8. Caspi, A., Elder, G. H., Jr., & Bern, D. J. (in press). Moving away from the world: Life-course patterns of shy children. Developmental Psychology. Cattell, R. B. (1965). The scientific analysis of personality. Chicago: Aldine. Cheek, J. M. (1982). Aggregation, moderator variables, and the validity of personality tests: A peer rating study. Journal of Personality and Social Psychology, 43, 1254-69. Cottrell, L. S. (1969). Interpersonal interaction and the development of the self. In D. A. Goslin (Ed.), Handbook of socialization theory and research. Chicago: Rand McNally. Endler, N. S., & Hunt, J. McV. (1966). Sources of behavioral variance as measured by the S-R inventory of anxiousness. Psychological Bulletin, 65, 336-46. Eysenck, H. J. (1970). Explanation and the concept of personality. In R. Borger & F. Cioffi (Eds.), Explanation in the behavioural sciences. Cambridge: Cambridge University Press. Gibbons, F. X. (1983). Self-attention and self-report: The "veridicality" hypothesis. Journal of Personality, 51, 517-42. Hartshorne, H., & May, A. (1928). Studies in the nature of character: Vol. 1. Studies in deceit. New York: Macmillan. Horney, K. (1945). Our inner conflicts. New York: Norton. Kelly, G. A. (1955). The psychology of personal constructs (2 vols.). New York: Norton. Kenrick, D. T., & Dantchik, A. (1983). Interactionism, idiographics, and the social psychological invasion of personality. Journal of Personality, 51, 286-307. Kenrick, D. T., & Stringfield, D. O. (1980). Personality traits and the eye of the beholder: Crossing some traditional philosophical boundaries in the search for consistency in all of the people. Psychological Review, 87, 88-104. Kohlberg, L. (1966). A cognitive-developmental analysis of children's sex-role concepts and attitudes. In E. E. Maccoby (Ed.), The development of sex differences. Stanford, CA: Stanford University Press. Lehmann, H. C , & Witty, P. A. (1934). Faculty psychology and personality traits. American Journal of Psychology, 44, 490. Magnusson, D. (Ed.), (1981). Toward a psychology of situations: An interactional perspective. Hillsdale, NJ: Erlbaum. Megargee, E. I. (1972). The California psychological inventory handbook. San Francisco: JosseyBass. Mischel, W. (1968). Personality and assessment. New York: Wiley. Mischel, W. (1970). Sex-typing and socialization. In P. H. Mussen (Ed.), CarmichaeVs manual of child psychology (Vol. 2). New York: Wiley. Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality. Psychological Review, 80, 252-83. Murray, H. A. (1938). Explorations in personality. New York: Oxford. Myers, I. B. (1976). Introduction to type. Palo Alto, CA: Consulting Psychologists Press. Olweus, D. (1977). A critical analysis of the "modern" interactionist position. In D. Magnusson & N. S. Endler (Eds.), Personality at the crossroads. Hillsdale, NJ: Erlbaum.

216

DarylJ. Bern

Patterson, G. R., Reid, J. B., & Dishion, T. J. (in press). Antisocial boys (Vol. 4, social learning series). Eugene, OR: Castialia. Peterson, D. R. (1968). The clinical study of social behavior. New York: Appleton-Century-Crofts. Rotter, J. B. (1954). Social learning and clinical psychology. Englewood Cliffs, NJ: Prentice-Hall. Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of genotype-environment correlations. Child Development, 54, 424-35. Skinner, B. F. (1935). The generic nature of the concepts of stimulus and response. The Journal of General Psychology, 12, 40-65. Snyder, M. (1983). The influence of individuals on situations: Implications for understanding the links between personality and behavior. Journal of Personality, 51, 497-516. Sullivan, H. S. (1953). The interpersonal theory of psychiatry. New York: Norton. Thomas, W. I., & Thomas, D. S. (1928). The child in America. New York: Knopf. Vernon, P. E. (1964). Personality assessment: A critical survey. New York: Wiley. Wachtel, P. L. (1977a). Psychoanalysis and behavior therapy. New York: Basic. Wachtel, P. L. (1977b). Interaction cycles, unconscious process, and the person-situation issue. In D. Magnusson & N. S. Endler (Eds.), Personality at the crossroads. Hillsdale, NJ: Erlbaum. Witkin, H. A., & Goodenough, D. R. (1981). Cognitive styles: Essence and origins. Psychological Issues, Monograph No. 51. New York: International Universities Press.

9

How genotypes and environments combine: development and individual differences

Sandra Scarr

Not long ago, most psychologists thought that the major purpose of their research was to discover eternal laws about human behavior - laws that applied in principle to all of the people all of the time. At the same time, a smaller group of psychologists focused their research on individual variation in people's responses to events in their lives-that is, on individual differences. The history of the two orientations can be traced to what Cronbach (1957, 1975) called "the two disciplines of scientific psychology" - to wit, research on species-typical or average trends in human behavior, and research on individual differences. Today, the two disciplines are merging in new theories based on concepts about the ways in which people are both typically human and uniquely so. In a fine paper, David Buss (1984) outlined the distinctive features of the two disciplines of scientific psychology, which, following Ernst Mayr (1963), he called "typological" and "population" approaches, as shown in Table 9.1. The former is concerned primarily with painting the human species in broad strokes, the latter with detecting fine variations within the species. Psychology's two scientific disciplines, to use Cronbach's terms, have their parallel in Mayr's contrast of typological and population approaches in biology. The former is philosophically Platonic; although there are variations within species, these are considered merely imperfect versions of the ideal type. What is important is to understand the true nature of the species, the ideal. Population thinking turns that idea on its head; its philosophical roots are in evolutionary theory and Darwin. In the evolutionary view, variation within and between species is what exists and must be understood; so-called species-typical traits are merely mental abstractions that provide a convenient shorthand for talking about average tendencies in species. Neither the typological nor the population approach alone is sufficient for our understanding of human nature and development, but both are necessary. Today we need a third kind of theory, one that addresses the ways in which genes and environments combine to produce both species-typical development and indi217

218

Sandra Scarr

Table 9.1. A comparison of typological and population approaches Typological

Population

Aims

To discover "human nature" To discover the adaptive significance of species traits To monitor the genetic basis of species behavior

To discover variation among species members To monitor the genetic basis of observed variation To discover the reasons for which variants are selected

Assumptions about variation

Variation is small and subsidiary in importance to species-typical characteristics

Variation is real and consequential

Primary methods

Comparative phylogenetic analysis

Quantitative and population genetics

Limitations

Variability of traits limits utility of postulating species-typical attributes Defined so broadly that any outcome of quantitative genetic research is compatible

Findings limited to extant population variation and to environments within which population is studied Will not discover species traits

Source: Buss (1984).

vidual variation. Unfortunately, as I argue in this chapter, we do not have the theories or the concepts to permit us to explore both average tendencies and variation within the same framework or to propose how they combine. A major problem preventing rapprochement between the two disciplines of scientific psychology is that they do not have a common language in which the word "cause" has the same meaning. The psychology of the average human prefers proximal, efficient, or situational causes to distal, evolutionary ones, whereas the psychology of individual differences prefers distal, traitlike causes with biological roots. A theory of human development and individual differences must incorporate both kinds of causes and specify their relationships if it is to provide a complete account. The theory I propose attempts to do that. Constructing theories from observations The contributors to this volume agree in principle, I think, that both the average tendencies of the human species and individual variations on that theme ought to be considered in formulating theories about human behavior. Further, most, if not all, participants would add that developmental changes in the human organism must be taken into theoretical account, as they are likely to affect the ways

How genotypes and environments combine

219

in which we can characterize both central tendencies and variation in the species. There is much goodwill on all sides. But there is not much integrated theory. In many ways, we talk past one another. I have argued (Scarr, 1985) that we must construct theories that account for agreed-upon observations, but that the same observations can become very different facts within different theories. Many versions of the truth are likely to coexist at any one historical time and cultural place. Instead of one Platonically ideal theory of everyone in every situation that can compel universal scientific belief, we are likely to have a population of theories (Toulmin, 1973) that cooperate by attending to different observations and compete to explain the same observations in only very limited ways. Thus, it is rather easy to avoid intellectual conflict within the scientific niche. Although I am undoubtedly a person of goodwill, I would like to invade the territories of a few others and try to explain that the facts of their theories are my theory's different facts. First, I present some observations on which we can all agree. Second, I contrast theories about these observations - theories that make facts of them and contrast the facts of others with my own. Last, I try to clarify why different causal models are more and less adequate to account for these observations. Naturally, I must select observations for consideration. Because this volume focuses on "persons in context," it seems appropriate to concentrate my remarks on human development in proximal and distal environments. My major question is, "How do organisms and environments combine to produce human development and the many variations on that theme?" Illustrations are taken from the intellectual realm, as the most relevant data are found there. Parents, children, and families We can all agree that parents and their offspring and biological siblings (especially identical twins) resemble each other in many behavioral ways. In fact, it is difficult to find any observations, based on varied samples, that show zero likeness of parents and their offspring or of biological siblings. I hope we can agree that any measured characteristics of parents and their children and of full siblings, past infancy, are likely to correlate positively (for references see Plomin, 1986; Scarr, 1986). Measures of the home "environment," such as the HOME Scale (Caldwell & Bradley, 1978) and the Family Environment Scale (Moos & Moos, 1981), have also been shown to correlate with children's behavioral characteristics such as intelligence and social adjustment. Parents who provide intellectually stimulating homes, who live in communities with "good" schools (which are better measured by the backgrounds of the other pupils than by what teachers try to do), safe streets, well-behaved peers, and so forth are likely to have intellectually

220

Sandra Scarr

more adept and less delinquent children than parents who provide a less advantaged environment. Thus, universal observations about differences among children reared in different home environments and ubiquitous correlations between parental and child characteristics are not in dispute. What is not clear is what the correlations mean. Theories of family effects on children most often focus on home environments as determinants of children's outcomes. Most psychologists interpret these correlations to mean that the environments provided by the parents, both at home and in the larger community, cause children to have favorable or unfavorable outcomes. But Beware! Just calling them measures of the home "environment" does not make them so. As Robert Plomin pointed out, Parental behavior is clearly involved in the two major "superfactors" of childrearing, parental love and control. Some parents hug and kiss their children whenever they are within reach; others rarely display physical affection. Parental behavior is also ultimately responsible for measures of physical aspects of the home environment. For example, the most widely used environmental measure in studies of mental development is the number of books in the home. But books do not magically appear on the shelves - parents put them there. (1986, p. 126) Parental behaviors underlie every aspect of the home environment and many characteristics of the neighborhood in which children grow up. Child-rearing styles are parental behaviors. Standards of living depend on parental educational and occupational achievements and family income. Aside from a small minority of families with inherited wealth, the style and advantages of children's environments depend largely on the parents' own characteristics. And parental behaviors are necessarily a function of their own unique gene-environment combinations that produce individual differences in talent, personality, and interests. Need I say, although many.of our colleagues seem to forget, that parents transmit these characteristics to their children genetically, as well as through behavioral interactions? The same observed correlations between parental-home characteristics and children's development can be fitted to many models (Scarr, 1986). Knowledge and intelligence Take, for example, a study that hypothesizes that parents' knowledge of child development causes better child development: Parents who know more about the course of development will provide more appropriate environments for their children (books, educational toys, stimulating activities) than parents with less knowledge of child development. The hypothesis also states that children reared by knowledgeable parents will develop better intellectual skills than children reared by less knowledgeable parents.

How genotypes and environments combine Parent's Knowledge of Child Development

221

Child — ^ Intellectual Development

Figure 9.1: The specific knowledge model of parent effects on child intelligence

Parents

Child

Knowledge

Knowledge

Figure 9.2: The general knowledge model of parent effects on child knowledge

Imagine a study in which parents were given a test of knowledge of contemporary facts about child development and their children were given a test of intelligence. Lo and behold, we find that knowledgeable parents have smarter children. The inference usually drawn from this result is that other parents should be taught what is currently believed about child development so that they may be better parents who promote the intellectual development of children. The specific knowledge model of parent-child effects is shown in Figure 9.1. But knowledge of one domain is positively correlated with knowledge of other domains. Knowledge of contemporary child development is doubtless correlated with knowledge of musical composers, world geography, and the engineering of bridges. I do not know of any theory of child development, however, that posits that parents' knowledge of composers, geography, or bridges per se is causally related to children's intellectual development, unless one concedes that general knowledge is important in child rearing. Indeed some investigators do. (These investigators usually ignore the implications of general cultural knowledge for a theory of general intelligence.) They highlight knowledgeable parents as (1) role models for, or (2) information transmitters to their children. In this model, and those to follow, single-headed arrows indicate causal paths in the direction of the arrowheads, and double arrow paths indicate correlations without causal direction. The model of such developmentalists is shown in Figure 9.2. Implicit in this model of knowledge transmission is a simple intervention: Give parents more knowledge and they will produce more knowledgeable children, a desirable outcome for everyone concerned. Another possible model of parent-child transmission of intelligence and general cultural knowledge stresses the knowledge pathway as a cause of parentchild correlations in intelligence, as shown in Figure 9.3. In this model, knowledge acquired by parents is transmitted to children; knowledge is the basis for inferences of intelligence in both parents and children; and therefore, the intelligence of parents and children is correlated. Because intelligence is merely an inference from general knowledge, the implicit inter-

222

Sandra Scarr Knowledgeable parents

Knowledgeable child

Intelligent parents

Intelligent child

Figure 9.3: The general knowledge model of parent effects on child intelligence

Intelligent parents

Knowledgeable parents

Intelligent *~ child

Knowledgeable child

Figure 9.4: The general intelligence model of parent effects on child intelligence

vention is still to help parents acquire knowledge that they will then transmit to their children, who will simultaneously become more intelligent. Problems with assumptions Most developmental investigators have a fundamentally different epistemology from that of investigators of intelligence. The former believe that information is discrete and environmentally transmitted from parents to children. Knowledgeable parents provide more information for their children to learn than do less knowledgeable parents. Intelligence experts believe that general cultural knowledge is one of the best indicators of some more general intellectual ability, which is why cultural knowledge of all sorts is sampled on most IQ tests. Intelligence researchers believe that knowledgeable parents are more intelligent and that more intelligent parents have more intelligent children, who learn more about their cultural environments, not only from their parents but from the world at large. One possible model is depicted in Figure 9.4. The implications of this model of intelligence for intervention are not encouraging; therefore, this is not the first approximation to the knowledge-intelligence connection for psychologists who want to intervene to improve people's educational and occupational functioning. Most psychologists recognize that improving general intelligence with deliberate interventions is a formidable task. Regardless of its ''truth" value, this model does not win popularity contests. All of these models and others can account for the common observations: To wit, intelligent parents tend to have intelligent children; knowledgeable parents

How genotypes and environments combine

223

tend to have knowledgeable children; intelligent parents tend to have knowledgeable children; and knowledgeable parents tend to have intelligent children. In any developmental model of parent-child effects, the nature of transmission and the nature of the intelligence-knowledge connection must be specified by theories. Choices among such partial models are not often empirically resolved, because all models are, in some theoretical contexts, true. How one construes the problems, accounts for the observations, and attempts to persuade others of one's model are the everyday problems of working scientists (Scarr, 1985). Other examples of items from IQ tests that are given causal status in developmental research can be found in language studies. Parents who speak to their children with larger vocabularies, in more complex sentences, with longer utterances, are often found to have more verbally complex and fluent children. Similarly, parents who use more sophisticated reasoning techniques with their children are said to induce more complex reasoning skills in their offspring. And so forth. In such studies, measures of parental and child behaviors taken in interactional situations are drawn from the same domain of behaviors that is sampled by items on tests of general intelligence. Therefore, it is hardly surprising that parental behaviors are found to correlate with child behaviors. What is surprising is the investigators' willingness to attribute causality exclusively to behavioral interactions between parents and their children.

