Rationalistic theories of the workplace and the claims typically made by organizations stress that an individual's access to the resources and advantages of an organization are determined by his or her qualifications and contributions to the collective enterprise, and that the payoffs for effort are essentially the same for all those doing similar work. However, as Jon Miller shows in this book, negotiating for workplace rewards is actually far more complicated than this model allows, and he demonstrates that access to networks of organizational communication is in fact fundamentally influenced by race and gender. Drawing on his study of American public service organizations, Professor Miller compares patterns of access to informal colleague networks and relations to the decision-making apparatus for white and nonwhite men and women. He shows that although no group monopolized the advantages of the workplace, and none was disadvantaged on all dimensions of work, no two race-gender groups faced the same set of reward allocation rules. Only white males experienced a fairly close correspondence between their bureaucratic "investments" and their workplace rewards, whereas for others more particularistic factors, such as age and ties to the external community, came to the fore. This revealing demonstration of the systematic and potentially divisive variations that exist in the ways in which qualifications and accomplishments are linked to the rewards enjoyed by individuals within the workplace will appeal to sociologists and other social scientists interested in formal organizations, as well as in the study of gender and race. It will also be of interest to readers concerned with organizational psychology and management studies.
The Arnold and Caroline Rose Monograph Series of the American Sociological Association
Pathways in the workplace
Other books in the series J. Milton Yinger, Kiyoshi Ikeda, Frank Lay cock, and Stephen J. Cutler: Middle Start: An Experiment in the Educational Enrichment of Young Adolescents James A. Geschwender: Class, Race, and Worker Insurgency: The League of Revolutionary Black Workers Paul Ritterband: Education, Employment, and Migration: Israel in Comparative Perspective John Low-Beer: Protest and Participation: The New Working Class in Italy Orrin E. Klapp: Opening and Closing: Strategies of Information Adaptation in Society Rita James Simon: Continuity and Change: A Study of Two Ethnic Communities in Israel Marshall B. Clinard: Cities with Little Crime: The Case of Switzerland Steven T. Bossert: Tasks and Social Relationships in Classrooms: A Study of Instructional Organization and Its Consequences Richard E. Johnson: Juvenile Delinquency and Its Origins: An Integrated Theoretical Approach David R. Heise: Understanding Events: Affect and the Construction of Social Action Ida Harper Simpson: From Student to Nurse: A Longitudinal Study of Socialization Stephen P. Turner: Sociological Explanation as Translation Janet W. Salaff: Working Daughters of Hong Kong: Filial Piety or Power in the Family? Joseph Chamie: Religion and Fertility: Arab Christian-Muslim Differentials William Friedland, Amy Barton, Robert Thomas: Manufacturing Green Gold: Capital, Labor, and Technology in the Lettuce Industry Richard N. Adams: Paradoxical Harvest: Energy and Explanation in British History, 1870-1914 Mary F. Rogers: Sociology, Ethnomethodology, and Experience: A Phenomenological Critique James R. Beniger: Trafficking in Drug Users: Professional Exchange Networks in the Control of Deviance Andrew J. Weigert, J. Smith Teitge, and Dennis W. Teitge: Society and Identity: Toward a Sociological Psychology
Pathways in the workplace The effects of gender and race on access to organizational resources
Jon Miller University of Southern California
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 London New York Melbourne Sydney
New Rochelle
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/9780521323659 © Cambridge University Press 1986 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 1986 A catalogue record for this publication is available from the British Library Library of Congress Cataloguing in Publication data Miller, Jon, 1940Pathways in the workplace. (The Arnold and Caroline Rose monography series of the American Sociological Association) Bibliography; p. Includes index. 1. Discrimination in employment — United States 2. Sex discrimination in employment — United States. 3. Race discrimination - United States. I. Title. II. Series. HD4903.5.U58M58 1986 331.13'3'0973 85-19550 ISBN-13 978-0-521-32365-9 hardback ISBN-10 0-521-32365-7 hardback Transferred to digital printing 2006
For Bob Hagedorn, Dick Ogles, and, of course, Sandy Labovitz
Contents
Preface Acknowledgments Introduction Research setting Preview of the findings 1 Rationality and equity in professional networks Gender and race as factors in organizational stratification Description of the organizations and specific research objectives Methods and measures
page ix xi 1 4 6 8 8 15 22
2 Ascription, achievement, and network centrality Stage 1: comparison of means and overall regression analyses Stage 2: the interaction of ascribed and achieved attributes Stage 3: considering the simultaneous effects of race and gender Exploring the importance of external activities A note on the ideal type: authority, expertise, and influence Summary
27 29
3 Access to the formal authority structure Stage 1: simple comparisons and overall regression analyses Stage 2: separate analyses for men and women, whites and nonwhites Stage 3: the simultaneous effects of race and gender Summary
61 62
32 38 42 56 59
66 69 78 vii
viii
Contents
4 Conclusions Summary of the findings: four modes of accommodation Conclusions and implications Notes References Index
83 85 97 100 107 113
Preface
My ambition in this work has been to give some preliminary but suggestive answers to the problem of internal organizational stratification. The questions that I have raised about race and gender differentiation and the tentative answers that I have offered should ramify in interesting directions; obviously, however, no single study can claim to resolve such a complex problem. In the analysis, I have concentrated most of my attention on the statistically detectable traces of organizational stratification. The measurement decisions that I have made, if not always optimal, are fairly straightforward; as a consequence, the results should be easily replicated. At the same time, I am very much aware that what people do with, to, and in spite of each other in organizations and how all of this relates to ascribed status differences must at some point be examined interpretively in order to place it in the context of a system of emergent intersubjective meaning. My hope is that the findings that I offer, and especially the gaps and uncertainties in the findings, will point to the kinds of questions that need to be asked in more fine-grained, qualitative studies in the future. On a more macro level, I have also been attentive to the community ties of the organizations and respondents in the study, and I have taken into account the broader features of the external labor market. In fact, such considerations provide the theoretical leverage for some of the interpretations that I have advanced. However, I have not directly addressed in any comprehensive way either the origins of human service organizations or the functions they serve for the larger structures in which they are embedded. Again, my hope is that what the findings show as well as what they are unable to show will provide some direction for other investigators. What is clearly needed is to connect the dynamics of internal stratification more systematically than I have been able to do to the larger political-economic realities in which organizations are embedded. In short, because of its objectives and its particular focus, this study should be seen as one that, more than anything else, calls for comparison, correcix
x
Preface
tion, and extension, both downward to the level of intersubjectivity and upward and outward to larger institutional structures. Two comments that reveal my personal biases are also in order in these opening remarks. First, I am concerned with the practical message, if any, that is taken from the results of this study. One important finding that I report is that access to a variety of interpersonal, informal network resources was roughly equal for white men, white women, nonwhite men, and nonwhite women, but that quite distinct pathways provided this access. It involves a judgment call that goes beyond what the data directly show, but my assessment is that such findings, and the explanations of them that I offer, reveal some of the things that traditionally excluded groups are able to do, or must do, in order to deal with the institutional obstacles they face in the world of work. The message is not that informal organizational mechanisms compensate for or neutralize these institutional limitations. Rather, I take the data as indicating that every category of organizational participants has a range of strategic options, including the use they make of their external attachments, with which to confront the internal system of reward and resource allocation. How rationally or equitably the organization behaves is in large part determined by where the organizational member is located in the larger ascribed social system outside the organization. Put simply, the more favorably the individual is placed externally, the stronger is his or her internal claim to rational or equitable treatment. Second, because of the topic I chose to address in this monograph, I have said next to nothing substantive about the clients who were "served" or "processed" or "treated" (the agency term is "deinstitutionalized") by the human service organizations in the survey. I am, in fact, acutely aware that many middle-class professionals - sometimes including social researchers - owe their employment and a large part of their occupational privileges to what society elects to do with various stigmatized groups. To me it is axiomatic that what happens to the people in people-processing organizations is of first importance, and this research should not be seen as elevating the problems of human service practitioners to a status above the problems of clients. Los Angeles
Jon Miller
Acknowledgments
James Lincoln and Jon Olson joined me in a very productive scholarly partnership in the early stages of this project, and they deserve a share of the credit for whatever I have accomplished here. The project has grown and changed enough since we worked together that they also deserve exemption from any problems or oversights that have survived the long process of review and revision. Peter Blau, Kurt Tausky, and several anonymous referees read earlier versions of the manuscript and oflFered a number of comments that have helped to shape and improve this final version. Malcolm Klein, Solomon Kobrin, and Elaine Corry provided the initial opportunity, the facilities, and the material resources to conduct the survey, and I am grateful to my research assistants, Margo Gordon, Larry Heck, and Sonya Miller, for their cheerful submission to that form of exploitation that we are fond of calling "learning from experience." Financial support for the analysis presented here was supplied by the U.S. Department of Justice, National Institute for Law Enforcement and Criminal Justice, grants 78NIAX-0135 and 80IJCX-0089, awarded to the Laboratory for Organizational Research, Department of Sociology, University of Southern California. The ideas and opinions expressed by me are entirely my own and in no way reflect or represent the official position or policies of that agency. Now I claim the pleasant privilege of expressing gratitude to my colleagues Herman Turk, Parvin Kassaie, Edward Ransford, Jennifer Glass, Robert W. Hodge and Barbara Laslett for being around, reading drafts, listening to arguments, or offering invaluable technical and conceptual advice (or all of these) throughout the evolution of the monograph. J. M.
XI
Introduction
The work experiences of men and women, whites and nonwhites, differ. On this much the evidence is clear. Whether these differences resolve into unmistakable patterns of privilege and deprivation involves harder questions, and here the answers are not as clear. The effects of gender and race have been investigated in some detail in labor market studies and surveys of occupational attainment and mobility; as a result, differences in such matters as average pay and aggregate career prospects that are justifiably called inequitable are well documented. But except for a handful of notable studies that I will discuss later, comparatively less systematic attention has been given to differences in experiences at the level of daily work activities. It remains to be seen whether the patterns that exist at this level parallel those that have been documented at higher levels of aggregation. Do levels and means of access to networks of organizational influence and communication differ predictably by race and gender? Are there systematic differences in access to the authority structure? Do the differences that emerge suggest that one group is consistently favored over others? What are the implications of such inquiries for traditional theories of organization? These questions are at the center of my concern here. It is in the immediate work arena that the impersonal forces of the occupational marketplace intersect with the structure of an organization. These two sets of elements, then, provide the backdrop for the personal encounters among individuals who are, in varying degrees, active participants in the construction of their own work realities, and who are in fact likely to use whatever resources are available to them to protect or further their own occupational interests. The social relationships in which different categories of participants find themselves involved are the end products of all of these forces, and they all deserve careful investigation. Evidence on these matters will be offered from just one kind of organizational setting. I studied six interorganizational human service networks made up of agencies that employed professional and semiprofessional practitioners such as clinical psychologists, youth counselors, and social workers. In the 1
2
Pathways in the workplace
delivery of social services, such coordinated arrangements are increasingly common. But this is not a setting that is representative of the work world in general. Moreover, the numbers involved are not large. The 256 respondents include 128 white women, 71 white men, 29 nonwhite women, and 28 nonwhite men. The claims and intentions of the study are shaped to fit these limitations. The value of the results lies in what they suggest about the range of patterns that are possible and likely when different interest groups work together in organizations, and what they suggest about the types of questions that need to be asked about the causal mechanisms that create differentiation among these groups. My primary theoretical concern has been to determine how closely the work experiences of white men, white women, non white men, and nonwhite women approximate the classical rational model of organizational participation. This model is the organizational counterpart of the human capital theory of work-force participation. It posits a consistent, direct relationship between individuals' investments of time, energy, training, and talent on the one hand and the formal and informal organizational rewards they enjoy on the other. The rationalism at the center of the model characterizes both the assumed frame of mind, or motivation, with which individuals approach the workplace and the procedures by which organizations distinguish among their members. I did not take this broad paradigm as my point of departure because I approached the study with a rationalistic bias or because I thought it generated hypotheses with a very high probability of confirmation. In fact, I began and finished the study convinced that the useful limits of such rationalistic theories are quickly reached. However, straw figures can be useful. On some crucial points, the rationalistic paradigm does offer a clear-cut set of definitions and predictions, and thus provides a kind of theoretical baseline against which empirical observations can be compared. The same cannot always be said about most other models of organizational activity, including those that I personally find more persuasive. For this reason, the idea of rationality is a pivotal concept and is given a prominent place in the discussions that follow. Note also that it is the simultaneous effect of race and gender that are of interest, not their effects taken separately or additively. Both the logic of my arguments and the findings I present converge on one perfectly simple point: Race makes a difference for both males and females, and gender matters a great deal for both whites and non whites. This truism has to be stressed repeatedly because the more usual approach has been to focus on either gender or race or to add the two variables separately (usually as controls or
Introduction
3
suspected "contaminants") to a statistical analysis. Both of these strategies force an artificial separation between two attributes that in fact always operate conjointly. Little progress can be expected toward understanding the impact of ascription on the workplace until the intersection of race and gender is routinely accommodated by the research designs of organizational studies. The participants in the study supplied a wealth of information about their relationships to the unofficial structures of communication, influence, and mutual assistance that tied together the interorganizational networks of which their agencies were a part. The nature of the work undertaken by these agencies required the members to supplement their intraagency, or local, colleague ties with professional linkages that crossed agency boundaries. Practitioners concerned about their work performance could not afford to take these contacts lightly. The ways they found to gain access to these informal interagency networks are the primary focus of this study. Similar information was also collected on the participants' relationships to the formal authority structure, so that patterns of access to the formal and the unofficial could be compared. After a theoretical discussion in Chapter 1, the presentation of these data proceeds in the following way. First, I present a sociometric analysis of the patterns of access to the informal colleague networks (Chapter 2). Five different dimensions of interaction and consultation are examined: 1. work contacts (patterns of day-to-day interaction) 2. influence (networks of informal decision making) 3. respect (hierarchies of professional esteem) 4. support (networks of trust) 5. assistance (networks of professional cooperation) Two broad questions guided this part of the investigation: 1. Did the levels of access to these informal networks differ markedly among the various race, gender, and race-gender categories? That is, did one category dominate these unofficial exchange structures to the exclusion of others? 2. Were the means of gaining access to central positions in these networks the same for each race-gender category? Here the intention was to determine the combination of variables for each category that provides the best understanding of how they found their way into the informal networks. The list of possible explanatory variables includes training, occupational identification, work assignment, experience, rank, age, and ties to the external community surrounding the workplace. After this sociometric analysis, Chapter 3 addresses a parallel set of questions about differences in access to the formal decision-making structure, this time concentrating on reports of the nature and frequency of contact with
4
Pathways in the workplace
supervisors and the extent of participation in the official decision-making process. Research setting A detailed description of the programs that provided the data for this study is given in Chapter 1; only a brief overview will be given here. The six service delivery systems included in the study were part of a national program, sponsored by the U.S. Department of Justice, that was designed to provide community-based treatment resources for juvenile offenders. Each of the systems was a clearly bounded network of cooperating agencies brought together to serve a single community or geographic region. The guiding assumption of the overall program was that such collections of agencies with their varying perspectives, skills, and resources would allow a more flexible response to the problems of troublesome young people than either the official justice system or isolated single agencies could provide.1 The separate agencies that were brought into a given network were bound to each other in some fairly specific ways by formal contractual arrangements; but, more importantly for their day-to-day activities, they were linked together functionally by the interpersonal ties among their members that developed across agency boundaries. Webs of professional interaction evolved around the exchanges of influence, information, professional respect, and assistance, and these emergent networks provided the mechanisms by means of which the activities of the systems were carried out. In the sociometric analysis in Chapter 2, these systems of interpersonal exchange are treated as if they were informal hierarchies, with organization members arrayed in positions from high to low according to their degree of network centrality. For an individual, a more strategic position in one of these interagency networks meant greater access to a range of professionally important and visible interpersonal resources. Because so much of the work with clients required the attention of personnel from different agencies, the jobs of very few individuals could be performed successfully in isolation from the larger interagency system. The division of labor among agencies and occupations placed loose constraints on the kinds of interpersonal ties that evolved, but these networks are still properly called "informal" because the precise nature of the linkages that developed was not officially prescribed in any detail. 2 There was leeway for both contention and cooperation among individuals in gaining strategic network positions for themselves. The examination in Chapter 3 of differences in relationships to the formal authority structure focuses on the ability of individuals to participate in the
Introduction
5
decision-making process and to interact to their benefit with their immediate supervisors. From the point of view of fulfilling work-role obligations, a favorable relationship to official superiors can be a significant instrumental resource, whereas isolation from this formal system of decision making can be a severe handicap for an individual. Together, these parallel investigations of informal and formal activities give a more complete picture of how ascribed status affects internal organizational differentiation. This approach also provides an idea of the degree of "coupling" of the formal and informal structures. This is assessed by determining whether access to the informal sociometric structure responds to the same or different sets of variables as access to the formal decision-making structure, and whether the extent and manner of the coupling vary from one member group, or race-gender category, to another. Predictions about the tightness of coupling vary depending on which of three possibilities is considered most likely: 1. 2. 3.
that formal and informal activities are different but complementary facets of a single overall complex of organizational activity; that the informal structure evolves as a reaction against the formal structure; that the informal structure is the actual structure, whereas the formal apparatus is largely irrelevant or epiphenomenal.
How vulnerability to formal and informal isolation differs by race and gender is an issue on which several theoretical perspectives converge. Such differences are a matter of interest for conceptions of social stratification, theories of labor market participation, and, more directly of concern here, models of organizational activity. Investigating the simultaneous impact of race and gender - as opposed to examining only one of the two or treating both simply as potential contaminants to be statistically controlled - has been largely ignored in all three of these areas. In the case of organizational theory, the oversight is especially troublesome. To be sure, not all organizations are characterized by pronounced racial or sexual differentiation. However, examining internal differentiation along race and gender lines promises to provide a better understanding of how members of different demographic categories make their way in organizations; conversely, it should also help to clarify how organizations deal with social and demographic cleavages among their work forces. Insights from the study of race and gender may suggest what to expect for other divisions such as age, nationality, ethnicity, family background, and the like. These are factors that vary greatly in the direct instrumental relationship they bear to the tasks an organization is attempting to accomplish. Yet they make a great deal of difference in the lives of the members in the larger world outside the organization, and they may break
6
Pathways in the workplace
through organizational boundaries to influence internal dynamics in profound ways. Preview of the findings The data in Chapter 2 show that no ascribed status group clearly dominated any of the five informal dimensions of activity. Mean levels of informal centrality in the interagency networks were remarkably constant across race and gender categories. However, when it came to a comparison of ways of gaining access to these informal networks, some striking differences were apparent. It was only for white males that avenues of access closely resembled the classical rationalistic bureaucratic conception. With some exceptions, the informal organizational advantages they came to enjoy were a function of their bureaucratic investments, their hierarchical positions, and their work-related resources. The same could not be said for white women, nonwhite women, or non white men. The experiences of the latter three groups were more likely to be linked to the nature of the clients they handled, their special ties to the surrounding community outside their programs, or their own personal, nonbureaucratic characteristics - age in particular. What these findings indicate is that social service organizations, like others, are characterized by a complicated process of negotiation for informal work-related advantages. To use as currency in these exchanges, participants had both internal bureaucratic resources and external ties to the larger community, and quite different combinations of these internal and external investments and contributions were brought into play by white men, white women, non white men, and non white women. When relationships to the formal structure are examined in Chapter 3, the pattern of equal levels of subgroup access that characterizes the informal structure is not apparent. Instead, the predominant finding is that non whites, especially nonwhite men, had closer relationships with the formal structure of authority than white men and white women did. Part of the explanation seems to lie in their superior ability to take advantage of their ties to the community outside the workplace. In addition to this surprising finding concerning race, the analysis revealed that the factors that seemed to work well for one group in the informal structure often turned out to be an advantage for another group in the formal structure. The indication is that tight informal-formal coupling was not characteristic of the organizations in the study. Given their level of professionalism, this may also be surprising, since professionalism can be seen as producing a comingling, or tight linkage, of informal and formal decision-making practices. To complicate the matter
Introduction
7
even further, in some cases factors that functioned as advantages for a group in the formal structure actually appeared as disadvantages for that same group in the informal structure. The presence of such trade-offs suggests that the organizations were "inversely coupled" rather than simply "decoupled." The implications that these results have for the classical stream of organizational theory are important. Narrowly conceived, the rationality principle at the heart of this body of theory suggests that the same reward allocation rules will function for all participants, and that where organizational exigencies alone are operating, ascribed factors such as race and gender will have a minimal impact on organizational functioning. The findings generally have a poor fit to this principle. The visible trappings of bureaucratic rationalism were certainly present in the organizations in the study, but patterns of daily formal and informal social relationships that actually conformed to the model were the exception and departures from the model were the rule. It is equally important to note, however, that the results are not very handily accommodated by any one of several theoretical alternatives to the rationalistic approach. The problem is that no theory has yet incorporated a very precise or comprehensive argument about how racial and sexual differentiation - or ascription in general - affects internal organizational arrangements. The findings previously summarized illustrate what this means. Some of the patterns follow the outlines suggested by a simple model of discrimination based on differences in power, but at the same time other findings suggest that persons with supposedly disadvantageous race-gender profiles could turn their ascribed identities to good interpersonal and instrumental advantage. The results of this analysis sometimes ran directly counter to what a simple discrimination hypothesis would suggest. In the face of such contrasts, I have been very selective in the use of the words "inequity" and "discrimination" to describe the findings. Parsimony in this area of investigation is difficult, and in the present analysis it takes a combination of insights gleaned from several perspectives to make sense of the findings. At the present level of theoretical development, this is a result that may well be typical of studies that try to deal in any systematic way with the effects of race and gender - or other kinds of internal divisions - upon organizational relations.
