Showing papers in "Social Forces in 1971"

Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States

Abstract: Summer 1994 272 Professor Hirschman wrote this book while in residence at the Center for Advanced Studies in the Behavioral Sciences at Stanford University. Reprinted by permission of the publishers from Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States by Albert O. Hirschman, Cambridge, Mass.: Harvard University Press, © 1970 by the President and Fellows of Harvard College.

The Logic of Naturalistic Inquiry

Abstract: A version of the research act, termed naturalistic behaviorism, is presented and compared to existing formulations of the method. A sensitizing framework for organizing naturalistic studies is presented. Special attention is given the problems of sampling, measurement, and causal analysis. Examples from an ongoing study of early childhood socialization are presented as tentative solutions to the sampling and analysis issues. Central weight is given to introspective-investigator accounts of social process. Existing formulations of naturalism as a distinct approach to empirical inquiry in the social sciences suffer from several overriding flaws. On the one hand, naturalistic theorists and practitioners have seldom been in agreement on what they mean by the method. For some (Catton, 1966), it is seen as rigorous positivism. For others (Matza, 1969), it is viewed as humanism in disguise. For still others (Barker, 1968; Hutt and Hutt, 1970; Willems and Rausch, 1969; Wright, 1967), it is compared to ecological psychology and/or ethology-a bare kind of behaviorism that studies people in their natural habitats. Here the naturalist, like the ethologist, makes little effort to record, probe, and study such socialpsychological processes as attitudes and definitions of the situation. There are those (Lofland, 1971) who view naturalism as a deep commitment to collect rich, often atheoretical ethnographic specimens of human behavior. These statements also suffer from a failure to specify the empirical phenomena to which the method is directed (e.g., if one observes behavior, what kinds of behavior?). Nor has there been any systematic attention given such traditional and perduring methodological problems as measurement, sampling, validity, reliability, and causal analysis. The basic unit of naturalistic analysis has never been clarified and the role of the naturalistic observer in his studies remains clouded. This conceptual diversity has led many to take a sceptical, if not irreverent, view of the naturalistic approach, viewing it as soft science or journalism. Perhaps the basic deficiency of prior naturalistic formulations has been the absence of a more general theoretical perspective that would integrate all phases of the sociological act. With few exceptions the dominant scientific paradigm has been imported from physics, chemistry, or biology.' In this article I offer a view of naturalism that takes as its point of departure the social behaviorism of Mead (1934; 1938) and the symbolic interactionism of Blumer (1969). I call this version of the research act naturalistic behaviorism and mean by the term the studied commitment to actively enter the worlds of native people and to render those worlds understandable from the standpoint of a theory that is grounded in the behaviors, languages, definitions, attitudes, and feelings of those studied. Naturalistic behaviorism attempts a wedding of the covert, private features of the social act with its public, behaviorally observable counterparts. It thus works back and forth between word and deed, definition and act. Naturalistic behaviorism aims for viable social theory, it takes rich ethnographic descriptions only as a point of departure. This version of behaviorism recognizes that humans have social selves and as such act in ways that reflect their unfolding definitions of the situation. The naturalist is thus obliged to enter people's minds, if only through retrospective accounts I am grateful to Philip Bechtel, Herbert Blumer, H. M. Blalock, Donald Dixon, John Lofland, Peter Manning, Anselm Strauss, and Clark McPhail for their comments and criticisms on earlier versions of this article. l The major exceptions here are Becker (1970a), Lofland (1971), and Schatzman and Strauss (in press).