Observations from behavior genetic studies To consider the merits of several possible theories about family resemblances and parents' effects on their children, consider the following additional observations about individual variation in intelligence from behavior genetic studies of families in which genetic and environmental parameters are varied. 1. In biologically related families, the correlations of IQ test scores of late adolescent and young adult siblings are approximately .50 (corrected for attenuation), whereas in adoptive families, siblings' IQ correlations are approximately zero. 2. Younger adopted siblings are more similar in intelligence than older adopted, siblings (.25-.39 versus zero). 3. In late adolescence and adulthood, IQ correlations of monozygotic (MZ) twins average .85, whereas the correlations for dizygotic (DZ) twins average .55. 4. In infancy and early childhood, intellectual correlations for MZ twins average .85, and correlations for DZ twins fall in the range of .60-.75. 5. Identical twins who were reared most or all of their lives in different families resemble each other intellectually to such an extent that their IQ correlations average .76 in several well-documented studies (Bouchard, 1984). The observations in Table 9.2 lead many reasonable people (e.g., Scarr, 1986) to conclude that 1. Genetic variation accounts for at least 60% of the individual differences in IQ test scores.

Sandra Scarr

224

Table 9.2. IQ and degrees of relatedness: similarities of genetically related and unrelated persons who live together and apart Relationship

Correlation

Number of pairs

Genetically identical Identical twins together Identical twins apart Same person tested twice

.86 .76 .87

1,300 137 456

Genetically related by half of the genes Fraternal twins together Biological sisters & brothers Parents & children together Parents & children apart

.55 .47 .40 .31

8,600 35,000 4,400 345

Genetically unrelated Adopted children together Unrelated persons apart

.00 .00

200* 15,000

"Based on data from Scarr and Weinberg (1978) and Teasdale and Owen (1985) on older adolescents who are comparable in age to other samples in this table. Younger adopted children resemble each other to a greater degree, with correlations around .24, according to samples of 800 pairs. Source: Adapted from Plomin and DeFries (1980).

2.

3.

Variations among home environments account for substantial individual variation in IQ scores in early development but little or no variation in IQ test scores in adolescence and adulthood. These results are obtained on families within a range of average to favorable environments (e.g., adoptive homes); no inferences can be made to severely disadvantaged environments. Nearly all of the environmental variation in IQ test scores of people reared in adequate to favorable environments arises from differences in the experiences of individual children within the same family.

The observations cited above should constitute a problem for developmental theories whose concepts are proximal and interactional. That they rarely do pose a problem is due to two characteristics of such studies: (1) the studies do not deal with enduring human characteristics, such as intelligence, but with interactional outcomes, which may or may not be proxies for intelligence; or (2) the investigators ignore these observations. An adequate theory of development and individual differences must deal, I think, with the ubiquitous correlations within biological families, as well as with the findings on adoptive families and twins of differing genetic and environmental relatedness. Theories of development Theories of behavioral development and individual differences have ranged from genetic determinism to naive environmentalism. Neither of these radical views

How genotypes and environments combine

225

has adequately explained the process of development or the role of experience in development. For the species, human experience and its effects on development depend primarily on the evolved nature of the human genome. In evolutionary theory, the two essential concepts are selection and variation. Through selection, the human genome has evolved to program human development. Phenotypic variation is the raw material on which selection works. Genetic variation must be associated with phenotypic variation, or there could be no evolution. It follows from evolutionary theory that individual differences depend in part upon genotypic differences. The process of development itself has been subject to evolution (Scarr, 1976). The timing of birth, the characteristics of the human infant, the requirements of parenting, and many other features of development are co-adapted complexes that must also be genetically variable within the species. The implications of this for a theory of development and individual variability are enbrmous. A theory of behavioral development must explain the origin of new behaviors. Because there is no evidence that new adaptations can arise out of the environment without maturational changes in the organism, genotypes must be the source of new structures. Maturational sequence is controlled primarily by the genetic program for development. But neither the genotype nor the environment is a fixed entity acting on the other. Both are dynamic, changing with development. Their interplay must also change, and observations cited earlier indicate that this is so. The challenge to developmental theory is to explain how genetically programmed development acts with changing environments to produce human beings who are both species typical and individually different. The dichotomy of nature and nurture has always been a bad one for two reasons. First, both a genotype and environment are required for development. This is so obvious that it requires no additional comment. Second, I think that a false parallel arises between the two concepts when they are juxtaposed. Development is indeed the result of nature and nurture, but I think that genes drive experience. Genes are components in a system that organizes the organism to experience its world. The organism's abilities to experience the world change with development and are individually variable. A good theory of development and individual differences that accounts for these observations can only be one in which experience is guided by genotypes that both facilitate and restrain experiences. A theory of genotype —> environment effects Each of us encounters the world in different ways. We are individually different in the ways we process information from the environment, which makes our experiences individually tailored to our interests, personality, and talents. Human beings are also developmentally different in our ability to process infor-

226

Sandra Scarr

mation from the environment. Each of us at every developmental stage will gain different information from the same environment, because we each attend to some aspects of our environments and ignore other opportunities for experience. I propose that these differences in experience - both developmental changes and individual differences - are caused by genetic differences. During the course of development, different genes are turned on and off, creating maturational changes in the organization of behavior, as well as maturational changes in patterns of physical growth. I argue that genetic differences among individuals are similarly responsible for determining what experiences people do and do not have in their environments. What people attend to and what they ignore are events that are correlated with individual differences in interests, personality, and talents. Environments to which a person is observed to be exposed are not good predictors of what the person actively experiences. Individual differences in cognitive processes, personality, and interests play roles in what and how much an individual records and how he interprets his environmental inputs. In addition, the relevance of environments changes with development. The toddler who has "caught on" to the idea that things have names and who demands the names for everything is experiencing a fundamentally different verbal environment from what she experienced before, even though her parents talked to her extensively in infancy. A model of genotypes and environments Behavioral development depends on both a genetic program and a suitable environment for the expression of the human, species-typical program for development. But differences among people can arise from both genetic and environmental differences. The process by which differences arise is better described as genotype —> environment effects. Figure 9.5 presents our model of behavioral development (Scarr & McCartney, 1983). In this model, the child's phenotype (Pc), or observable characteristics, is a function of both the child's genotype (Gc) and her rearing environment (E c ). There will be little disagreement on this. The parent's genotypes (Gp) determine the child's genotype, which in turn influences the child's phenotype. Again, there should be little controversy over this point. As in most developmental theories, transactions occur between the organism and the environment; here they are described by the correlation between phenotype and rearing environment. In most models, however, the source of this correlation is ambiguous. In this model, both the child's phenotype and rearing environment are influenced by the child's genotype. Because the child's genotype influences both the phenotype

How genotypes and environments combine

227

and the rearing environment, their correlation is a function of the genotype. The genotype is conceptually prior to both the phenotype and the rearing environment. Developmental changes in the genetic program prompt new experiences through maturation. Before the full phenotype is developed, the child becomes attentive to and responsive to aspects of the environment that previously were ignored or had other meanings. Infants are responsive to speech sounds and intonations but they do not yet have a use for the syntax or semantics of language. Toward the end of the first year, infants seem to become more attentive to words as names for things. Before the development of productive speech, the child attends to the language environment in a receptive fashion. Changes in what is ' 'turned on" in the genotype affects an emerging phenotype both directly through maturation (Gc to Pc) and through prompting new experiences (Gc to E c ). It follows from the preceding argument that the transactions we observe between phenotypes and environments are merely correlations, determined by developmental changes in the genotype. The correlation of phenotype and environment is represented by the double arrows from Ec to P c . The theory states that developmental changes in phenotypes are prompted both by changes in the effective genotype and by changes in the salience of environments, which are then correlated. We recognized that this is not a popular position, but we proposed it to account for the observations from family studies. The path from the Gc to P c represents maturation, which is controlled primarily by the genetic program. New structures arise out of maturation, from genotype to phenotype. Behavioral development is elaborated and maintained, in Gottlieb's (1983) sense, by the transaction of phenotype and environment, but it cannot arise de novo from this interaction. Thus, in this model, the course of development is a function of genetically controlled, maturational sequences, although the rate of maturation can be affected by some environmental circumstances, such as the effects of nutrition on physical growth (Watson & Lowrey, 1967). Behavioral examples include cultural differences in rates of development through the sequence of cognitive stages described by Piaget and other theoretical sequences (see Nerlove & Snipper, 1981). Separation of genetic and environmental effects on development At any one point in time, behavioral differences may be analyzed into variances that can be attributed more or less to genetic and environmental sources (see Plomin, DeFries, & Loehlin, 1977; Scarr & Kidd, 1983). Depending on how much genetic variation and how much environmental variation there are for the characteristics of interest in the population studied, sources of individual dif-

228

Sandra Scarr

Figure 9.5: A model of behavioral development

ferences can be estimated. But such an account does not describe how those differences developed. The theory that we proposed does not attempt to allocate variance but to describe how individuals evoke and select their own environments to a great extent. There may appear to be arbitrary events of fate (Bandura, 1982), such as accidentally meeting a theater producer on the commuter train taking you to acting class or entering a sweepstakes and winning a free trip to Hawaii, but even these may not be entirely divorced from personal characteristics that have some genetic variability. The budding actress did talk with the producer, she did not hide in her newspaper; the sweepstakes winner did enter the contest. Experiences are likely to be correlated to some extent with the personal characteristics of individuals; they are not likely to be entirely random. But please understand that I do not mean that one's environmental fate is entirely determined by one's genotype - only that some genotypes are more likely to receive and select certain environments than others. The theory is not deterministic but probabilistic (see Scarr, 1981). Genotype —» environment effects Plomin et al. (1977) described three kinds of genotype-environment correlations that form the basis for a developmental theory. The theory of genotype —» environment effects we proposed has three propositions: 1.

2.

The process by which children develop is best described by three kinds of genotype —» environment effects: a passive kind whereby the genetically related parents provide a rearing environment that is correlated with the genotype of the child (sometimes positively and sometimes negatively); an evocative kind whereby the child receives responses from others that are influenced by his genotype; and an active kind that represents the child's selective attention to and learning from aspects of his/her environment that are influenced by his genotype, and indirectly correlated with those of his biological relatives. The relative importance of the three kinds of genotype —» environment effects

How genotypes and environments combine

229

changes with development. The influence of the passive kind declines from infancy to adolescence, and the importance of the active kind increases over the same period. 3. The degree to which experience is influenced by individual genotypes increases with development and the shift from passive to active genotype —» environment effects, as individuals select their own experiences. The first type, the passive genotype —> environment effect, arises in biologically related families and renders all of the research literature on parent-child socialization uninterpretable. Because parents provide both genes and environments for their biological offspring, the child's environment is necessarily correlated with his/her genes, because his/her genes are correlated with her parents' genes, and the parents' genes are correlated with the rearing environment they provide. It is impossible to know what about the parents' rearing environment for the child determines what about the child's behavior, because of the confounding effect of genetic transmission of the same characteristics from parent to child. Not only can we not interpret the direction of effects in parent-child interaction, as Bell (1968) argued, but we cannot interpret the cause of those effects in biologically related families. The second kind of genotype —> environment effect is called evocative because it represents the different responses that different genotypes evoke from the social and physical environments. Responses to the person further shape development in ways that correlate with the genotype. The third kind of genotype —> environment effect is the active niche-picking or niche-building sort. People seek out environments they find compatible and stimulating. We all select from the surrounding environment some aspects to which to respond, learn about, or ignore. Our selections are correlated with motivational, personality, and intellectual aspects of our genotypes. The active genotype —» environment effect, we argue, is the most powerful connection between people and their environments and the most direct expression of the genotype in experience. Developmental changes in genotype —> environment effects The second proposition is that the relative importance of the three kinds of genotype —» environment effects changes over development from infancy to adolescence. In infancy much of the environment that reaches the child is provided by adults. When those adults are genetically related to the child, the environment they provide in general is positively related to their own characteristics and their own genotypes. Although infants are active in structuring their experiences by selectively attending to what is offered, they cannot do as much seeking out and niche building as older children; thus, passive genotype —> environment effects are more important for infants and young children than they are for older children, who can extend their experiences beyond the family's