1. Rationality and equity in professional networks
Few would doubt that race and gender have an influence on patterns of interaction in the workplace; but, as the discussion in this chapter will show, consensus has not been reached on just how such ascribed factors affect organizational activities. Some argue that organizations are essentially evenhanded in their treatment of different categories of personnel and that they do not add to the disadvantages that certain groups bear because of their location in systems of privilege and deprivation outside the organization. Others maintain that organizations duplicate and even intensify the patterns of advantage and disadvantage that exist externally. In the present investigation, the picture is further complicated because the organizations had some features that clearly worked against pronounced race and gender differences in work experiences, but other features that suggested that the competition for professional advantages and personal rewards was intense as well as uneven. I will review some of the important studies that deal with ascription in the area of work and then turn to a description of the rationalistic and nonrationalistic characteristics of the programs in the present study. The chapter concludes with an explanation of the measurement techniques and analytic strategies that were used. Gender and race as factors in organizational stratification The concept of rationality comes to us from Weber in several variations. It captures the shift from ascription to achievement in the allocation of positions and resources and from particularism to universalism in the evaluation of performance [cf. Parsons (1950,1966); Kalberg (1980)]. Through this set of meanings, the term "rationality" is closely identified, historically and theoretically, with the increasing reliance on complex, bureaucratized forms of organization. For Weber (1978), bureaucratic rationality was indexed by the ways in which access to formal positions and other resources were linked to expertise and by the presence of performance criteria based on the achievement of technical role obligations. From this principle of rationality, he de-
Rationality and equity in professional networks
9
duced that organizational practices would have a leveling effect on differences ascribed in the external society. This would be a natural consequence of the attempt to eliminate irrelevant or counterproductive considerations from the search for and utilization of talent (Gerth and Mills 1946:224ff; Weber 1978:225). Questions about the accuracy and generalizability of this rationalist argument run throughout the organizational literature.1 However, the idea that organizations are in principle universalistic and achievement oriented - or "intendedly rational," to borrow the language of March and Simon (1958: 169ff) - is still a common assumption in theoretical discussions and is certainly part of the normative view of themselves that most organizations encourage, however vigorously they may reject the bureaucratic label. In fact, the basic elements of this thinking are not confined to discussions of business, government, and political organizations (with which classical theories were most directly concerned). They have also come to represent important questions in a wide range of other types of organizations, as elements of the bureaucratic way of doing things have found their way into more and more areas of collective activity. In stating this, I am not overlooking the large literature on "bureaucratic" versus "professional" organizations [two particularly relevant sources are Bucher and Stelling (1969) and Lawler and Hage (1973)]. The point debated in much of this literature is whether the self-directed behavior of professionals or semiprofessionals (the latter being predominant in the present research) is compatible with the formal, rule-oriented nature of bureaucracy. Bureaucratic principles, it is insisted, do not apply where professionals are responsible for the organization's central activities. This debate is not directly relevant in this study. "Bureaucracy" and "bureaucratic rationality" are not used here as synonyms for rigid formalization or strict centralization and discipline, but rather in the simpler (and what I take to be the more Weberian) sense of the coupling of rewards and advantages to expertise and performance. Defined in this special way, "bureaucracy" or "rationality" can be said to be the centerpiece in the image of themselves that professional organizations like to project.2 In this image, the positions individuals attain and the rewards and advantages they enjoy are directly traceable to the skills, experience, and energy they bring to their work. Questions concerning the limits of these ideal principles are as relevant in professional and semiprofessional organizations as in any others. The relationships involving gender and race under investigation here provide one way of assessing the extent to which such rationalistic claims and intentions are actually applied in practice. In a narrow reading of Weber,
10
Pathways in the workplace
wherever rational mechanisms are fully operational, differences by race and gender in gaining access to professional and interpersonal rewards ought to be minimized. It is where rationality breaks down that particularistic and ascribed elements will find a point of entry.3 Considering the importance of this issue for organizational theory, curiously little of the existing literature has directly scrutinized the effects of ascribed characteristics on the internal allocation of system resources, rewards, and advantages. To be sure, there is ample evidence that women and nonwhites are at a disadvantage in finding their way into organizations and that they are disproportionately found in lower-level occupations and in positions that convey fewer rewards and less chance for individual progress (Siegel 1965; Rossi 1965; Treiman and Terrell 1975; Kluegel 1978; Treas 1978; Wolf and Fligstein 1979a). In fact, much of the research on labor market mechanisms is quite explicit about the existence of discrimination [for an excellent review of this work, see Martin (1980)]. However, what is unresolved is whether the work experiences of women and nonwhites parallel those of their male and white counterparts within a given organizational position or at a given level of skill - in other words, once the entry-level barriers are passed. Some labor market studies have made use of organizational variables in attempting to explain race and gender differences in income and to account for differential access to decision-making authority (Kluegel 1978; Wolf and Fligstein 1979a,b; Halaby 1979; Grandjean 1979). These studies offer useful macroscopic generalities but little specific guidance for my purposes because they are not actually conducted in organizations. Typically, they rely on a small number of items of organizational information gathered in a sample survey [see, e.g., Wolf and Fligstein (1979a,b)], and they are unable to specify in any detail what the actual organizational practices are that would account for their findings. What is needed is a more direct look inside the organizational "black box" at the mechanisms by which individual attributes and resources, or investments, are actually translated into the daily work experiences of respondents. Considering these mechanisms, a compelling argument is that in a culture characterized by sexism and racism, discriminatory practices must penetrate the boundaries of organizations; must, like a genetic code, be present in every part of the overall structure; and must operate even after differences in skill and credentials are taken into account. In order for this not to be true, there would have to be a marked disparity between the race and gender practices of organizations and what we know about patterns of ascription in the surrounding culture upon which, after all, they have to rely for their support. However, a parallel argument can be constructed that disputes the discrimi-
Rationality and equity in professional networks
11
nation hypothesis. The clients of most people-processing organizations are differentiated by race and gender, as are many of the other important community actors with whom such organizations must deal. Because of this, the ascribed attributes of their members may be of great importance to organizations for instrumental reasons. An argument that I draw upon in this analysis, for example, is that there are task-related advantages to be gained when there is a correspondence between the race and gender of practitioners and those of their clients. Similar advantages are likely to be gained if the ascribed characteristics of practitioners parallel those of important decision makers in the surrounding community. When factors such as race and gender become directly involved in an organization's activities in these essentially technical ways, the strict leveling principle - which implies the eventual elimination of differentiation based on ascribed characteristics - may not be in evidence. But this may be for reasons that do not imply inequitable and illegitimate differentiation. The difficulty in separating out the effects of these two sets of processes, one discriminatory, the other far less clearly so, is what makes it problematic to use the language of inequity to describe all race and gender differences in organizations [for a similar argument, see Perrow (1979)]. The direction of the evidence from the few studies that have addressed such issues in a straightforward way is still unclear. To take one illustration, Kanter's thesis concerning gender differences (Kanter 1977a,b) is that differentiation is to an important extent a matter of minority-majority proportions. Many women in organizations have to function alone or in small, disconnected numbers surrounded by men. They have experiences that are different from, and less favorable than, those of women who are more able to form alliances with other women. The point is to escape becoming individually visible tokens. Kanter is certainly sensitive to the relative power of men and women as it derives from their placement in the larger social system and to related considerations concerning competing vested interests and direct discrimination. Nevertheless, it is interesting that her influential paper on this topic appears in condensation (Kanter 1979) with the subtitle "Tokenism, not sex discrimination."4 Other studies have reached conclusions that differ substantially from those of Kanter. For example, in an earlier investigation (Miller, Labovitz, and Fry 1975), I surveyed several small, highly professionalized bureaucracies in which the numbers of men and women were fairly evenly balanced. For men, a rational reward allocation pattern existed: The greater the individual's investment or contribution, the greater the subjective and social relational rewards (measured by job attitudes and sociometric status, respectively). For
12
Pathways in the workplace
women, on the other hand, the greater the investment, the greater was the disparity between their rewards and those of men. This finding was referred to as "compartmentalized rationality" to convey the image of a seemingly rational mechanism (for men) embedded in a larger nonrational structure. It provides a strong argument for the operation of differential reward allocation mechanisms in the internal workings of organizations, but the imputation of discrimination is still essentially speculative. Corroborating evidence for this pattern of differentiation is to be found in studies by Fox, by Halaby, and by Wolf and Fligstein. Fox (1981b) has shown that, in academic settings, achievement determines salary rewards, but that the rate of return to women is consistently lower than that to men. Halaby's evidence (Halaby 1979) indicates that male-female income differences are strongly influenced by the fact that the higher organizational levels are closed to women. Discrimination by employers in establishing different promotional structures is again the suggested explanation. Wolf and Fligstein (1979a; see also 1979b) offer similar evidence, suggesting that employers establish routes to positions of authority that are also differentiated by gender. None of these studies of gender differences actually demonstrates discrimination by offering direct observation of practices intended to preserve the advantages of men, but all of them argue strongly for its probable operation.5 A study that a colleague and I recently conducted (Olson and Miller 1983) failed to find a male advantage in moving into positions of authority, but in the same study we reported a very clear male advantage in converting authority into other work-related advantages, including centrality in informal influence structures. Turning to race, even less research has dealt directly with the experiences of nonwhites in organizations. It could be argued that one dimension of ascription will produce differences in advantages and disadvantages much like those of another dimension and that the mechanisms for race and gender differentiation will therefore be similar. For example, Epstein's (1970) study of black professional women suggests that Kanter's hypothesis about the relative numbers of men and women could be extended to cover racial proportions as well. An alternative approach closer to the hypothesis of discrimination is implicit in Blau and Duncan's (1967) study of differential attainment by race in the larger occupational structure. The racial differences they observed persisted when other achievement differences were statistically controlled [more recently, see Parcel (1979)]. Organizational variables did not enter directly into Blau and Duncan's analysis. Nevertheless, their findings on differences in occupational experiences were probably the product not just of external labor market mechanisms but also of systematic racial differentiation within the organizations to
Rationality
and equity in professional
networks
13
which those occupations were tied. The fact that the occupational deficits faced by nonwhites were greater the more they invested in education, for example, is very similar in form to the result I reported for women in the study referred to previously (Miller et al. 1975). In other words, the deficits observed by Blau and Duncan may well have been a reflection, in large part, of promotional practices at the level of the organization. Kluegel's (1978) research on inequitable black-white income differentials fits the same pattern and has the advantage of including some contextual information on the size and industrial category of the employing organization. Finally, Butler (1976) has brought this question of racial inequity even closer to the intraorganizational sphere in a study of differential promotion rates in the army. Independent of ability and training (as measured by the army), race was shown to have a direct effect on the difficulty of promotion, and the delays experienced by nonwhites were more pronounced in the higher than the lower levels of the enlisted structure. Except for the work of Kanter, all of the studies mentioned stress the discrimination hypothesis. At best, they argue, women and nonwhites receive proportionally less return on their investments in the workplace (measured by weaker correlations between investments on the one hand and income, authority, desirable social relationships, and career mobility on the other). At worst, they face a net loss in rewards and advantages when they raise the level of the contributions they make to the collective enterprise. These relative deficits can persist and, in some cases, can even be intensified by access to the resources that are thought to be the keys to success in the workplace. In short, even though many theoretically important details about the mechanisms remain unclear,6 it is possible to piece together an impression of ascription carried over from the larger society that contradicts the "leveling" argument and thereby directly challenges the principles of organizational rationality and equity.7 Put another way, the segmentation that characterizes the labor market (Edwards, Reich, and Gordon 1975) finds direct parallels at the level of organizational structure, with favored groups in the rational core of the structure and other groups relegated to the less predictable and less equitable, therefore less rational, periphery of the organization. When apparently inequitable processes do occur, the explanation most often advanced is one involving disparities in power. Being a member of a race or gender category with greater control in the society at large is an advantage in virtually every area of social activity, and if the supposedly blind rationality and equity of bureaucratic procedure are to be deflected or compromised in any specific instance, power in the defense of vested interests is the variable most likely to be proposed as the cause. However, a counterargument is sometimes advanced that the work experiences of women
14
Pathways in the workplace
follow a different logic from that of men. Women approach work in a different frame of mind, so the argument goes, and therefore behave differently in work-related matters. They expect different levels and kinds of payoffs and, as a consequence, their job rewards, especially nonmaterial ones, are not responsive to utilitarian investments in the same way as those of men are. However, the argument continues, this is not, strictly speaking, a matter of inequity. Rather, it is the reflection of contrasting sets of job expectancies. A strong critique of this presumption is offered by Grandjean and Bernal (1979), who argue that no evidence for such motivational differences exists. The argument that I advanced earlier (Miller et al. 1975) also disputes the notion that the work motives of women are fundamentally different from those of men, a critical position that can also be inferred from the findings reported by Wolf and Fligstein (1979a,b), Feldberg and Glenn (1979), Brief and Aldag (1975), Brager and Michael (1969), and Alves and Rossi (1979). Whatever the weaknesses of the motivational argument, however, caution is still advisable in advancing an argument predicated solely on gender and race differences in power. It can easily be shown that when the power argument is stated in very broad terms, it is not especially useful for explaining the daily reality of the workplace. The difficulty comes in specifying which race and gender categories are dominant across a variety of societal and organizational circumstances. That men and whites are dominant in general, whereas nonwhites and women are subordinate, of course, comes almost reflexively to mind, but whether these patterns actually hold in all work settings is not as obvious as it may seem. In the present study, for example, it is significant that the human service agencies were staffed by representatives of the helping professions such as social work, counseling, and the like. Some have argued that these are female-dominated occupations, whereas others have noted that they are staffed predominantly by women but often controlled predominantly by men [see Montagna (1977) and Wolf and Fligstein (1979a)]. Similar considerations could be raised about race. In community-based treatment programs that deal with juvenile offenders, the surrounding community is often disproportionately minority, and even where this is not the case, nonwhites are usually well represented both among clients and among the actors in the community with whom members of the treatment agencies must deal. As was pointed out earlier, certain categories of organizational members can be brought to the fore simply by the demographics of the surrounding situation, and the fact that their prominence is directly tied to their race is not by itself evidence of nonrationality in the reward allocation process. What appears particularistic may simply reflect the fact that racial identity can be an important technical resource in some circumstances. In fact, I
Rationality and equity in professional networks
15
will offer evidence that this demographic contingency can produce differentiation by race, but of a sort that departs radically from the configuration of racial-ethnic domination that is characteristic of the society as a whole. As a result, the balance of power between majority and minority racial factions within a program may be such that a quick and sure answer to the question "Who dominates?" is simply not possible. This issue becomes even more complicated when race and gender are taken together. It is curious that, with very few exceptions (Treas 1978; Hogan and Pazul 1981), race and gender are not dealt with simultaneously in studies of work. Yet their intersection creates four unique race-gender categories, the experiences of which can only be obscured by separating the effects of race and gender or by combining them in an additive fashion for purposes of statistical analysis. [For a general discussion of the intersection of ascribed attributes, see Jeffries and Ransford (1980).] Beyond the easy impulse to rank white men first (which itself may be a serious substantive mistake in some situations), a little reflection shows that the ordering among the remaining three categories is very uncertain. Do white women rank more or less favorably than nonwhite men? And what of nonwhite women, whose status could be seen as the superimposition of one form of subordination upon another (Dahrendorf 1959), but who by other accounts are sometimes better integrated into the work force than are nonwhite men (Ransford and Miller 1983)? When to this is added the fact that clients and community actors, like program practitioners, are also subdivided into the same four race-gender categories, the result is a very complicated power equation, one that cannot easily be portrayed in terms of simple dichotomies. Previous labor market studies and organizational research are almost completely silent on this matter. With these cautions in mind, I approached this investigation as a series of strategic questions rather than a set of clear-cut, formal predictions. The basic task was to determine the importance of race and gender in shaping access to the informal and formal structures of the organizational systems included in the survey, and to use this as a way of probing (to the extent possible with one study) the limits of the rationalist perspective as a model for accounting for day-to-day organizational processes. Description of the organizations and specific research objectives To repeat a point made earlier, the programs in the survey were complicated interorganizational structures. They evolved as part of a national effort to provide noninstitutional service and treatment facilities for juvenile offend-
16
Pathways in the workplace
ers. These were young people, technically called "status oflFenders," whose offenses (running away, truancy, and the like) are troublesome for parents and authorities but would not usually be illegal for adults. Status oflFenders are often handled, along with delinquents, by the official juvenile justice system, but recent legislation has called for community-based alternatives to this practice. Beginning in 1976, the U.S. Department of Justice funded a national pilot project to develop such alternatives, and this survey was part of the evaluation of that project. The six programs covered by the survey8 ranged in size from 20 to 90 members and from 7 to 25 participating agencies. During the period covered by the evaluation (mid-1976 through 1978), nearly 5,000 juvenile cases were processed. The programs were widely dispersed geographically, so that every major region of the country was represented. Of the 358 practitioners employed in the programs, 261 (73%) were included in the survey. Of these, 199 (77%) were white, with a range among the programs from 60 to 100%. Such a pronounced imbalance did not occur for gender. Overall, 60% of the practitioners were female, with a range among the programs from 48 to 70%. The most characteristic occupations were social worker and youth counselor, categories that are certainly not the privileged preserve of men. Nor was the frequent observation that most supervisors in these kinds of occupations are men (Montagna 1977:280) borne out. Eighty-five of the respondents reported having some supervisory responsibilities. This proportion (about one-third) did not vary greatly in the breakdowns for men (37%) and women (31%), whites (34%), and nonwhites (30%). 9 It is clear from these breakdowns that white male dominance is not to be taken for granted in these programs. Rationalistic features of the program The organizational guidelines provided for these programs by the Justice Department were neither detailed nor restrictive. However, there was a clear mandate to enlist and coordinate the efforts of previously separate community agencies as a step toward moving the treatment of status oflFenders out of the juvenile justice system and into the community. To accomplish this objective, each program was structured as a network of interacting but formally independent agencies. All status oflFenders in the surrounding area were then to be remanded to this interagency network. The planners of this interorganizational strategy presumed that a practitioner in one of the agencies in such a network would not only be able to draw on his or her own expertise but could also call on the resources of colleagues in the same agency and members of the other agencies in the network as a reservoir of helpful contacts,
Rationality and equity in professional networks
17
information, and assistance. As a result, a well-connected practitioner should have had professional advantages that would be denied to an individual who was isolated or poorly placed within this network,10 and the quality of service delivered to the client was expected to be demonstrably superior to what could be offered by the juvenile justice system or by independent practitioners or isolated agencies. In each of the programs, one agency was the official recipient of the Justice Department grant. It became the fiscal and administrative center, with the responsibility for establishing and overseeing the interagency service delivery network. The agencies that were brought into these networks had to agree contractually to function as part of such an effort in order to receive their share of the grant funds. Of the six programs covered in this study, one relied almost exclusively on private agencies, two relied primarily on established and newly created linkages among public agencies, and the other four brought together a mixture of public and private agencies. A roster of participating organizations and their members specified the boundaries of each of the networks. Within these boundaries, the agencies in a program shared a common source of funding, a common set of official reporting requirements, a common client pool, and a common relationship to the organization occupying the recognized fiscal and administrative center of the program. The agencies were not all expected to deliver the same services. Some were involved primarily in diagnosis and classification, others provided residential services, and still others concentrated on various kinds of treatment. The agencies included in a network therefore occupied complementary positions in an interagency division of labor. As a result, they were bound together to some extent by the reciprocal obligations that this functional interdependence implied. Some of the features of these programs clearly represented bureaucratic responses, in the sense in which I have used this word, to the problem of administrative coordination. In part because of their federal government sponsorship, they recruited individual practitioners and agencies according to publicly stated criteria of expertise; they had intricate record-keeping systems and routing procedures; they had official intra- and interagency hierarchies of authority that were not elaborate but were clearly articulated; and, although there were profound disagreements over the particulars of proper treatment practice, there was general acceptance of the overall goal of helping status offenders in whatever way possible. Finally, not the least important feature of these programs was their visibility. They were publicly chartered and funded; they had responsibility for dealing with a visible social problem; and their members were publicly committed to social justice and profession-
18
Pathways in the workplace
alism in the delivery of services. An agency representative would be unlikely to use the word "bureaucracy" in describing his or her program's activities, but this same person would be very likely to embrace the ideal principles that I have specified as the conceptual core of that term. In particular, the emphasis on expertise and universalistic reward allocation rules are strong elements in the legitimating symbolism used by systems such as these (see Meyer and Rowan 1977). Without adding to the description given so far, it could be argued that the rationalist model should be a relatively accurate guide to the activities and relationships that took place within these social service delivery systems. Certainly if the legitimating symbolism is taken at face value, radically different kinds of experiences for the different race and gender categories should be rare. But the accuracy of the legitimating symbolism is precisely the issue. These public features of the programs must not be allowed to screen out an equally compelling list of characteristics that would tend to pull them in other directions. To these I now turn. Divisive features of the program A perplexing difficulty faced by organizations that must try to alter human behavior is the impossibility of establishing unequivocal standards of professional performance for practitioners. Dealing with status offenders is a good illustration. There is little consensus on how to define and characterize status offenders, and distinctions that make good theoretical sense break down in practice. Status offenses, by definition, are not the same as delinquent offenses, but youngsters persist in mixing the two. The point at which a status offender becomes a delinquent is virtually impossible to specify, yet millions of dollars are spent on programs that assume that the differences are clear. There is still less agreement on what causes the troublesome behavior of offenders. (The same could be said for adult criminality, alcoholism, mental illness, or almost any other problem behavior.) Consequently, practitioners and researchers alike know that conflicts among individuals and between agencies over incompatible treatment philosophies and strategies are common, although they are seldom documented in the literature. In the best of circumstances, the criteria used to move a case through a course of treatment are ill-defined, and what constitutes a success is by no means straightforward. This problem is essentially technological in the sense that Perrow (1967) uses this term. The raw material is variable, unstable, and ill-understood; exceptional cases are frequent, and solutions to their special problems are
Rationality and equity in professional networks
19
elusive. In the language of "garbage can" theory (Cohen, March, and Olsen 1972), what is to be done, how, and by whom is always uncertain, and this has profound behavioral consequences for an organization. One crucial consequence is ambiguity in what expertise means, and expertise is a factor of central importance in classical bureaucratic theory. Educational levels, credentials, and certification requirements can be established as initial screening devices. However, given the level of technical uncertainty just described, it is difficult to evaluate the practical, daily expertise that Weber (1947:339) placed alongside training and credentials as an important rational bureaucratic element. These problems are characteristic of people-processing organizations in general, but there were several other features of the programs that exacerbated them. No agency was compelled to join any of the six programs, and the central coordinating agency in each program attempted to recruit organizations that were philosophically and technically compatible. Nevertheless, funding for small, community-based treatment agencies, especially private ones, is always problematic, and agencies were frequently willing to suppress differences of philosophy and technique in order to be able to tap into the source of reliable federal funds that participation in the program promised. To illustrate, some agencies defined status offenders as victims of their families and other social institutions, and therefore saw them as in need of protective advocacy. Yet they participated in programs dominated by strategies that stressed adjustment and therefore concentrated on individual psychological counseling. The reverse kind of compromise also occurred. Practitioners who held juveniles responsible for most of their own problems or who saw behavioral problems in clinical, psychogenic terms sometimes found themselves in programs publicly committed to advocacy or to structural or institutional solutions. As a result, interagency systems that looked (on paper) like consensual unions based on common definitions of problems were frequently characterized by contention and profound differences of philosophy. In this regard, the programs resembled what Benson (1975) has called "political economies," where the emphasis is on competition for influence, definitions of legitimacy, and money. Finally, perhaps the most critical problem that these programs had to deal with was the very high probability of program failure. On one level, they were defined as successful if they simply mounted a treatment alternative that caused status oflFenders to be moved out of the official juvenile justice system and into community-based facilities. On a deeper level, there were always implicit claims by the programs that the particular treatment strategy they had chosen would be effective in reducing the level of problem behavior
20
Pathways in the workplace
among the clients served. If aggregate recidivism among clients is taken as the criterion, then programs such as these rarely succeed11 and most of those who are direct participants in them - or evaluators of them - know that this is the case. From the point of view of the organization, this is a drastic and melancholy condition, for it means that a major program goal toward which activity ought to be directed is, for all realistic purposes, unattainable. Because of this, and despite the energy expended to create the contrary public impression, there is reason to argue that the programs were not "about" resocializing, saving, reintegrating, or curing clients any more than prisons are "about" rehabilitation. The survival of human service organizations, however, need not be predicated on the attainment of their publicly stated goals (Fry and Miller 1975; Rooney 1980). Because of this, on a dayto-day basis, they can come to be "about" something other than program success. In an earlier study of service delivery agencies (Fry and Miller 1974), a colleague and I described a situation in which the sorting out of interpersonal relationships among the practitioners came to take precedence over interaction with clients. Taken to the extreme, this would represent the final decoupling of organizational action from organizational goals. How race and gender differences figure into the final equation in such circumstances, which already involve several overlapping layers of potential conflict, would be doubly interesting to determine. Potentially divisive factors such as these are not aberrations unique to the programs I studied; many of them are endemic to interorganizational arrangements (Turk 1973). In social service delivery systems, they are especially common (Hasenfeld 1972). It would be shortsighted to expect the public insignia of expertise, the public symbols of professionalism, and the publicly visible formal system of authority to account fully for what actually went on in these programs. The drawing of neat causal arrows from expertise to performance to goal attainment or, more importantly for our purpose here, from expertise to performance to favorable job evaluation to individual rewards may not be possible - either for the observer or, more critically, for any category of participants. In such circumstances, it is unlikely that any potential resources that participants command, including those that are associated with status in the system external to the workplace, would be kept out of their attempts to structure acceptable work situations for themselves. In developing the "coalition view" of organizations, Pfeflfer (1981:235) puts the issue this way: ". . . when consensually agreed upon, objective standards are missing from the situation, social power and influence affect the outcome of decisions." To this I would add that the specific forms of power and influence that may come
Rationality and equity in professional networks
21
into the balance may be externally derived and not confined to those predicated upon the resources of the organization. This, in turn, provides an opening for the intrusion of nontechnical, possibly nonrational factors into the daily round of interpersonal relations. It is virtually assured that race and gender would be among those factors. Finally, the demographic composition of the programs studied requires a further comment. In most workplaces, the likelihood of white male domination is increased because traditional patterns of control as well as sexually and racially imbalanced labor force demographics have favored it. It may still be useful to regard such settings as competitive arenas, but the ultimate outcome of the struggles between the races and genders is not often in doubt. White males dominate and gain more substantial returns on their investments, just as they do in the society at large. The organizations I studied are different in this regard. The delivery of social services to juveniles is not a major area of traditional white male domination (it is not part of the economic core of the labor market), nor did the demographics of race and gender favor them, since white males made up less than a third of the work force of the programs. The result is that contests were possible in which women and racial minorities were not irretrievably on the defensive. In other words, the power imbalance at the outset was not so great as to predetermine the final result. In such a setting, it may be possible to learn more about the factors that the participants use to structure their work to their satisfaction than could be learned in more grossly imbalanced situations. To summarize, the organizations in this study provided both protections and restraints to the white and nonwhite men and women who were the respondents. No group could prevail by sheer weight of numbers, and the bureaucratic and quasi-bureaucratic principles that were endorsed by the programs, if they had been truly operative, would have lent a leveling impartiality to the social relations that evolved among their members. Moreover, the helping professions that were represented in the programs have an ideological commitment to equity and a generally liberal public image. Both of these factors should have worked against polarization and inequality in access to resources and rewards. On the other side of the balance must be counted certain features of the larger environment in which the programs existed. Whites and nonwhites, men and women, contend for advantages and rewards in most social arenas, and there is a persistent level of intergroup distrust. Against this backdrop, it must be added that in social service delivery systems, monetary and professional rewards are generally in short supply. The scarcity of professional rewards is even further aggravated if there is a low probability of program success and if there is, in addition, a lack of
22
Pathways in the workplace
career opportunities for the participants. Contention may be especially intense where the stakes are small and the rewards are most elusive. Methods and measures Dependent variables: informal centrality and access to authority There are two categories of dependent variables in this analysis: 1. Centrality in the networks of professional interaction is a measure of integration into the nonformal dimension of work-related activities. 2. Relationships with people in the hierarchy of official authority represent access to the formal decision-making structure. Network centrality will be explored first and will occupy the largest share of our attention. It is important to explain this measure in detail. Network centrality: Each system in the survey was characterized by a network of professional ties linking together the members of its separate constituent agencies. These ties were charted using sociometric measures of the participants' work contacts and the patterns of influence, respect, support, and professional assistance that had evolved among them. My purpose is to try to expose the mechanisms that individuals could use to gain central positions in these interorganizational networks. In order to do this, my primary strategy is to compare the effects of race and gender with the effects of several indicators of professional and bureaucratic achievement, including education, experience, organizational rank, and occupation. Then, other variables are brought into the analysis in an exploratory way for purposes of clarification. In this way, the list of independent variables is eventually expanded to include measures of the practitioners' community ties and the characteristics of the client population. Integration into the structures of interaction and professional exchange was crucial to the participants' work-role performance. The pool of resources available for dealing with a client's problems was significantly expanded for the practitioner who was able to rely not only on his or her own skills and those of colleagues in the same agency, but who was also strategically placed in these interagency networks [see Miller (1980) for a corroboration of this]. Defined in this way, centrality in a network is not a return on investment of the same type as more familiar job rewards such as income, rank, or occupational mobility. In particular, it does not share their more easily recorded, quasi-contractual status, and it has not appeared among the dependent variables included in labor market and status attainment studies.
Rationality and equity in professional networks
23
Nevertheless, there is a wide scholarly consensus that centrality in either intra- or interorganizational networks is a valuable generalized resource that can be converted readily into an array of other concrete advantages and rewards (Benson 1975; Aldrich 1976; Galaskiewicz 1979). Furthermore, taking network integration as a dependent variable is consistent with a focus on organization-level processes, for although it conveys advantages to individuals, the overall distribution of centrality becomes a structural property, one that rationalist theories would claim should be responsive primarily to the achievement-based attributes of organizational members. That is, directing communication channels through persons with high training and experience is ostensibly conducive to the achievement of organizational goals, whereas assigning persons to central network positions on the basis of ascriptive traits is likely to impair goal attainment. The consideration of network centrality, then, has implications both for processes of resource competition among individuals and for theories of organizational structure. The interpersonal connections that linked together the agencies in each program were plotted from participants' reports of their closest work contacts and four other sociometric dimensions, including influence (who determined how the work of the respondents was done on a day-to-day basis), respect (whose professional opinion was most highly regarded), informal support (who was considered supportive in times of contention), and assistance (who was a good source of professional advice).12 Respondents were asked to include in their reports ties with others throughout the overall interagency program, rather than to confine their choices to the members of their own agency. The nominations that were in fact interagency as opposed to intraagency ranged from 36 to 60% on the work contacts question, with similar proportions on the other four sociometric items. "Centrality" is the graph-theoretic measure of network location proposed by Harary, Norman, and Cartwright (1965:188).13 Consider a sociometric matrix P, any element ptj of which is the smallest path distance (i.e., the smallest number of links in the chain) from actor / to actor j in the network of a given type of tie in a program. The centrality measure C is then defined as follows:
c= where / =f=j and TV is the total number of actors in the network. In other words, all the distances in the graph are summed and then divided by the sum of
24
Pathways in the workplace
distances to a particular individual, j . Centrality in this sense is often interpreted as the access by an actor to information and resource flows from other actors, since high centrality implies short communication channels converging on j (Lin 1976:345). This interpretation is appropriate in view of the conceptualization of centrality itself as a generalized network resource. The data from the six programs were pooled for the purpose of the regression analyses. For this reason, centrality measures were standardized by calculating each person's score as a proportion of the highest score achieved in the program in which he or she participated (individual centrality + maximum centrality). Thus, the unstandardized regression coefficients that are reported have a direct interpretation as percentage increments toward the highest centrality score in the individual's program.
Access to the hierarchy of authority: By definition, supervisors are expected to control important organizational resources at the same time that they oversee the processes of performance evaluation and reward allocation. The ability to interact on favorable terms with official decision makers, then, can clearly be an advantage and exclusion a handicap. The key here is the phrase "on favorable terms." In professional and semiprofessional organizations especially, there is an important distinction to be made between having access to a supervisor, which I take to be a valuable advantage, and domination by a supervisor. Both access and domination may result in frequent supervisorsubordinate contact, but the underlying implied continua are quite diiferent: Access is the opposite of isolation, and domination is in contrast to autonomy. A simple measure of the frequency of contact, used by itself, would lead inescapably to ambiguous findings. With this in mind, two questions were used to assess superior-subordinate contact. The first recorded the subjects' frequency of interaction with their immediate supervisors: "How often in the course of your work . . . do you have contact with the person who supervises the work you do for the program?" Replies were recorded on a six-point scale ranging from "less than once a week" to "almost constantly." The second question asked: "How would you describe the time you spent with the person who supervises your work on the . . . program?" Replies ranged from a nine-point scale from "almost never helpful" to "almost always helpful." A third item [adapted from George Miller (1967)] provided more information on the style of authority relations: Which of the following statements best describes your relationship with the person who supervises the work you do . . . ?
Rationality and equity in professional networks 1.
25
We discuss things a great deal and come to a mutual decision about the task at hand. 2. We discuss things a great deal and the supervisor's decisions are usually adopted. 3. We discuss things a great deal and my decisions are usually adopted. 4. We don't discuss things very much but usually come to a mutual decision. 5. We don't discuss things very much and the supervisor's decisions are usually adopted. 6. We don't discuss things very much and my decisions are usually adopted. Among several possibilities, three dummy codes for this item were considered. One of these (1 = 1; others = 0) separates participation, in the sense of discussion and mutual decision making, from nonparticipation; a second makes a similar but less restrictive distinction ( 1 , 4 = 1; others = 0); and the third (1, 2, 3 = 1; others = 0) separates much discussion from little discussion between supervisors and subordinates. For convenience, I have labeled these three measures "participation," "involvement," and "discussion," respectively. A case could be made for using any one of the three as a measure of individuals' relationships to the decision-making process, but of the three I would argue that the first, participation, presents the strongest evidence of meaningful sharing of decision making, since it requires both consultation and mutual decisions. This is a restrictive criterion for participation because it does not call participatory the pattern (number 4) that involves mutual decision making without much discussion. However, it seems to me that this pattern of decision making is more clearly a workplace advantage than either of the other patterns. Another more technical consideration that also influenced my decision is that participation is the only one of the three measures that produced an acceptable marginal distribution (no more skewed than about 75%/25%) for all categories of participants. Each of the other two was too skewed for at least one of the different race and gender categories. Therefore, only participation was subjected to the multivariate analyses in the following chapters, although descriptive data are presented for all three measures. The questions that I asked about the frequency and quality of contact with supervisors are parallel to those asked about network centrality: Are gender and race major determinants of these dimensions of activity, or do achieved qualifications and more obviously technical considerations predominate? Does access to hierarchical superiors vary according to individuals' extraorganizational resources and ties, and, if so, do the patterns appear the same for the different race-gender categories?
26
Pathways in the
workplace
Independent variables: measures of achieved status In this category of variables, "education" refers to the number of years of formal schooling; professional "experience" records the number of years the individual had worked at his or her occupation by the time of the survey; and formal "rank" (also referred to as "authority") was measured by a dummy variable based on whether the respondent was in a position with official supervisory responsibility.14 A measure called "workplace," also a dummy, reflects whether the respondent's own employing agency was or was not the grantee agency, that is, the one with responsibility for the overall coordination of the program. Almost by definition, members of these administrative centers were likely to have greater centrality in the networks of interaction,15 and it was necessary to allow for this in the analysis. In addition to social workers and counselors, these programs included court personnel (primarily probation officers), staff and technical consultants (testing experts, psychologists, statisticians, etc.), individuals with administrative but not client-contact responsibilities, and clerical personnel. For purposes of the regression analyses, a dummy variable called "job type" was created that regroups the occupational variables according to amount of client contact. Social workers, counselors, and probation officers who worked directly with clients were coded 1 and other occupations that were more peripheral to client treatment (clerical, administrative and consultant positions) were coded 0. It is appropriate to note here that, based on the results of a preliminary analysis (Miller, Lincoln, and Olson 1981), professional experience (or, more precisely, its inverse, inexperience) will actually be interpreted as "recency of acquiring professional skills." This is an important consideration, given the emphasis in the programs on young practitioners collaborating to deal with youthful clients. This is the interpretation that I have stressed throughout the analysis. This emphasis on recent training, in turn, suggested that the age of practitioners might also be important, and it, too, was included in the analyses as a control variable.
2. Ascription, achievement, and network centrality
There are three stages in the analysis of network centrality. 1. In the first stage, the effects of gender and race are compared with the effects of those variables that were used as indicators of achievement or formal organizational position, including formal training, professional experience, occupation, workplace location, and formal authority, or rank. The effects of age are also taken into account. The hypothetical, ideal-typical reference point here is simply stated: In a purely rational situation, the advantages of the work setting would not be allocated differentially along the lines of gender and race, but rather according to objective professional investments and contributions to the system's activities. Therefore, the effects of achievement-related variables such as those I have included ought to outweigh by far the effects of ascription. A comparison of mean network centrality across race and gender categories, together with a simple regression of centrality on race, gender, and the various indicators of objective achievement, addresses this proposition. 2. By itself, this first step in the analysis is insufficient. Simply indicating whether women or men, whites or nonwhites have greater network access does not rule out the possibility that different mechanisms for achieving centrality are in operation for different categories of participants. Regardless of whether a woman has more or less access to informal channels, what she has to do to gain that access may be very different from what a man does; similarly, whites may draw upon strategies that are quite different from those available to non whites. In other words, ascribed attributes may condition the effects that achieved characteristics have on determining the advantages gained by individuals. Such a possibility is the subject of the second stage of the investigation, which compares the configurations of findings from regression analyses performed separately, first for men versus women and then for whites versus non whites. 3. A third concern involves differences among the four ascribed categories that are formed by the intersection of gender and race. Any racial differences in the effects of the achievement variables that are observed in the second 27
28
Pathways in the workplace
stage of the analysis could be complicated further when gender is taken into account; conversely, gender differences could be complicated in additional ways when whites are separated from nonwhites within each gender category. Here again, the results of four regression analyses are presented, but in this case they were performed separately for the naturally occurring categories of white men, white women, nonwhite men, and nonwhite women. This final breakdown produces subgroup sizes that are uncomfortably small for the nonwhites. The need for replication of the findings I report for them is therefore especially urgent. From one perspective, this third stage of the analysis is the decisive one, and it could be seen as superseding the other two. Although it violates parsimony, all three levels of analysis are reported to provide a direct illustration of the consequences of either ignoring or oversimplifying the effects of ascribed factors on organizational behavior. Each successive stage or layer clarifies, extends, or further complicates the issues raised in the preceding ones. A brief preview of the findings from each level will facilitate the discussions that follow. The first stage indicated that informal network centrality was distributed with surprising equality across race and gender categories. No group dominated the center of the networks, and none was consistently excluded from favorable positions. The second stage, however, revealed that the experiences of women in gaining network access differed significantly from those of men, and the experiences of whites departed substantially from those of nonwhites. The argument that the basic pathways to success are the same for all is simply not borne out. However, the differences discovered to exist between the races and between the genders do not reveal a clear-cut, segmented system of privilege/deprivation, because both handicaps and advantages were revealed for each of the categories of men, women, whites, and nonwhites. The third stage offered some clarification of these patterns but raised still another set of questions. The racial differences revealed in the second stage were complicated further by gender, and the gender differences were complicated further by race. What I have called rationalistic resource allocation mechanisms seemed to work reasonably well for white men, less well for white women and non white women, and least well for non white men. In exploring these differences further, data on participants' frequency of contact with outside organizations, their community activism, and the characteristics of their clients were brought into the analysis. Thefindingsindicate that such factors did sometimes function as professional assets in the internal networks, but that the specific ways in which they entered the picture also varied from one race-gender category to another. Clearly, the boundaries of these
Ascription, achievement, and network centrality
29
organizational systems were penetrated and their internal practices altered in several ways by the manner in which their members were tied into the surrounding community. Finally, at the end of the chapter, I will take a close look at a supplementary part of the analysis that dealt with the connections among just three variables - authority, sociometric influence, and expertise - for each of the four race-gender categories. I will argue that these three variables and their interrelations comprise the conceptual core of the rationalistic model. Although it represents something of a digression from my central theme, the same rationale that led me to ask about race-gender differences in network access aroused my curiosity about this issue as well.1 Stated simply, the rationalistic conception stipulates that the connections among authority, expertise, and influence should be close and consistent across diflFerent groups in an organization. In the present case, this means that those who had authority ought to have been individuals with superior expertise who were also recognized by others as influential in the daily affairs of the programs. The findings failed to confirm this ideal-typical formulation for any of the four racegender categories, but in the process of discovering this result, some important additional insights into the experiences of the white men in the study became apparent. I think this result justifies the space given to this discussion. Stage 1: comparison of means and overall regression analyses No category of individuals controlled the center of the professional networks and no group was consistently relegated to the periphery. Mean centrality on the five network dimensions varied little from one category to another (Table 2.1), and neither race nor gender was a significant factor in explaining the variance in centrality (Table 2.2).2 Much of the variance in centrality was traceable to authority (rank) and, as expected, to central workplace location (i.e., being a member of the grantee agency). In addition, a pattern of greater network integration on the dimensions of respect and assistance appeared for those in the high client-contact occupational category (that is, those who were scored 1 on the variable job type). Relative isolation on the support and assistance measures characterized practitioners with greater experience, afindingthat might seem paradoxical but, as was indicated previously, is consistent with the fact that such programs rely upon the collaboration of newly trained practitioners to carry out their client-related activities. The statistically significant inverse relation
Table 2.1. Means and standard deviations for the variables used in the analysis B. Breakdowns by race- gender
A. Breakdowns by race and gender separately
White
Gender* Color* Experience Education Rank* Age Workplace* Social work* Counselor* Court work* Staff/consulting* Administration0 Other occ* Work contacts Influence Respect Support Assistance
Women (#=159)
Nonwhites (# = 57)
Men (# = 71)
s.d.
Mean
s.d.
Mean
s.d.
Mean
s.d.
Mean
s.d.
Mean
s.d.
.48 — 5.43 2.09 .48 8.90 .45 .46 .39 .25 .24 .41 .35 .14 .13 .13 .15 .12
.49 — 5.23 15.88 .30 31.11 .18 .46 .14 .05 .04 .11 .16 .74 .74 .74 .78 .77
.50 — 4.18 2.90 .46 8.49 .38 .50 .35 .23 .19 .31 .37 .16 .16 .16 .19 .15
6.71 17.30 .38 32.77 .25 .25 .24 .07 .07 .23 .10 .73 .76 .77 .78 .79
6.58 2.04 .49 8.07 .44 .44 .43 .26 .26 .42 .30 .15 .13 .13 .16 .13
5.16 16.32 .32 31.05 .30 .32 .16 .06 .05 .20 .17 .73 .75 .76 .78 .80
4.60 2.03 .47 9.31 .46 .47 .37 .24 .23 .40 .38 .14 .13 .13 .14 .11
4.48 16.25 .36 29.36 .25 .39 .14 .11 .07 .11 .14 .75 .75 .75 .79 .77
3.00 3.24 .49 4.96 .44 .50 .36 .32 .26 .32 .36 .16 .17 .16 .19 .16
5.96 15.52 .24 32.79 .10 .52 .14 .00 .00 .10 .17 .73 .73 .73 .77 .76
5.02 2.54 .44 10.69 .31 .51 .35 .00 .00 .31 .38 .16 .16 .16 .19 .15
Whites (#=199)
All (# = 261)"
Men (#=100)
Mean
s .d.
Mean
s.d.
Mean
s .d.
Mean
.39 .78 5.58 16.49 .33 31.57 .26 .33 .18 .06 .05 .19 .15 .73 .75 .76 .78 .79
.49 .42 5 .15 2 .29 .47 8 .81 .44 .47 .38 .24 .23 .39 .36 .15 .14 .14 .16 .13
_ .72 6.08 16.99 .37 31.81 .25 .29 .21 .08 .07 .19 .11 .73 .75 .76 .78 .79
.45 5.87 2.46 .49 7.42 .44 .46 .41 .27 .26 .39 .31 .15 .14 .14 .17 .14
.82 5.29 16.16 .31 31.42 .27 .35 .16 .05 .04 .19 .18 .73 .74 .75 .77 .79
.39 4 .65 2 .14 .46 9 .59 .45 .48 .37 .22 .21 .39 .38 .15 .13 .14 .15 .12
.36 — 5.71 16.67 .34 31.66 .28 .30 .19 .07 .06 .21 .15 .73 .75 .76 .78 .79
Nonwhite Women (#=128)
Men (#=28;)
Women (# = 29)
"Small discrepancies in #'s are due to missing values on gender and color. These are dummy variables, where male, white, supervisor, and member of the agency charged with overall coordination were coded 1; for the occupational variables, membership in the category was coded 1.
Table 2.2. Regression of five measures of network centrality on eight indicators of ascription and achievement Work contacts*
Color* Gender* Experience Education Age
Rank* Workplace* Job type* R2 (R2)
Intercept
Support
Respect
Influence
Assistance
b
seZ?
b
seZ?
b
seZ?
b
seZ?
b
seZ?
-.015 .002 -.002 -.004 .003** 067*** .106*** .009
.021 .018 .002 .020 .001 .020 .020 .018
.010 .014 -.003 -.004 -.002 .058** 076*** .007
.020 .017
.017 .011 -.003 -.005 -.002 .078*** .084*** .037*
.020 .017 .002 .004
.000 .017 -.007 -.007 -.002 064*** 064*** .005
.023 .019 .002 .004 .001 .021 .021 .019
.026 .003 - .006*** -.001 .000 .050** .068*** .032*
.018 .016
.20 (.17) .866
14 (.11) .822
.002 .020 .001 .019 .019 .017
.17 (.15) .820
.001 .019 .018 .017
.14 (.12) .943
.002 .004 .001 .017 .017 .015
.16 (.14) .757
Note: */?<.05, **/?<.01, ***/?<.001. °In this and subsequent tables, b refers to unstandardized regression coefficients andseZ? refers to their standard errors. *Color, gender, rank, workplace, and job type are dummy variables where 1 refers, respectively, to white, male, supervisory rank, central workplace, and high client-contact occupation. These distinctions are used throughout the analysis in this and the following chapters.
32
Pathways in the workplace
between age and centrality in the work contacts networks is not inconsistent with this observation, given that younger practitioners were thought to be more in touch with the problems of juveniles. Finally, an unexpected result is that education had no appreciable impact on network integration. There is virtually nothing in this stage of the analysis to indicate that considerations of either race or gender dominated the way these programs were structured. In fact, using the strategy of many past studies, it would probably be reported that ascribed differentiation had been neutralized, much in the manner suggested by the leveling hypothesis. However, the following sections will make it clear that such a conclusion would have been premature.
Stage 2: the interaction of ascribed and achieved attributes The analysis in this section addresses whether there were important race and gender differences in the patterns of effects for experience, education, rank, workplace, occupation, and age. The approach used was to compare the results of separate analyses for males and females, whites and nonwhites. Three types of assessments are made. First, an overall test of the equality of regression slopes across pairs of categories (men versus women, whites versus nonwhites) was performed. The purpose of this global test was to determine whether the separation of the population of respondents into subcategories significantly improved the explained variance in the dependent variables over what was accomplished in stage 1 of the analysis, where race and gender were added separately to the zero-order regression equation. A second kind of comparison focuses on male-female and white-nonwhite differences in the regression coefficients associated with particular variables. Sixty such comparisons are possible: comparisons by race, then by gender, for the effects of six independent variables on five dependent variables. Statistical tests (r-tests) for all of these comparisons have been computed but are not presented in the tables. They will be referred to in the text only when a specific race or gender difference is singled out for discussion. The information needed to perform these tests (regression coefficients and standard errors) is given in the tables. 3 Finally, the general configuration of findings for each of the different subcategories is examined. Here the focus is on the differences in the combinations of variables that seemed to be operating for each subgroup. The global tests for race and gender interaction effects are presented formally as hypotheses 1 and 2 in Table 2.3. The test statistic in each case is an F-ratio of the form:
Ascription, achievement, and network centrality
33
Table 2.3. F-tests for race and gender interactions'1
Hypothesis 6
df
Work contacts
Influence
Respect
Support
Assistance
1. pm=£py
6 ,242 6 ,242 6 ,231
4 .24*** 2 .17* 4 .36***
4 .78*** 2 .66* 5 .91***
3 .40*** 2 .71** 5 .21***
4 .44*** 2 .09 5 .66***
3 .46* .54 5 .68*
3
ft -/-ft P«H™ ^
Pmv/
Note: > < . 0 5 , * > < . 0 1 , **><.001. "Cell entries are F-statistics. b fim is the population vector of six slope parameters for males. $f is the analogous vector for females. $w and $nw are the same vectors for whites and nonwhites. p wm, P>v/> Pnwm> anc * P«w/ a r e t n e vectors of six slope parameters for equations within racegender subgroups. =
(q.N - P)
ESS C - ESSu/
ESSC is the error or residual sum of squares from the constrained regression of centrality on the six independent variables. The constraint is that the vector of slopes is presumed to be constant across race and gender classes, although the intercepts of the equation may vary. The unconstrained ESSU is the residual sum of squares from an identical equation, except that intercepts and slopes are allowed to vary. In the equation, q is the number of constraints, N is the number of observations, and p is the number of parameters estimated in the unconstrained regression. 4 The first row of Table 2.3 shows that all of the relevant F-ratios concerning gender are significant at conventional probability levels, indicating that for all five centrality dimensions, gender-specific effects differ by more than sampling error. The F-ratios for race are significant for three of the five network dimensions, namely, work contacts, influence, and respect. These results constitute strong statistical evidence that, on the whole, race and gender strongly affect the patterns of relations between the independent variables and network centrality. Having established this, the next step is to examine the detailed within-group regression results that appear in Table 2.4. Comparisons of men and women It would not have been surprising to find that men and whites were better able to translate job-related qualifications into network advantages. There is
Table 2.4. Regression of network centrality on experience, education, rank, age, workplace, and occupation, performed separately for men, women, whites, and nonwhites Work contacts seb
A. Men Experience Education Rank Age Workplace Job type R2 (R2)
Intercept
B. Women Experience Education Rank Age Workplace Job type R2 (R2)
Intercept
C. Whites Experience Education Rank Age Workplace Job type R2 (R2)
Intercept D. Nonwhites Experience Education Rank Age Workplace Job type R2 (R2)
Intercept
Influence
Respect seb
Support seb
Assistance seb
seb
.003 .003 .026*** .006 .101*** .030 .001 .030 .120*** .031 .076** .029 .31 (.27) 1.040
005 .003 024*** .006 118*** .028 004 .002 087** .029 073** .027 30 (.25) .953
004 .003 022*** .006 115*** .028 .002 004 093** .029 097** .027 31 (.27) .924
-.008* .004 -.027*** .007 .140*** .035 .004 .003 .053 .036 .075* .034 .26 (.21) 1.060
.009** .003 .015** .006 .084** .029 .004 .002 .047 .029 .081** .027 .26 (.21) .894
.002 .003 .011* .005 .053* .025 .004** .001 .096*** .024 .022 .024 .25 (.22) .654
.002 003 011* .005 023 .023 002 .001 072** .023 .021 026 18 (.15) .648
.002 002 .005 007 .024 059* .001 002 079*** .023 .021 004 18 (.15) .672
- .007** .003 .007 .005 .013 .025 -.002 .001 .068** .025 -.031 .022 .19 (.15) .758
.004 .002 .013** .004 .024 .021 .000 .001 .082*** .020 .003 .018 .20 (.17) .570
.001 .006
.002 .005 .092*** .020 .004*** .001 .116*** .020 .015 .019 .28 (.25) .894
.002 001 .004 004 074*** .019 .001 002* 081*** .019 .018 010 18 (.15) .838
.002 000 .004 006 094*** .019 003* .001 082*** .019 046* .018 21 (.18) .863
- .004 .002 -.008 .005 .076*** .022 -.003* .001 .059** .022 .012 .020 .15 (.13) .988
.002 .005** .001 .004 .063*** .018 .000 .001 .068*** .017 .033* .016 .19 (.16) .785
.006 .008 .008 .001 .058 .023 .003 .001 .060 .122* .049 .025 .13 (.03) .795
.006 014* .008 001 .056 000 .003 000 .058 099 .048 018 19 (.10) .837
014* .005 001 .008 .054 010 000 .003 115* .055 002 .046 .22 (.12) .827
- .019* .006 .000 .009 .003 .063 .001 .003 .112 .065 - .036 .053 .24 (.15) .865
.011 .006 .002 .008 .013 .055 .001 .003 .095 .056 .014 .047 .15 (.05) .791
Note: *p<.05, **p<.0l,
***p<Ml.
36
Pathways in the workplace
ample evidence for a superior return on investment in the literature on status attainment and mobility, and it is a reasonable working hypothesis that it would also be apparent for informal integration in the workplace. This inference is only partially supported. Different combinations of positive and negative factors did in fact go into the determination of network centrality for men (panel A) and women (panel B). For both genders, workplace had an important positive influence; for men the coefficients were significant on three dimensions of centrality and for women on all five. For education, however, a crucial difference is apparent that contradicts what is usually reported in comparisons of men and women. For women, this variable contributed significantly and positively to centrality on three of the five network dimensions, but for men the result was reversed: The effect of education was significant and strongly negative across all five dimensions. When the education coefficients for men are compared to those for women, all the differences are statistically significant at the .01 level. Less pronounced but nevertheless interesting contrasts appeared in the effects of rank. The payoff for high rank was consistently positive for both men and women, but the increments to centrality were generally far greater for men. [For a similar finding in a very different organizational context, see Olson and Miller (1983).] All of the specific differences between the genders again were significant at or beyond the .05 level, but the contrasts were particularly great on the dimensions of influence and support. This is evident in the comparison of the unstandardized coefficients: on influence, .118 for men versus .023 for women (p<.05); and on support, .140 for men versus .013 for women (/?<.01). Recall from the previous chapter that centrality was standardized by dividing the individual's own score by the highest score attained in his or her program. Therefore, the regression coefficients in this table represent percentage increments toward these top scores. When it came to influence and support, in other words, men who had authority had network gains equal to about 12 and 14 percent compared to centrality gains of roughly 2 and 1 percent for women with authority. Statistically significant contrasts by occupation are also apparent. For men, having a high client-contact occupation (that is, being a counselor, social worker, or court employee) was associated with greater centrality, a pattern that was entirely missing for women in the same occupations. Although men were not numerically dominant in these high client-contact categories (see Table 2.1 A), it is apparent that they fit into the professional exchange networks in different ways than their female colleagues did. Finally, the effect of age was also variable by gender, but only on the work contacts dimension. For women, the relationship was negative (as expected),
Ascription, achievement, and network centrality
37
but for men age was not an important factor. The male-female difference is significant at the .05 level. Comparisons of whites and nonwhites Panels C and D of Table 2.4 compare the outcomes for whites with those for nonwhites. By far the most important differences between the races, as between the genders, involves formal rank. This achievement-related mechanism for gaining network access conveyed distinct advantages for whites, a result that is conspicuously absent for nonwhites. The differences in regression coefficients between the races on this variable are all significant at the .01 level. On the other hand, the advantages of recent professional training tended to favor nonwhites, although the differences are not as large as in the case of authority (three of the five differences - those for influence, respect, and support - were statistically significant). To recapitulate briefly, in the diverse array of findings from this stage of the study, several gender- and race-linked influences on centrality are indicated. However, the data do not conclusively demonstrate systematic inequity in the patterns of reward and resource allocation. Each race and gender category had both advantages and disadvantages that were either weaker or missing altogether for its counterpart. Clear race and gender polarities did not emerge and, as a result, a number of plausible scenarios could be advanced to account for the findings. One scenario could be built around the fact that women commanded greater network centrality on the basis of their superior educational credentials. This is an apparently rational process that calls directly to mind Weber's discussion of the bureaucratic importance of certification (Weber 1947:333). Men, by comparison, consistently had to pay a price in the form of isolation for the educational investments they had made. Focusing on this contrast points to what may be an inequity directed against men, and could be used to lend credibility to the idea that the social service delivery area has leveled, in fact, in part reversed, the general societal pattern of male domination. In contrast, however, a scenario that focuses on the advantages of rank would suggest a very different conclusion. The fact that men and whites who had formal supervisory authority consistently found themselves close to the center of the exchange networks signals their ability to exploit this crucial organizational resource, the formal structure, to their benefit. Their rank advantages in fact were sufficient in magnitude to more than offset whatever disadvantages they faced where other variables were concerned. In this connection, it is of particular importance that for women and nonwhites, author-
38
Pathways in the workplace
ity carried with it significantly less claim to influence over co-workers than it did for men. In the Weberian scheme, the ability to translate official authority into actual influence is a major indicator of bureaucratic success (Miller 1970; Miller and Fry 1973). As a whole, the pattern observed here is definitely not consistent with the leveling effect suggested by the rationalist model. The data supporting this second scenario are more consistent with what previous research has shown. From the perspective of organizational theory, it is especially important to have found that the power of office was clearly less effective for women and nonwhites than for men and whites. Nevertheless, the two scenarios based on education and authority do in fact undermine each other. Without more information than that presented so far, it is futile to search for a pattern that conclusively supports or unequivocally refutes the notion of domination by whites and males. The data simply do not offer this certainty. They are better suited to an argument that posits an underlying process of interpersonal negotiation in which each category of participants had strong suits with which to advance its own position and liabilities for which it had to compensate. The third stage of the analysis, which adds information on the effects of the intersection of gender and race, will provide a better understanding of the forms that this process might have taken. Stage 3: considering the simultaneous effects of race and gender In this section the focus is on whether white men, white women, nonwhite men, and non white women differed in ways that add substantially to what was previously reported. Hypothesis 3 in Table 2.3, which posits different regression parameters across the four race-gender subgroups, is strongly confirmed for all five types of network ties. 5 The detailed results appear in Table 2.5. Some variables, such as workplace and experience, had essentially similar effects for all four categories (although, given the variations in subgroup size, not all of the coefficients were statistically significant). On other variables, however, the differences among the four groups were pronounced and the patterns of relative advantage and disadvantage across ascribed categories now become more apparent. Compared to others, white men profited above all from occupying positions of authority and from having high client-contact occupations. Their coefficients on these two variables were strong and were consistently different from those for the other three race-gender categories. Education, on the
Ascription,
achievement,
and network centrality
39
other hand, came through as a strong, significant, negative factor for white men on all but one of the five centrality measures (assistance). This is a pattern that they shared with nonwhite men but that clearly set them apart from both categories of women. A closer look at the overall configuration of findings for white men shows that, although not all of the results confirmed the prediction derived from the rationalist model, every one of the variables measuring their bureaucratic standing (education, experience, authority, occupation, and workplace) had a statistically significant impact on one or more aspects of their integration into the professional networks. Their experiences, in other words, were largely defined, one way or another, by their bureaucratic investments and contributions. This cannot be said for any of the other three race-gender categories. To be more specific, unlike white men, white women did not have authority in their favor on the three dimensions of influence, support, or assistance. The regression coefficients are significantly lower than those for men and not significantly different from zero. Authority facilitated their attainment of centrality on the work contacts and respect measures, but even here the payoflF was significantly less (/?<.O5) than for white men. Education had virtually no clear payoflF for white women, a surprising finding given the positive effect reported earlier for the larger aggregate of all women. Another unexpected finding had to do with age, or more correctly youth, which contributed significantly to four out of five aspects of centrality. This result was confined entirely to this one practitioner category. For nonwhite women, superior education was a comparatively effective credential factor upon which to base a claim to network centrality. In fact, this was the only group for whom education had a consistently positive impact. The coefficients are significantly different from zero and in general are significantly different from those for the other three race-gender categories. On the other hand, the effects of authority were consistently lower than those for either white men or white women, indicating that on this factor nonwhite women resembled neither whites in general nor women in general. The absence of any noteworthy effect of age distinguishes them further from white women. Finally, in the configurations of effects for nonwhite men, none of the variables in the equation reached the .05 level of significance on any of the five centrality dimensions. In the consistently negative effect of education (which approached the .05 level on four of the five dimensions of centrality), they were like white men and unlike both categories of women. Their coefficients for workplace and job type were also generally similar to those for