230

Sandra Scarr

influences and create their own environments to a much greater extent. Thus, the effects of passive genotype —» environment effects wane when the child has many extrafamilial opportunities. In addition, parents can provide environments that are negatively related to the child's genotype to compensate for the child's perceived deficiencies or problems. Parents with a child who is aggressive may attempt harsher discipline to curb the child's unacceptable behavior. Parents with a child who does not learn to read easily may provide compensatory practice in reading. Although parents' genotypes usually affect the environment that they provide for their biological offspring, it is sometimes positive and sometimes negative and, therefore, not as direct a product of the young child's genotype as later environments will be. Thus, as stated in proposition 3, genotype —•» environment effects increase with development, as active forms replace passive. Genotype —> environment effects of the evocative sort persist throughout life, as we elicit responses from others on the basis of many personal, genotyperelated characteristics from appearance to personality and intellect. Those responses from others reinforce and extend the directions our development has taken. High intelligence and adaptive skills in children from very disadvantaged backgrounds, for example, evoke approval and support from school personnel who might otherwise despair of the child's chances in life (Garmezy, Masten, & Tellegan, 1984). In adulthood, personality and intellectual differences evoke different responses in others. Similarities in personal characteristics evoke similar responses from others, as shown in the case of identical twins reared apart (Bouchard, 1984). These findings are also consistent with the third proposition. Accounting for observations from family research In an earlier report (Scarr & McCartney, 1983), we reviewed evidence about how monozygotic twins come to be more similar than dizygotic twins, and biological siblings more similar than adopted siblings on nearly all measurable characteristics, at least by the end of adolescence (Scarr & Weinberg, 1978). We also reviewed the evidence on declining similarities between DZ twins and adopted siblings from infancy to adolescence. Most fascinating to laypeople are the unexpectedly great similarities between identical twins reared in different homes. Neither extreme genetic determinism nor naive environmentalism can account for these observations from research on twins and families. Our theory of genotype —> environment effects can account for these data by predicting the degree of environmental similarity that is experienced by the co-twins and sibs, whether they live in the same family or not. The expected degree of environmental similarity for a pair of relatives can be thought of as the product of a person's own genotype —> environment path and

How genotypes and environments combine

231

path b

path x

path x

path a Figure 9.6: A model of environmental similarity based on genetic resemblance

the genetic correlation of the pair. Figure 9.6 presents a model of the relationship between genotypes and environments for pairs of relatives who vary in genetic relatedness. Gj and G 2 symbolize the two genotypes, Ex and E 2 their respective environments. The similarity in the two environments (path a) is the product of the coefficient of each genotype with its own environment (path JC) and the genetic correlation of the pair (path b). On the assumption that individuals' phenotypes are equally influenced by their own genotypes, the similarity in the environments of two individuals becomes a function of their genetic correlation. This model can be used to describe the process by which MZ twins come to be more similar than DZ twins, and biological siblings more similar than adopted siblings. For identical twins, for whom b = 1.00, the relationship of one twin's environment with the other's genotype is the same as the correlation of the twin's environment with her own genotype. Thus, one would certainly predict what is often observed: that the hobbies, food preferences, choices of friends, academic achievements, and so forth of the MZ twins are very similar (Scan* & CarterSaltzman, 1980). Kamin (1974) proposed that all of this environmental similarity is imposed on MZ co-twins, because they look so much alike. We proposed that the home environments provided by the parents, the responses that the co-twins evoke from others, and the active choices they make in their environments lead to striking similarities through genotypically determined correlations in their learning histories. The same explanation applies, of course, to the greater resemblance of biological than adopted siblings. The environment of one biological sib is correlated to the genotype of the other as one half the coefficient of the sibling's environment to her own genotype, because b = 0.50, as described in Figure 9.2. The same is true for DZ twins. There is a very small genetic correlation between adopted siblings in most studies that arise from selective placement of the offspring of similar mothers in the same adoptive home. More important for this theory, however, is the selective placement of adopted children to match the intellectual characteristics of the adoptive parents. This practice allows adoptive parents to create a positive, passive genotype-environment correlation for their adopted

232

Sandra Scarr

children in early childhood, when the theory asserts that this kind of correlation is most important. In fact, the selective placement estimates from studies by Scarr & Weinberg (1977) can account for most of the resemblance between adoptive parents and their children. In addition, adoptive parents, like their biological counterparts, can provide negative genotype-environment correlations that ensure that their several children will not differ too much on important skills, such as reading. The most interesting observation is the unexpected degree of resemblance between identical twins reared in different families, and most often in different communities and even different cultures. With the theory of genotype —» environment effects, their resemblance is not surprising. Given opportunities to attend selectively to and choose from varied opportunities, identical genotypes are expected to make similar choices. They are also expected to evoke similar responses from others and from their physical environments. The fact that they were reared in different homes and different communities is not important; differences in their development could arise only if the experiential opportunities of one or both were very restricted, so that similar choices could not have been made. According to previous studies (Newman, Freeman, & Holzinger, 1937; Juel-Nielsen, 1980; Shields, 1962) and the recent research of Bouchard and colleagues at the University of Minnesota (Bouchard, 1984), the most dissimilar pairs of MZs reared apart are those in which one was severely restricted in environmental opportunity. Extreme deprivation or unusual enrichment can diminish the influence of genotype on environment and therefore lessen the resemblance of identical twins reared apart. Changing similarities among siblings The second discrepant set of observations from previous research concerned the declining similarities of dizygotic twins and adopted siblings from infancy to adolescence. It is clear from Wilson's (Matheny, Wilson, Dolan, & Krantz, 1981; Wilson, 1983) longitudinal study of MZ and DZ twins that the DZ correlations for intelligence of .60-.75 are higher than genetic theory would predict in infancy and early childhood. For school age and older twins, DZ correlations were the usual .55. Similarly, the intelligence correlations of two samples of late adolescent adopted siblings were zero compared to the .25-.39 correlations of the samples of adopted children in early to middle childhood (Scarr & Weinberg, 1978). How can it be that the longer you live with someone, the less like them you become? The theory put forward here predicts that the relative importance of passive versus active genotype-environment correlations changes with age. Recall that passive genotype-environment correlations are created by parents who

How genotypes and environments combine

233

provide children with both genes and environments, which are then correlated. Certainly in the case of DZ twins, whose prenatal environment was shared and whose earliest years are spent being treated in most of the same ways at the same time by the same parents, the passive genotype —> environment effect is greater than that for ordinary sibs. Biological and adopted siblings do not, of course, share the same developmental environments at the same time because they differ in age. The passive genotype-environment correlation still operates for siblings, because they have the same parents, but to a lesser extent than for twins. MZ twin correlations for intellectual competence do not decline when active genotype-environment correlations outweigh the importance of the passive ones, because MZ co-twins typically select highly correlated environments anyway. DZ pairs, on the other hand, are no more genetically related than sibs, so that as the intense similarity of their early home environments gives way to their own choices, they select environments that are less similar than their previous environments and about as similar as those of ordinary sibs. Adopted sibs, on the other hand, move from an early environment, in which parents may have produced similarity, to environments of their own choosing. Because their genotypes are hardly correlated at all, neither are their chosen environmental niches. Thus, by late adolescence, adopted siblings do not resemble each other in intelligence, personality, interests, or other phenotypic characteristics (Grotevant, Scarr, & Weinberg, 1977; Scarr & Weinberg, 1978; Scarr, Weber, Weinberg, & Wittig, 1981). The early environments of biological siblings, like those of adopted children, lead to trait similarity as a result of passive genotype -» environmental effects. As biological siblings move into the larger world and begin to make active choices, their niches remain moderately correlated, because their genotypes remain moderately correlated. There is no marked shift in intellectual resemblance of biological sibs as the process of active genotype —» environment influence replaces the passive one. Summary In summary, the theory of genotype —> environment effects proposed by Scarr and McCartney (1983) describes the usual course of human development in terms of three kinds of genotype-environment correlations that posit cooperative efforts of nature and nurture. Both genes and environments are constituents in the developmental system, but they have different roles. Genes determine much of human experience, but experiential opportunities are also necessary for development to occur. Individual differences can arise from restrictions in environmental opportunities to experience what the genotype would find compatible. With a rich array of opportunities, however, most differences among people arise

234

Sandra Scarr

from genetically determined differences in the experiences to which they are attracted and that they evoke. In addition, the theory accounts for observations from research on twins and families that no other developmental theory can. How do genotypes and environments combine? An argument for correlations over interactions One must distinguish environmental events that on the average enhance or delay development for all children from those that account for variation among children. There can be "main effects" that account for variation among groups that are naturally or experimentally treated in different ways. Within these groups of children there still remain enormous individual differences, some of which arise in response to the treatment. It is rare that the variation between groups approaches the magnitude of differences within groups, as represented in the pervasive overlapping distributions of scores. In developmental psychology, we have usually been satisfied if the treatment observed or implemented produced a statistically reliable difference between groups, but we have rarely examined the sources of individual variability within the groups. Two major hypotheses represented in this volume by Bronfenbrenner and his group, and by the present chapter, disagree about the principal sources of individual variability within groups. Bronfenbrenner proposes that genotype-environment interactions play an important part in explaining variability. I propose that g-e interactions may be present, but of minor importance, and that g-e correlations have greater effects on development and individuality. Linear and nonlinear effects Most often, the same treatments that alter the average performance of a group seem to have similar effects on most members of the group. Otherwise, we would find a great deal of variance in person-treatment interactions; that is, what's sauce for the goose would be poison for the gander. For the kinds of deprivation or interventions studied most often in developmental psychology, the main effects seem not to change the rank orders of children affected. The main effects are real, but they are also small by comparison to the range of individual variation within groups so treated or not. Some children may be more responsive than others to the treatment, but I doubt that there are many situations in which disordinal interactions are the rule. Very few children lose developmental points by participating in preschool programs or gain by being severely abused in adolescence. The search for aptitude-treatment interaction (Cronbach & Snow, 1971) and genotype-environment interactions (ErlenmeyerKimling, 1972) has not produced dramatic or reliable results (Plomin, 1986).

How genotypes and environments combine

235

In studies of adoptive and biologically related families, the correlation of children's IQ scores with the educational level of biological parents is about .35, whether or not the parents rear their children (Scarr & Weinberg, 1983). Adopted children on the average have higher IQ scores than their biological parents as a result of the influence of their above-average adoptive parents. Taken together these findings support the claim that treatments can have main effects without overcoming genetic differences in children's responsiveness to those environments. Adopted children have IQ scores above those of their biological parents, yet the correlations of adopted children are higher with their biological than adoptive parents (Scarr & Weinberg, 1977, 1978, 1983). The average effects of treatments, such as adoption, seem to increase the mean IQ scores, but they do not seem to alter the rank order of individual differences, on which correlations depend. In the theory just presented, emphasis was placed on the correlation of genotypes with their environments, on the nonrandomness of people, places, and experiences. Other theorists, such as Bronfenbrenner (in this volume) prefer to think in terms of person-situation interactions, such that different people respond differently to different events in their lives. Similarly, Elder and Caspi (in this volume) emphasize the different responses of men to stresses, depending on their prior characteristics. The theoretical preference for what I will call genotype-environment (g-e) interactions has a long and honorable history. There just is not much empirical support, however fondly one looks (Plomin, 1986; Erlenmeyer-Kimling, 1972). The quest for aptitude-treatment interactions has a similarly disappointing history (Cronbach & Snow, 1977). Few consistencies can be found unless one is willing to deprive bright children of suitable instruction and bore them to death with task-analyzed learning that is appropriate for the retarded. Under those circumstances, which no real-life teacher would implement, one can find consistent decrements in the performance of bright children and consistent increments in the learning of dull ones (Snow, 1982). I submit that is not the result we were seeking. Under normal instructional conditions, bright children learn more than dull ones; the main effect of intellectual differences overpowers any interactional one. Yet, one is intuitively drawn to the idea that different genotypes have differential responsiveness to different environments. Perhaps the problem is that in ordinary life people selectively expose themselves to correlated environments, thereby largely obviating any necessity to demonstrate g-e interactions. Children who are terrible at sports do not usually go out for the varsity team. Children who will make the varsity do not usually place themselves in remedial physical education. Thus, we fail to observe the interaction that could occur if varsity athletes were demoralized by slow walks around the soccer field and the phys-

236

Sandra Scarr

ically ungainly were required to run 20 laps around the field. There must be something to the appealing idea of interactions; perhaps they rarely occur in the real world of self-selection and natural assortment. In animal studies where genotype-environment interactions are regularly reported, investigators use highly inbred strains (and their crosses) and restricted versus more normal environments in their tests. It may be that there are no human analogues to such animal studies, so that, conversely, the animal results do not apply straightforwardly to human populations. In human behavior genetic studies, estimates are seldom made of genotypeenvironment interactions or correlations. In large part, this is because there are no powerful tests of such terms. Twin designs are not suitable for estimating g-e interaction, because, even if it exists, it could be assigned to either the genetic or environmental term. If both MZ and DZ co-twins react equally similarly (or equally dissimilarly), any g-e interaction will be part of the environmental estimate. If the greater genetic resemblance of MZ twins leads them to have more similar reactions to the same events than DZ twins, any g-e interaction will be included in the genetic term (Plomin, 1986). Because we do not know in advance if g-e interaction exists or how it works on individual differences, we cannot use the twin method to find out. Adoption studies offer a more favorable but still weak test of g-e interaction. Unlike animal studies, where the genotypes can be controlled through generations of selective breeding and environments manipulated at the will of the investigator, human populations do not lend themselves to such tidy analyses. And the adoption study is only a partial solution to the problem of estimating g-e interactions in human populations. Here are the reasons - first, the traditional ones from quantitative genetics, then my developmental ones. Where's the interaction? In biologically related families, the resemblance between parents and children is potentially composed, in traditional variance terms, of g, e, cov g-e, and g-e interaction. In adoptive families, the variance terms are limited to e alone; because there is no g-connection, there also cannot be g-e connections between parents and adopted children. Measures of the natural or birthparents of adopted-away children correlate with their offspring because of shared genotypes only. Because there is no e connection, there can be no g-e connections either. Thus, a cross-tabulation of IQ values for adoptive and natural parents can potentially estimate the main effects of g and e and reveal any effects of g-e interaction by nonlinearities in the combined main effects. The intellectual genotypes of adopted children can be very roughly estimated

How genotypes and environments combine

237

Table 9.3. Adopted children's IQ scores by adoptive parent IQ and natural parent education (Transracial Adoption Study) Adoptive midparent IQ Low Mid High S.D.