Table 2.5. Regression of network centrality on experience, education, rank, age, workplace, and occupation, performed separately for white men, white women, nonwhite men, and nonwhite women
A. White men Experience Education Rank Age Workplace Job type R2 (R2)
Intercept
Work contacts
Influence
b
b
seb
- .004 .003 -.025*** .008 .127**** .033 .001 .003 .134**** .034 .074** .033 .37 (.31) 1.021
Respect seb
b
Support seb
b
Assistance seb
b
seb
- .006* .003 -.019*** .007 .138**** .029 .005* .002 .090*** .030 .075** .029 .38 (.32) .847
-.005* .003 - .017** .007 .127**** .029 .004* .002 .085** .031 .100*** .029 .36 (.29) .825
-.009 ** .004 - .024** .009 .150**** .038 .005 .003 .042 .040 .074* .038 30 (.23) 1.001
-.012**** .003 -.009 .007 .101 .029 .004* .002 .026 .031 .067** .029 .36 (.29) .805
B. White women .000 .003 Experience .004 .006 Education .075*** .027 Rank -.005**** .001 Age Workplace .108**** .026 Job type - .011 .024 .31 (.28) R2 (R2) Intercept .772 C. Nonwhite men Experience .003 .013 Education -.022* .013 Rank .005 .094 .000 .007 Age Workplace .115 .087 Job type .043 .077 .26 (.04) R2 (R2) 1.056 Intercept
-.000 .003 .003 .006 .042 .026 -.004** .001 .083**** .025 .011 .023 .20 (.16) .781
.002 .003 -.002 .006 .079*** .027 -.004*** .001 .081*** .025 .027 .024 .21 (.17) .841
-.003 .003 -.001 .006 .033 .023 -.004 *** .001 .077 *** .027 -.004 .025 ;20 (.15) .916
-.002 .002 .004 .005 .034 .028 .000 .001 089**** .022 .016 .020 .18 (.14) .705
-.007 .013 -.023* .013 .028 .092 - .002 .007 .126 .086 .037 .077 .31 (.11) 1.147
-.007 .012 -.023* .012 .026 .087 -.017 .007 .161* .081 .042 .072 .37 (.19) 1.124
-.011 .015 -.023 .015 .060 .111 -.002 .008 .120 .102 .045 .091 .:25 (.04) 1.195
.000 .012 - .022* .012 -.003 .084 .000 .007 .134 .082 .073 .055 .31 (.1 1) 1.059i
D. Nonwhite women -.010 .007 Experience .031 .013 Education .033 .073 Rank .002 .003 Age Workplace .164 .100 -.063 .065 Job type .43 (.27) R2 (R2) .242 Intercept
-.015 ** .006 .029 ** .012 .037 .068 .003 .003 .103 .092 -.060 .060 50 (.36) .292
.006 015** .011 028** .064 048 002 .003 093 .089 045 .051 .51 (. 37) .315
020*** .006 029** .013 021 .073 003 .003 097 .100 106 .064 .56 (.43) .387
.006 -.013** .031*** .011 .047 .063 .002 .003 .087 .086 -.022 .056 .50 (.36) .274[
Note: *p
42
Pathways in the workplace
white men. But the most important finding for this category is the fact that official authority displayed consistently weaker effects than for all three of the other categories - significantly weaker, in fact, than the coefficients for both white men and white women. In short, the repertoire of negotiating resources for nonwhite men was comparatively meager. This is as close as the data came to showing a clear pattern of inequity, and it could be argued that the principle of bureaucratic rationality had broken down or was simply not in operation for this group of practitioners. This brings to mind the internally segmented arrangement referred to in my earlier study (Miller et al. 1975) as "compartmentalized rationality," a situation in which nonrational patterns for one group were seen to exist alongside more rational-appearing patterns for another group. The weakness of this interpretation for the present case is that nonwhite men were not, on the average, disproportionately isolated (refer again to Table 2. IB), and therefore cannot be said to have been deprived of the benefits of social integration with their colleagues. Nevertheless, their unique configuration of findings raises the possibility that, for them, alternative resource and reward allocation mechanisms not yet tapped by this analysis were in operation. Exploring the importance of external activities A special effort was made to assess this possibility by asking in what additional ways nonwhite men might have differed from other practitioners. Were any interpersonal strategies or unique contributions open to them that were less available to others? One reasonable hypothesis, given the circumstances in which the service delivery programs were created, is that they gained network access by relying upon the fact that they were precisely who they were: nonwhite men in programs dealing with fairly large numbers of minority and especially minority male clients in largely minority communities. As I suggested in Chapter 1, in such a context their contributions to the programs may have been both uniquely valuable and highly visible, and for this reason they could not, in good sense, be held apart from the informal mainstream of program activity. Perrow (1979:7-12) makes a similar point when he argues that what appears highly particularistic (and therefore suspect from the point of view of impersonal bureaucratic universalism) may in fact not be irrational when the environmental demands on the organization are considered. He uses the example of an organization giving preference to Masons as boundary personnel because Masons are known to occupy dominant positions in the local power structure. Although such a procedure is clearly particularistic from the per-
Ascription,
achievement,
and network centrality
43
spective of equity and the assumed leveling process, from the perspective of goal attainment it does not violate the principle of rationality, if rationality is taken simply to mean the deliberate allocation of resources by an organization for the accomplishment of its technical objectives. Conversely, an individual Mason who took advantage of such a situation, whether to perform his job better or to gain personal advantage, would elicit little surprise. Stated more generally, there are three parts to this argument. First, organizations can be expected to use whatever resources are at their command to further their interests, including, if need be, the ascribed or demographic characteristics of their members [see Olson and Miller (1983)]. Second, it would be consistent with this proposition if the membership of an organization responded positively, collectively, to the useful contributions made by co-workers, even where the ability to make these contributions is clearly tied to ascribed status differences and/or to connections in the external system. Finally, in their own interest, individuals could also be expected to be aware of and to make conscious use of whatever benefits these first two processes might bring to them. Persons who gain increased access to professional interaction and enhanced professional regard in any of these ways, especially where more institutionalized internal reward mechanisms offer them less promise, could only be seen as making sensible use of all of their available personal resources. Client relations, external contact, and community activism To explore the hypothesis that such considerations affected the centrality of practitioners in this survey, particularly the nonwhite men, three sets of factors not previously considered were brought into the analysis. The first, referred to as "client-practitioner homophily," has to do with matching the race and gender of the practitioner with those of the clients processed by the programs. The second, "external contact," deals with six indicators of the interaction that practitioners had with organizations officially outside the interagency treatment network but that were directly important to the functioning of the programs. The third, "community activism," refers to four items measuring special activities undertaken by practitioners in the larger community. Given the nature of the programs and the communities in which they were placed, all of these factors might be expected to work disproportionately in favor of nonwhite male practitioners [cf. Kluegel (1978:286)]. The fact that these men were present in the programs in relatively small numbers could also have amplified their importance as individuals and improved their bargaining position as a result.6
44
Pathways in the workplace
Only client-practitioner homophily had an effect on centrality that confirms the prediction. It will be discussed in some detail, followed by shorter discussions of the measures of external contact and community activism. Effects of client characteristics: In the delivery of personal services, it is often argued that matching clients and practitioners by race and gender should have significant treatment advantages for the clients. A corollary to this argument is that the position of any practitioner ought to be enhanced, even at the price of an increased work load, by the presence of a large number of clients in his or her own race-gender category. A strong argument can be made that this corollary should apply with special force to nonwhite men. To support this, in Table 2.6 the juvenile offenders remanded to the programs are compared by mean number of officially recorded offenses, which in turn are subdivided first into "status" (technically nondelinquent) versus legally delinquent offenses and then into "serious" and "nonserious" offenses according to the Rossi scale of seriousness [see Rossi et al. (1974)]. There is little doubt that nonwhite male offenders had a more serious behavioral profile than other categories of clients, since they had committed an above-average number of nondelinquent status offenses (second to those of nonwhite females) and the largest number of delinquent offenses. They had also committed twice as many serious offenses (those rated above the Rossi midpoint) as any other category. Where nonwhite male clients were especially numerous, my hypothesis is that this troublesome behavioral profile would enhance the importance of nonwhite male practitioners and, as a result, increase their professional integration into their programs. Since the clients in other race-gender categories had less troublesome behavioral profiles, such a result was less to be expected for the other race-gender categories of practitioners, although in principle the proposition - the more clients, the greater the practitioner centrality - ought to apply in varying degrees to all four race-gender categories. To test this hypothesis all practitioners' data records were extended to include the proportion of clients remanded to their program whose race and gender were the same as their own. This measure of homophily is indirect because it does not record actual client-practitioner interaction. The bias, if any, created by this should have the effect of decreasing the correlations between homophily and centrality, contributing to a conservative test of the hypothesis.7 Statistically, the procedure I followed was to add homophily to the subgroup regression equations that produced the basic findings in Table 2.5. This provided an estimate of the net effect of homophily when all the other
Ascription, achievement, and network centrality
45
Table 2.6. Offense patterns ofjuvenile clients
All
clients
Males
1.35
1.34
.81
.80
Mean number of status (non-delinquent) offenses: Priors Subsequents
.63 .43
Mean number of delinquent offenses: Priors Subsequents Number of offenses above Rossi seriousness scale midpoint: Priors Subsequents
Mean number of officially recorded offenses (all types):
Priors 0 Subsequents*7
Number of offenses below Rossi seriousness scale midpoint: Priors Subsequents a
Nonwhite
White Females
Males
Females
.78 .60
2.50 1.15
1.58
.40 .30
.51 .41
.79 .35
1.01
.68 .33
.91 .48
.25 .15
1.64 .71
.49 .18
.19 .11
.25 .17
.05 .04
.51 .26
.14 .05
1.16
1.09 .63
.73 .56
1.99
.70
.73 .92
.89
.98
.69
Priors are those offenses recorded in clients' files before they entered the programs. Subsequents refers to the number of offenses officially recorded within six months after leaving the programs.
b
46
Pathways in the workplace
independent variables were controlled. To simplify the presentation, Table 2.7 displays only the regression coefficients and contributions to R2 for the homophily variable. For nonwhite men, there is clear confirmation of the homophily hypothesis. The R2 increments (that is, the variance explained beyond that reported in Table 2.5) ranged from 7% to a very substantial 26%, and on three of the five dimensions (influence, respect, and support) the changes were significant at or beyond the .05 level. The increment for assistance closely approached this criterion (/?<.10), but the relationship to work contacts was the weakest of the five (p>\0). Among the other three race-gender categories, only one R2 increment, involving white women and the sociometric dimension of support, achieved statistical significance. With this single exception, the composition of the client population did not exert an important influence on the standing of any of these other categories of practitioners. The importance of this result is easy to assess. Client-practitioner homophily is the only variable brought into the analysis so far that suggests a uniquely positive way for nonwhite men to improve their standing in the informal networks of their programs. It is likely that the intersection of their race and gender identities in this case indexes their superior ability to carry out an activity that was crucial to program success. Therefore, the unique association between client-practitioner homophily and social-relational rewards can be seen as the sensible exploitation, both by the organization and by the nonwhite male practitioners themselves, of this special technical capability. Note, however, that even this variable failed to have a significant effect on their actual daily working interaction (work contacts) with other practitioners. It is important, too, that the homophily mechanism turns on a factor - the nature of the client pool - the importance of which was almost certainly recognized by program participants and decision makers alike, but that was at the same time largely beyond the control of both of them. Keep in mind that variables that would seem to be more readily subject to selfinterested manipulation by the individual, such as experience, occupation, education, and rank, failed to have any significant positive effects for nonwhite men. Contacts with agencies outside the network: Because of the nature of the offenses committed by their clients (the most common status offenses, ranked by the level of trouble they were seen as causing, were runaway, ungovernability, truancy, curfew violation, and possession of alcohol), the practitioners in the programs had to interact frequently with community agencies that were not part of the contractual program network but existed just outside its
Table 2.7. Effects of client-practitioner homophily on the centrality of program participants Unstandardized regression coefficients0 and increments6 to explained variance for proportion same race-gender clients Nonwhite
White Women
Men b
Work contacts
-.069
R2
Influence R2
Respect R2
Support R2
Assistance R2
Change
b
.011
.000
.37
.030
.002
113 ***
.482**
b
Change
.071
-.053
.379**
.049
.664***
.179
-.011
.113
.012
-.066 .52
.256
.008
.066 .57
.097 .41
.000 .50
.51
.342*
.011 .45
.48
.000 .18
Change
.48
.25
.005
R2
.33
.22
.001 .36
.297
.005
.036
.30
.023
.000 .22
.000 .36
b
.018
.063
.38
.000
Change .31
.000
-.011
R2
R2
R2
Dependent variable
Women
Men
-.143
.053 .56
Note: */?<.10, **/?<.05, ***/?<.01. a The subgroup regressions from which these values are taken include seven regressors: client-practitioner homophily plus the six independent variables (experience, education, rank, age, job type, and workplace) that appear in Table 2.5. b 2 R refers to the total explained variance in centrality in the seven-variable solution and R 2 Change refers to the net contribution of homophily to this total.
48
Pathways in the workplace
Table 2.8. Differences among four race—gender categories on six measures of contact with important agencies outside the official program network and four measures of activism in the community A. Measures of external contact Nonwhite
White
Frequency of contact with: Law enforcement agencies Schools Religious groups Private service agencies Courts Public service agencies
All
Men
2.50 2.66 1.74 2.52 3.43 3.05
2.35 2.61 1.58 2.57 3.39 3.03
Women 2.42 2.61 1.76 2.36 3.35 3.00
Men
3.00 2.68 1.57 3.29 4.25 3.25
Women 2.69 2.97 2.21 2.31 3.11 3.15
B. Measures of community activism Frequency of activities designed to: Increase cooperation of community organizations Increase community financial support Influence community policies for juveniles Improve treatment resources
2.20
2.28
1.95
2.82
2.43
2.95
3.07
2.63
3.69
3.32
2.83 3.21
3.13 3.35
2.58 2.77
3.32 4.57
2.68 3.36
boundaries. The federal sponsor encouraged the development of a strong community base in the programs, and this reinforced the incentive to develop ties with these outside agencies. The practitioners provided information on their contact with schools, law enforcement agencies, courts, religious organizations, and public and private service delivery agencies outside the program. Since these external contacts were both officially mandated and unofficially encouraged by the nature of the work, it was assumed that practitioners would have more reason for cultivating them than for avoiding them. The hypothesis was that those who were most effective in developing these external ties would find their internal standing among their co-workers enhanced. The levels of external contact for the four subgroups are reported in Table 2.8A. Because my point of reference in this part of the analysis is nonwhite
Ascription, achievement, and network centrality
49
men, it is important to point out that they had noticeably more contact than the other three race-gender categories with four of the six external agencies, including law enforcement agencies, the courts, and both public and private service delivery agencies. In addition to the differences reported in the table, nonwhite men also reported more frequently than the other practitioners that their external contacts were directly related to the needs of clients, and they placed the highest evaluation on the actual client benefits produced by their contacts. For five of the six external organizations they rated the level of cooperation in the contacts higher than others did, and infiveout of six cases they reported more frequent contact with persons in positions of authority.8 Overall, this profile suggests that nonwhite males were more deeply involved in extraprogram activity and, judging from their own evaluations, that they interacted with these external organizations on a more favorable basis than other practitioners did. Did this level of external activity work disproportionately to their benefit in preventing isolation from the informal networks? To test this possibility, the six external contact measures were added to the six independent variables in the basic regressions in Table 2.5. The regression coefficients and contributions to explained variance for just the external contact measures are given in Table 2.9. In examining these data, note that for nonwhite men and women the use of 12 independent variables encroaches dangerously on the number of cases (28 and 29). For this reason, this supplementary analysis is best seen as suggestive and exploratory. The justification for proceeding is that a potentially unstable estimate presented for discussion here can be substantiated, modified, or discarded in subsequent studies, whereas a potentially fruitful insight abandoned now might be more difficult to regenerate in the future. Only the highlights of this analysis will be summarized here.9 Unlike what was found for client-practitioner homophily, no special advantage from external organizational ties was confirmed for nonwhite men. The overall increase in explained variance is substantial, but in looking at the details of the analysis, it can be seen that negative effects were as much in evidence as positive ones. To be specific, only one variable, external contact with religious groups, had a consistently positive impact on any dimension of network centrality, and this was not one of the external groups with which nonwhite men reported above-average contact. For work contacts and assistance, the contributions to R2 (12.9 and 9%, respectively) were statistically significant (/?<.O5), and for each oif the remaining three centrality dimensions, the contributions just missed significance at conventional probability levels (/?<.1O). Just as important as this finding, however, is the fact that contact with external public agencies showed even stronger and consistently significant negative effects on centrality. Here the R2 increments ranged from
Table 2.9. Effects of six types of external contact" on five dimensions of network centrality for four race-gender categories Nonwhite
White Women
Men
b
Work contacts Law enf. orgs. Schools Relig. orgs. Private agencies Courts Public agencies Total/?2 = Increment = Influence Law enf. orgs. Schools Relig. orgs. Private agencies Courts Public agencies Total/?2 = Increment = Respect Law enf. orgs. Schools Relig. orgs. Private agencies Courts Public agencies Total/?2 = Increment = Support Law enf. orgs. Schools Relig. orgs. Private agencies Courts Public agencies Total R2 = Increment = Assistance Law enf. orgs. Schools Relig. orgs. Private agencies Courts Public agencies Total R2 = Increment =
Men
Women
R2
R2
Change
Change
b
Change
b
Change
.000 .002 .000 .028 .000
-.004 .021 .088** -.006
.001 .013 .129 .002
.028 .030 .009 -.010
.021 .038 .004 .007
.021 -.072**
.023 .226
-.027 .009
.033 .003
.003 .005 .019 .022
.000 .001 .008 .018
-.014 .003
.014 .000
b
.002 .006 .003 .021** -.002 -.016*
.019
R2
.37 .05
.43 .06
.018 .008 .034 .080 .040 .000
-.019 .012 .035 ** .042 *** -.020** .001
.002 .004 -.004 .032**** -.001 -.013
R2
.59 .33
.000 .001 .001 .088 .000 .017
-.014 -.005 .077* .005 .007 - .058**
.65 .22
.007 .001 .092 .001 .003 .143
.051** .013 .014 -.002 -.040** .002
.57
.31
.64
.75
.19***
.11**
.34*
.25*
-.018 .016 .045 ** .032 ** - .024 ** .008
.016 .014 .057 .047 .056 .004
.002 .001 -.001 .031**** -.006 -.016*
.56
.32
20***
.10**
-.022 .004 .010 .062! '** -.012 -.009
.017 .001 .002 .111 .009 .003
.013 .004 -.010 .033**** -.002 -.021**
.44
.33
.15**
.13**
.000 .000 .000 .078 .000 .023
-.012 .011 .072* -.004 .008 -.057**
.005 .004 .084 .001 .003 .138
.011 .001 .005 .079 .001 .038
-.006 -.015 .087* .013 -.001 -.061*
.74 .23
.001 .005 .090 .007 .000 .120
.061** .014 .003 -.001 -.055** .002
.59 .34
22**
.004 .001
.002 .003
.002 .001
-.013 .016
.006 .008
-.003 .027** .003 .008
.000 .034 .001 .004
.000 .017** -.003 -.018**
.000 .034 .001 .043
.073** -.001 .009 - .066**
.090 .000 .005 .198
.27
.67
.08*
.36**
.082 .007 .000 .000 .108 .000
.78
-.009 .003
.42 .06
.059 .005 .019 .000 .051 .000
.043* .010 .019 -.002 -.031* .002
.66 .29
.074 .008 .010 .001 .076 .000
.042* .005
.057 .001 .018
.018 -.005 -.024 .001
.002 .030 .000 .70 .19
Note: *p<.10, **p<.05, ***p<.01, ****p<.001. "The coefficients in this table are derived from solutions in which the six external contact measures were added to the six independent variables included in Table 2.5. Total R2 is the R2 from the 12-variable solution and increment refers to the difference between the R2 from the 12-variable solution and that from the 6-variable solution in Table 2.5. R2 Change is the contribution to the total R2 in the 12-variable solution that is attributable to each of the six contact measures. (The SPSS routine that was used reports the R 2 contribution for each independent variable when added to the equation in last position.)
52
Pathways in the workplace
12% to nearly 23%. Since this is one of the external organizations with which nonwhite men reported above-average contact, this result is directly contrary to the hypothesis. On the whole, the argument that external organizational ties would translate into internal benefits is not a strong one for this group. The other side of the question, of course, is how these external contacts affected the network standing of the other three race-gender categories. For both white men and white women, consistent although not always pronounced R2 increments - ranging from about 2% to about 11% - were attributable to external contact with private social service agencies. For white women, the effects were significant across all five network dimensions, and for white men on the dimensions of influence, respect, and support. These results were not in evidence at all for either category of non whites. Like nonwhite men, white men appeared to profit, slightly but significantly, from external religious contacts on two dimensions of centrality, influence (R2 increment = 3.4%) and respect (R2 increment = 5.7%). However, they showed decreasing centrality on the same two dimensions if they had frequent external contact with courts (increments of 4 and 5.6%, respectively). Finally, nonwhite women were more central in the networks of influence and support if they had frequent contact with law enforcement agencies (R2 increments of 7.4 and 8.2%, respectively) but were less central if they had frequent contact with courts (R2 increments of 7.6 and 10.8%, respectively). To summarize, there is evidence to support the proposition that some kinds of external contacts translated into internal network advantages, since every practitioner category was able to profit in at least a small way from at least one of the six external organizational contacts. However, for every group there is almost as much evidence in the opposite direction. And, most importantly for the hypothesis with which I began, even though nonwhite males were generally engaged in useful external activities at a higher than average rate and in a way that was qualitatively more favorable than was reported by their co-workers, their payoff for cultivating these ties was no less mixed than for any other group. Effects of community activism: In addition to querying external organizational contacts, the survey asked the respondents to report on four types of activism in the larger communities in which their programs were located. These four areas involved efforts: 1. to elicit the cooperation of other community organizations that were not officially part of the status offender programs; 2. to raise the level of financial support in the community for juvenile programs;
Ascription, achievement, and network centrality 3. 4.
53
to influence community policies with regard to juvenile offenders; to improve the treatment resources available for dealing with the problems of youth.
Compared to the measures of external organizational contact, these kinds of activities are more diffuse and less likely to be directly required by the daily work of most practitioners. Therefore, they represent external liaison efforts at a farther remove from the core of the program itself and, for this reason, may have been more under the control of the individual practitioners. Nonwhite males scored significantly higher than the mean (p<.05) on all four of these items (see Table 2.8B). Again, the hypothesis was that their unusually high level of community activity would translate into a larger increment in centrality than would be the case for the other three race-gender groups. The investigation provided only very limited confirmation of this prediction and one strong and unexpected contrary finding. The data are reported in Table 2.10.10 None of the increments to R2 was significant, even at the .10 level, for white men, white women, or nonwhite women, but for nonwhite men there were quite large and significant effects for all five dimensions of centrality. On the positive side, attempts to increase the level of cooperation of community organizations added significantly to centrality on the dimensions of influence (the increment to R2 was 9.3%) and support (about 15.5%). However, these positive effects were far outweighed by the one consistent and strong negative effect, associated with attempts to bring in additional financial support for the programs. For this variable, the R2 increments, which ranged from 15.9 to 30.9%, were significant for all five dimensions of centrality, but, of course, the direction of the relationships is the opposite of what was predicted. The more nonwhite men were involved in this community activity, the less their network centrality. To summarize this supplementary analysis, the findings on external contacts and community activism further underscore the special situation of nonwhite men among the practitioners in this study, but they do not add up to a consistent explanation of how these respondents found their way into the informal professional networks. On some of the external contact measures the results were stronger for white men, white women, or nonwhite women than for nonwhite men, and for all four groups positive effects were offset by negative effects. The evidence presents a picture that is, at best, mixed. Clearly, nonwhite men needed to be mindful of their external contacts and community activities, but the process by which these resources were turned to internal advantage must have been quite complicated. Why they would persist in external activities that produced negative internal payoffs is a ques-
Table 2.10. Effects offour types of community activism" on centrality for four race-gender categories Nonwhite
White Women
Men R
Work contacts Org. coop.* Financial* Policies6 Resources* Total R2 = Increment = Influence Org. coop.* Financial* Policies* Resources* Total R2 = Increment = Respect Org. coop.* Financial* Policies* Resources* Totalfl2 = Increment = Support Org. coop.* Financial* Policies* Resources* Total/?2 = Increment = Assistance Org. coop.* Financial* Policies* Resources* Total/?2 = Increment =
Women
Men
2
R
2
R
2
R2
b
Change
b
Change
b
Change
b
Change
-.031* .007 .022 .006
.029 .003 .019 .001
.010 .013 -.005 -.006
.004 .007 .001 .002
.019 - .062** .001 .028
.013 .159 .000 .025
.006 .002 -.030 .044
.002 .000 .023 .056
.41 .04
.33 .02
.019 .006 .013 .001
-.022 .009 .017 .005
.23 .03
.011 .001 .015 .000
-.017 .004 .018 -.003
.021 .001 .002 .040
-.029 .005 -.009 .039* .34 .05
.41 .05
.22 .03
.057 .239 .004 .004
.078*** -.101**** .027 -.013
.155 .309 .026 .004
.56 .06
.039 .260 .004 .005
.67
.35***
.001 .004 .018 .008
.005 .013 .029 -.019 .63 .07
.46*** .032 -.078*** .009 .013
.004 .009 .000 .000
.009 .017 -.003 .001
.71
.001 .016 .012 .002
.004 .015 -.011 -.004
.57 .06
33***
.23 .04
.023 .000 .008 .039
-.024 .000 -.013 .030*
.040* - .076*** .009 .012
.001 .006 .002 .000
.005 .015 .009 -.003
.70
.000 .012 .000 .001
.002 .016 -.002 .004
.093 .276 .014 .000
.39***
.26 .04
.38 .02
.052** -.083*** .017 -.001 .69
.003 .014 .022 .000
.008 .017 -.018* -.002
.55 .12
.25*" .000 .020 .003 .000
.003 .020* -.007 -.002
.41 .03
.51
.008 .003 .004 .002
.012 .009 -.012 .007 .55 .05
Note: *p<.\0, **p<.05, ***p<.0\,****p<.00l. "The coefficients in this table are derived from solutions in which the four community activism measures were added to the six independent variables included in Table 2.5. Total R2 is the R2 from the 10-variable solution and increment refers to the difference between the R2 from the 10-variable solution and that from the 6-variable solution in Table 2.5. R 2 Change refers to the contribution to the total R2 in the 10-variable solution that is attributable to each of the four activism measures. (The SPSS routine that was used reports the R2 contribution for each independent variable when added to the equation in last position.) b Org. coop, refers to attempts to elicit the cooperation of community organizations that were not part of the program; financial refers to efforts to increasefinancialsupport from the community; policies refers to efforts to influence community policies regarding juvenile offenders; and resources has to do with efforts to improve treatment facilities for dealing with offenders.