<11

N

12

N

>12

N

X

S.D.

99.8 106.2 106.5 104.2 (9.51)

16 14 8

107 107.8 106.6 107.1 (8.02)

22 33 31

106.7 105.8 110.2 107.6 (9.06)

18 12 18

104.5 106.6 107.8 106.3 (8.66)

(9.08) (8.23) (8.69)

Note: F Natural parent education = 2.58, p<. 10; F Adoptive parent IQ = 2.07; F natural X adoptive = 1.50.

from the educational phenotypes of their biological parents, who have had no postnatal environmental influence on them. Adopted family environments are roughly estimated from the phenotypic characteristics of the adoptive parents, who are not genetically related to the children. No one is really satisfied with these estimates, but they often show main effects for both genotypes and environments. Thus, one can assemble a cross-tab table with cells arrayed on dimensions from low to high values for both genetic and environmental components. An interaction could be found, if the high-high or low-low groups differ from expected values, based on linear combinations of the two sets of values. (Of course, one could find a curvilinear interaction, where the middle-middle group differed from others, but this is never predicted for g-e interactions..) Table 9.3 gives IQ data of young children from the Minnesota Transracial Adoption Study, arrayed to test for g-e interaction (Scarr & Weinberg, 1983). There is no interaction shown in this table, only a marginal main effect for estimated genotypes. Children with more highly educated birthparents score higher on IQ tests than those with less bright birthparents. But there are no g-e interaction effects such that the high-high or low-low groups deviate from linear prediction. The same lack of results was reported by Plomin, DeFries, and Loehlin (1977) in their reanalysis of the Skodak and Skeels adoption study data and by Plomin (1986, ch. 5) for the Colorado Adoption Project data. With older adolescents and young adults who were adopted in infancy, only genetic differences show main effects on their IQ scores. There are marginal environmental main effects and g-e interactions within the range of families sampled in the Minnesota Adolescent Adoption Study (Scarr & Weinberg, 1978, 1980). As Richard Weinberg and I have stressed repeatedly, however, the So-

238

Sandra Scarr

Table 9.4. Adopted children's IQ scores by adoptive parent IQ and natural parent education (Adolescent Adoption Study) Adoptive midparent IQ Low Mid High S.D.

>11

N

12

N

<12

N

X

S.D.

102.3 99 111.7 102.9 (9.9)

28 14 8

106.6 107.8 106.9 107.14 (2.9)

22 28 28

106.6 107.7 108.7 107.8 (9.1)

18 10 25

104.8 106.2 108.4

(8.8) (9.7) (8.0)

Note: F Natural parent education = 3.49, p = .033; F adoptive parent IQ = 2.81, p = .063; F natural X adoptive = 2.37, p = .054; R = .11; R = .07.

cioeconomic Status (SES) range of environments represented in this sample covers about three quarters of the U.S. white population. By the end of the childrearing period, differences among family environments have very slight, if any, effects on individual difference in intelligence test scores, as shown in Table 9.4. One can readily object to the nature of this test for g-e interaction and the lack of positive results. One major objection is that the effective environment is not measured, only indexed by parental phenotypes (Wachs, 1983). Although I agree, I remind critics that the correlation between adopted parents and their adopted children is an estimate of the total environmental effect of being reared in one family rather than another. For adolescents and young adults, correlations in IQ, personality, and interests between adopted children and their adoptive parents are very low. Because we know from other evidence (e.g., other adoption studies and studies of biological families) that parental IQ correlates more highly with child IQ than other measures of the home environment correlates with child IQ, adopted parents' IQ scores can be seen as a valid index of the quality of the home environment for intellectual development. If there is little correlation between parental and child IQ scores, is there much hope of finding other aspects of the home environment that correlate more highly? If environmental differences among families contribute so little to intellectual variation among genetically unrelated children, is there much hope for g-e interactions of the sort studied in quantitative genetics? A second major objection is that the kind of g-e interaction tested by quantitative genetic models is not what we had in mind. Developmentalists do not think in terms of genotype but of phenotypes, or observable characteristics. A child's observable characteristics could well influence the treatment he or she receives from parents and significant others, even without any genetic relatedness of the parties. Since the child's phenotype bears some relationship to his/her

How genotypes and environments combine

239

u

1 Figure 9.7: Genotype-environment interaction created by phenotype-environment interaction

genotype, the evocative and active kinds of genotype -» environment effects can generate phenotype-environment correlations and, therefore, g-e correlations. One can also imagine the differential responsiveness of different phenotypes to different treatments, which would presumably generate g-e interactions. In the absence of a clear methodology to study such effects, I can only speculate on what a model might look like. Such a model of disordinal interaction in shown in Figure 9.7. Because there is a less-than-perfect correlation between phenotypes and genotypes, owing to unpredictable developmental events, the measured interaction between phenotype 1 and 2 and treatments 1 and 2 should yield lesser degrees of interaction between genotypes 1 and 2 and the same treatments. If strong phenotype-environment interactions could be proven, then there would be presumptive evidence for some degree of g-e interaction. Unfortunately, the magnitude of the g-e interaction could not be estimated for normal behavioral characteristics. We have no methods by which to do so. The theory McCartney and I proposed accounts for individual and developmental differences in responsiveness to environments - differences that are not primarily interactions of genotypes and environments but roughly linear combinations that are better described as three types of genotype-environment correlations with deviations around a common regression line. Surely, some gen-

240

Sandra Scarr

otypes are more responsive than others to some treatments or events than others. Common sense, if not data, tell us so. But disordinal interactions seem to be rare in the natural world. Differences in the slopes of effects are seemingly more common but difficult to replicate from study to study. We prefer to retain the idea of g-e correlations as the more general and powerful phenomenon. Uncorrelated environments and correlated experiences The units to which the variances are assigned are probably wrongly conceived for a proper understanding of developmental processes. Most typically, the total phenotypic variance in a measured characteristic, such as IQ or personality test scores, is assigned to one of four major components - within family, between family, genetic, and environmental components. In addition, the interaction and correlation of genotypes and environments may be considered, if the data set permits (see Plomin, 1986). Unfortunately, as noted earlier, tests for genotypeenvironment correlations and interactions are neither easy nor powerful in human data. More important, if one believes as I do that genes and environments always combine to produce development and individuality, then the terms of the analysis are not designed to answer the developmental questions one wants to ask. Rather, we should ask how different people with their different genotypes negotiate their environments and glean experiences from them. What shapes experience? To what extent are environments correlated with individual differences and developmental status? I will now leave the world of variance analysis and enter the world of psychological theory. Bandura (1982) has argued that many important events in one's life are uncorrelated or random happenings. I have argued that virtually no experiences are uncorrelated with one's genotype. The slippery distinction in the two statements is between observable events and experiences. Experiences are cognitively processed events. That cognitive processing is a property of the individual and his/her developmental status is hardly arguable today. The components of experience, then, are exposure to events and their cognitive processing. One can, of course, think of catastrophes, such as the 1985 Mexico City earthquake and World War II, that are surely unconnected to one's individuality or period in development. Similarly, one does not choose to be born into the eighteenth or twentieth centuries, in India or Chile. Historians, anthropologists, and some sociologists are rightly interested in differences among people that are associated with different historical times and cultural places. Most psychologists assume the era in which they live, and most often the culture in which development takes place, as givens - contexts within which they study average human

How genotypes and environments combine

241

development and individual variation. It is within the current times and our own places that Bandura and I disagree about the importance of random versus correlated events and experiences.

Macro- versus micro-environments Earthquakes, wars, depressions, and other large-scale environmental events are uncontrollable, personally uncorrelated events in the lives of individuals. Even so, I do not agree with Bandura that many important experiences in one's life are random with respect to one's individual characteristics or developmental status. Experiencing such events is not uncorrelated to personal characteristics and personal resources. The impact of such large-scale events on experiences can be quite different for different people (see Elder & Caspi, this volume) and can be correlated with their genotypes. Even macro-environmental events can lead to correlated experiences. Experiences gleaned or constructed from immediate, face-to-face environments seem clearly correlated with one's cognitive processing and developmental status, as I proposed in earlier sections of this chapter. People actively select some events and selectively ignore others. Some events have personal meaning; others do not. The phenomenology of experiences is, I argue, correlated with the genotype of the individual perceiver and processor. The perception of meaning in events is a function of the intelligence, personality, and interests of the observer and has great influence on individual differences and developmental changes in behaviors. There are micro-environments with important "contingencies" and sequences of interactions that are so brief as to go undetected by the participant. Patterson (this volume) studies such sequences of coercive interplays between aggressive children and their beleaguered parents. The parry and thrust of these interactions are so rapid and automated that the participants seem largely unaware of what hit them. More important, the parents seem unaware of the impact their retreats in the face of child misbehavior have on the probability of the recurrence of such misbehavior. The child is also probably unaware of why the misbehavior "works" to reduce parental demands and interference. In such cases, the participants' phenomenology of events may be connected to their behaviors (He's a rotten kid; I can't handle him anymore), but they are powerless to change their behaviors without micro-contingency management.

Treatment versus etiology Treatment effectiveness, of which I have no doubt, is largely irrelevant to the question of how the child became aggressive in the first place, or how the child

242

Sandra Scarr

has created a family environment of discord and coerciveness that is surely correlated with the child's aggressive personality. One can argue, and I will agree, that the child's aggressiveness is exacerbated and reinforced by the unfortunate contingencies set up in the home, but I do not believe we can ignore the child's genotype in assessing the origins of the coercive home environment. Note that the definitions of cause are very different in the behaviorist and evolutionary traditions, a matter to which Buss (1984) addressed himself, as cited in Table 9.1. In the behaviorist tradition, questions about how the second-by-second interchanges of the aggressive child and his or her parents cause aggressiveness in the child are answered after the aggressive behavior is first emitted. From that point in time on (after the horse has left the barn?), contingency management is the treatment of choice. But what caused the child to be aggressive in the first place? In the evolutionary tradition that question is addressed by taking into account selection and genetic variability. Not every child is capable of setting up such exquisite torture for other family members, as Patterson would agree, I believe. Parents, too, vary in their abilities to manage an aggressive child, with and without tutelage. Some children are difficult for any parent to manage, but some manage far better than others. Some parents are so inept as to make problems of children whom most parents could manage with fair ease. Like children, parents' genotypes are correlated with their own environments.

Conclusions To understand development and individuality, we need a theory about the ways in which people determine their own experiences, such as the one we (Scarr & McCartney, 1983) have proposed. Such a theory must include both distal and proximal causes of behavior, but will favor the former as explanations of enduring developmental changes and individuality. Evolutionary views can subsume behaviorist ones by simply positing susceptibility to various forms of contingency as part of human nature (indeed, part of the nature of all sensate species). Behaviorist explanations are incomplete, as they fail to consider the enduring characteristics of the human species, the genetically programmed developmental process, and the ways in which individuality manifests itself in all areas of life. An adequate theory of development and individual differences must take into account the observations that everyone agrees upon from studies of families - the ordinary biological observations and the challenging data that come from studies of twins, adoptees, and twins reared apart.

How genotypes and environments combine

243

References Bandura, A. (1982). The psychology of chance encounters and life paths. American Psychologist, 37, 747-55. Bell, R. Q. (1968). A reinterpretation of the direction of effects in studies of socialization. Psychological Review, 75, 81-95. Bouchard, T. J. (1984). Twins reared together and apart: What they tell us about human diversity. In S. W. Fox (Ed.), Individuality and determinism. New York: Plenum. Buss, D. M. (1984). Evolutionary biology and personality psychology: Toward a conception of human nature and individual differences. American Psychologist, 39, 1135-47. Caldwell, B. M., & Bradley, R. H. (1978). Home observation for measurement of the environment. Little Rock: Arkansas. Cronbach, L. J. (1957). The two disciplines of scientific psychology. American Psychologist, 12, 671-84. Cronbach, L. J. (1975). Beyond the two disciplines of scientific psychology. American Psychologist, 30, 116-27. Cronbach, L. J., & Snow, R. E. (1977). Attitudes and instructional methods. New York: Irvington. Erlenmeyer-Kimling, L. (1972). Gene-environment interactions and the variability of behavior. In L. Ehrman, G. Omenn, & E. Caspair (Eds.), Genetics, environment and behavior. New York: Academic Press. Garmezy, N., Masten, A. S., & Tellegan, A. (1984). The study of stress and competence in children: A building block for developmental psychopathology. Child Development, 55, 97-111. Gottlieb, G. (1983). The psychobiological approach to developmental issues. In M. M. Haith & J. J. Campos (Eds.), Handbook of child psychology: Infancy and developmental psychobiology (Vol. 2). New York: Wiley. Grotevant, H. D., Scarr, S., & Weinberg, R. A. (1977). Patterns of interest similarity in adoptive and biological families. Journal of Personality and Social Psychology, 35, 667-76. Juel-Nielson, N. (1980). Individual and environment: Monozygotic twins reared apart. New York: International Universities Press. Kamin, L. J. (1974). The science and politics oflQ. Potomac, MD: Erlbaum. Matheny, A. P., Jr., Wilson, R. S., Dolan, A. B., & Krantz, J. Z. (1981). Behavioral contrasts in twinships: Stability and patterns of differences in childhood. Child Development, 52, 57998. Mayr, E. (1963). Animal species and evolution. Cambridge, MA: Belknap Press. Moos, R. H., & Moos, B. S. (1981). Family environment scale manual. Palo Alto, CA: Consulting Psychologists Press. Nerlove, S. B., & Snipper, A. S. (1981). Cognitive consequences of cultural opportunity. In R. H. Munroe, R. L. Munroe, & B. B. Whiting (Eds.), Handbook of cross cultural human development. New York: Garland. Newman, H. G., Freeman, F. N., & Holzinger, K. J. (1937). Twins: A study of heredity and environment. Chicago, IL: University of Chicago Press. Plomin, R. (1986). Genetics, development, and psychology. Hillsdale, NJ: Erlbaum. Plomin, R., & DeFries, J. C. (1980). Genetics and intelligence: Recent data. Intelligence, 4, 15-24. Plomin, R., DeFries, J. C , & Loehlin, J. C. (1977). Genotype-environment interaction and correlation in the analysis of human behavior. Psychological Bulletin, 84, 309-22. Scarr, S. (1976). An evolutionary perspective on human intelligence. In M. Lewis (Ed.), Origins of intelligence: Infancy and early childhood. New York: Plenum. Scarr, S. (1981). Comments on psychology: Behavior genetics and social policy from an antireductionist. In R. A. Kasschau & C. N. Cofer (Eds.), Psychology's second century: Enduring issues. New York: Praeger.