56
Pathways in the workplace
tion that will be partly answered when their ties to the formal authority structure are examined in the next chapter. There it will be shown that isolation from the formal structure was substantially reduced for nonwhite men who were active in fund-raising eflForts in the community. It will be argued that isolation from the informal structure could be traded off against greater access to the formal structure, suggesting that an activity will be continued so long as it has a "net balance of favorable consequences" (to take a Mertonian term out of its intended context) for the individual. In contrast to the ambiguous picture that emerged for external contacts and community activism, considerations involving clients followed a pattern much closer to that predicted for nonwhite men. Client-practitioner homophily did not affect their daily work contacts significantly, but it did change the picture appreciably on the other centrality measures. Its effect was especially pronounced on the dimension of support, where it increased the explained variance by nearly 26%. The suggestion that nonwhite men were relied upon by others more frequently, seen as more influential, and accorded more respect because of their ability to deal with nonwhite male clients received strong indirect support from this pattern of findings. A note on the ideal type: authority, expertise, and influence Before moving to the next stage of the analysis, a special comment - in fact, a digression - is in order concerning the relationship between possessing official authority and exercising day-to-day influence over program activities. How members gain access to power is always a crucial question in organizational analysis, and of course, the connection between authority and influence lies at the theoretical center of the Weberian ideal-typical model. An authority figure without recognized influence cannot be effective in implementing organizational objectives; conversely, exercise of control over organizational activities without the official authority to do so violates the commitment to hierarchical arrangements. As I argued in the Introduction, bureaucratic arrangements are distinguished from other organizational forms primarily by a reliance on technical expertise as the key to establishing a coherent system of control. This is especially characteristic of organizations that place a premium on professionalism as an organizing principle [on this point, see Miller (1970) and Miller and Fry (1973)]. In Weber's model, the mechanism is clear: It is the visible expertise of superordinates that encourages subordinates to legitimate their right to make decisions and issue binding directives. Empirically, it follows that, in organizations that closely approximate the ideal type, influence will be a function of both authority and expertise. Or, to put it the other way but in terms of
Ascription, achievement, and network centrality
57
Table 2.11. Summary of relations among authority, influence, and education Relationship between Authority and education Authority and influence Education and influence
Nonwhite
White Men
Women
Men
Women
Positive Positive Negative
Positive Neutral Neutral
Positive Neutral Negative
Neutral Neutral Positive
more direct interest here, organizations, particularly professional organizations, can be said to approach the ideal type to the extent that influence is coupled to both authority and expertise. Did this configuration of relationships actually obtain in the organizations studied here, and did it vary from one race-gender category to another? The findings that relate to this matter indicate that the ideal-typical gestalt was absent for all four race-gender categories. As the summary of net effects in Table 2.11 shows, at least one of the crucial linkages is missing for each group: In the figure, the relationships between authority and influence and education and influence are derived from the data in Table 2.5. "Positive" indicates a significant net positive effect, "negative" a significant net negative effect, and "neutral" the absence of a significant effect. The relationships between authority and education were assessed by an analysis in which authority (official rank) was regressed on education, experience, occupation, workplace, and age.11 To be sure, there are scattered elements of the classical bureaucratic model here, seen in the education-authority linkages for three of the four groups, the positive association between influence and authority for white men, and the positive relationship between influence and education reported for nonwhite women. Expertise, authority, and influence are therefore not completely disconnected. Nevertheless, the relationships across the four racegender categories do not add up to a coherent bureaucratic system of control because no category fit the ideal-typical configuration completely. It is also possible to approach this issue in a somewhat more complicated way. Ideally, authority and expertise ought to show an interactive, or multiplicative, relationship to influence. Specifically, the Weberian argument stipulates that for those with authority, expertise is vital for the exercise of actual interpersonal influence. Therefore, there should be an especially strong relationship between expertise and influence where authority is high; among
58
Pathways in the workplace
those without authority, the connection of influence to education ought to be attenuated in comparison. To test this proposition, the term "rank times education" was added to the basic regression equations for influence that were used in Table 2.5. Again, none of the results confirmed the ideal-typical prediction.12 For white women, nonwhite women, and nonwhite men, the interaction term did not significantly improve the prediction of influence. For white men, the interaction term did contribute positively and significantly (/?<.001) to the influence equation, indicating that the effect of education on influence was different for those with and without authority. But this result does not offer support for the ideal-typical pattern. Recall that for white men the basic relationship between education and influence was strongly negative (see the summary in Table 2.11 and the appropriate data in Table 2.5). What produced the significant effect of rank times education is a situation in which education had virtually no effect on influence for those with authority but a strong negative effect for those who lacked authority. This was shown directly by performing separate regressions for white men with and without official authority.13 For those with authority (ft = 27), education had no effect on influence (b= .0004; seft = .009); for those without authority (ft = 44), education and influence were inversely related ( £ = - . 0 2 1 ; set =.008; p<.05). From this result, the best that can be said for those white men with authority is that education had no implications for the influence they were able to exercise; on the other hand, for those without authority, it continued to function as a handicap. Taking the analysis one step further, the relationships between education and the other four dimensions of centrality - work contacts, respect, support, and assistance - were also examined separately for those with and without authority. (These additional data are not shown.) The results paralleled those for influence very closely: Among those with authority, education had little effect on any dimension of centrality; among those without authority, education was inversely and significantly related to all but the assistance dimension of centrality. In short, highly educated white males without authority were isolated from the informal exchange networks with remarkable consistency. Apparently, education activated a social distance mechanism that deserves to be investigated more thoroughly than I have been able to do here. I will have more to say about how this distancing mechanism may have operated in the concluding discussion in Chapter 4. In interpreting these results, it is important to keep in mind one final complicating element. Although the authority-influence configuration for white men was not dependent upon expertise in the way anticipated by the ideal-
Ascription,
achievement,
and network centrality
59
typical model, those with authority did not lack the approval of the others around them. After all, the data in Table 2.5 show that authority, net of education, was an excellent predictor not just of influence but also of increased work contacts and increased standing on the dimensions of professional respect, support, and assistance. In other words, although white men with authority enjoyed an influence advantage that was unavailable for all the other race-gender categories, it would not appear that this caused pronounced disapproval from those around them. Summary The data from the first stage of this analysis suggested that no group monopolized the informal networks and that race and gender accounted for very little of the variance in network centrality. The second stage revealed important differences between men and women, whites and nonwhites, and suggested alternative pathways to organizational advantages and alternative scenarios to account for the results that were reported. The third and most crucial stage showed that, with the exception of workplace, the effects of virtually all the variables in the analysis were specified jointly by race and gender. Rank is perhaps the most crucial example. Its positive effects were largely confined to whites, and within this racial category it was clearly differentiated by gender. Similarly, the effects of education ranged from positive (for nonwhite women) to negligible (for white women) to strongly negative (for white and nonwhite men). Looking at the data from a different angle, it should be stressed that every race-gender category had an almost exclusive claim to at least one factor among those examined. White males were distinguished from all others by the consistently reliable informal payoff for official supervisory rank and by stronger effects associated with having a client-contact occupation; white women were set apart by the modest but persistent advantage/disadvantage of youth/age; nonwhite men alone profited from client-practitioner homophily; and nonwhite women were the only ones who experienced a consistently high return on investment in education. We will discuss what each of these factors means in more detail in the final chapter. The data also suggest an interesting hierarchical ordering among the racegender categories. When attention is confined to the internal variables (that is, those defining individual investments or official placement in the structure of the programs), it is evident that white men enjoyed a clear precedence over the other three categories. Their experiences most closely approximated those of a classical bureaucratic career pattern. Although education did not
60
Pathways in the workplace
work at all in their favor, rank clearly did, and so did having a client-oriented (that is, high client-contact) occupation. Whether white women or nonwhite women ought to be placed second is more difficult to determine. White women actually profited more than white men from their physical placement (workplace) in the program, but the effects of rank were very much attenuated for them, age was a handicap, and no advantages of occupation were in evidence. On the other hand, nonwhite women were even less able than white women to profit from rank or workplace, but superior education and recent occupational entry were strong points for them. Thinking just in terms of the human capital argument (which stresses return on investment), it would have to be concluded that nonwhite women had the edge over white women. Finally, the paucity of strong positive effects other than workplace for nonwhite men suggested that they were caught up in a set of activities and evaluations separate from those of all of the other participants. This collection of results demonstrates quite well the error that can be made in assuming that the bureaucratic imperative evens out the reward and resource allocation process in a way that assures that all participants will confront essentially the same organizational reality. I will return to a discussion of this diverse set of findings in the concluding chapter. In the next chapter, however, I shift attention away from network centrality and focus instead on practitioners' relationships to the structure of authority in their programs.
3. Access to the formal authority structure
The discussion in the preceding chapter started with the observation that the four race-gender categories had approximately equal access to the informal networks but quite distinct combinations of advantages and disadvantages, both internal and external, that helped explain their ability to gain that access. On the basis of this conclusion, it was possible to argue that four distinct pathways to network integration existed, in contrast to the single allocation rule stressed by rationalist theories. In this chapter, the focus is on relationships to the formal authority structure, as measured by the ability of different groups to participate in the decision-making process, the frequency of their contact with their supervisors, and their evaluations of the usefulness of that contact. The analysis here parallels the three stages employed in the analysis of network centrality. A simple regression analysis is followed by separate analyses for race, gender, and race-gender subgroups; later, the measures of client-practitioner homophily and external ties will also be brought in to clarify the findings. The results are complicated, but it will quickly become apparent that the data follow a pattern quite different from the one that characterized access to informal network resources. As a preview, I shall cite four major substantive differences. First, compared to the analysis of informal centrality, much of the variance in access to authority is left unexplained, both in the overall regression analysis and in the various subgroup analyses. The addition of client-practitioner homophily, external contact, and community activism improved some of the predictions considerably, but until those variables are brought in, the results here are less decisively drawn than were those for informal centrality. Second, favorable relationships with authority, unlike informal network access, were not equal for the different categories of participants. To the contrary, nonwhite men and (closely following them) nonwhite women appeared to be in the most favorable positions, with white men clearly occupying the least favorable configuration. Consistent with this finding, the overall regression analysis showed race to be the strongest predictor of how 61
62
Pathways in the workplace
the participants fared in this area of formal organizational activity. In the subgroup regressions, age also emerged as a crucial factor, but primarily for nonwhite women. In comparison to these race and age effects, the variables representing individuals' professional investments - education, occupation, and experience - were of very little use as explanatory factors. Participants' ascribed identities, to put it succinctly, had a stronger impact on their formal experiences than their bureaucratic capital did. The third and most important difference is that the specific combinations of variables that came into play for the subgroups in determining their access to the formal structure do not resemble at all the patterns reported for informal network access. It is what these contrasts imply for the linkage between formal and informal areas of organizational activity that constitutes the most significant result to emerge from this part of the investigation. In several instances, factors that had represented informal network disadvantages for a subgroup either disappeared or appeared here as formal advantages for that same category. For example, it will be recalled that the financial ties that nonwhite men developed with the external community worked against them in their relations to the informal structure. Thefindingshere show that these same ties represented a strong positive factor for them where access to their supervisors was concerned. This suggests that the peers and superiors of the members placed very different evaluations on their extracurricular activities. As a result, many practitioners were faced with a situation in which retreat in one dimension of work (access to informal exchanges) could be, or had to be, traded off against progress in the other sphere of activity (access to the formal structure). Finally, in another kind of formal-informal contrast, variables that were shown to work in favor of one race-gender group in the area of informal relations became advantages for another group in gaining access to the formal structure. For example, the beneficial effect of client-practitioner matching that was so important for nonwhite men on the informal dimension was apparent only for white men on the formal dimension. Similarly, the advantage of youth (or the disadvantage of age) that was apparent for white women on the informal dimension was strongly in evidence for nonwhite women in the formal dimension of activity. Stage 1: simple comparisons and overall regression analyses The racial differences in access to authority are first apparent in Table 3.1 A. On the average, non whites reported more favorable relations to their supervisors on all of the measures, and women were somewhat more favorably
Table 3.1. Means and standard deviations for the variables used in the analysis of access to authority B. Breakdowns by race- gender
A. Breakdowns by race and gender
White
Participation* Discussion* Involvement* Interaction with supervisors Evaluation of supervisors
All (W = 202)*
Men (Af = 7())
Mean
Mean
s.d.
s.d.
Mean
s.d.
Mean
Men
Nonwhites (N = 47;)
Whites (AT =155)
Women (N=\26)
s. d.
Mean
Nonwhite
(N = 5\)
s. d.
Mean
s.d.
Woment (N=\04)
Men (N = 25)
Mean
Mean
s.d.
Women (N = 22) s .d.
Mean
s.d.
.62 .72 .79
.49 .45 .41
.57 .69 .75
.50 .47 .43
.66 .73 .81
.48 .44 .39
.59 .68 .77
.49 .47 .42
.74 .81 .87
44 40 34
.47 .61 .69
.50 .49 .47
.64 .72 .81
.48 .45 .40
.76 .84 .88
.44 .37 .33
.73 .77 .86
.46 .43 .35
3.68
1.86
3.54
1.83
3.76
1.87
3.42
1.85
4.54
1. 58
3.06
1.53
3.60
1.71
4.48
1 .47
4.61
1.46
7.05
2.15
6.97
1.93
7.10
2.28
6.89
2 .20
7.55
1. 97
6.59
1.70
7.04
2.11
7.76
1 .55
7.32
1.99
°The number of cases in the analyses in this chapter (202 compared to 261 in Chapter 2) is decreased by the number of supervisors who themselves had no superordinates and therefore did not report on their own access to authority. *These are dummy variables. Participation is scored 1 if the respondent indicated both participation in decisions and much discussion. Discussion is scored 1 if the respondent indicated much discussion, whatever the precise division of decision making. Involvement is scored 1 if the respondent indicated participation in decisions, whether or not much discussion was involved.
64
Pathways in the workplace
situated than men. The apparent gender difference, however, is shown by Table 3. IB to be limited largely to whites, or more specifically, to the advantages of white women over white men; nonwhite women and men were quite similar to each other in their scores, although the men were slightly better off on four of the five measures. If these racial differences had been confined to the measure of interaction with supervisors, an argument might have been made that non whites were simply under closer surveillance than whites. But the fact that the result was repeated for the three measures of "participation," "discussion," and "involvement," and especially, for the measure called "evaluation," favors a more positive interpretation, that is, that non whites had an advantage over whites in their relationship to formal organizational activity1 The comparison between the two categories of men, who represent the extremes, is particularly striking. The location of white men in the least favorable position on all of the measures indicates that they were less able than other categories of participants to structure their relations to supervisors to their own advantage. Yet, recall from the previous chapter that white men who themselves had formal decision-making responsibilities were distinguished precisely by their ability to translate this activity into favorable positions in the informal structure; authority was strongly correlated with network centrality. In other words, although white males who were supervisors realized disproportionate informal advantages, it is simply not the case that white males who were subordinates realized disproportionate advantages in their relationships to the formal authority structure. The symmetry across the formal-informal distinction is broken, and even though white men occupied slightly more than their share of supervisory positions in the programs (refer back to Table 2. IB), it is not justifiable to conclude that the authority structure was "captured" by white male vested interests.2 Non white men represent essentially the opposite situation. The data on access to the informal networks indicated that, unlike white men, they were unable to convert a position of supervisory responsibility into informal advantages on any of the dimensions of centrality, including, most crucially, informal influence. But they are shown here to have experienced, as subordinates, very favorable relationships with those in authority. Again the symmetry is broken, and again it is impossible to conclude that one race-gender group's interests consistently held sway over those of the others in the programs. The impression of which group is favored changes when attention shifts from the informal to the formal arena. For this reason, the arguments and interpretations that helped to make sense of the informal structure will obviously not be directly transferable to the formal structure.
Access to the formal authority structure
65
Table 3.2. Regression of access to authority on seven indicators of ascription and achievement Measures of access to the formal authority structure Participation in
Independent variables Race Gender Experience Education Age
Workplace Job type R2 (R2)
Intercept
decision making seb
b
.084 .071 .008 .015 .005 .078 .070 .06 (.03)
-.189* -.104 -.006 .012 -.006 .072 -.036
83
Frequency of interaction with supervisor b
seb
-1.177** .302 -.274 .258 -.013 .028 -.034 .056 .016 -.029 .663* .280 .172 .252 .13 (. 10) 5.99
Evaluation of time spent with supervisor b
sefr
-.703 .373 -.168 .319 -.045 .035 .026 .069 -.022 .020 .346 .169 .004 .312 .05 (. 02) 8.09
Note: *p<.05, **/?<.001; no values significant at .01.
Attention turns now to a consideration of the regression analysis that appears in Table 3.2. This table reports the results of the regression of "participation," "interaction," and "evaluation" on the list of ascribed and achieved attributes that includes gender, race, education, age, experience, occupation, and workplace. (The two dummy variables called "discussion" and "involvement" that appeared in Table 3.1 had marginal distributions that were too skewed for regression purposes, and they will not appear in the remainder of the investigation. Note also that formal authority, measured by the possession of supervisory rank, is dropped from this part of the analysis because access to those in authority is the dependent variable.) The race effect that was noted in Table 3.1 is again in evidence here when the other variables in the equation are controlled. Although all of the R2 values are small, being nonwhite is significantly related to two of the three measures of supervisor-subordinate relations. The strongest effect is apparent for the measure of interaction with supervisors. Here the result is significant beyond the .001 level and the regression coefficient is nearly twice as large as that for the second strongest effect, which was recorded for work-
66
Pathways in the workplace
place. For participation, the effect of nonwhite identity is the only statistically significant factor among the seven examined; and for evaluation the effect of race is the strongest of those reported (/?<.1O), although none of the results in this part of the table is significant at conventional probability levels. The results for gender in this table consistently favor women, but none of the coefficients is statistically significant.3 This set of outcomes leaves a lot of variance unexplained, but it does show that relationships to the formal structure were influenced more by race than by the rationalistic factors represented by education, experience, workplace, and occupation. In fact, if no allowance is made for the special technical advantages that attach to being nonwhite in social service delivery systems (see the discussion of this argument in the previous chapter), then this finding would work strongly against the idea of universalism in the allocation of professional advantages and disadvantages. Part of the explanation may lie in the participants' generally high mean levels of education and experience (refer back to Table 2.1), which may have attenuated the effects of these two variables, leaving more latitude for the operation of ascribed factors. However, even if this factor did influence the results, it still does not explain why the findings violate the conventional wisdom that racial differentiation will favor whites. Stage 2: separate analyses for men and women, whites and nonwhites Table 3.3 presents the analyses conducted to determine whether avenues of access to the formal structure differed by gender and race. The significance tests for the overall ascription-achievement interaction effects are given in Table 3.4. The logic of the analysis here is the same as that reported in Table 2.3. The objective is to determine whether the achievement measures on the whole operated differently for men and women or for whites and nonwhites. The results are mixed. Comparisons of men and women In Table 3.3A and B, what stands out is the absence of statistically significant effects for either men or women. It is true that the R2 values for "interaction" and "evaluation" were significantly stronger for men than for women (p<.05 in both cases), but the dominant impression for both categories replicates what was shown in the overall regression in Table 3.2: Relationships to the formal structure were not strongly differentiated by training, occupation, workplace, experience, or age.
Table 3.3. Regression of access to authority on experience, education, age, workplace, and occupation, performed separately for men, women, whites, and nonwhites Measures of access to the formal authority structure
decision making
Frequency of interaction with supervisor
Evaluation of time spent with supervisor
b
b
b
Participation in
A. Men Experience Education Age
Workplace Job type R2 (R2)
Intercept B. Women Experience Education Age
Workplace Job type R2 (R2)
Intercept C. Whites Experience Education Age
Workplace Job type R2 (R2)
Intercept D. Nonwhites Experience Education Age
Workplace Job type R2 (R2)
Intercept
sefr
seb
sefr
.014 .024 .011 .134 .122 .06 (-.003) 1.105
- .033 .047 - .075 .083 - .056 .038 .672 .459 .376 .419 .16 (.11) 6.410
-.062 .060 -.079 -.292 -.505
-.008 .010 .016 .020 -.005 .005 .143 .097 .080 .088 .05 (.01) .502
.015 .040 - .042 .079 -.023 .019 .497 .380 .117 .344 .03 (-.01) 4.881
-.017 .049 .025 .097 -.007 .024 .364 .467 .189 .423 .01( -.03)
-.020 * .009 .025 .019 .003 .005 .056 .088 -.087 .082 .05 (.02) .204
-.015 .033 - .007 .072 - .023 .020 .710* .326 .109 .301 .05 (.02) 4.096
-.060 .056 -.008 .197 -.058
.025 .015 -.019 .020 -.030 ** .008 .132 .152 -.032 .124 .30 (.21) 1.834
-.085 .057 - .099 .074 -.041 .029 .827 .565 .539 .460 .24 (.15) 7.330
-.050 .077 .007 .101 .060 .039 .770 -.229 .005 .627 .10( -.01)
-.005 -.002 -.011 -.049 -.194
Note: */?<.05, **p<.001; no values significant at .01.
.049 .087 .040 .477 .435 .19 (.13) 9. 206
6. 797
.039 .086 .024 .390 .361 .03 (.00) 6. 540
9. 849
68
Pathways in the workplace
Table 3.4. F-Tests for race and gender interactionsa
Hypothesis^
df
Participation in decision making
1. 2. 3.
5,243 5,243
1.53 3.00*
0.59 0.81
2.83* 1.42
5,232
0.46
1.09
1.20
Pw^Pnw Pwm^Pwf Pnwm ^ Pnwf
Contact with supervisors
Evaluation of supervisors
Note: */?<.01. a Cell entries are F-statistics. b $m is the population vector of five slope parameters for males. pf is the analogous vector for females. (3W and Pnw are the same vectors for whites and nonwhites. |3wm, Pwf, pnwm, and pnwf are the vectors of five slope parameters for equations within racegender subgroups. The tests for interaction effects in Table 3.4 indicate that the regression solution for men was significantly difiFerent from that for women only where the evaluation of contact with supervisors was concerned. Even here the detailed results in Table 3.3 were far from compelling. The largest male-female differences involved workplace and occupation, each of which showed positive (but nonsignificant) effects for women (Table 3.3B) and negative (also nonsignificant) effects for men (Table 3.3A). The F-ratios in Table 3.4 for the other two measures of access to authority, "participation" and "interaction," did not attain significance. Note, however, that the effects of workplace and job type on "participation" again differed significantly between men and women (p<.05 in each case). Comparisons of whites and nonwhites When separate analyses were conducted for whites and nonwhites (Table 3.3C and D), only three results were statistically significant. Among whites, participation in decision making was inversely related to professional experience (the coefficient is significantly different from that for nonwhites p<.05) and interaction with supervisors was positively related to placement in the programs' administrative centers (here the coefficients for whites and nonwhites are not significantly different). Eclipsing both of these findings, however, is the fact that for nonwhites there was a strong negative effect of age on "participation." Here the advantage of being young (or the handicap
Access to the formal authority structure
69
of age) by itself explained 27.4% of the variance (/?<.0001), and the difference between the white and nonwhite coefficients is significant beyond the .01 level. It was largely because of this pronounced contrast that the overall regression solution for nonwhites was significantly different from that for whites (see Table 3.4). Neither of the two remaining F-ratios, which compare the white and nonwhite solutions for interaction, was significant, although the R2 for nonwhites in the case of "interaction" was considerably larger than that for whites. To recapitulate the more important results presented up to this point, nonwhites unexpectedly had more liberal access to those in authority and evaluated that access on the average quite favorably. Among the objective investments that were thought to have influenced the participants' access to the formal structure, only experience (or, more correctly, recent training) was at all important; its effect was apparent only for whites and was confined to participation in decision making. On the other hand, age emerged as a powerful predictor of participation, but for nonwhites only. The effect of this variable can be seen even more clearly in the finer breakdowns that follow. Stage 3: the simultaneous effects of race and gender Assessment of the basic model for the four race-gender categories In this section, access to authority is examined separately for each of the four race-gender combinations. The strong differences among the subgroups that the comparable analysis of informal centrality produced in Chapter 2 are not repeated here. None of the relevant statistical tests comparing the overall regression solutions for the four race-gender subgroups is statistically significant (hypothesis 3 in Table 3.4). Moreover, it is evident from Table 3.5 that, for all four race-gender categories, favorable access to the formal authority structure was for the most part not based on considerations emphasized in rationalist human capital or investment/payoff arguments. In fact, the largest unadjusted R2 value, .55, is largely attributable to the handicap that age represented for nonwhite women. Examining the details in Table 3.5 first for white men and white women, only one regression coefficient for each subgroup is significantly different from zero. In each case, it involves participation in decision making. Among white men, those in occupations with infrequent client contact (the variable "job type") had an advantage over those in occupations that required high client contact. Recall that on the informal sociometric dimensions (Table 2.5)
Table 3.5. Regression of access to authority on experience, education, age, workplace, and occupation, performed separately for white men, white women, nonwhite men, and nonwhite women Measures of access to the formal authority structure Participation decisiori making
Evaluation of time spent with supervisor
b
b
in
A. White men Experience Education Age
Workplace Job type R2 (R2)
Intercept
sefr
-.015 .015 .051 .035 -.003 .013 -.196 .159 -.304** .148 .15 (.05) .024
B. White womeni Experience -.024** .012 Education .026 .024 Age .006 .006 Workplace .105 .169 Job type .035 .098 R2 (R2) .07 (.03) Intercept .096
-.054 .084 -.065 -.293 -.260
se&
.058 .132 .048 .607 .562 16 (.07) 7.834
Frequency of interaction with supervisor seb
b
.052 .117 .042 .539 .500 .]17 (.09) 2.093
-.034 .142 -.048 .524 .313
.057 -.034 .115 .090 .028 .010 .505 .342 .474 -.052 .01 (-.03) 5.376
.046 .027 -.034 .093 .023 -.013 .648 .409 -.028 .384 .03 (-.01) 4.237
.038 .033 .022 .227 .217 .12 (-.12) 11.649
.125 -.205 .107 .180 * .072 -.104 -.230 .743 .709 -.507 33 (.15) 9.196
-.093 -.170 -.012 .917 .907
D. Nonwhite women Experience .017 .026 Education -.011 .035 Age -.033*** .009 .131 .281 Workplace Job type .015 .173 R2 (R2) .55 (.41) ]i.822 Intercept
-.004 .116 -.185 -.167 .062 -.065 -.009 1.961 .795 1.209 .14 (-.14) 11.529
-.084 .009 -.047 1.000 .184
C. Nonwhite men Experience .028 Education -.033 Age -.016 Workplace .179 Job type -.085 R2 (R2)
Intercept
Note: *p<.10, *^*/?<.05, ***/?<. 01.
.122 .104 .070 .722 .689 .:30 (.11) 7.201 .073 .014 .039 1.228 .757 .:30 (.09) 6.286
Access to the formal authority structure
71
this same variable had just the opposite effect; the advantage in informal centrality went to those with high client contact. The difference here between the coefficients for white men and white women approached statistical significance (/?<. 10), but the differences between white men and the two nonwhite categories did not. For white women, some advantage in participation in decision making came with recent professional training. Although statistically significant, this effect was not large, and it distinguished white women significantly only from nonwhite men (/?<.05). Turning to nonwhite men, the variance in "participation" explained was small (the unadjusted R2 was .12) and none of the individual effects of the different variables was statistically significant. The explained variance was substantially larger for the measures of "interaction" (R2= .30) and "evaluation" (R2= .33). Although none of the separate results here was significantly different from zero, the education slopes (negative for "interaction" and positive for "evaluation") approached statistical significance [each coefficient was more than 1.5 times its standard error, a defensible criterion for significance, according to Heise (1969) and Lincoln and Zeitz (1980)]. It is also of interest that these two education coefficients were reversed in sign and were significantly different from those for nonwhite women (in each case, p<.05). Looking finally at the details for nonwhite women, these respondents were clearly set off from the other three subgroups by the very strong negative effect of age on their ability to participate in decision making. Evidence of such an age effect first appeared in Table 3.3 for the larger aggregate of nonwhites; it can now be seen as largely confined to nonwhite women. By itself, this variable accounted for almost 50% of the variance in "participation," a result that differs significantly (at the level of .05 or beyond) from the corresponding age effects for each of the other three race-gender categories. If the cutoff for what constitutes a significant effect is dropped to 1.5 times the standard error, one additional variable, "experience," had an effect for nonwhite women. I mention this because the direction of the effect - the greater the chronological distance from professional training, the greater the participation in decision making - suggests that it was youth, per se, and not youth as a correlate of recent training that worked to their advantage. Stated in a way that emphasizes the negative side, this finding for age is simple: Older nonwhite women were disproportionately likely to be excluded from participation in the decision-making process and none of their resources, investments, or credentials served reliably to counteract this effect. Stated positively, younger nonwhite women were virtually guaranteed participation, even when the effects of their resources and qualifications were controlled.4 It is especially interesting that the advantage of youth that these participants exhibited with respect to access to the formal structure was in
72
Pathways in the workplace
evidence only for white women in relation to the informal structure (see the relevant data in Table 2.5.) I will have more to say about this in the concluding chapter. To recapitulate this part of the analysis, the similarity of the findings from one race-gender category to another is very slight. White men whose occupations did not involve high client contact had greater access to the decisionmaking process. Among white women, only recent training made a significant difference. For nonwhite men, education had effects that approached statistical significance on the two variables called "interaction" and "evaluation," but the former coefficient was positive and the latter negative. Finally, nonwhite women who were young experienced increased participation in decision making, whereas those who were older were largely excluded. This age effect is the strongest single finding in this part of the analysis. When it is added to the race effects reported earlier in the chapter, it adds to the impression that access to the formal structure was shaped in important ways by ascriptive factors but was comparatively less responsive to considerations involving occupation, credentials, or work location. Expanding the model: variables reflecting external activities As in the previous chapter, the analysis here was expanded to include the level of client-practitioner homophily (a single indicator), the frequency of practitioners' contact with agencies in the community surrounding the programs (six measures), and the level of their community activism (four measures). The results for this part of the analysis appear in Tables 3.6, 3.7, and 3.8. Again, I must stress the tentative nature of these explorations. Client-practitioner homophily: The presence of same-race and -gender clients (Table 3.6) had little impact on interaction with supervisors or evaluation of that interaction for any of the participants. For white women, nonwhite women, and nonwhite men, essentially the same result extended to the third dependent variable, "participation in decision making." For white men, however, a positive effect was revealed for participation that added 23% to the variance explained for that variable. Since client-practitioner matching had so little effect on the participation of the other three groups, it is remarkable that it was the single most important factor for white men, outweighing all of the other variables in the regression combined. This result takes on added interest when it is placed alongside certain data from Chapter 2. Recall from that part of the study that client-practitioner matching (dealing with white male clients) had virtually no effect on the
Table 3.6. Effects of client-practitioner homophily on the access by participants to the formal authority structure Unstandardized regression coefficients0 and increments6 to explained variance for proportion samerace-gender clients Nonwhite
White
Dependent variable Participation in decision making R2 Interaction with supervisors R2 Evaluation of interaction with supervisors R2
b .887* .276 1.492
Men
Women
Men
38 18 22
R2 Change
R2 Change
b
.231
-.030
.012
-.134
.064
.260
.07 .04 .01
Women R2 Change
b
.003
.304
.004
-.724
.000
-.014
.12 .32 .35
R2 Change
b
.000
.077
.018
.337
.019
-.105
.61 .32 .16
.058 .018 .019
Note: *p<.0001. a The subgroup regressions from which these values are taken include six regressors: client-practitioner homophily plus the five independent variables (experience, education, age, job type, and workplace) that appear in Table 3.5. b 2 R here refers to the total explained variance in access to authority in the six-variable solution and R 2 Change refers to the net contribution of homophily to this total.