244

Sandra Scarr

Scarr, S. (1985). Constructing psychology: Facts and fables for our times. American Psychologist, 40, 499-512. Scarr, S. (1986). Protecting general intelligence: Constructs and consequences for intervention. In R. J. Linn (Ed.), Intelligence: Measurement, theory and public policy. Champaign, IL: University of Illinois Press. Scarr, S., & Carter-Saltzman, L. (1980). Twin method: Defense of a critical assumption. Behavior Genetics, 9, 527-42. Scarr, S., & Kidd, K. K. (1983). Developmental behavior genetics. In M. M. Haith & J. J. Campos (Eds.), Handbook of child psychology: Infancy and developmental psychobiology (Vol. 2). New York: Wiley. Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of genotype —> environment effects. Child Development, 54, 424-35. Scarr, S., Weber, P. L., Weinberg, R. A., & Wittig, M. A. (1981). Personality resemblance among adolescents and their parents in biologically related and adoptive families. Journal of Personality and Social Psychology, 40, 885-98. Scarr, S., & Weinberg, R. A. (1977). Intellectual similarities within families of both adopted and biological children. Intelligence, 1, 170-91. Scarr, S., & Weinberg, R. A. (1978). The influence of "family background" on intellectual attainment. American Sociological Review, 43, 674-92. Scarr, S., & Weinberg, R. A. (1980). Calling all camps! The war is over! American Sociological Review, 45, 859-64. Scarr, S., & Weinberg, R. A. (1983). The Minnesota adoption studies: Malleability and genetic differences. Child Development, 54, 260-7. Shields, J. (1962). Monozygotic twins brought up apart and brought up together. London: Oxford University Press. Snow, R. E. (1982). Education and intelligence. In R. E. Sternberg (Ed.), Handbook of human intelligence. Cambridge, MA: Cambridge University Press. Teasdale, T. W., & Owen, D. R. (1985). Heredity and familial environment in intelligence and educational level - a sibling study. Nature, 309, 620-2. Toulmin, S. (1973). Human understanding. Chicago, IL: University of Chicago Press. Wachs, T. D. (1983). The use and abuse of environment in behavior-genetic research. Child Development, 54, 416-23. Watson, E. H., & Lowrey, G. H. (1967). Growth and development of children. Chicago, IL: Year Book Medical Publishers. Wilson, R. S. (1983). The Louisville twin study: Developmental synchronies in behavior. Child Development, 54, 298-316.

Name index

Acland, H., 7 Acock, A. D., 62, 63 Adams, G. R., 174 Ageton, S. S., 143 Allen, A., 181, 204 Allen, L., 90 Allen, V. L., 154 Allport, G. W., 14, 205, 206 Alwin, D. F., 8 Applebaum, M. I., 12 Arnold, J., 137 Arnold, L., 198 Arthur, J., 123 Aschenbrenner, B. G., 10 Bachman, J. G., 172 Bailey, S., 103 Baldwin, A. L., 41, 42 Baldwin, J. M., 184 Baltes, P. B., 40-1 Bandura, A., 69, 102, 183, 190, 213, 228, 240, 241 Bane, M. J., 7 Bank, L., 137, 138, 139, 143 Barker, R. G., 41-2, 51, 52, 66, 67 Barrett, J. E., 93 Bartholomai, F., 28 Baszormenyi-Nagy, I., 53, 65 Bates, J. E., 144 Baumrind, D., 8 Bayley, N., 31^2 Beattie, M., 6 Beck, S. J., 196 Becker, H., 155 Beckmann, M., 44 Belsky, J., 10, 60, 95 Bell, R. Q., 13, 33, 229 Bern, D. J., 14, 15, 73, 98, 181, 204, 205, 210-11, 212 Bern, S. L., 208, 212

245

Bengston, V. L., 62, 63 Bentler, P. M., 137, 152, 164 Berger, P. L., 58 Bergeron, G., 181 Best, J. A., 173 Biglan, A., 123 Binet, A., 181 Bleuler, M., 81 Block, J., 87, 89-90, 95, 192, 214 Block, J. H., 95, 214 Blyth, D. A., 88, \64nl Bott, E., 53 Botvin, G. J., 174 Bouchard, T. J., 223, 230, 232 Bourdieu, P., 58«1 Bowers, K. S., 181 Boyd, E., 121 Bradley, R. H., 219 Brake, M., 155 Brandtstadter, J., 153, 154 Brehm, J. W., 108 Brehm, S. S., 108 Brody, G., 10 Bronfenbrenner, U., 1-2, 3, 5, 9, 12, 28-9, 32, 33, 35, 36, 38, 40, 42, 50, 52, 66, 71, 73, 89, 93, 145, 154, 155, 188, 203, 206, 234, 235 Brooks-Gunn, J., 17, 188 Brown, K. S., 173 Buckley, S., 155 Bugenthal, D. B., 71 Bullivant, B., 56 Burns, A., 55, 56, 59, 65, 66, 67, 72 Burton, L. M., 4 Burton, R. V., 137, 181, 182, 191 Bush, D. M., 88 Buss, D. M., 14, 217, 218/, 242 Cairns, B. D., 15, 16, 55, 181, 184, 191, 194, 195, 196, 198

Name index

246 Cairns, R. B., 15-16, 17, 55, 118, 120, 181, 182, 184, 187, 191, 194, 195, 196, 198, 199 Caldwell, B. M., 219 Calkins, V., 198 Campbell, D. T., 20 Campbell, J. D., 137, 181, 182, 191 Capaldi, D., 134 Caplan, G., 92 Carlton-Ford, S., 164«1 Carter-Saltzman, L., 231 Carver, C. S., 153 Cashmore, J., 54, 57, 63, 71 Caspi, A., 4, 6, 15, 36, 44, 46, 51, 60, 81, 82, 87, 92, 94, 97, 98, 99, 100, 132, 163, 210-11, 212, 241 Cattell, R. B., 209 Cauble, A. E., 91 Cavan, R., 92 Chamberlain, P., 141 Cheek, J. M., 204 Chess, S., 9 Child, I. L., 89 Clarke-Stewart, A., 10 Clausen, J. A., 62 Cobb, J. A., 121, 134 Cobb, S., 9 Cochran, M., 35 Cohen, D., 7 Cohen, J., \12n5 Cohen, P., \12n5 Cohen, S., 164 Coie, J. D., 125-6 Coleman, J. S., 7 Cook, T. D., 20 Cooley, C. H., 183, 185 Cottrell, L. S., 211 Covey, M. K., 122 Coyne, J. C , 11 Crockenberg, S. B., 37-8 Cronbach, L. J., 217, 234, 235 Cross, C , 92, 94 Crouter, A. C , 28-9, 33, 38, 42, 89, 154, 155 Daniels, D., 12, 13 Dantchik, A., 204 Darley, J., 14 Darwin, C , 217 Dasen, P., 69 D'Avernas, J. R., 173 Dawes, R. M., 120 DeFries, J. C , 15, 224t, 227, 228, 237 deLone, R. H., 38 Delugach, J. D., 17

Dengerink, H. A., 122 Deur, J. L.,94 Diener, E., 14 Dishion, T. J., 115, 120, 130, 133/, 134, 138, 139, 144/, 210 Dodge, K. A., 17 Dolan, A. B., 232 Douglas, M., 70 Downey, G., 36, 44, 46, 94, 99, 100 Dreher, E., 156, 157 Dreher, M., 156, 157 Duner, A., 152 Dunn, J., 10, 13 Durkheim, E., 106 Dusek, J. B., 172 Dwyer, J. H., 136, 137, 138-41, 143, 165 Easterbrooks, M. A., 10 Eckenrode, J., 37 Edwards, M., 53 Eichorn, D. H., 82 Eick, K., 155 Eigenbrodt, J., 155 Elder, G. H., Jr., 3, 4, 6, 15, 36, 42, 43, 44, 46, 51, 60, 77, 79, 81, 82, 84, 85-6, 87, 89, 91, 92, 94, 97, 98, 99, 100, 103, 132, 210-11, 212, 235, 241 Elliott, D. S., 143 Emery, R. E., 91, 95 Emmons, R. A., 14 Endler, N. S., 209 Engfer, A., 44 Entwisle, D. R., 55 Epstein, J. L., 9 Epstein, S., 181, 184, 192 Erlenmeyer-Kimling, L., 234, 235 Eron, L. D., 62 Esveldt-Dawson, K., 196, 198 Eyferth, K., 152, 155, 156, 162 Eysenck, H. J., 209 Farrington, D. P., 17, 145 Fazio, R. H., 14 Featherman, D., 105 Feldman, K. A., 91 Feldman, S. S., 10 Fend, H., 169 Ferguson, L. L., 191 Ferguson, T. J., 194 Fischer, A., 172 Fischer, R. C , 172 Fisher, R. A., 30 Flaherty, J. F., 172 Flay, B. R., 173 Fleischman, M. J., 137

247

Name index Foner, A., 80, 84 Forehand, R., 196 Forgatch, M. S., 131, 132 Foucault, M., 58 Frame, C. M., 17 Freeman, F. N., 232 French, J. P., 9 Freud, S., 213 Friedman, L., 123 Fuchs, W., 172 Fuhrer, U., 155 Funder, D. C , 181 Furstenberg, F. F., Jr., 13, 17-19, 61, 62, 63 Galton, F., 32 Gamble, W., 60 Garbarino, J., 66, 71 Gardner, K., 120 Gardner, W., 147 Gariepy, J.-L., 191 Garmezy, N., 81, 88, 95, 230 Gewirtz, J. L., 121, 181 Gibbons, F. X., 204 Gibson, J. J., 155 Gintis, H., 7 Glueck, E., 94 Glueck, S., 94 Goldberg, W., 10 Goode, W. J., 95 Goodenough, D. R., 212 Goodnow, J. J., 8, 54, 55, 56, 57, 59, 63, 65, 66, 67, 71, 72 Gottlieb, G., 227 Gottman, J. M., 94, 118, 123, 124, 141 Grabow, S., 155 Grajek, S., 12 Green, J. A., 182 Green, K. D., 196 Greenberg, J., 183 Greenwald, A. G., 182, 184 Griffin, W. A., 147 Groeneveld, L. P., 20 Grotevant, H. D., 233 Guiller, H., 146 Gulotta, T., 174 Gunnel, P. K., 121 Gurin, G., 6 Gurin, P., 6 Guttman, L. A., 126 Handel, G., 62 Hannan, M. T., 20 Hansen, D. A., 8-9 Hareven, T., 53 Harms, G., 155, 162

Harter, S., 134, 182 Hartmann, H., 53 Hartshorne, H., 181, 182, 205, 212 Hauser, R. M., 8 Havighurst, R. J., 153, 156, 157 Hayduk, L. A., 55 Heath, S. B., 55 Heider, F., 53, 185 Henri, V., 181 Herjanic, B., 198 Hess, R. D., 62 Heyns, B., 7 Hinde, R. A., 10, 11, 53, 93 Hineline, P., 121 Holahan, C. J., 155 Holzinger, K. J., 232 Honzik, M. P., 90 Hood, K. E., 16 Hops, H., 123, 134 Homey, K., 212 Huba, G. J., 152, 164 Huesmann, L. R., 62 Huizinga, D., 143 Hunt, J. McV., 209 Jackson, B., 68 James, W., 102 Jaworski, B., 91 Jencks, C . , 7 Jennings, M. K., 62 Jessor, R., 152, 154, 164, 174 Jessor, S. L., 154, 164 Johnson, L. D., 172 Johnson, M., 80, 84 Jones, D. R. W., 155 Jones, E. E., 181 Joreskog, K. G., 42, 165 Joy, C. B., 91 Juel-Nielsen, N., 232 Kagan, J., 31«2, 88 Kain, E. L., 81 Kamin, L. J., 231 Kaminski, G., 155 Kandel, D. B., 152, 164 Kantor, B., 20 Kaplan, H. B., 163 Kastner, P., 153, 157 Kazdin, A. E., 196, 198 Kelley, H., 181 Kelly, G., 208 Kendrick, C , 10 Kendrick, D. T., 182, 204 Kendzierski, D., 14 Kenny, D. A., 139