74
Pathways in the workplace
informal network placement of white males (See Table 2.7) but that having a job with high client contact was consistently advantageous (See Table 2.5). Here, paradoxically, an essentially reversed configuration is apparent. Matching (the presence of white male clients) was associated with greater sharing in the formal decision-making process, but simply having a high client-contact occupation, which meant working with large numbers of clients in general, was associated with less participation in decision making (see Table 3.5). In the latter case, note that a factor that had a strong positive payoff in informal integration at the same time served to isolate the practitioner from the formal decision-making process. For nonwhite men, an equally important but very different contrast can be seen: For them, dealing with non white male clients was shown in Chapter 2 to be very important for informal centrality (the data are in Table 2.7) but was apparently of very limited importance for participation in formal decision making (Table 3.6). Thus, among male practitioners, the effects of client contact and client demographics vary between the formal and informal arenas and change with the race of the practitioner. These contrasts among the male respondents are important for a number of reasons. The variation across the formal-informal distinction is an indication that these two areas of instrumental activity followed divergent rather than parallel lines of determination. The variation by race, on the other hand, reinforces what I said in Chapter 2 about differences in the personal work strategies available for white and nonwhite men. Client-practitioner homophily, which produced informal integration for non whites, led to markedly better formal relations for whites and the reverse for non whites. However, there is one sense in which the two male groups are similar: A factor beyond their direct control, namely, the ascribed characteristics of clients, figured prominently in the organizational experiences of both, albeit in contrasting ways. Contact with organizations outside the program. The key to this part of the analysis again lies in the contrasts between the formal and informal spheres of activity. Judging from the increments to explained variance (see Table 3.7), nonwhite men were the most affected, sometimes adversely, by their contacts with organizations outside their programs. Their contacts with law enforcement agencies were the most important. This activity had no appreciable effect on their informal centrality (see the discussion in Chapter 2), but it detracted significantly from their level of participation in decision making and from their evaluations of time spent with supervisors; the contributions to R2 for "participation" and "evaluation" were 22.6 and 16.9%, re-
Access to the formal authority structure
75
spectively. (For "participation" the coefficients for nonwhite males were significantly larger than those for white women and nonwhite women; for "evaluation," only the comparison between nonwhite men and white women was significant.) The effect of law enforcement ties on the frequency of interaction with supervisors, also negative, closely approached significance (/?<. 10; R2 increment = 12.2%), but here none of the comparisons between subgroups was significant. On a positive note for nonwhite men, evaluations of supervisors were significantly higher (p<.05) for those with frequent external contact with private agencies; however, the between-subgroup comparisons for this variable were not significant. Turning to nonwhite women, relations with the authority structure were not substantially affected by external ties. None of the results for them reached even the .10 level of significance. Most importantly, the positive payoff that contacts with law enforcement agencies had for them in terms of informal centrality (see Table 2.9) is not at all in evidence in these findings for access to authority. Finally, only two significant effects were apparent for white men and white women in Table 3.7. For white men, interaction with supervisors was increased appreciably by frequent external contact with court personnel. Among white women no single external tie was strongly indicated, but adding the six external contact variables together to the regression equation significantly increased the R2 for participation in decision making (R2 increment = 12%; p<.05). Again, certain comparisons with the results for informal centrality are important. For both categories of whites, the positive effect of external contact with private agencies (refer back to Table 2.9) has disappeared and for white men the positive effect of contact with religious groups is no longer apparent. In addition, for white men the effect of external contact with courts has changed; the positive effect it had upon two aspects of informal centrality has given way to a very strong negative effect on interaction with supervisors. Strategically, then, external actions taken by a white male to increase his informal integration could have produced estrangement from the formal structure, and steps taken to improve relations with supervisors could lead to isolation from colleagues. Community activism: Of the three sets of external variables added to the analysis, community activism proved on the whole to be the least useful for understanding the patterns of access to the authority structure. The increases in R2 that resulted from adding these four variables to the regression equations were generally not large, and only two of the specific results were
Table 3.7. Effects of six external contact measures" on access to authority for four race-gender categories Nonwhite
White Men
Men
Women R2 Change
R2 Change Participation in decisions Law enf. orgs. - .098 Schools - . 109 Relig. orgs. .158 Private agencies .000 Courts .048 Public agencies .065 Total R2 = .28 Increment = .13 Contact with supervisors Law enf. orgs. Schools Relig. orgs. Private agencies Courts Public agencies Total R2 = Increment = Evaluation of supervisors Law enf. orgs. Schools Relig. orgs. Private agencies Courts Public agencies Totalfl2 = Increment =
.033 .043 .049 .000 .015 .018
Women R2 Change
R2 Change
.043 .010 .062* .024 .087* .033 .030 .006 .010 .001 .035 .008 .19 12**
.208** .226 .151 .101 .138 .047 .074 .046 .002 .000 .139* .125 .47 .36**
.013 .021 .082 .032 .006 .006
- .344 .033 -.308 .028 .045 .000 .463* .053 .576*** .181 .103 .004 .43 .25**
.197 .255* .093 .160 .066 .027
.545* .467 .514 .140 .420* .419* .64 .34*
.137 .303 .215 .191 .104 .063
— .309 - .389 -.037 . 140 .112 .059
.056 .221 .394* .277 .181 .105
.23 .07
.022 .036 .000 .004 .005 .001
.09 .06
.014 .027 .002 .010 .002 .000
.001 .013 .030 .020 .002 .10 .08
.122 .075 .051 .013 .118 .089
.676** .169 .250 .020 .589 .060 .426** .106 .025 .000 .185 .016 .73 .41**
.001 .003 .043 .009 .000 .000 .64 .09 .008 .053 .023 .026 .005 .002 .51 .21
.171 .351 .362 .172 .242 .315
.006 .036 .033 .011 .015 .026 .30 .17
Note: *p
78
Pathways in the workplace
significant at the .05 level (Table 3.8). However, in both of these cases, the contrasts with what was found for the informal dimension are important. One of these significant results involved white women. Participation in decision making was enhanced for those of them who reported frequent attempts to engage the cooperation of community organizations (/?<.05), but the same variable had no appreciable effect on their informal centrality (see Table 2.10). The second strong finding involved nonwhite men. Contact with supervisors was increased for those who reported frequent external financial efforts in the community. This pattern is of particular interest because this same external activity, which involved work undertaken to improve the financial viability of the programs, was a persistent handicap for non white men in the informal networks. It is tempting to see this as an indication that peers took a dim view of certain kinds of external activities, in particular those that were most likely to be appreciated by formal supervisors. Involvement in such community efforts was frequently up to the discretion of the individual practitioner. It is possible, therefore, that non white men were willing to pay a price (or had to pay a price) in the form of informal isolation for the superior access to authority that they gained as a result of their external financial activities. Such trade-offs, in which a positive effect on one side of the formal-informal distinction actually becomes a negative effect on the other, are not in evidence in Table 3.8 for the other race-gender categories. Summary In general, access to the formal authority structure was not distributed in the same way as access to the informal networks. In the first place, the baseline patterns were quite different. Men differed from women and whites from nonwhites, but it was the latter, racial difference, that was most compelling.5 Surprisingly, nonwhite men and women were more firmly and more favorably tied to the authority structure than whites were. The contrast found here between nonwhite and white men is especially remarkable. It represents a direct reversal of the relative abilities of these two groups, reported in the previous chapter, to use authority to their advantage in gaining access to the informal networks. In the detailed regression analyses designed to detect subgroup variations, there were important differences from one race, gender, and race-gender category to another, but the results were neither as striking nor as consistent as in the case of network centrality. Most of the significant results were confined to participation in decision making, but no obvious pattern stands out
Access to the formal authority structure
79
to reveal the underlying logic that might account for how advantages and disadvantages were distributed among the four race-gender categories. For nonwhite males, it is important to point out that statistically significant results appeared on all three measures of superordinate-subordinate relations, but all of these significant coefficients, both positive and negative, represented activities undertaken outside the programs. In this very general sense - that is, in terms of vulnerability to the effects of external, not internal, activities - the results here resemble those for these same men on the informal centrality measures, although, crucially, the specific details of the two sets of data are as often as not reversed. For the other three race-gender categories, the collection of significant results is quite variable and there is no commonality of findings from one category to another. It is important to note that where participation in decision making is concerned, the two strongest effects both involved ascribed considerations. To be specific, client-practitioner homophily was a crucial advantage for white men, and age was a strong liability for non white women. The client effect for white men is of special interest because, as was pointed out in the previous chapter, on the informal centrality measures this variable made a difference only for non white men. This reinforces the impression of contrasting career strategies that separate these two categories. The other powerful effect, involving the age of non white women, was so strong as to suggest a combination of age-upon-race-upon-gender effects of the sort that Dahrendorf (1959:213-18) has called "superimposition" and that he described as an important element in the generation of overt social conflict. It is perhaps crucial, however, that this superimposition for non white women was apparent only for participation in formal decision making. The same pattern of disadvantages did not generalize across the formal-informal distinction, and this may have mitigated whatever its harsher effects might have been. From the viewpoint of organizational theory, the importance of the contrasts between this analysis and that in the previous chapter lies in their implications for the degree of formal-informal coupling in the programs. From the viewpoint of individual participants, the contrasts hint at a far more intricate web of relationships to be dealt with than most organizational theories indicate. Techniques for negotiating the informal networks appear to have been complicated enough without adding a separate and sometimes contradictory set of contingencies to be dealt with in relations to the formal structure. It is likely that some hard choices had to be made by individuals, since no category of participants was able to use a single strategy to maximize access to both the formal and informal structures.
Table 3.8. Effects of four community activism measures" on access to authority for four race-gender categories Nonwhite
White Men
Participation in decisions Org. coop.* Financial* Policies* Resources* Total R2 = Increment = Contact with supervisors Org. coop. Financial Policies Resources Total R2 = Increment = Evaluation of supervisors Org. coop. Financial Policies Resources Total R2 = Increment =
Men
Women
Women 2
2
2
R Change
b
R Change
b
R2 Change
.063 .011 .081 .012
.059 -.038 .074 -.044
.026 .005 .019 .009
b
R Change
.009 .043 .010 -.019
.000 .009 .000 .001
.116** .054 -.036 .005 - .079* .031 .018 .002 .16 .08*
-.096 .037 .107 .048
.005 .001 .025 .012
.360* -.041 -.209 -.243 .10 .07
.035 .000 .014 .019
-.450* .532** -.021 .153 .51 .21
.108 .175 .000 .010
-.167 .796* -.324 -.142 .48 .18
.016 .175 .028 .007
.027 .050 .004 .031
.404* .035 -.030 -.207 .05 .04
.029 .000 .000 .009
-.334 .157 .191 .008
.054 .014 .018 .000
-.209 .673 -.301 .213
.013 .064 .013 .008
.157 .057 -.311 .225
.16 .01
.21 .04
-.395 -.395 * -.139 .413 .27 .11
b
.36 .25
.42 .09
.62 .07
.23 .09
Note: *p<.10, **/><.05. "The coefficients in this table are derived from solutions in which the four community activism measures were added to the five independent variables (experience, education, workplace, occupation, and age) included in the regressions in Table 3.5. Total R2 is the R2 from the nine-variable solution and increment refers to the difference between the R2 from the nine-variable solution and that from the five-variable solution in Table 3.5. R2 Change refers to the contribution to the Total R2 that is attributable to each of the activism measures. (The SPSS regression routine that was used reports the R2 contribution for each independent variable when added to the equation in last position.) b Org. coop, refers to attempts to elicit the cooperation of community organizations that were not part of the program. Financial refers to efforts to increase the financial support from the community. Policies refers to efforts to influence community policies regarding juvenile offenders. Resources has to do with efforts to improve the treatment facilities for dealing with offenders.
82
Pathways in the workplace
To conclude, still on a theoretical note, the results in this chapter fall far short of confirming the rationalist model of formal organizational participation. When it came to gaining favorable access to superiors and participation in the decision-making process, indicators of achievement were far less important than the ascribed attributes of race, age, and client demographics. At the same time, the data are equally far from confirming a hypothesis based on starkly polarized vested interests and clear mechanisms of discrimination. If any group was formally favored, it was nonwhite males, but some of their advantages were not shared by non white females. Therefore, it was not simply a matter of racial differentiation, whether of the traditional or "reverse" form. The fact that white males were the least well-situated group rules out the interpretation of traditional male dominance. What survives is a picture of a less than clear-cut set of arrangements that were negotiated by the participants and an impression that age, race, and external ties figured prominently, but variously, in the give-and-take that went on among them.
4. Conclusions
I began this work by contrasting the rationalist conception of how resources and advantages are distributed in organizations with recent perspectives that are critical of the rationalist approach. I used these two opposing sets of assumptions to bracket the range of opinions about what is to be expected in the study of reward allocation mechanisms. My objective has been to see where race and gender differences in informal network integration and access to formal authority fall along this rational-nonrational continuum. The findings have not been simple, nor have they been decisively in favor of any single perspective. Mechanisms generally regarded as rational were operating alongside some that are commonly seen as nonrational, and, moreover, these contrasting sets of mechanisms operated in different ways for different categories of members, and in different ways in the informal and formal arenas. The results were further complicated by the overlap between internal relations and activities located in the extraorganizational environment. Interpreted with appropriate caution, the results provide a good point of reference for speculating about the effects of ascribed status on organizational practices. The data suggest three general conclusions for the organizations in the present study. All three could reasonably be restated as hypothetical guidelines for future investigations. The first has to do with the relationship between the formal and informal dimensions of activity: Ways of gaining access to the networks of collegial ties did not resemble the means of gaining access to the formal authority structure.
Informal ties were basically equal by race and gender. Formal connections were not; they varied by gender and were even more strongly affected by race in ways that are not anticipated by theories of organizational or racial stratification. Moreover, no single category of personnel was disproportionately integrated or disproportionately isolated informally, but there was a rank order among the participants in access to the formal decision-making structure. Nonwhite males were the most favorably and white males the least favorably situated. It is clear from the data that the formal and informal structures were 83
84
Pathways in the workplace
not just different facets of a single overall system of activity, but neither were the two structures entirely decoupled in the usual sense of this term. There were instances in which they seemed to indicate alternative - although sometimes incompatible - styles of organizational involvement. The notion of "inverse coupling," which conveys the idea that what happens in one area can circumscribe the options available in another one, best describes these patterns. The second conclusion concerns the degree of support found for the rationalistic organizational model: Although there were points of confirmation, the configurations of effects suggested by the rationalist model were generally not in evidence. Neither the formal nor the informal dimensions of activity consistently conformed to its predictions.
To further sharpen this conclusion, there was little evidence for the segmented outcome that some previous research has reported, in which a rational pattern for one group exists alongside nonrational patterns for others. It could be argued that the patterns found for white men came closest to the ideal-typical configuration, but both consistencies and important departures from the model were evident for them, much as they were for everyone else. In short, a consistent relationship between individuals' bureaucratic investments and contributions on the one hand and their formal and informal organizational payoffs on the other was not found for any of the four categories of participants. In my opinion, most of the criticisms leveled against the human capital model of labor force dynamics apply with equal force to the rationalistic conception of intraorganizational dynamics. At the same time, the data do not bear any closer correspondence to any of the more extreme unstructured or nonnormative perspectives that were discussed. There were a number of strong empirical findings in the data, some of them involving internal processes, some involving clients and ties to the external community, and some involving age, which indicate that the programs were not without structure or predictability for their participants. It is clear that the structure and predictability did not evolve out of purely bureaucratic considerations, but it certainly goes too far to characterize the programs as "organized anarchies" (Weick 1976). Stated another way, within broad limits, individuals and categories of individuals in the study could make instrumental use of their assets and circumstances in a variety of ways. The third conclusion is implicit in this finding but warrants a separate statement: There was no single set of strategies for gaining instrumental advantages that worked for all participants. Instead, four different types of techniques for dealing with the work setting are suggested, one for each race-gender category.
Conclusions
85
These distinct modes of accommodation provide a convenient framework for summarizing the results of the study. Summary of the findings: four modes of accommodation At the expense of some oversimplification, Tables 4.1 to 4.4 display the stronger findings from the previous chapters for each of the four race-gender categories. The variables listed are those that showed reasonably clear-cut and consistent effects on either the informal or the formal dimension of activity. I will discuss what these configurations imply for white men, white women, nonwhite women, and nonwhite men in turn. The resources and handicaps of white men From the data in Chapters 2 and 3, it cannot be said that white men monopolized the formal dimension of activity or that they dominated the informal networks. As subordinates, their access to the authority structure was extraordinarily low and their evaluations of the contact they had with those in authority were the least favorable of the four personnel categories surveyed. Their level of informal integration was no more than average, but they had a wider variety of factors working in their favor in this arena than the other three categories did (see Table 4.1). They realized clear informal benefits from such bureaucratic factors as the possession of authority, a strategic workplace, and a high client-contact occupation. They also benefited informally from external contacts with private agencies and religious groups. None of these variables improved their access to the formal authority structure. Instead, on that dimension there was a strong positive effect associated with the presence of large numbers of white male clients, and some advantage was associated with frequent contact with court personnel. With only these positive data to go on, the rationalist contributions/inducements formula appears to describe the circumstances of white men reasonably well. Unexpectedly, education broke up this pattern by showing a strong negative relationship to informal network access. However, I showed in Chapter 2 that this reversal from the rationalist prediction was confined to those without authority. For those with authority, the negative relationship between education and centrality disappeared, and their experiences closely resembled what would be expected in a normal bureaucratic career pattern. Some insight into this set of results can be gained by speculating about what it means, in a motivational sense, to be a white male involved in the
86
Pathways in the
workplace
Table 4 . 1 . White men: abbreviated summary of positive and negative effects on informal network centrality and access to authority Informal network centrality
Access to formal authority
Advantageous factors
Disadvantageous factors
Advantageous factors
Disadvantageous factors
1. Possessing authority 2. Central workplace 3. Client-contact occupation 4. Recent training 5. Contact with private agencies 6. Contact with religious agencies
1. Superior education 2. Contact with courts
1. Same racegender clients 2. Contact with courts
1. Client-contact occupation
kinds of work relationships that were represented in these social service delivery systems. Compared to women and nonwhites, college-trained white men generally have open to them a wide range of primary-sector options by which is meant relatively well paid, relatively prestigious white-collar jobs with relatively clear career trajectories. A social service job such as those held by most of the respondents could easily come to be viewed as a career trap, especially by persons who make invidious comparisons between their own work and what others like themselves are able to accomplish in other professional fields. Professional education, usually at least a bachelor's degree, is required to gain entry to social service occupations, but beyond the entry point further education can lead to minor and uncertain career progress at best. It never guarantees job security. The pay is mediocre and the work enjoys comparatively little community wide regard (Ritzer 1977: 178ff.). In describing a similar situation, Pfeflfer (1981:231) sees the outcome this way: "members . . . are continually calculating whether to remain in the organization, or whether they might fare better if they altered their participation." The most hoped-for scenario for many human service practitioners, particularly those who do not work in large, relatively stable public bureaucracies, is to get away from direct client-contact responsibilities and to move up one or perhaps two levels into positions of limited supervisory authority. However, even for those who are successful, the point is quickly reached where
Conclusions
87
little further movement is likely, and an acceptable subjective balance between the credentials and contributions that are required and the rewards that the system has to offer is difficult to strike. A well-educated person in such a situation who has not succeeded in achieving authority is in a particularly awkward position. Failure to achieve even the limited upward mobility permitted by the system could produce an acute sense of relative deprivation and a weakening of what Barnard (1938) called the "willingness to cooperate." Translated into the present context, such a person would be less willing to be caught up in "bartering" the technical expertise that superior education is supposed to represent for workrelated informal involvements that, over time, have become less intrinsically gratifying to him. In light of this interpretation, white males without authority may have disengaged themselves partially from the informal exchange networks as a reaction against unfavorable comparisons between their own work experiences and the opportunities that they perceived as being available to their peers in other careers. In addition, it is possible that those who saw themselves as trapped in the manner I have described exhibited their alienation and resentment visibly and, because of this, were avoided unless they had other claims to the attention of their co-workers. As Witte (1980:147) puts it: "Losers will always outnumber winners, and workers quickly become aware of the long odds against advancement. The young and ambitious seldom accept their lot with equanimity." [See also Kanter (1977a:416) for a similar argument.] One of these reactions implies withdrawal by the alienated and disappointed, and the other implies their rejection by co-workers. Either outcome would help to account for the finding that educated white males without authority were relatively isolated from the networks of interaction. For those who had managed to achieve positions of authority, the negative effects of education on informal centrality were not in evidence. Moreover, as was shown in Chapter 2, white men with authority gained more informally from their possession of high rank than any other category of participant. They were consistently more favorably situated (indicating more willingness to cooperate?) on all five informal sociometric dimensions. To summarize, it seems likely that the career expectations and actual experiences of white males affected the way they brought their potential bureaucratic resources into play in order to gain integration into the informal networks. Career expectations, in turn, are strongly influenced by where the members of a given race-gender category actually stand in the larger labor force or class structure outside the workplace. I assume that the fact that white men in the aggregate generally enjoy advantages in these larger external structures affects their perceptions of how the system works, and this, in
88
Pathways in the workplace
turn, influences how they as individuals evaluate their own career experiences. The interplay of these two factors - structural realities and individual evaluations - in turn strongly affects how they will become integrated into the daily round of work activities in an organization. To put this same social psychological proposition in a slightly more provocative way, it is their belief in the general rationality of the external labor structure that may make white men more vulnerable to an accumulation of interpersonal disadvantages when the internal system, insofar as it concerns specifically them, does not meet their expectations. Ironically, then, if this speculation is correct, the results of having external labor market priority might include a withdrawal of legitimacy from some kinds of organizations and a degree of isolation from the interpersonal resources of the workplace. It should be apparent that the argument I have advanced here for white men applies with much less force to the other three race-gender categories. Their labor market positions are in general inferior and perceived as inferior to those of white men, and when they calculate whether to remain in the organization or when they strike a balance in their willingness to participate on a continuing basis, they have fewer external options to consider. Therefore, having a social service occupation is less likely to create the same degree of relative deprivation, and the logic that leads to interpersonal isolation is not put into action. This argument echoes the reasoning of exchange theory, and especially the concept of "alternative opportunities" [see Blau (1964:31-2) and elsewhere]. However, to me it is just as profitably seen as an example of the micropolitics of the workplace [see Burawoy (1979)] or as an illustration of how individuals participate in the social construction of both the positive and negative features of their own work experiences [see Willis (1981)]. One final pattern involving white men requires comment. For this group, no variable had the same eflFect on both the informal and formal dimensions of activity (see Table 4.1). For example, client-practitioner homophily, the variable with the strongest eflFect on relationships to the formal hierarchy, had virtually no eflFect on informal integration. Conversely, the comparative wealth of positive resources that was visible in the informal arena was totally missing from the formal dimension. A closer look suggests that informal involvements were in some instances an alternative to formal involvement. This would indicate formal-informal coupling, but of an inverse sort. Two findings support this interpretation. First, a high client-contact occupation produced closer ties with colleagues but more distant relations with supervisors. Second, contact with courts was a benefit in gaining access to authority but a handicap in gaining informal access to co-workers. Such direct con-
Conclusions
89
trasts imply that there were definite consequences in one sphere for actions taken in the other. This would not be evident in a truly decoupled system in which formal and informal elements have few, if any, implications for each other. In short, white men who wished to participate fully in the informal exchanges in their programs apparently could use a wide range of strategies. In many ways, the profile of their experiences came closest to the rationalist conception of a normal bureaucratic career pattern. They could exploit their rank if they occupied a position of authority; similarly, they could benefit from recent training, from their occupation, from working in a central location, and from their contacts with other community agencies, particularly private agencies and religious organizations. Their one clear handicap among the bureaucratic variables may have been superior education, and if my speculation about this is correct, this was in part a reflection of the favorable placement of white men in the larger occupational structure. Caution is advisable, however, before concluding decisively that these respondents were the most favored group in the programs. In general, as subordinates, their relationship to the formal authority structure was definitely not as good as that of other participants. Moreover, in some instances, what it took to gain a favorable relationship to the official hierarchy was incompatible with what it took to gain informal network integration. From the standpoint of the job, both types of integration were desirable because both co-workers and superiors represented useful work-related resources for the participants. But the two were not always simultaneously possible. This implies a certain dialectical tension between formal and informal activities [cf. Benson (1977)], and for individuals confronted with this tension, it suggests that some strategic shrewdness was called for in finding a workable overall accommodation. The dominance that they as white men might be able to count on to overcome this ambiguity in other areas of the workforce was by no means guaranteed here. The situations of white women Up to a point, the profile offindingsfor white women resembles that of white men (see Table 4.2). To be specific, the women in the aggregate were no less favorably situated in the informal networks than the men, and the possession of authority was also of some informal benefit to them. As with white men, there were also advantages to be gained from having a strategic workplace and from frequent contact with external private agencies. But here the similarity ends. The informal effects of authority were considerably weaker for
90
Pathways in the
workplace
Table 4.2. White women: abbreviated summary of positive and negative effects on informal network centrality and access to authority Informal network centrality
Access to formal authority
Advantageous factors
Disadvantageous factors
Advantageous factors
Disadvantageous factors
1. Possessing authority 2. Central workplace 3. Contact with private agencies
1. Age
1. Recent training
None
white women; in addition, the effects of occupation, recent training, and* contact with religious groups - all of which were advantageous for white men - were not in evidence at all. On the whole, then, the probability of an informal return for a positive contribution to the program's activities was lower for white women than for white men, and the range of workable strategies for gaining advantages was much narrower. It is also significant that white women did not share the inverse effect of education that was apparent for white men. This variable bore no relationship, positive or negative, to their informal experiences. They are the only group about which this can be said, and if the absence of a negative can be considered a positive, then perhaps white women had the advantage over white men in this area. However, if the labor market segmentation thesis is correct, it is unlikely that white women would have a comparable set of objective external peer comparisons in mind. For this reason, they may have responded in a diflFerent way to the relatively meager internal career opportunities available to them. This is not to imply that their orientations to work or their definitions of an appropriate return on investment were fundamentally diflFerent from those of men, but only that they would perceive less relative deprivation than white men in the same situation. As a consequence, the particular logic of resentment that I argued might have eroded the colleague relations of white men would be less likely to occur. Put another way, the withdrawal/rejection syndrome would be less likely to affect the women. However, in making this comparison between white men and white women, it must be reemphasized
Conclusions
91
that the women were not benefited by superior education. In this strict sense, the investments/payoff balance was also violated for them on this variable, although in a less extreme fashion than for white men. Of equal interest for white women is the unique effect of age on their informal integration. No other category was either handicapped or benefited by youth on the informal dimensions of activity, although there was a similar effect for nonwhite women on the formal dimension. Two different but complementary interpretations of this finding are possible, but each requires an assumption that I was unable to test directly. One interpretation would argue that older women are more likely to be passed over in the informal networks of communication and collaboration [see Miller et al. (1975)]. Several reasons for this are possible, with varying levels of discriminatory implications. As one example, older women may be less sought after because they are seen as less desirable interpersonal contacts than younger women, or they may be avoided because they are thought to be less vulnerable, that is, better able to resist unwanted contacts. In either case, exclusion from some ordinary colleague relationships would be the result, causing them to face a barrier to effective job performance that no other category has to confront. In this argument, the experiences of younger women are taken as the implicit baseline, and the stress is on what it is presumed is done to relatively older women to isolate them from others in their work setting. The other interpretation, which changes the emphasis but does not contradict the direction of the first argument, concentrates more directly on female attractiveness. Attractiveness is a correlate of youth in the eyes of men and, perhaps, women as well. It can be used as an exchange resource or source of power for lower organizational participants [for a suggestive argument along these lines, see Mechanic (1962); French and Raven's (1960) discussion of "referent power" is also relevant]. In this line of reasoning, the focus is specifically on the ability (or necessity) to manipulate youth as an interpersonal resource. It therefore stresses what is done by (or must be done by) a category of women in order to advance (or protect) their interests relative to others in the same work setting. It is important to note that what one observer (probably a male?) would see as a unique advantage to be exploited by women, another observer (a woman?) would see as a more defensive strategy required by an inequitable situation. This accounts for the ambiguity in the way I have advanced the argument. The unstated and untested premise throughout this discussion is that other participants prefer to interact with younger women for one reason or another. This is an assumption about the culture of the workplace that has wide cur-
92
Pathways in the workplace
rency, but it should not be taken for granted. A logically related but also untested supposition is that it was men who withdrew from informal relations with older women and migrated to the younger ones. Finally, I have not explained why the pattern of informal isolation does not appear for older nonwhite women. Nevertheless, if the assumption that youth attracts is tenable, then the finding I have reported indicates a second specific way (the first involved the labor force advantages of white men) in which the boundaries of organizations are penetrated by male-female relations that predominate in the larger society. The leveling hypothesis that is such an integral part of the rationalist and human capital traditions does not fare well in the face of such findings. Turning briefly to the formal authority structure, white women benefited only from having been recently trained. On the whole, their access to the hierarchy was neither particularly favorable nor as unfavorable as that of white men, and for the most part it was unaffected by their educational credentials, occupational identities, or external contributions to program activities. Again the interpretation can be pushed either in a negative direction (their access to their supervisors was largely beyond their influence) or in a positive direction (no clear barriers that were not apparent for other groups blocked their access). To summarize, the options open to white women were not asflexiblystructured as those for white men. In the informal structure, possessing authority and having a favorable workplace functioned to their advantage, but for those who could not draw upon these two bureaucratic factors, only external contact with private agencies moved them closer to the working center of their programs. Age was a negative factor that had to be dealt with. Its influence brought up for the first time the possibility of informal barriers consisting of a combination of sexism and ageism imported from the outside culture. On the other hand, white women did not face the formal-informal trade-offs (in which an advantage in one dimension became a disadvantage in the other) that were in evidence for other members. In fact, there was no indication that their relationship to the formal hierarchy was affected at all by their own activities, since only the recency of their training had any impact. This could simply reflect an uncomplicated relationship to the hierarchy guided primarily by the technical requirements of the job. It could also mean that their relationship to the hierarchy was largely initiated and structured by their supervisors. If the latter is correct, this strengthens the implication that women were constrained to a passive role in this part of their work activities.