Name index

248 Kersell, M. W., 173 Kessler, R. C , 11, 164 Kidd, K. K., 227 Knight, R., 57, 71 Knorr, G. L., 134 Kohlberg, L., 183, 208 Kohn, M., 4, 5, 6, 89, 110 Kraines, R. J., 62 Krantz, J. Z., 232 Kumka, D., 46 Kuo, Z. Y., 184 Kupersmidt, J. B., 126 Lamb, M. E., 118, 120 Langer, E. G., 121 Larson, R. J., 14 Lautrey, J., 57, 69 Lav, R. C , 6 Ledingham, J. E., 181 Lee, C. L., 144 Lefkowitz, F., 62 Lehmann, H. C , 204 Lerner, J. V., 172 Lerner, M., 53 Lemer, R. M., 172 Levenson, R. W., 123, 124 Levin, H., 137 Levine, A., 137 Le Vine, R. A., 2 Levinger, G., 53 Lewin, K., 1, 13, 25, 26/zl, 27, 28, 31, 40 Lewis, C , 56 Lewis, M., 73, 188 Liddle, C , 17, 102 Liebermann, M. A., 93 Liker, J. K., 91, 92, 94 Linville, P. W., 182 Lippmann, Walter, 78 Lipsett, L. P., 41 Loar, L. L., 196, 198 Loeber, R., 138, 139 Loehlin, J. C , 15, 227, 228, 237 Logan, J. A., 152 Lorber, R., 125 Lorence, J., 46 Lowrey, G. H., 227 Lynd, H. M., 78 Lynd, R. S., 78 Lytton, H., 15-16, 182 McAvay, G., 31 McCall, R. B., 12 McCall, R. R., 16 McCartney, K., 15, 210, 226, 230, 233, 239, 242 Maccoby, E. E., 11, 12, 66, 137

McCord, J., 94 McCubbin, H. I., 91 Macfarlane, J. W., 84, 90 McLeod, J. D., 11 Madge, N., 88 Magnusson, D., 152, 154, 183, 199, 209 Margulies, R., 164 Markus, H., 182 Marlowe, J. H., 137 Marsden, D., 68 Martin, J. A., 12 Martin, S. S., 163 Masten, A. S., 230 Matheny, A. P., Jr., 232 Maughan, B., 7 May, A., 205, 212 May, M., 155 May, M. A., 181, 182 Mayr, E., 217 Mead, G. H., 183 Mead, M., 55 Medrich, E. A., 155 Megargee, E. I., 213 Mehan, H., 73 Menaghan, E. G., 93 Merton, Robert, 79, 80 Michelson, S., 7 Mills, C. W., 3-4, 110 Mills-Byrd, L., 199 Minuchin, P., 10 Mischel, W., 181, 204, 206-7, 213 Moen, P., 6, 66, 71, 81 Moorehouse, M., 5 Moos, B. S., 219 Moos, R. H., 219 Morgan, S. P., 17 Morrison, A., 95 Mortimer, J. T., 46 Mortimore, P., 7 Moss, H. A., 31^2 Muchow, H., 155 Muchow, M., 155 Muchowski, L., 157 Mullen, J. T., 93 Murray, H. A., 209 Myers, I. B., 212 Nash, S. C , 10 Neckerman, H. J., 191 Needle, R. H., 91 Nerlove, S. B., 227 Neugarten, B. L., 62 Newcomb, T. M., 91 Newman, B. ML, 156, 157 Newman, H. G., 232 Newman, P. R., 156, 157

249

Name index Niemi, R. G., 62 Nisbett, R. E., 121, 181 Noack, P., 14, 70, 105, 154, 156, 157, 160, 162 Nurius, H., 182 Olweus, D., 145, 209 O'Malley, P. M., 172 Osteen, V., 123 Ouston, J., 7 Owen, D. R., 224t Parke, R. D., 10, 73, 94 Passeron, J., 59 Patterson, G. R., 11, 15, 16-17, 94, 96, 99, 115, 116, 118, 120, 121, 122/, 123, 124, 127, 130/, 131, 132, 133/ 134, 135/ 136, 137, 138, 139, 141, 142, 143, 144/ 145, 146, 210, 211, 213, 241, 242 Patterson, J. M., 91 Payne, R. T., 155 Pearlin, L. I., 93 Perrin, J. E., 198 Perry, C. L., 174 Pervin, L. A., 9 Peterson, A. C , 153 Peterson, D. R., 204 Piaget, J., 184, 227 Pillegrini, A. D., 10 Plomin, R., 12, 13, 15, 219, 220, 224r, 227, 228, 234, 235, 236, 237, 240 Poole, M., 56, 68 Preissing, C , 155, 162 Pysczczynski, T., 183 Quinton, D., 17, 102 Ranck, K., 92 Reese, H. W., 41 Reich, W., 198 Reid, J. B., 115, 118, 120, 125, 130/ 133/ 137, 141, 144/ 210 Reitzle, M., 165, 169 Richtermeier, A., 155, 162 Riley, D., 37 Riley, M. W., 3, 80, 84 Robbins, C , 163 Robins, E., 60 Robins, L. N., 90, 97, 102 Robinson, J. P., 126 Rockwell, R., 103 Rodgers, W., 9 Rogosa, D., 139, 142 Roizen, J., 155 Rollins, B. C , 62 Rosenbaum, J. E., 8

Rosenfeld, H. M., 121 Rotter, J. B., 213 Rubin, V., 155 Rubin, Z., 66 Rule, B. G., 192 Russell, Graeme, 54 Rutter, M., 7, 17, 87, 88, 95, 102, 146 Ryan, K. B., 173 Ryder, N., 79-80 Sackett, G. P., 120 Salkind, N. J., 155 Sameroff, A. J., 145 Sarason, S. B., 19 Scarr, S., 12, 15, 31, 105, 182, 209-10, 219, 220, 223, 224r, 225, 226, 227, 228, 230, 231, 232, 233, 235, 237-8, 242 Scarr-Salapatek, S., 43 Schacter, F. F., 10 Schaefer, W. S., 31«2 Schaffer, R., 92 Schaie, K. W., 40-1 Scheier, M. F., 153 Schlundt, D. G., 17 Schnedler, R. S., 122 Schneewind, K. A., 44 Schneirla, T. C , 184 Schocken, I., 17 Schoggen, P., 52 Schooler, C , 4, 89, 110 Schroer, S., 169 Schuck, J. R., 137 Schuhler, P., 157 Schulz, R., 19 Schwabe, H., 28 Schwartzman, A., 181 Schwarzer, R., 169, 172 Sears, R. R., 137, 182 Sears, S., 181 Seaver, W. B., 10 Seifer, R., 145 Sewell, W. ML, 8 Sharrock, W. W., 58 Shennum, W. A., 71 Sheridan, A., 58 Sherman, L., 123 Shields, J., 232 Shweder, R. A., 57, 58 Sidman, M., 141 Sigel, I. E., 10 Silbereisen, R. K., 14, 70, 105, 152, 153, 154, 155, 156, 157, 162, 163, 165, 169 Simmons, R. G., 88, \(An\ Skinner, B. F., 120, 207 Skinner, M., 131, 132 Sloman, J., 66

Name index

250 Smith, M., 7 Smith, R., 145 Smith, R. S., 88 Smith, T. E., 63 Snipper, A. S., 227 Snow, R. E., 234, 235 Snyder, J. J., 123 Snyder, M., 14, 204 Sorbom, D., 42, 165 Sorenson, A., 105 Spark, G., 53, 65 Spencer, M., 56 Spiga, R., 153 Spyrou, S., 120 Stattin, H., 154 Stephenson, G. R., 118, 120 Stevenson-Hinde, J., 10, 11 Steward, M., 61, 64-5, 70 Stodolsky, S., 58, 67 Stone, C. P., 41-2 Stouthamer-Loeber, M , 134, 136, 138 Stringfield, D. O., 182, 204 Strober, M. H., 53 Sullivan, H. S., 211 Suomi, S. J., 118, 120 Swann, W. B., Jr., 14 Syme, S. L. , 164 Tal, M., 11 Taplin, P. S ., 125 Teasdale, T. W., 22 Tedin, K. L . , 6 3 Tellegan, A,., 230 Tesser, A., 10 Thibodeaux, S., 10 Thomas, A.., 9 Thomas, D. L., 62 Thomas, D. S., 209 Thomas, W. I., 60, 78-9, 106, 107-8, 209 Thorsen, C , 123 Thrasher, F. M., 155 Toulmin, S., 219 Troll, L. E., 62 Tulkin, S. R., 34-5, 45, 46 Tuma, N. B., 20 Turnbull, J., 11 Uchtenhagen, A., 173 Valsiner, J., 67 Van Cleave, E. F., 88

Van Nguyen, T., 36, 46, 51, 60, 81, 82, 87, 92, 132 Vaughn, R., 164 Vernon, P. E., 204 Viken, R., 131 Volkart, E. H., 78 von Bertalanffy, L., 1 Vosk, B., 196 Wachs, T. D., 6, 238 Wachtel, E. F., 19 Wachtel, P. L., 14, 19, 210, 211 Wallwork, E., 184 Watson, E. H., 227 Weber, P. L., 233 Weinberg, R. A., 224f, 230, 232, 233, 235, 237-8 Weinrott, M., 137 Werner, E. E., 88, 145 West, D. J., 17, 145 Whiting, B. B., 66 Willis, P., 155 Willoughby, T. L., 198 Wills, T. A., 164 Wilson, R. S., 232 Wilson, T. D., 121 Witkin, H. A., 212 Wittig, M. A., 233 Witty, P. A., 204 Wohlwill, J. F., 6, 155 Woo, D., 122 Woolever, F., 35 Wortman, C. B., 11 Wright, H. F., 51, 52, 66, 67 Wright, S., 42 Wrong, D., 13 Yarrow, M. J., 137 Yarrow, M. R., 181, 182, 191 Younger, A., 181 Youniss, J., 117 Zahn-Waxier, C , 11 Zank, S., 165, 169 Zanna, M. P., 173 Zetterboom, G., 152 Zigler, E., 89 Zimmer-Hofler, D., 173 Zinnecker, J., 172 Znaniecki, F., 78, 107-8

Subject index

abstract thinking, 187, 189, 192 academic failure, 131, 134-6, 142, 143 accentuation hypothesis, 60, 91 access, in description/perception of settings, 67-70; meanings of, 66-7; observations prompting questions about, 64-6; perceptions of, 68, 70, 71; and qualities of people, 70-3 access to information, 8, 58, 67; blocked by adults, 58; effect on development, 58 achievement: prospects for, 98, 103; training for, 89 action-reaction sequences, 117, 118, 120 action-theoretic approach, 153 adaptation(s): in attempts to regain control, 108, 109; to economic hardship, 82, 96, 107; new, 225; options and resources in, 85-6; problem behavior as, 90 adaptive functions: of self cognitions, 184, 190, 199-200; of self evaluations, 191 adaptive resources: marital integration as, 92 adjustment, 212, 219-20 adolescence: change in, 187-9; constructive role of problem behavior in, 152-80; insider status in, 58; self concepts in, 194-5; selfother ratings in, 192-3; leisure time as context of development in, 154-6; effects of Depression in, 84, 87-8; as mother, 17-18; and peer group, 117; role in shaping own development, 153 adopted siblings, 223, 230-3, 235 adoption studies, 12, 236-8, 242 adoptive families, 224; parent-child IQ correlations in, 235; parent-child resemblance in, 236-8 adulthood, 42; transition to, 97, 103, 104, 152, 154 adults: vested interests in advice of, 56, 57 age, 100; and effects of economic hardship on development, 86-8; and perceptions of set-

251

tings, 71-2; as personological construct, 40-1; in society and history, 79, 80; social meaning of, 77 age-graded roles, 97-9 age strata, 80 aggression, 95; etiology/treatment of, 241-2; irritability and, 96; microsystem and intervention in, 11-12; self-other ratings of, 194, 197; social networks and, 16-17 alcohol use, adolescent, 160, 161; and selfesteem, 167, 169-73 alienation, 173-4 anarchy progression, 11 antisocial behavior, 94; determinants of, 142; discipline and, 146-7; feedback loops in, 142-3; and homework completion, 134; in school, 142-3; stages in, 129-36, 210; temperament and, 145; training for, 132-4, 138; see also aggression; coercion process; problem behavior antisocial traits: progression in, 125-8, 129 aptitude-treatment interaction, 234, 235 arbitrariness: in parenting, 94, 96 aspirations, 106-7, 108 attitudes: and behavior, 14; intergenerational transmission of, 91, 99-102 authoritarian personality, 208 authority, 211 autonomy values, 4-5 aversive behavior: learned, 129, 211; see also antisocial behavior; problem behavior aversive stimuli, 120-1, 122 behavior(s): effect of economic hardship on, 91; as function of person and environment, 203, 208-9; generalized from family to peer group, 116-17; intergenerational transmission of, 91, 99-102; across life span, 9 9 100, 115-16; nomothetic classification of, 207-8, 212, 213; school, 7-8