Conclusions
93
Table 4.3. Nonwhite women: abbreviated summary of positive and negative effects on informal network centrality and access to authority Informal network centrality
Access to formal authority
Advantageous factors
Disadvantageous factors
Advantageous factors
Disadvantageous factors
1. Superior education 2. Recent training 3. Contact with law enforcement agencies
1. Contact with courts
None
1. Age 2. Recent training0
1
Coefficients for this variable usually fell just short of the .05 significance level.
The experiences of nonwhite women Both superior education and recent training worked to the informal benefit of nonwhite women (see Table 4.3). If one expects that the disadvantages of being nonwhite and female will accumulate in a simple additive fashion, then the first of these findings is especially surprising. It could mean that this one rational, human capital factor structured their daily work experiences in a way that was not evident for any other group. If so, they would be a group for whom an important part of the bureaucratic model had strong applicability. To shift the angle of vision, however, the finding could also mean that nonwhite women were the only ones who needed to use their credentials to validate their claim to positions in the informal networks. In other words, they were in danger of being ignored unless they used their expertise to command attention. If this was the case, it would be evidence of a gatekeeping barrier that was not in effect for the other three categories. With only the data at hand, there is no sure way to choose between these interpretations, but the latter is more consistent with what previous studies involving women and nonwhites have reported. The informal advantage of recent training was one that nonwhite women shared with white men. There were also informal advantages associated with external contact with law enforcement agencies, and these were unique to nonwhite women. However, they were counterbalanced by disadvantages as-
94
Pathways in the workplace
sociated with another external activity, contact with courts (again, a pattern shared with white men). This collection of findings suggests a complicated balance that nonwhite women had to negotiate in the informal structure, but a balance that was responsive, either positively or negatively, to their objective qualifications and their actual performance on behalf of their programs. If they were well and recently trained and had useful ties to the law enforcement community, they were in a strong position to profit from close ties to colleagues. In the formal structure, however, none of these variables was of any consistent benefit. In fact, recent training now appeared as a handicap, and age introduced a unique barrier for these women that seriously compromised their relations with the official hierarchy. As was suggested previously when discussing a similar pattern for white women on the informal dimension, this strong age effect can be taken two ways. Emphasis coulcl be placed on the possibility that younger nonwhite women could use their youth to advantage in their relations with superiors, or it could be stressed that older women were passed over by supervisors in the daily routine of work. What distinguishes this situation from that discussed for white women is that relationships to formal superiors in the official hierarchy are at stake. There is therefore a stronger and more consequential indication that the ascribed characteristics of participants interfered with supposedly normal bureaucratic processes. It is possible, of course, that supervisors worked more closely with younger nonwhite women because their youth bore some direct functional relationship to the work being done with clients. I pointed out in Chapter 1 that the younger practitioners may have been better able to relate to the youthful clients served by the programs. If this were true, supervisors may simply have been directing their attention to the more effective female practitioners. However, why this should have influenced supervisors' relations with only one of the four personnel categories is not easy to understand, and it is my strong impression that this interpretation is not correct. An explanation based directly on sexism - male supervisors drawn to young female subordinates and away from older ones - is another possibility, but again this suggests that it was male supervisors whose reactions were affected by their subordinates' ages. Subordinates were not asked to indicate the gender of their supervisors, and this possibility could not be investigated directly. However, since female supervisors actually outnumber males (59 to 37), it is unlikely that the actions of the males could completely account for the observed age-biased outcome. What is apparent is that, whatever the source, nonwhite women had to
Conclusions
95
contend with one complication in their relationship to authority that was not in evidence for any other group. The effect of this complicating factor was so strong as to suggest that it dominated this aspect of their work experience, and in a way that no technical job description would ever predict. Nonwhite men The situation of nonwhite men has already been discussed at length. What remains is to speculate about the strategy that their position suggests for negotiating a favorable relationship to co-workers and supervisors. Above all, their chances of achieving a favorable informal position were enhanced by the presence of nonwhite male clients (see Table 4.4). This indicates that they were rewarded for a contribution they could make to their programs, and it makes good technical sense. As I noted earlier, however, the composition of the client pool is a factor beyond the immediate control of the practitioner himself. Those nonwhite men who could not rely on this factor, that is, those in programs with very few nonwhite male clients, must have been comparatively empty-handed in the competition for the informal resources represented by network centrality. Contact with external religious organizations produced some benefit, but workplace was the only intraorganizational variable that had a sizable positive (but nonsignificant) effect on their standing with their co-workers. Neither education, experience, nor occupational identity - all of which represent attributes of the participant or factors more or less under his control - promised a favorable informal return. In several instances, just the opposite was the case. Education, external contact with public agencies, and external financial activities, for example, decreased the likelihood of informal integration. Although the negative effect of education was a handicap that they shared with white men, they lacked the counterbalancing positive effect of authority that white men enjoyed. In short, compared to the other participants, there is less evidence that nonwhite men could manipulate their occupational qualifications, their organizational positions, their work activities, or their age in order to gain more favorable informal ties with colleagues. They could turn their race-gender identity to advantage where nonwhite male clients were numerous, but the nature of the client pool was beyond their control. Aside from this, they must have relied upon resources and personal attributes that remained beyond the reach of this survey. In relations with supervisors, education was again a disadvantage for nonwhite men, the only group for whom this was true. Having been recently trained, on the other hand, was advantageous, although not always strongly so. The most interesting effect here, however, had to do with the payoff
96
Pathways in the workplace
Table 4.4. Nonwhite men: abbreviated summary of positive and negative effects on informal network centrality and access to authority Informal network centrality
Access to formal authority
Advantageous factors
Disadvantageous factors
Advantageous factors
Disadvantageous factors
1. Workplace0 2. Same-racegender clients 3. Contact with religious agencies0
1. Superior education 2. Contact with public agencies 3. External financial activities
1. External financial activities 2. Recent training0
1. Superior education0
°Coefficients usually fell just short of the .05 level of significance.
associated with external financial activities. Making this important effort for the program's benefit was associated with closer supervisor-subordinate relations, but, as I have pointed out, this appeared to come at the expense of less favorable informal relations with colleagues. In the final accounting, nonwhite men succeeded as often as anyone in achieving informal integration, and their relationship to the formal authority structure was extraordinarily favorable. Nevertheless, an effective, proactive career strategy is difficult to visualize because so few traditional bureaucratic factors functioned for them in traditional bureaucratic ways. It is interesting to trace out what these patterns would mean for a newly recruited nonwhite male practitioner entering one of the programs for the first time. Unlike a member of every other race-gender category, he apparently could not make any safe assumptions at the point of entry about how to gain access to the interpersonal rewards and professional instrumentalities the programs had to offer. In fact, if the pattern for white men came closest to the rational model, that for nonwhite men departed furthest from it. For a nonwhite male who had learned how things worked and who had an active interest in maintaining close collegial ties, determining the appropriate strategy must have been difficult, since education was a persistent problem, occupation and experience had little bearing, and external organizational contacts, unless they involved religious groups, were just as well avoided. For the person more interested in achieving favorable relations with formal superiors,financiallyproductive efforts in the community might have the desired effect, but at the expense of
Conclusions
97
less favorable informal ties. What the informal and formal areas of activity had in common is that both were responsive primarily to external factors, client characteristics in one case and community activism in the other. The fact that these mechanisms were available to nonwhite men must certainly be counted as advantages for them, but it is very likely that the part of their career success that was due to their own eflForts had to be very carefully choreographed. Their circumstances bore little resemblance to a clearly structured field of opportunity. A minefield might be a more apt analogy. Conclusions and implications Coherence and coordination were not absent in the programs in this survey, but these organizations were not the kinds that have undisputed objectives or unambiguous systems of goal-directed activity. Because consensus on what the programs were about was lacking, no single set of technical criteria served as a benchmark for judging the formal and informal experiences of all the participants. In fact, in the face of competing sets of evaluative criteria, the possibility that different groups would confront different kinds of experiences was increased. To some extent, this kind of pluralism is endemic to most people-processing organizations. In such circumstances, the limits of the rationalistic model are quickly reached because the model assumes a high level of clarity and a close articulation of technical requirements, bureaucratic inducements, individual contributions, and career rewards. Not one of the four race-gender patterns I have reported is directly deducible from such a narrow characterization of the rationalist approach, and the members of none of the four categories could have assumed, at their point of entry into the programs, that a clear-cut set of bureaucratic expectations would govern their professional experiences. Neither could they have assumed that what produced favorable relationships to their superiors would pay off in their relationships with co-workers. At the same time, it cannot be said that any of the categories faced a completely unpredictable situation. The perceptive members of even the most ambiguously situated category, which I take to be nonwhite men, could learn in time that certain factors could be turned to advantage and others could not. Put another way, there were regularities for every group that, expressed statistically, reduced the unexplained variance or, stated in personal terms, reduced the uncertainty they faced in their daily work. The theoretical difficulty with this state of affairs is that no one organizational model clearly and convincingly incorporates all of the findings. The nearest thing to an overarching explanation - and it is far from complete - is
98
Pathways in the workplace
that internal organizational exigencies intersect with the configurations of external race and gender relations and that together these form the context for the confrontations that take place among individuals. However, the clear lines of cleavage that characterize the external system, or at least appear to do so when only race or gender and not race-gender are considered, are not duplicated internally. At this point, details about the specific organizational context, and especially its internal micropolitics, must be brought into the picture. It is precisely here that ethnographic data based on close and direct observation would exercise the greatest leverage in developing explanations for observed patterns of organizational behavior [see Emerson and Warren (1983)]. The fine details of the findings and interpretations that I have reported here may not generalize far beyond this immediate research setting. The collective tasks in which the programs in the survey were involved, the specific demographic configurations they displayed, and the community contexts in which they were placed are unlikely to be duplicated. For this reason, it is not claimed that they provide a standard against which other types of organizations are to be compared point by point. I would never suggest, for example, that nonwhite men typically have greater access to organizational authority structures, that white men typically have to contend with the "handicap" of superior education, or that the disadvantages associated with age in the workplace are typically confined to women. Obviously, in other organizational contexts, very different patterns may be encountered. However, to look for exact replication would be, in a very important sense, to miss the point. The more general implications of the findings are likely to have wide applicability. Social service delivery systems are themselves increasingly common organizational arrangements, and many of the strategic contingencies concerning race, gender, environmental and cultural influences, and the distribution of rewards that are discussed here are faced in one form or another by organizations of widely differing types. So little is known at present about how organizations deal with problems involving the intersection of ascription and achievement that any data will stand out against a largely blank background. The best starting point for approaching the study of these and related problems is to bring to center stage what is implicit in almost every conception of organizations: Official mandates, official tasks, and publicly expressed goals typically oversimplify the diversity of objectives that are actually being pursued. They set rough limits on organizational arrangements, but within these limits processes of competition and negotiation among different categories of personnel are sure to come into play (Benson 1977; Edwards 1979). As Blau
Conclusions
99
(1977) has pointed out, when the workforce is highly differentiated by race and gender, these ascribed attributes take on added significance in these organizational processes. Race and gender groups differ in their placement in the larger labor market and in their locations in the external structure of social power. They bring these advantages and disadvantages with them into the workplace. Moreover, the kind of work the organization is involved in will determine how ascribed identities fit into the specific tasks that participants are called upon to perform, and this will be another source of variation among them. Human capital considerations, technical imperatives, and bureaucratic leveling effects are not sufficient to neutralize these differences, which continue to be reflected in the expectations and behavior of organizational members. The most general conclusion that can be derived from this study logically follows: Single rules of reward allocation break down or simply never come into play when organizations have multiple and competing sets of objectives. The picture is further complicated when the personnel are highly differentiated, whether along lines of race, gender, and age or according to any other set of factors that separate the members into categories but that bear an imperfect relationship to task performance. Rationalist theories are most applicable where there is, in effect, a single source of goals, a single reward structure, and a single category of participants (white males, for example) who share, in the aggregate, more or less equal potential access to crucial external and internal resources. This, of course, is a rare - perhaps increasingly rare - configuration. Where there is diversity, universalistic rules must coexist in altered form alongside other systems of allocation that follow other sets of distributive principles. Some of these sets of principles may be functional, in the sense of technically sensible, adaptations to problems and situations that are not effectively covered by formalized bureaucratic arrangements. Others may involve inequitable and discriminatory patterns that carry over from the larger culture. How "rational" a given individual's work experiences are, then, is determined by the convergence for that person of all the ascribed and achieved interpersonal, organizational, interorganizational, and community handicaps and resources.
Notes
Introduction 1 This was the stated rationale and official mandate of the program, which was committed to the process of deinstitutionalization, that is, moving juvenile status offenders out of the official juvenile justice system and into community-based treatment centers. Critics of this legal movement make a strong case that it is the initial stage in the gradual turning over of previously public responsibilities to the private - and increasingly profit-oriented - sector [see Warren (1981)]. 2 I have used the word "informal" throughout to describe the sociometric data, although "nonformal" might be a better choice to reflect the intended meaning. "Informal" sometimes connotes simple conviviality; often it refers to exchanges that have developed independent of or as a reaction against the work being demanded by an organization. By contrast, the networks I have called informal are task-related arrangements sustained by the members in the interest of task performance, but without clear formal guidelines or regular formal surveillance.
1. Rationality and equity in professional networks 1 Many of these arguments were anticipated by Weber. See the discussion by McNeil (1978), in which Weber's awareness of inequities and irrationalities is documented. 2 Criticisms of Weber's "rationalism" miss the point when they confuse his treatment with later and far less perceptive accounts. In an otherwise provocative article on the limits of rationalist theories, Brown's (1978) identification of Weber's model with Taylorism is a good illustration of this confusion. Although I am occasionally guilty of using "Weberian" and "rationalist" as if they were synonyms, I am aware that much that is typically called rationalist is only vaguely Weberian. 3 It is important to note that in Weber's discussion of bureaucracy, ascription referred primarily to such attributes as social class and family background, and not to gender and race. Weber discussed how ascription, especially ethnicity, could complicate systems of stratification (Gerth and Mills 1946:188; Hechter 1976; Weber 1978:342), but the bureaucracies in his experience would have been staffed almost completely by males who shared the same racial/ethnic background, and his theorizing must have been constrained accordingly. Had the presence of women and nonwhites in the labor force been the pressing and potentially divisive issue then that it is now, there is little doubt that the ideal type would have had to be formulated differently, or at least qualified in important ways. In a strict sense, then, to demonstrate that race and gender cause systematic departures from strict rationality does not disprove Weber's notion of leveling, but it does indicate strongly that the ideal type is historically and culturally bound in important ways. 4 Limited support for Kanter's viewpoint appears in Spangler et al. (1978), and some theo-
100
Notes to pp. 12-17
5 6
7
8
9
10
101
retical support for her concentration on numbers as the key to intergender relations is found in Blau (1977). See also Lorber (1979). Fox (1981a) explored the effects of male-female proportions on the salary levels of the two categories and concluded that the numbers effects are largely indirect, by way of their impact on aggregate levels of achievement (training and experience). Arguments by Acker and Van Houten (1974) and Stern, Gove, and Galle (1976) are also relevant here, although neither article offers empirical support for its conclusions. To illustrate the lack of clarity, three of the five organizations in my earlier study [Miller et al. (1975)] were involved in scientific research, and highly qualified women with professional credentials or administrative positions were few compared to the number of similarly qualified men. The disadvantages of the women might have been tied to this skewness as much as to active discrimination. Note also in Butler's (1976) army data that blacks were disadvantaged relative to whites but that the system actually worked rationally for neither category. This is in contrast to the Blau and Duncan (1967) finding that investments paid off for both whites and nonwhites, only more so for whites. A note of caution in reading the evidence on organizational stratification comes from, of all places, a study of baseball players. Abrahamson (1979) reports that rewards on baseball teams are structured in a manner consistent with the functional theory of stratification (Davis and Moore, 1945): Salaries are the outcome of an interaction between functional importance and uniqueness of contribution, two principles that are quite compatible with the notion of rationality as I have discussed it here. Although he implies that this pattern will generalize across a wide range of organizational types, Abrahamson presents no data on racial differentiation, and for obvious reasons the distribution of rewards to baseball players has nothing to contribute to the understanding of male—female differences. In a parallel to the idea of compartmentalized rationality previously mentioned, his data do suggest that in studies in which race is ignored and women are absent, reward allocation procedures may be seen to develop along lines suggested by rationalist theories. It is interesting to remember that the Weberian model was developed in similar circumstances: few women, few or no representatives of racial minorities. This is not the place for an extended discussion, but the functionalist formulation is also compatible with a line of thinking that predates by many years the point at which the massive and indispensable movement of women and nonwhites into the urban labor force was recognized as irreversible. This makes the perspective historically bound in a way that I do not believe has been acknowledged. There were actually eight programs in the project, but one was delayed in starting and another did not participate in the part of the survey that provided the data on interagency relations. Findings reported by Wolf and Fligstein (1979a) and Kluegel (1978) that indicate that women and blacks are excluded from authority are not directly duplicated in this study. Overall access to decision-making authority was not very different between subgroups, although white men were somewhat more numerous among agency directors and the other three race-gender groups were more numerous in the intermediate levels of authority. Moreover, the means of gaining authority were not fundamentally different (see Table 2.1 and note 11 in Chapter 2). However, in many organizational contexts, how people gain authority may be no more interesting than what they are able to do with it, either for themselves or for the work they are charged with doing. The validity of this proposition has long been apparent in the organizational literature, even if rarely stated in network terms. Blau's (1955) study of consultation patterns is a good example, as is Homans's (1961) discussion of interaction patterns in organizational surroundings. More recent examples include the work of Rogers and Agarwala-Rogers (1976) and Miller (1980). The proposition is also supported by a wide range of other studies,
102
11 12
13 14
15
Notes to pp. 20-38
including studies of job-finding networks (Granovetter 1974), networks that mediate access to mental health services (Horwitz 1977), and those that define the structure of community power (Perucci and Pilisuk 1970; Laumann and Pappi 1976). On this point, see Lipton, Martinson, and Wilks (1975), Lerman (1975), and Lundman and Scarpitti(1978). The method I used involved asking the respondents for three nominations on each of these sociometric questions, an approach that is still used in important studies (Laumann and Pappi 1976; Burt 1977), but that has been criticized (Holland and Leinhardt 1973) in part on the technical grounds that it may be subject to error as a result of underreporting of ties. The forced-choice format that I used has the advantage of confronting respondents with a standard stimulus, and it can be used to minimize the reporting of ties that might be unimportant. In this study, no participant in any of the programs was excluded, that is, neither chose nor was chosen because of this method; therefore, a meaningful range of centrality scores was available that included all the respondents. Centrality scores were generated with Peter Marsden's DIGRAPH program. Simple hierarchies in the programs justified this dichotomy. None of the programs had more than three levels, including, in a fairly typical arrangement, a director, one or more supervisors, and the ordinary practitioners. The proportion of all work contacts exchanged that were directed to the members of the administrative center averaged 36% for the six programs, with a range of 16 to 83% (the latter figure representing the smallest program, with only 20 participating members). Similar proportions appeared for the other four sociometric questions.
2. Ascription, achievement, and network centrality 1 I have explored this issue in two previous studies. See Miller (1970) and Miller and Fry (1973). 2 Separate zero-order regression analyses for the six programs in the study (not shown) produced results that were essentially the same as those reported in Tables 2.1 and 2.2 for the combined programs. The amount of variance explained differed but the effects of gender and race were similar, that is, quite small. Because of the small number of nonwhite men and women, it was not possible to perform the later analyses that involved first- and secondorder interaction effects separately for the six programs. 3 The significance of the difference between regression coefficients was determined by the formula:
Vse/?, 2 + seb22 where df = Nx + N2 - 4 [see McNemar (1962:143)]. Pedazhur suggests that the level of significance for comparing regression coefficients ought to be set at .10 or even .25 [see Pedazhur (1982:438-42)]. With a few exceptions, I have stayed with the more conventional .05 level. 4 This method for detecting interactions is described in more detail in Hanushek and Jackson (1977:124ff.); see also Weisberg (1980). 5 The computation of the test statistic in this case is more complex than the previous one and warrants more discussion. The residual sum of squares from the constrained model is computed as follows:
Notes to pp. 43-49 N
6
103 6
6
ESSC = ^[Y{ - (a + YjbkXik + cMi + dWt + ^2jfkMXik + ^gkWXik i=i
l
k=i
k=i
+ hMW{)]2
k=i
where Y,- is the ith value of the dependent variable; Xik is the ith value of the kth regressor; M and W are dummy variables for male and white, respectively; MXik, WXik, and MW( are product terms; and a, bk, c, d,fk, gk, and h are estimated regression parameters. The residual variation from the unconstrained model is then:'
ESSU = 2 2 [Ytj - (a
6
7
8
9
where y» is the ith value of Y in theyth gender-race subgroup; Xijk is the /th value for the7th subgroup of the kxh regressor; n- is the size of the/th sbugroup; and a and ^ are estimated regression parameters. To illustrate this point, some of my colleagues in public school systems report that minority male professionals are sometimes shifted so quickly into prominent positions in times of acute minority-majority crisis that there is insufficient time to assess their preparation for the demands of the assignment. Over an extended period, their centrality would be greatly increased, while at the same time their objective contributions (credentials, specialized expertise, experience, and the like) could come to bear only a tenuous connection to their positions in the system's hierarchy. Corroboration of this phenomenon is offered by Gross and Trask (1976), who report that women are "drafted" into positions of responsibility in schools four times as often as men. Kanter (1977a), of course, provides a detailed exploration of the more clearly negative and isolating consequences of grossly imbalanced gender proportions in the workplace. Of the 4,849 clients remanded to the programs during the period of the study, 1,788 (36.9%) were white females, 1,346 (27.8%) were white males; 1,010 (20.8%) were nonwhite females, and 705 (14.5%) were nonwhite males. The average proportions of same race-gender clients remanded to the practitioners' own programs were as follows: For white women, the average proportion of white female clients was 37.8%; for white men, the average proportion of white male clients was 30.12%; for nonwhite women, the average proportion of nonwhite female clients was 24.8%; and for nonwhite men, the average proportion of nonwhite male clients was 20%. These additional data on the quality of the external ties are not shown. The scales used ranged over nine points from "contacts never produce benefit for . . . clients" to "contacts always produce benefits for . . . clients"; from "contacts never produce closer cooperation" to "contacts always produce closer cooperation"; and from "contacts are never with persons in positions of authority . . . " to "contacts are always with persons in positions of authority." The possible responses for reasons for the external ties were: (1) to clarify the needs of individual clients; (2) to encourage change in the way juveniles were handled; (3) to get the resources that clients need; and (4) to encourage respect for the client as a person. As an alternative way of examining the effects of these external ties, each of the six external contact variables was also added by itself to the basic equations for each subgroup. This method creates less of a problem with degrees of freedom for the smaller subgroups because on a given pass through the data it increases the number of independent variables by only one. However, it does not produce estimates of the net effects of each external contact
104
Notes to pp. 53-58
variable. Not surprisingly, when the two methods are compared, this uncontrolled approach produces a larger number of significant increments than the analysis presented in Table 2.9 and summarized in the text. However, the same basic configurations of positive and negative effects for the external contact measures emerge from both methods. Another way to alleviate the degrees of freedom problem is to combine the external contact measures into a scale. The variables are substantially intercorrelated and do in fact scale reliably ( a = .77). When this scale is added to the original list of independent variables, the R2 values are increased substantially for all groups, particularly the two nonwhite ones, but the substantively important mixture of positive and negative effects on the dependent variables is completely obscured. For this reason, I prefer to concentrate on the findings presented in the text, even though in some instances they must be taken very cautiously. 10 As with the external contact measures, each community activity measure was added by itself to the basic subgroup regressions. The results closely replicated those described in the text and reported in Table 2.10. Similarly, as in the case of the external contact measures, the activism items form a reliable scale (a= .88) that, added to the basic regression equations, increases the R2 values substantially. However, since the effects of the individual activism items are again a mixture of positives and negatives, I decided against using this scale in the text. 11 The results, with authority the dependent variable, are as tabulated. The education-authority connections shown here are essentially unchanged when influence is also added to these equations.