252 behavior genetic studies, 10, 12, 15; family resemblances/parent-child effects, 223-5; genotype-environment interactions, 236 behaviorism, 206-8, 213, 242 beliefs: and family management skills, 145-6; about self and reality, 189-91 Berkeley Guidance Study, 15, 78, 84-5, 8 5 9, 89-105 Berlin Longitudinal Study on Adolescent Development, 155-6 bidirectional influences, 42-3; in coercive exchanges, 132, 147; person/context, 163-4 biological famlies: correlations within, 224; genotype —> environment effects in, 229; parent-child IQ correlations in, 235; parentchild resemblance in, 236 biophysical change, 187-8 catastrophes, 240-1 cathartic activities, 163, 174, 175; as mediator between self-esteem and substance use, 164, 167, 169-72 causal attribution, 181 causal models, 42-3, 79; effects of economic hardship on development, 87, 92; and theory construction from observations, 219 cause/causality: in genotype/environment combination, 220, 221, 223 Centre for Contemporary Cultural Studies, Birmingham, 155 change, 3, 91, 93; in development, 40, 44, 229-30; family and, 114, 138; in life experience, 109; microsocial mechanisms for, 118-36; in personality, 97; process of, 114, 115-16; social /individual, 79; see also social change Chicago School of Sociology, 78 children: access to wider society, 65, 66; as agents of society, 65; in Depression, 86-9; difficult, 144-5, 241-2 (see also aggression); effect of stress on, 81-9; and genotype-environment combination, 219-20; influences on family relations, 93-4; relationships with families and communities, 50-76 cigarette use, 160, 161, 162, 164; and selfesteem, 167, 169-73, 174, 175 class; see social class class-theoretical models, 27-32 coercion process, 8, 115, 116-17; dyadic exchanges in, 118, 147, 241-2; mechanisms in, 117; microsocial structure and, 123-5, 128-9; production of change in dyadic exchanges, 120-3; stages in, 120, 123-5, 129-36, 142; testing stage model of, 136— 43

Subject index cognition: in self-organization, 183; see also self cognitions cognitive development, 69, 184, 187, 189; cultural differences in, 227; Piagetian theory of, 57; sex typing in, 208 cognitive style, 212 cohort(s), 110; as concept for studying life course, 80; confounding effects of age and, 41; and historical events in life course, 84 Colorado Adoption Project, 237 communication: changes in abilities in, 187, 188; self cognitions in, 184, 185 community(ies), 2, 66-7; description of, 52; and parent-child interactions, 64-5; relationships with children and families, 50-76 competence, 85, 91; in stressful situations, 81 competition, 51, 53, 65-6 conduct disorders, 116-17; see also antisocial behavior; problem behavior context(s), 2; family in, 143-6; hierarchy of systems of, 38-9; in human development, 77; and interventions, 20; in life-course perspective, 77, 79, 85, 100; process and, 207-9; in process-person-context model, 36, 38 continuity: in development, 40, 44; interactional/cumulative, 98-9; in personality, 97 control, 59, 106-8; over access, 51; force in, 94; locus of, 37; perceived, 19 control cycles (model): and evolving life course, 105-9 control potential, 109; loss of, 106-7 coping mechanisms: social support and, 164 coping strategy, 96; problem behavior as, 152, 154, 174-5 cross-sectional research design, 41-2 cultural differences: in cognitive development, 227 cultural values: transmission of, through parents, 61-4 culture, 50, 240; continuity in, 51; description of, 52; influence on development, 3-6; macrosystem as blueprint of, 39; and personality variables, 212-13; relation of family to, 114; of schools, 7 cumulative continuity, 98-9, 210 danger: perceptions of, 69-70 defense mechanisms, 212, 213 delinquency, 78, 116, 131, 134, 138; risk of, 145-6 depression, 143; effects on family relations, 11; in mothers, 123; rejection and, 131, 136; self appraisal of, 183 deprivation, early, 85; effects of, 17, 77; fami-

Subject index ly, 82-4, 87-8, 90-105; see also economic hardship determinism (genetic), 224-5, 230 development: agentic conceptualization of, 14; and average tendencies/individual variation, 218, 225; defined, 58; delayed effects in, 142-3; effect of maternal employment on, 5-6; family income loss and, 82-4; family relations and, 10, 11, 12-13; genetically programmed, 242; genotype —* environment effects in, 226-8, 229, 233-4, 240; holistic models of, 184-5; and individual differences, 217-44; and individual differences in information processing, 225-6; macrosocial influences on, 3-6; opportunities for, 55; parental knowledge about, 220-3; and problem behavior, 153-6, 163; proximal influences in, 6, 7-13; realignment of, in self concepts, 186-9; self-esteem in, 163; separation of genetic and environmental effects on, 227-8; as series of behavioral reorganizations around developmental tasks, 100; shifts in perceptions of settings as form of, 70; stability of, and enhancement of self, 193-4; substance use in, 173-4 development as action in context (concept), 152, 153, 154 developmental change, 40, 44; in genotype —> environment effects, 229-30 developmental integration, 191, 199; self cognitions in, 184-5, 189-90 developmental outcomes, 31; connections between settings in, 55, 59; defining, 44-6; effect of environment on, 34, 50; participation and, 52; research paradigm and, 26 developmental status: correlation of environments with, and individual differences, 2402; measures of, 47 developmental tasks, 97, 100, 153; adolescent, 156, 157; substance use in mastering, 174-5 developmental theory, 1-3, 182-3, 224-6, 228-9, 242; genotype—environment correlations in, 228-9 deviance, 131; in defense of self, 163 discipline, 127, 134, 141; and child behavior, 136-7, 138, 146-7; consistent, 94-5; effects of changes in, 136; effects of stress on, 132; inept, 131-2, 242; marital discord and, 96 disposition(s), 116; and behavioral style, 99; and economic hardship, 91, 93; generality of, 181; and selection of environment, 15210 division of labor: in children-families-communities relationship, 65-6; in family, 53-4 divorce, 80, 98, 105, 107, 146

253 dizygotic (DZ) twins, 15, 223, 230-3, 236 dreams: self cognitions in, 184, 186 drugs: adolescent use of, 157, 160, 163, 173 dyadic exchanges: as mechanism for change in families, 118-20, 147; microsocial structure and, 123-5; production of changes in, by coercion process, 116, 120-3, 128-9, 131

ecological perspective, 1-2, 3, 9, 12, 47, 93 economic conditions, 30; contextual analysis of, 85; effect on family functioning and child development, 51, 53, 82-4, 96; stress imposed by, 59 economic hardship: and development, 44, 89105; linkages to children's behavior, 90, 94-6; pre-crisis histories and, 90, 91-4 education, 7, 8; in life course, 97, 98, 102, 103; and military service, 105 ego: compromising, 189-91, 199 ego defense, 213 emotional support, 60, 61 enhancement model, 50-1, 58-61, 64, 65, 66, 70-1 environment(s), 10, 12, 31, 50; in behavior, 203; in control of self concepts, 183; correlation with individual differences and developmental status, 240-2; and development, 2, 3, 11-13, 32-3, 34, 226, 227-8; and family relationships, 10, 12-13; genotypes and, 15-16, 209-10, 217-44 {see also genotype —» environment effects); individual's effects on, 13-16; and intelligence, 224; linkages between levels of, 2; persons in reciprocal transaction with, 21011, 214; in process-person-context model, 38-40; representations of, 1-13; in research paradigms, 26; shaped and selected by individuals, 99, 210, 214, 2^8, 229, 231, 2356; in social address model, 29; uncorrelated, 240 environmental change, 88, 89, 110; children's response to, 90; differential effects of, 8 5 6; individual in, 105-6; vulnerability to, 86-8 environmentalism, naive, 224-5, 230 environmental similarity: model of, 230-2 equivalence classes, 204, 205, 206-7, 212, 213, 214 escape-avoidant conditioning, 121 events: random vs. correlated, 240-2 evolutionary theory, 217, 225, 242 exosystem, 2, 38-9, 47 experience(s): change in, 109; correlated, 240; directed by genotypes, 15-16, 225, 226, 227, 229, 233-4; events shaping, 240-2;

254 experience (cont.) individual differences in, 225-6; role in development, 225, 233 family(ies), 2, 3, 10, 39; in context, 143-6; as developmental context, 3, 9-13; and educational/occupational achievement, 9, 31; and genotype-environment combination, 219-20; single-parent, 132, 146; social position of, 5; and state, 65-6; study of change in, 138; as unit(s), 50, 52-4; variable responses to economic hardship, 82-4, 90; see also adoptive families; biological families; parent-child interactions family dependency, 83, 86, 87, 96 family deprivation: effects on development, 82-4, 87-8; in Great Depression, 90105 family discord, 84, 87 family disorganization, 78, 91-2; economic hardship and, 93-4, 96 family interactions: aversive, 94, 100, 101, 102; destructive behavior patterns in, 99, 211; developmental research on, 32; as predictor of grade point average, 8-9; see also parent-child interactions family management, 94, 96, 131; beliefs, values, neighborhood and, 145-6; causal status of, 138-9, 141; and child behavior, 136-7, 146-7; temperament and, 145 family networks, 53 family relationships, 53-4; change in, in response to economic hardship, 83, 84; child influences on, 96; with children and communities, 50-76; effect of economic hardship on, 91-2; interdependence of, 10-12, 16; problem behavior in, 100-2; see also family interactions family research, 12, 229; accounting for observations from, in genotype —» environment effects theory, 227, 230-2, 234, 242 family resemblances, 223-5 family roles, 90 family stress, 93 family systems, 65; analysis of, 53 fathers, 54; and effects of economic hardship, 92-3; influence of work experience of, on intellectual functioning of sons, 89; role in mother's treatment of child, 10; role of, 5, 90 feedback loops, 117, 142-3; positive, 131, 138 fictions: adaptive, 186, 199; self-enhancing, 189 Field Dependence/Independence, 212 field-theoretical formulations, 27, 28 fighting, 127-8, 129, 138

Subject index "fit": person/context, 163, 172; person-environment, 9 functional equivalents: for problem behaviors, 174-5 gender constancy, 188-9 generalization/generalizability: boundaries of, 84; of processes, 38; of scores in stage models, 137-8; of social patterns, 116 genetic differences: and individual differences in experience, 226 genetic effects in development: separation of environmental effects from, 227-8 genotype-environment (g-e) correlations, 234, 239-40 genotype —> environment effects: and changing similarities among siblings, 232-3; developmental change in, 15-16, 229-30; and family research, 230-2; model of, 226-7; propositions in, 228-9; theory of, 225-34 genotype-environment (g-e) interactions, 23440; objections to theory of, 236-40 goals, adolescent, 153, 157-8; leisure-time activities in, 174-5; related to developmental issues, 175-6; pursued in leisure settings, 162; strategies in attainment of, 158— 60; substance use and, 163, 174 goals: self cognitions in, 184, 186 group constraints: and interventions, 19 health: effect of economic stress on, 88; impact of social networks on, 164; views about, 70 health services, 37-8 historical change: influence on development, 3-6 historical contexts, events, 2; cohorts in, 84; effect on life course, 79, 110; life patterns in, 79-80; timing of, 88 identical twins; see monozygotic (M2) twins idiographic approach: content in, 207, 208, 212, 213 idiographic balance: self concepts, 194-5 idiographic/morphogenic approaches to personality, 204-6 individual(s): as agent of change, 114; contribution to own development, 3, 20, 38; effects on environment, 13-19; selection of environment(s), 99, 210, 214, 228, 229, 231, 235-6 individual differences, 183, 206; in coping with stress, 91; correlation of environments with, and developmental status, 10, 227-8, 240-2; development and, 217-44; in evolutionary theory, 225; genetically determined,

Subject index 227-8, 233-4; in intelligence, 223-5; in overcoming limitations of problematic childhood, 102; in perceptions of world, 71-3; predictability of, 16; in response to drastic change and stress, 93; in self-other ratings, 192-3; stability of, 118; theory of, 242 individual-differences approaches to personality, 205-6 individual-family functioning: interactive view of, 100-2 infants, infancy, 227; genotype —> environment effects in, 229-30 information domains: open/restricted, 58 information: managed (prepackaged), 57-8, 60, 63-4, 67; routes to gaining, 60; see also access to information information processing: individual differences in, 209-10, 212, 225-6 instrumental support, 60, 61 intellectual flexibility: work and, 4-5 intelligence, 230; environment and, 219-20; individual differences in, 223-5; parental knowledge and, 220-3 interactional approach: life/social history, 7 9 80; to personality, 203, 208-11; to study of families and individuals, 106 interactional continuity, 98-9, 211 interaction: meanings of, 182, 209, 210-11; phenomenological approach to, 209 interactions: between features of individuals/families and settings, 70; community/individual, 51; face-to-face, 2, 50; interpersonal, 32; intrapsychic, 213-14; see also family interactions intergenerational transmission of behavior and attitudes, 91, 99-102 internal synchrony, 184, 186-7 intervention(s), 3; improving intelligence, 222; individual-environment effects and, 11-12, 19-20 irritability, 116; in family exchanges, 143-6; parental, 96, 132, 147; and response to economic hardship, 91, 93-6 knowledge: openness of, 58; of parent, and intelligence of child, 220-3; and power, 58; relevance of, 58; tagged, owned, 58 legal policies: effect on family, 51 leisure activities: substitutive, 174-5 leisure settings: developmental issues in, 15662; role in mastery of developmental tasks in adolescence, 152, 163-73 life course: age-graded, 108; alteration in, 109; childhood behavior and adult, 90, 97-9; control cycles and, 105-9; families in, 88;

255 military service in, 102, 103-5; occupational attainment in, 46; study of development in, 42-3, 44; turning points in, 91, 102-5, 214 life-stage hypothesis, 80, 84, 85-6; in comparison of Oakland and Berkeley cohorts, 85-9 life transitions, 91, 102-5, 214; in chronosystem paradigm, 42; see also transitions linkages: between antecedents and outcomes, 89-90; between developmental contexts, 8 9; between economic hardship and children's behavior, 90, 93-5; between social change and human development, 77, 78; between social change and life patterns, 110; family patterns as, 83/ longitudinal studies, 31, 77, 78-9, 88; active role of, in life-course perspective, 80, 85; individual's effect on environment, 15; juvenile delinquency, 116, 134; occupational attainment, 46; person-context link in, 16, 17-19; problems of testing stage models in, 136-43; self concepts, 191-8; randomized, 141-2 macrosocial level, 2, 117, 146; behavior changes at, 120; influences on development, 3-6; integration with microsocial, 131, 147 macrosystem, 3, 39 marital discord, 98, 105; in linking economic hardship to children's lives, 94-5; problem behavior and, 100, 101, 102 marital relationships, 10, 143; child influences on, 96; economic hardship and, 84, 91-2 marriage: idealization in, 107; in life course, 97, 98, 102 maturation, 105, 115, 225, 227 measurement, 118; in stage models, 137-8; in study of self, 199, 200 measurement nonconvergence, 181-2 measurement theory, 182-3 menarche, 41-2, 154 mesosystem, 2, 9, 38, 39, 47 Michigan Panel Study of Income Dynamics, 79 micro-environments, 241 micromodels, 82-4 microsocial analysis, 120 microsocial environment: integration with macrosocoial, 131-4, 147 microsocial exchanges: changes in, 120-3; patterns of, 147 microsocial mechanisms for change, 118-36 microsocial processes: effect of discipline on, 136-7 microsocial progression: integration with trait