Nonwhite
White
b Experience Education Age Workplace Job type R2 (R2)
Women
Women
Men seZ?
-.001 .012 .064** .028 .010 .004 .129 -.097 -.254** .120 .14 (.07)
b
seb
.009 .001 .075**** .019 .014*** .005 .085 .158 .079 - .140 .21 (.17)
b
Men seb
.038** .016 .041 .035 -.009 .009 .282 .049 -.274 .174 .28 (.12)
b
se&
.025 .028 .080*** .024 .003 .016 .484*** .108 -.325* .160 .55 (.45)
Note: *p<.10, **p<.05, ***/?<.01, ****p<.001. 12 For white women, the unstandardized regression coefficient for the interaction term, "rank x education," was - .004, with a standard error of .013; for nonwhite women, the corresponding figures are - . 0 3 1 and .031; and for nonwhite men, - . 0 5 2 and .033. None of these results approaches the .05 or even the .10 level of significance. For white men, in contrast, the regression coefficient was .057 and the standard error .010 (p<.001). 13 The complete regression, where influence is the dependent variable, is as follows:
Notes to p. 64
105 White men Without authority
With authority b
.0004 - .0003 .002 -.002 .187** .66 (.57)
Education Experience Age Job type Workplace R2(R2)
sefc
b
seb
-.021* -.006 -.005 .118** .048 .31 (.22)
.009 .003 .002 .027 .031
.008 .004 .003 .041 .041
Note: */?<.O5, **/?<.001.
3. Access to the formal authority structure 1 There is further indirect support for this interpretation in the pattern of correlations of "participation" and "interaction" with "evaluation." The coefficients are weaker for nonwhite men, but nevertheless all four categories tended to place a favorable evaluation on contact with supervisors:
Evaluation
Participation Interaction
White men
White women
Nonwhite men
Nonwhite women
.54 .39
.55 .39
.27 .25
.55 .65
Two other items bear on this issue. The survey included afive-itemscale that measured the work strain the members experienced concerning amount of job autonomy and a single item that asked them to assess the effectiveness of their own professional efforts on the job. The correlations between interaction with supervisors and the measure of work strain are — . 34 and —.30 for nonwhite men and women, respectively, whereas the correlations between their interaction with supervisors and evaluation of their own effectiveness are + .29 and + .36, respectively. For whites the correlations are weaker and, in fact, for white men there is a positive although nonsignificant correlation (.13) between interaction with supervisors and work strain. It is also of interest that for nonwhites there were strong negative correlations between evaluation of time spent with supervisors and work strain (r = - .72 and - .43 for men and women, respectively), accompanied by strong positive correlations (.63 and .54) between evaluation of supervisory contact and assessments of own professional effectiveness. Again, the corresponding correlations were weaker for the white men and women. If "close surveillance" were the correct construction of "frequent interaction with supervisors," then the interpretation that nonwhites had an advantage in terms of access to au-
106
2
3
4
5
Notes to pp. 64-78
thority would have to be reconsidered. The differences in the configurations of empirical findings reported in this chapter would of course remain, but the interpretations of what the differences mean would have to be altered. Close contact might be a disadvantage, and the factors that produce it might need to be seen as handicaps, not assets. There is no conclusive evidence in the survey data to rule out this competing interpretation, but neither is there a compelling a priori reason to accept it, and the data just presented indicate that the assumption that contact with supervisors is an advantage is in fact more strongly supported for nonwhites than for whites. In fact, I much prefer the interpretation of this matter offered in the text because the alternative pushes me to assume that for many of the respondents, and especially the nonwhites, approval or perhaps even gratitude was being exchanged for being held under close but helpful scrutiny. I would find this notion unlikely as a general proposition because it conflicts with the findings of most studies of participation in decision making, and I find it especially unpersuasive when applied to young nonwhite practitioners in public service organizations. The gender and race of each practitioner's supervisor were not recorded, and it was not possible to determine whether white male practitioners fared any better with white male supervisors than they did with supervisors in general. Participation in decision making is a categorical variable, but in the tables it was analyzed using multiple regression (ordinary least squares) so that the form of the findings would parallel that for the other, continuous dependent variables. The skewness of "participation" was well within accepted limits for this purpose, but it can be argued that logistic regression, which is specifically intended for categorical dependent variables, is a preferred procedure. Accordingly, the analysis was repeated with this alternative method. The same basic patterns of significant and nonsignificant effects emerge from the two methods of analysis. The only important difference is that in the logistic regression, workplace is added to the list of variables that bear a significant positive relationship to participation in decision making. Among the 22 nonwhite women without supervisory responsibilities, 11 were 30 years of age or younger and 11 were over 30. All (100%) of those in the younger group reported participation in the decision-making process; among those over 30, only five (45%) reported participation. The corresponding contrasts for the other race-gender categories were considerably less dramatic. For white women, the differences in rates of participation for those 30 and under versus those over 30 were, respectively, 66% versus 60%; for white men, 50% versus 44%; and for nonwhite men, 78% versus 7 1 % . To reinforce this impression, the 0-order correlation between age and participation for nonwhite women was - .68, compared to values of —.01, —.09, and —.17 for white men, white women, and nonwhite men, respectively. This extraordinary age effect also continued to be strong and significant after the measures of client-practitioner homophily, external contact, and community activism were statistically controlled (data not shown). In fact, when all of the variables considered in this chapter, both those that comprise the basic model and all three sets of external factors (client-practitioner homophily, external contact, and community activism), were entered simultaneously into a regression equation for the total population of program participants (data not shown), the effects of race remained. For participation the effect was significant at .05, for contact with supervisors at .0001, and for evaluation of supervisors at .10. In the last case, no other variable reached even this level of significance.
References
Abrahamson, Mark, 1979. "A Functional Theory of Oganizational Stratification." Social Forces 58 (September): 128-45. Acker, Joan, and Donald R. Van Houten, 1974. "Differential Recruitment and Control: The Sex Structuring of Organizations." Administrative Science Quarterly 19 (June): 152—63. Aldrich, Howard E., 1976. "Resource Dependence and Interorganizational Relations." Administration and Society 1 (February):419—54. Alves, Wayne M., and Peter M. Rossi, 1979. "Who Gets What? Fairness Judgments of the Distribution of Earnings." American Journal of Sociology 84 (November): 561-4. Barnard, Chester, 1938. The Functions of the Executive. Cambridge, Mass.: Harvard University Press. Benson, J. Kenneth, 1975. "The Interorganizational Network as a Political Economy." Administrative Science Quarterly 20 (July):229-49. 1977. "Organizations: A Dialectical View" Administrative Science Quarterly 22 (March): 121. Blau, Peter, 1955. The Dynamics of Bureaucracy. Chicago: University of Chicago Press. 1964. Exchange and Power in Social Life. New York: Wiley. 1977. Inequality and Heterogeneity: Toward a Primitive Theory of Social Structure. New York: Free Press, and Otis D. Duncan, 1967. The American Occupational Structure. New York: Wiley. Brager, George, and John A. Michael, 1969. "The Sex Distribution in Social Work: Causes and Consequences." Social Casework 50 (December):595-601. Brief, Arthur P., and Ramon J. Aldag, 1975. "Male-Female Differences in Occupational Attitudes within Minority Groups." Journal of Vocational Behavior 6:305-14. Brown, Richard Harvey, 1978. "Bureaucracy as Praxis: Toward a Political Phenomenology of Formal Organizations." Administrative Science Quarterly 23 (September):265— 382. Bucher, Rue, and Joan Stelling, 1969. "Characteristics of Professional Organizations ."Journal of Health and Social Behavior 10 (March): 3-15. Burawoy, Michael, 1979. Manufacturing Consent. Chicago: University of Chicago Press. Burt, Ronald, 1977. "Positions in Multiple Network Systems, Part Two: Stratification and Prestige among Elite Decision Makers in Altneustadt." Social Forces 56 (December):551— 75. Butler, John S., 1976. "Inequality in the Military: An Examination of Promotion Time for Black and White Enlisted Men." American Sociological Review 41 (October):807—18. Cohen, Michael D., James G. March, and Johan P. Olsen, 1972. "A Garbage Can Model of Organizational Choice." Administrative Science Quarterly 17 (March): 1—25. Dahrendorf, Ralf, 1959. Class and Class Conflict in Industrial Society. Stanford, Calif.: Stanford University Press. Davis, Kingsley, and Wilbert E. Moore, 1945. "Some Principles of Stratification." American Sociological Review 10 (April):242-9.
107
108
References
Edwards, Richard, 1979. Contested Terrain. New York: Basic Books. Michael Reich, and David M. Gordon, 1975. Labor Market Segmentation. Lexington, Mass.: D.C. Heath. Emerson, Robert M., and Carol A.B. Warren (eds.), 1983. Trouble and the Politics of Control in Social Control Institutions. A special issue of Urban Life: A Journal of Ethnographic Research 12, 1 (October). Epstein, Cynthia, 1970. "Encountering the Male Establishment: Sex Status Limits on Women's Careers in the Professions." American Journal of Sociology 75 (May):965-82. 1973. "Positive Effects of the Multiple Negative: Explaining the Success of Black Professional Women." American Journal of Sociology 78 (January) :912-35. Feldberg, Roslyn L., and Evelyn Nakano Glenn, 1979. "Male and Female: Job versus Gender Models in the Sociology of Work." Social Problems 26 (June):524-38. French, J.R.P., Jr., and B.H. Raven, 1960. "The Bases of Social Power," in D. Cartwright and A. Zander (eds.), Group Dynamics: Research and Theory, 2nd ed. New York: Row, Peterson, pp. 607-23. Fox, Mary Frank, 1981a. "Sex Segregation and Salary Structure in Academia." Sociology of Work and Occupations 8 (February):39-60. 1981b. "Sex, Salary and Achievement: Reward-Dualism in Academia." Sociology of Education 54 (April):71-83. Fry, Lincoln J., and Jon P. Miller, 1974. "The Impact of Interdisciplinary Teams on Organizational Relationships." The Sociological Quarterly 15 (Summer):417—31. 1975. "Responding to Skid Row Alcoholism: Self-Defeating Arrangements in an Innovative Treatment Program." Social Problems 22 (June):675-88. Galaskiewicz, Joseph, 1979. "The Structure of Community Networks." Social Forces 57 (June): 1346-64. Gerth, Hans, and C. Wright Mills, 1946. From Max Weber: Essays in Sociology. New York: Oxford University Press. Grandjean, Burke, 1979. "History and Career in a Bureaucratic Labor Market." Presented to the Education and Work Group, National Institute of Education, Washington (October), and Helen H. Bernal, 1979. "Sex and Centralization in a Semi-Profession." Sociology of Work and Occupations 6 (February):84—102. Granovetter, Mark, 1974. Getting a Job. Cambridge, Mass.: Harvard University Press. Gross, Neal, and Anne E. Trask, 1976. The Sex Factor in the Management of Schools. New York: Wiley. Halaby, Charles N, 1979. "Job-Specific Sex Differences in Organizational Reward Attainment: Wage Discrimination versus Rank Segregation." Social Forces 58 (September): 108-27. Hanushek, Eric A., and John F. Jackson, 1977. Statistical Methods for Social Scientists. New York: Academic Press. Harary, Frank, Robert L. Norman, and Dorwin Cartwright, 1965. Structural Models: An Introduction to the Theory of Directed Graphs. New York: Wiley. Hasenfeld, Yeheskel, 1972. "People Processing Organizations: An Exchange Approach." American Sociological Review 37 (April):256-63. Hechter, M., 1976. "Response to Cohen: Max Weber on Ethnicity and Ethnic Change." American Journal of Sociology 81 (March): 1162—8. Heise, David R., 1969. "Problems in Path Analysis and Causal Inference," in E.F. Borgatta (ed.), Sociological Methodology. San Francisco: Jossey-Bass, pp. 38-74. Hogan, Dennis P., and Michele Pazul, 1981. "The Career Strategies of Black Men." Social Forces 59 (June): 1217-28. Holland, Paul W, and Samuel Leinhardt, 1973. "The Structural Implications of Measurement Error in Sociometry." Journal of Mathematical Sociology 3 (January): 1-27.
References
109
Homans, George, 1961. Social Behavior: Its Elementary Forms. New York: Harcourt Brace. Horwitz, Allan, 1977. "Social Networks and Pathways to Psychiatric Treatment." Social Forces 56 (September):86-105. Jeffries, Vincent, and H. Edward Ransford, 1980. Social Stratification: A Multiple Hierarchy Approach. New York: Allyn and Bacon. Kalberg, Stephen, 1980. "Max Weber's Types of Rationality: Cornerstones for the Analysis of Rationalization Processes in History." American Journal of Sociology 85 (March): 1145— 79. Kanter, Rosabeth Moss, 1977a. Men and Women of the Corporation. New York: Basic Books. 1977b. "Some Effects of Proportions on Group Life: Skewed Ratios and Responses to Token Women." American Journal of Sociology 82 (March):965-90. 1979. "Effects of Proportions on Group Life: Tokenism, not Sex Discrimination." Sociological Inventory (Sample Issue):5—8. Kluegel, James A., 1978. "The Causes and Costs of Racial Exclusion from Job Authority." American Sociological Review 43 (June):285-301. Laumann, Edward O., and Franz U. Pappi, 1976. Networks of Collective Action: A Perspective on Community Influence. New York: Academic Press. Lawler, Edward J., and Jerald Hage, 1973. "Professional-Bureaucratic Conflict and Powerlessness among Social Workers." Journal of Sociology and Social Welfare 1 (Fall):92—102. Lerman, Paul, 1975. Community Treatment and Social Control. Chicago: University of Chicago Press. Lin, Nan, 1976. Foundations of Social Research. New York: McGraw-Hill. Lincoln, James R., and Gerald Zeitz, 1980. "Organizational Properties from Aggregate Data: Separating Individual and Structural Effects." American Sociological Review 45 (June):391-408. Lipton, Douglas, Robert Martinson, and Judith Wilks, 1975. The Effectiveness of Correctional Treatment. New York: Praeger. Lorber, Judith, 1979. "Trust, Loyalty, and the Place of Women in the Informal Organization of Work," in Jo Freeman (ed.), Women: A Feminist Perspective, 2nd ed. Palo Alto, Calif.: Mayfield, 371-81. Lundman, Richard J., and Frank R. Scarpitti, 1978. "Delinquency Prevention: Recommendations for Future Projects." Crime and Delinquency 24 (April):207-20. March, James G., and Herbert A. Simon, 1958. Organizations. New York: Wiley. Martin, Patricia Y , 1980. "Women, Labour Markets and Employing Organizations: A Critical Analysis," in David Dunkerly and Graeme Salaman (eds.), The International Yearbook of Organization Studies 1980. London: Routledge and Kegan Paul, pp. 128-50. McNeil, Kenneth, 1978. "Understanding Organizational Power: Building on the Weberian Legacy." Administrative Science Quarterly 23 (March):65-90. McNemar, Quinn, 1962. Psychological Statistics. New York: Wiley. Mechanic, David, 1962. "Attractiveness as a Source of Power of Lower Participants." Administrative Science Quarterly 7 (June):349-64. Meyer, John W , and Brian Rowan, 1977. "Institutionalized Organizations: Formal Structure as Myth and Ceremony." American Journal of Sociology 83 (September): 340-63. Miller, George, 1967. "Professionals in Bureaucracy: Alienation among Industrial Scientists and Engineers." American Sociological Review 32 (October):755-67. Miller, Jon, 1970. "Social-Psychological Implications of Weber's Model of Bureaucracy: Relations among Expertise, Authority, Control and Legitimacy." Social Forces 49 (September):91-102. 1980. "Access to Interorganizational Networks as a Professional Resource." American Sociological Review 45 (June):479-96.
110
References
and Lincoln Fry, 1973. "Social Relations in Organizations: Further Evidence for the Weberian Model." Social Forces 51 (March):305-19. Sanford Labovitz, and Lincoln Fry, 1975. "Inequities in the Organizational Experiences of Women and Men." Social Forces 54 (December):365-81. James Lincoln, and Jon Olson, 1981. "Rationality and Equity in Professional Networks: Gender and Race as Factors in the Stratification of Interorganizational Systems." American Journal of Sociology 87 (September):308—35. Montagna, Paul D., 1977. Occupations and Society. New York: Wiley. Olson, Jon, and Jon Miller, 1983. "Gender and Interaction in the Workplace," in Helena Z. Lopata and Joseph H. Pleck (eds.), Research in the Interweave of Social Roles: Families and Jobs. Greenwich, Conn.: JAI Press, pp. 35-58. Parcel, Toby L., 1979. "Race, Regional Labor Markets and Earnings." American Sociological Review 44 (April):262-79. Parsons, Talcott, 1950. The Social System. Glencoe, 111.: Free Press. 1966. Societies: Evolutionary and Comparative Perspectives. Englewood Cliffs, N.J.: Prentice-Hall. Pedazhur, Elazar J., 1982. Multiple Regression in Behavioral Research. 2nd ed. New York: Holt, Rinehart and Winston. Perrow, Charles, 1967. "A Framework for Comparative Analysis of Organizations." American Sociological Review 32 (April): 194-108. 1979. Complex Organizations: A Critical Essay, 2nd ed. New York: Scott, Foresman. Perucci, Robert, and Mark Pilisuk, 1970. "Leaders and Ruling Elites: The Interorganizational Bases of Community Power." American Sociological Review 35 (December): 1040-57. Pfeffer, Jeffrey, 1981. "Who Governs," in Oscar Grusky and George A. Miller (eds.), The Sociology of Organizations. New York: Free Press, pp. 228—47. Ransford, H. Edward, and Jon Miller, 1983. "Race, Sex and Feminist Outlooks." American Sociological Review 48 (February):46-59. Ritzer, George, 1977. Working: Conflict and Change. Englewood Cliffs, N.J.: Prentice-Hall. Rogers, Everett M., and Rekha Agarwala-Rogers, 1976. Communication in Organizations. New York: Free Press. Rooney, James F., 1980. "Organizational Success through Program Failure: Skid Row Rescue Missions." Social Forces 58 (March):904-24. Rossi, Alice S., 1965. "Barriers to the Career Choice of Engineering, Medicine and Science among Women," in J. Mattfeld and C. Van Aken (eds.), Women and the Scientific Professions. Cambridge, Mass.: MIT Press, pp. 51—127. Rossi, Peter, Emily Waite, Christine E. Bose, and Richard E. Berk, 1974. "The Seriousness of Crimes: Normative Structure and Individual Differences." American Sociological Review 39 (April):224-37. Siegel, Paul M., 1965. "On the Cost of Being Negro." Sociological Inquiry 35 (Winter):4157. Spangler, Eve, Marsha A. Gordon, and Ronald M. Pipkin, 1978. "Token Women: An Empirical Test of Kanter's Hypothesis." American Journal of Sociology 84 (July): 160-70. Stern, Robert N., Walter R. Gove, and Omer R. Galle, 1976. "Equality for Blacks and Women: An Essay on Relative Progress." Social Science Quarterly 56 (March):664—73. Treas, Judith, 1978. "Differential Achievements: Race, Sex and Jobs." Sociology and Social Research 62 (April):387-400. Treiman, Donald J., and Kermit Terrell, 1975. "Sex and the Process of Status Attainment." American Sociological Review 40 (April): 174-200. Turk, Herman, 1973. "Comparative Urban Structure from an Interorganizational Perspective." Administrative Science Quarterly 18 (March):37—55.
References
111
Warren, Carol A.B., 1981. "New Forms of Social Control: The Myth of Deinstitutionalization." American Behavioral Scientist 25 (July-August):724-40. Weber, Max, 1947. The Theory of Social and Economic Organization. New York: Free Press. 1978. Economy and Society. Berkeley, Calif.: University of California Press. Weick, Karl, 1976. "Educational Organizations as Loosely Coupled Systems." Administrative Science Quarterly 21 (March): 1-20. Weisberg, Sanford, 1980. Applied Linear Regression. New York: Wiley. Willis, Paul, 1981. Learning to Labor. New York: Columbia University Press. Witte, John F., 1980. Democracy, Authority and Alienation in Work: Workers' Participation in an American Corporation. Chicago: University of Chicago Press. Wolf, Wendy C , and Neil D. Fligstein, 1979a. "Sex and Authority in the Workplace." American Sociological Review 44 (April):235—52. 1979b. "Sexual Stratification: Differences in Power in the Work Setting." Social Forces 58 (September):94-107.
Index achieved status, 26 achievement, 8 age, 3, 26; special effects for nonwhite women, 71, 94; special effects for white women, 39, 91-2 ascription, 3, 5, 6, 8—11; see also age, gender, race assistance, see informal networks authority, 3,13, 24-6, 56-9; see also achieved status authority structure, 1, 3, 5, 6; access to, 246,61 autonomy, see job autonomy Barnard, Chester, 87 Benson, Kenneth, 19 Blau, Peter, 88, 98 Brown, Richard Harvey, lOOnl bureaucracy, 17—18, 19; and professionalism, 8-11; and race-gender differences, 8-11; Weber's ideal type, 6, 56-9, 100n3 bureaucratic imperative, 60 bureaucratic leveling effect, 11, 13, 21-2, 32, 38, 92 career expectations: and motivation, 86-7, 89; of men and women, 13—14 career mobility, 13, 21, 87, 90 careers, 86, 96 centrality, see network centrality class structure, 87 client-practitioner homophily, 11, 88; and access to the formal structure, 72-4; and network centrality, 43—6 clients, 6, 14, 43-6; see also juvenile offenders, status offenders coalitions, 20 community activism: effects on access to authority structure, 77-8; effects on centrality, 43, 52-6, 75, 78; measurement, 52-3 community-based treatment, 4, 14, 16, 1920, lOOnl community ties: and access to formal authority structure, 74-7; and network centrality, 6, 43, 46-52; measurement, 47-9 competition, 22, 23, 95, 98 conflict, 81 contact with supervisors, 3, 4, 5, 24, 61; differences by race and gender, 62-6; determinants for nonwhite men and women,
71-2; determinants for white men and women, 68-7lp core-periphery distinctions, 13,21 Dahrendorf, Ralf, 81 decision-making structure, see authority structure deinstitutionalization, 1 OOn 1 demography, 5, 14—15; and white male domination, 21; of client pool, 42-6 discrimination, 7, 10, 11, 12, 91, 92, 94, 99; see also inequity division of labor, 4 domination, 15, 21, 24, 38, 82; see also power, subordination education, see achieved status effectiveness, 105nl evaluation of contact with supervisors, 61; comparison by race and gender, 62—6; determinants for nonwhite men and women, 71; determinants for white men and women, 69-71 exchange theory, 88 experience, see achieved status expertise, 8, 18-19, 56-9 external contacts, see community ties external status system, 20-1; see also class structure formal structure, see authority structure Fox, Mary Frank, 12 gender, 1, 2-3, 5, 6, 8-12; and access to formal authority structure, 62-6; and network centrality, 59—60; and power, 14— 15 Halaby, Charles N. homophily, see client-practitioner homophily income, 13 inequity, 7, 13, 37, 91, 92, 94, 99; and bureaucratic rationality, 11, 21—2, 37, 42; see also discrimination influence, see informal networks informal networks, 1, 3, 4, 6, 100n2; of assistance, 3, 4, 23; of influence, 3, 4, 23, 56-9; of respect, 3, 4, 23; of support, 3, 23; of work contacts, 3, 23; see also informal structure
113
114
Index
informal structure, 5, 6; see also informal networks interaction with co-workers, see informal networks, work contacts interaction with supervisors, see contact with supervisors interorganizational networks, 15-16 interpersonal ties, see informal networks investments, 13, 27 job autonomy, 24, 105nl job type, see achieved status juvenile justice system, 16-17, 19, lOOnl juvenile offenders, 4, 14, 15—16, 44; see also clients, status offenders Kanter, Rosabeth Moss, 11-13, 87, 100n4, 103n6 labor market, 5,21, 87-8, 98; segmentation of, 13, 90 legitimacy, 18, 56-9 leveling effect, see bureaucratic leveling effect March, James G. (and Herbert A. Simon), 9 Miller, Jon, 11-12, 13 motivation, see career expectations negotiation, 6, 38, 93, 98 network centrality, 4, 6; and age, 41; correlates of, 29—32; determinants for nonwhite men, 39—40; determinants for nonwhite women, 41; determinants for white men, 38, 40-1; determinants for white women, 41; measurement, 22-4, 26, 36, 102nl3; see also informal networks occupation, 3; see also achieved status occupational structure, 1; see also labor market organizational coupling, 5, 6, 20, 62, 71—2, 79, 88 organizational networks, see informal networks organizational symbolism, 18 organizational theory, 5, 38, 79, 82, 83-5; and the concept of bureaucracy, 7; functional theory, 101n7; garbage can theory, 19; human capital theory, 2, 60, 69, 92; rationalist theory, 2, 60, 79, 82, 89, 92, 96, 97 participation in decision making, 4, 5, 25, 61; comparisons by race and gender, 626; determinants for nonwhite men and women, 71—2; determinants for white men and women, 69-71; see also authority structure, access to particularism, 42 people-processing organizations, 11, 19, 97
periphery, see core-periphery distinctions Perrow, Charles, 18-19, 42 Pfeffer, Jeffrey, 20-1,86 power, 13—14, 91, 98; and authority, 56; imbalance of, 21; and race and gender, 1415, 21; see also domination, subordination professionalism, 9, 56 race, 1, 2-3, 5, 6, 8-11, 12-13; and access to formal authority structure, 62-6, 6872; and network centrality, 59—60; and power, 14—15; see also ascription racism, see discrimination, inequity rank, see authority, achieved status rationality, 2, 8-11, 37, 42-3; compartmentalized, 12, 42, 100n7; see also organizational theory relative deprivation, 87-8, 90 respect, see informal networks Ritzer, George, 86 Rossi scale of seriousness, 44-6 sexism, see discrimination, inequity social isolation, 24, 37, 91; of white men, 58 social service delivery systems, 17, 20—1, 66, 86, 98 social service organizations, 6, 14, 20 sociometric analysis, 3, 102nl2 statistical interaction: tests for, 32-3, 68-72, 102nn3-5 status offenders, 15-17, 44, lOOnl; and delinquents, 18; see also clients, juvenile offenders stratification, 5 subordination, 15; see also domination, power superimposition, 81 support, see informal networks surveillance, 64, 105, 106nl technology, 18 training, see achieved status treatment philosophy, incompatibilities in, 18-19 United States Justice Department, 4, 16 universalism, 18, 42, 66, 99 Weber, Max, 8, 100nnl-3 willingness to cooperate, 87; differences by race and gender, 87-8 Willis, Paul, 88 Witte, JohnF., 87 work contacts, see informal networks work place, see achieved status work strain, 105nl youth: and ability to deal with clients, 26; and social attractiveness of nonwhite women, 94; and social attractiveness of white women, 91—2