256 microsocial progression (cont.) progression, 128-9; target child and sibling exchange, 123-5; traits define, 125-8 microsocial structure, 123-5 microsystem, 2, 38, 39 microsystem models, 11, 32-3 military service, 97, 98, 102, 103-5, 109 Minnesota Adolescent Adoption Study, 237-8 models: class-theoretical, 27-32; joint functions in, 26-7; parent effect on child intelligence, 222-3; see also under individual models, e.g., process-person-context model moderating variable strategy, 204-5, 206 monozygotic (MZ) twins, 15, 219, 223, 2303, 236; reared apart, 230, 232 morphogenic approach to personality, 204-6 mother(s), 54, 118; employment of, 5-6; middle class, 34; parenting skills, 145; single, 132, 147; teenage, 17-18, 37 mother-child relations: and effect of economic hardship, 93-4 mother—infant interactions, 34, 35; support networks in, 37 motivation, 187-8 multimethod research, 182; in developmental psychology, 198, 199; in personality, 182-3 multiple settings: joint effects of, 8-9 naturalistic field study(ies): developmental issues in leisure places, 156-62 negative reinforcement, 121 neighborhood(s), 3; and family management skills, 145-6 nomothetic approach to personality, 204, 2057; process in, 207-8, 212, 213 noncompliance, 126, 127, 128-9; and academic achievement, 134; generalization to school, 129-31 nonnormative events, 41 normative experiences, 41, 42 Oakland Growth Study, 78, 81, 82-4; Berkeley cohort compared with, 85-9 observation (method), 32; in field studies of adolescent leisure settings, 156, 160-2; and self-ratings, 181, 182, 199; theory construction from, 218-33 observational learning: in sex typing, 207 occupational attainment, 6, 46; education in, 103 occupational conditions, 4-5 options: adaptive, 85-6, 107; in control cycle, 109; prior to economic loss, 81-2, 91, 93; in reaction to environmental change, 85-6; of school underachievers, 143

Subject index parental competence: contextual theory of, 60-1 parent-child effects, 219-25, 229 parent-child interactions, 3; across social addresses, 6; community in management of, 64-5; and impending arrival of sibling, 41, 42; in process models, 33 parent-child relationship: and agreement on values, 62, 63; and effect of economic hardship, 93-4; effect of marital discord on, 95; mediating stress through, 85; models of, 9— 10, 13, 16; see also family relationships parent hostility, 100, 101 parent training, 137, 141 parenting: and child behavior, 136-7, 139; inept, inadequate, 98, 146, 242; punitive, 94, 95, 100 parenting skills: with difficult children, 145, 242; institutional rearing and, 17; outside stressors and, 131-2 parents, 5, 53-4, 65; adoptive, 231-2; as agents of society, 13, 51, 61-4; and aggressive children, 242 (see also coercion process); distancing, 127-8; and genotype/environment combination, 219-20; perceptions of outside world, 71; rejection of coercive child, 130, 131, 134-6; transmission of knowledge and intelligence, 220-3 pathologic processes: in families, 115, 123 peer contact: and self-esteem, 163; target and frequency of, 156, 157, 158, 160-2 peer group(s), 55, 116-17; deviant, 143, 155; function of, 117; rejection of coercive behavior by, 117, 126, 129-30, 131, 134-6, 142, 143 peer group integration, 156, 157, 162, 174, 175; strategies for, 176-7; substance use in, 172-3 perception(s): of access, 68, 70, 71; of agreement in values (parent-child), 62-3; of power of parents, 65; of settings, 67-70, 71-3 personal attributes model, 29-30, 31, 46 personality, 226, 230; and academic development, 9; continuities in, 97, 102, 210, 21112; idiographic/morphogenic approaches to, 204-6; interactional approach to, 208-11; measurement theory and, 182-3; nomothetic principles in, 204, 205, 208, 212, 213; research paradigms in, 203-16; and selection of environments, 14 person(s): as agent and consequence of changing life paths, 77; and behavior, 209; in reciprocal transaction with situations, 210-

Subject index 11; situation as function of, 209-10; see also individual(s) person/behavior equivalence classes, 204 person-context link: across time, 16-19; and interventions, 19-20 person-context model, 31-2, 46 person-environment models: applied to multiple environments, 8-9 person-situation interaction, 235 persons-in-situations: as unit of analysis, 209 phenotype(s): and environment, 226-7; and genotype(s), 231, 238-9; variation in, 225 phenotype-environment correlations, 239-40 phylogenesis, 182 physical world: self cognitions and, 184, 186 Piagetian theory, 56-8, 69 plans, 108-9; self cognitions in, 184, 186 Pollyanna bias, 188 population approach, 217-18 positive feedback loops, 131, 138 positive reinforcement, 136 poverty, 36-7, 79; see also economic hardship predictability: in developmental paradigms, 29-30 prejudice, 56 prevention: of problem behaviors, 174 problem behavior, 100; and adult life course, 90, 97-9; childhood stress and risk of, 8 1 9; constructive role of, 152-80; as coping strategy, 152, 154, 174-5; development and, 153-6; as effort to master age-typical difficulties, 154, 173-5; environmental selection and, 15; and response to economic hardship, 85, 89, 90, 94, 97; stressful antecedents of, 78; and unstable family relationships, 100-2 problem behavior proneness, 154 process: analyzing: nature of access, 664-73; analyzing: questions about enhancement, 58-61; analyzing: nature of transmission, 61-4; concept of, 129; and content, 207-9, 212-13; defined, 114; in developmental paradigm, 27; family, 114-51; generalizability of, 38; in human development, 77; multiple, 114-15; in past events in evolution of patterns of psychological functioning, 45-6; in process-person-context model, 36; in relationships among children, families, communities, 50-1; short-/long-term effects, 117, 136; in social address model, 29; study of, 146; varies as joint function of context, 36 process/content distinction, 207-8 process-context models, 33-5, 39 process paradigms: human development research, 32-40

257 process-person-context models, 35-40, 42, 43, 47; defining properties of, 36 process-person paradigms, 39 prosocial skills, 137 protective factors, 65, 85, 88, 93; mother as, 92 proximal influences, 6, 7-13, 32-3 psychoanalytic theory: sex typing in, 208 psychobiology, 184-5 public spaces, 155 punishment(s), 121, 132 rational expectations model, 20 rationalization, 187, 189 reactance behavior, 108 reality: self cognitions and, 183-4, 185, 18990, 191, 195-6, 198-9 reason: self cognitions and, 186 reinforcement, 60, 120, 123, 147; of destructive behavior, 211; negative, 121; positive, 136; in sex typing, 207 reinforcement trap, 121-3 relationships, 53; consensus/disagreement about settings in, 70; see also family relationships resilience, 51 resources: claims and, 106-7; prior, 81-2, 91, 93; in reaction to environmental change, 85-6 response-style variables, 212-13; person-centered theory of, 214 responsibility: in children-families-communities relationship, 65-6 risk factors, 85, 88 risk: perceptions of, 70 roles: age-graded, 97-9, 214; allocation, 80; family, 90; of fathers, 5, 90 schema theory, 208 school(s), 3, 10, 55-6, 59, 66, 72; and academic development, 9; age-graded, 106; as developmental context, 3, 7-8; and family/community factors, 8; introduction to, 52; interactional style in, 8-9; and student participation, 9 schooling: state and, 65 school performance: maternal employment and, 5 school reform, 105 selection of environments, 99, 210, 214, 228, 229, 231, 235-6; and individual development, 13-16 selection factors: in school effects, 7-8 selective attention, 228, 229, 232 self-concepts, 69, 70; functional analysis of

258 self-concepts (cont.) properties of, 181-2; functions of, 199-200; inflated, 190; sociogenesis of, 181-202; stability of, 14; therapeutic strategies in, 198— 9; unrealistic, 19, 186 self cognitions: adaptive, 199; developmental and evolutionary functions of, 181-2; enhancement of, 193-4; functions of, 183-6; multiple masters in, 189-91 self-direction, 13-14 self-esteem, 55, 213; leisure activities and, 175; low, 142, 143; rejection and, 131, 134; substance use and, 163-73 self-organization, 183 self-other ratings, 192-3, 197 self: public/private, 196-7; stability of, 200; variables involving, 213 (see also self cognitions) self-ratings: compensatory and self-balancing, 194-5; divergence in, 192-3; and ratings from peers and teachers, 181, 183, 191, 192-8, 199-200; and social consensus, 182-3, 189, 199; stability in, 193-4 self-regulation: of development, 153, 173-4 self-system, 213; construction of, 183 settings, 52; access notions in description/perception of, 58, 67-70; compatibility among, 8-9; community, 66-7; "continuity is good" value judgment, 55; developmental, 154-5; joint effects of, 9; moments of entry/exit, 66; nonopen, 57; qualities of individuals and, 71-3; as units, 50, 52-4; variance in, 58; "wider is better" value judgment, 54-5; "world is open" value judgment, 55-8; see also leisure settings sex typing, 207-8, 212 sexual dimorphism, 187-8 siblings, 10, 15; adopted, 223, 20-3, 235; biological, 230-3, 235; changing similarities among, 232-3; correlations among, 12-13, 230-3 single-parent families, 132, 146 situationism, 206, 207, 208 situations: and behavior, 209; as function of person, 209-10; idiographic classification of, 207; persons in reciprocal transaction with, 210-11 social address model, 28-9, 30, 31, 45-6, 155 social adjustment, 219-20; maternal employment and, 5 social change, 55; human develompent in, 3 6, 77-113; and individual development, 8 1 2; study of lives in, 78-9, 80 social class, 34, 39, 43, 84, 110; and cognitive functioning, 89-90; and effects of

Subject index Depression, 87; effects on development, 4, 6; and life experience, 90; of women, 6 social cognitions, 57, 188, 199; function of, 186; and social maps, 197-8 (see also self cognitions) social consensus, 185; and self ratings, 191, 195-6, 199 social development: measurement theory and, 182-3 social ecology, 187, 188-9 social institutions: effect on life course, 79, 109; as macrosystem, 39 social interactional approach, 118, 120 socialization, 5, 105; age and, 80; effect of socioeconomic status on, 34; in family, 9, 13, 114; family responses to economic hardship and, 94; parent-child, 229; perception of access and, 71; school, 7-8; setting selection in, 66; unstable family relations and, 101-2; work and, 6 social maps, 197-8 social network: and behavior, 16-17; impact on health, 164; and individual development, 11; and stress, 37 social policy: family in, 53; relationships in, 65-6; world views and, 55-6 social skills, 17, 117, 122, 127 social structure(s), 2; age-related change in, 188; of children and adolescents, 197-8; historical context and, 110; influence on development, 3-6; research on, 6 social support, 61 stage models: causal status in, 138-42; coercive process, 129-36; multiple indicators in, 137-8; problems in testing, 136-43 stealing, 126, 127-8, 129, 205 stress, 61; and behavior change over time, 116; economic conditions and, 60; effect on childhood, 81-9; effects on discipline, 132; and family dynamics, 93-6; parenting behavior and, social networks and, 37; vulnerability to, 51 substance use in adolescence, 176; constructive role of, 152, 173-4, 175; context of, 1546, 158, 160, 162; in mastering developmental tasks, 174-5; as problem behavior, 154; roles of, 153; and self-esteem, 163-73; stages in, 173 supervision of children, 93, 128 synchronistic model, 139, 141, 143 temperament, 212; difficult, 144-5, 242 temper tantrums, 98, 210, 211, 212; in trait progression, 126, 127, 128-9 theory construction: from observations, 21823

Subject index therapy: family, 10; regarding self concepts, 198-9; see also intervention(s) time/temporality: in delayed-incremental processes, 142; in effects of economic deprivation in childhood, 84, 88; historical/life, 80; in human development, 40-4, 77; in lifecourse perspective, 77, 79, 85, 100; of life events, 104-5 trait approach to personality, 204, 205, 206 trait progression, 125-8; integration of microsocial progressions with, 128-9 trait psychology, 208 traits, 212; progression in antisocial, 125-8, 129 trait similarity: in siblings, 233 transitions, to adulthood, 97, 103, 104, 152, 154; home-to-school, 35-6; life course, 91, 102-5, 107-8, 214; major, 109; into settings, 52 transmission: of behavior and attitudes, 91, 99-102; nature of, 61-4 transmission model, 5 0 - 1 , 61-4, 65, 66, 71 treatment(s): average effects of, 234, 235; vs. etiology, 241-2; phenotypes and, 238, 239; and within-group variability, 234, 235 turning points, 91, 102-5, 214; see also transitions

259 twins, 224; reared apart, 242; see also dizygotic (DZ) twins; monozygotic (MZ) twins twin studies, 12-13, 15-16, 234, 236, 242 typological approach, 206, 217-18 urban life-space, 154-5 value judgments: in children-family-community relations, 50, 54-8 values, 5, 110; and family management skills, 145-6; intergenerational agreement on, 623; in occupational attainment, 46; and self cognitions, 186; substance use and, 174 violence: family, 11 vulnerable child (the), 81 well-being, sense of, 143 withdrawn behavior, 210, 211, 212; and adult life course, 97-9; and selection of environment, 15 women, 83; see also mother(s) work: and development of conformity/autonomy, 4-5; in life course, 97, 98 world view(s), 55-60 youth gangs, 155