Critique of Ordinal Variables

Abstract: The formal properties of ordinal variables are examined in relation to the requirements for (1) empirically testing models formulated in terms of interval variables and (2) stating substantive propositions within a theory. It is found that, strictly speaking, ordinal variables cannot be used in a rigorous way in either context.. This implies, on the one hand, that the measurement problem is the critical issue in scientifically oriented sociology, since in that case the emphasis must be on rigorous theory and precise methodology. On the other hand, the use of ordinal variables in substantively oriented sociological research can be viewed as having important heuristic and metaphorical functions, provided the limitations on taking the results literally are observed. A recurrent methodological theme in sociology has been the effort to formulate theoretical ideas in mathematical terms. Although some attention has been given to the mathematical description of structural patterns (e.g., Davis and Leinhardt, 1968; Oeser and Harary, 1962, 1964), and discrete phenomena (Coleman, 1964a; 1964b), the major emphasis recently has been on formulating substantive models employing systems of real variables (e.g., Berger et al., 1966; Blalock, 1969a; Boudon, 1965; Coleman, 1968; Stinchcombe, 1968). The virtues of mathematical methods in theoretical work are well known: clarity, precision, capacity to analyze complex relations, and the like. However, the power and elegance of mathematical methods also impose stringent requirements on the empirical measurement of variables entering into theoretical models, and the measurement problems become particularly acute in the case of models formulated in terms of real variables. As Blalock (1965; 1969c) has emphasized, uncontrolled or inadequately assessed random measurement error can seriously weaken conclusions concerning the fit between a model and the data, and when undetermined measurement error is nonrandom, the situation is nearly hopeless. The purpose of this paper is to show, first, that the level of measurement is no less important: when the variables in a model are conceptualized as real numbers, measurement must be at least at the interval level, and when only ordinal measurements are possible, only very weak conclusions, if any, can be drawn from the data relative to a proposed model. Second, it will be shown that the formal properties of strictly ordinal variables are too weak to permit their use in the formulation of theoretical propositions about substantive phenomena. In the past two decades, apparently only Simon (1957:Part II) has attempted to come to grips in a serious way with the limitations of ordinal measurement. Drawing on econometric methods, he has shown that under certain circumstances even with ordinal measurement some inferences can be made concerning the behavior of systems expressed in terms of differential and difference equations, but these inferences are relatively weak and require strong faith in the causal structure posited in the model. More frequently in sociological research, the causal structure itself is uncertain, and what one is seeking is evidence concerning the usefulness of some particular set of assumptions about the causal structure. This, for example, is the paradigm situation in path analysis studies. Other authors (Labovitz, 1967, 1970; Labovitz and Lubeck, 1969:4; Gold, 1969:43; Boyle, 1970) have questioned the importance of the level of actual measurement and have argued for the use of intervallevel statistical techniques even when the data are clearly at best ordinal. While this position may be somewhat defensible with univariate * I wish to thank Morris F. Friedell and Bruce C. Straits for discussing the argument with me and for their useful comments on earlier versions of the manuscript.

Behavior and Adaptation in Late Life.

Abstract: • Dr. Busse, a well-known geriatrician, psychiatrist, and author, is no doubt one of the best qualified pioneers in gerontological research. As chief editor-together with Dr. Pfeiffer, of Duke University's Center for the Study of Aging-he achieves a rather thorough gathering of papers dealing with the psychological and socioeconomic aspects of the aging process. The chapters are written by members of the Duke faculty, with three exceptions: Carl Eisdorfer, Ethel Shanas, and the late Alan Goldfarb. Some chapters are of a high level, especially those concerned with psychology, physiology, and psychiatry. Others, treating such subjects as retirement, public policy, geriatric nursing, housing, and institutional care, are useful and informative. Included, no doubt for completeness, are studies on sexual behavior of older persons and attitudes held by the elderly toward death. I read with interest the chapters on \"Brain and Time,\" by F. S. Vogel, \"Functional Psychiatric Disorders in Old Age,\" by Busse and Pfeiffer, \"Health Experience in The Elderly,\" by E. H. Estes, Jr., and \"Organic Brain Syndrome,\" by H. S. Wang. The chapter, \"Theories of Aging,\" by Busse, however, though it is both informative and interesting, seems out of place in this particular book. Drs. Busse and Pfeiffer discuss competently the biological, social and psychological factors that de-

The Indiscriminate State of Social Class Measurement

Abstract: Two frequently used composite indices of social class, the Hollingshead Two Factor index and the Duncan SEI, are critically analyzed, both conceptually and methodologically, using data secured from two national samples. Although both indices are based on occupation, comparison of their scores shows several types of work are differentially evaluated. The findings suggest that the Duncan SEI suffers from weaknesses which undermine its usefulness, while the Hollingshead index requires updating. Thus there is a need for a valid index based on the theoretical distinction between class and status, and adapted to current structural realities. Although social class is often a critical factor in sociological research, either as a major analytic variable or as a principal control, there is no measure of class generally agreed upon or utilized by American sociologists. An informal survey of articles from 1963 through 1968 in the American Sociological Review, for example, reveals that many researchers evaluated socioeconomic distinctions by a single blue-collar/ white-collar, or middle-class/ working-class dichotomy. Others made various groupings of census occupation categories. Another type of analysis developed some contrived index, often combining income, rent, education or occupational levels on an ad hoc basis. The two most frequently utilized composite indices were (1) the Hollingshead Two Factor Index, which appeared six times in its original form and made five additional appearances in a modified form, and (2) the Duncan Socio-Economic Index (SEI), which turned up ten times. This variety is at least partly a consequence of the fuzzy conceptualization of class which has plagued the discipline. The Marxian view that class position is a function of relation to the means of production has been rejected as * Revision of papers read at the annual meeting of the Ohio Valley Sociological Society, South Bend, Indiana, April 1967, and the annual meeting of the American Sociological Association, Boston, Massachusetts, August 1968. Project supported by RSA RD-1726-P-68-63. This content downloaded from 157.55.39.151 on Thu, 07 Jul 2016 05:01:20 UTC All use subject to http://about.jstor.org/terms

The Development of Class Awareness in Children

Abstract: Class awareness on three dimensions (cognitive, behavioral, and evaluative) was tested among 216 children in grades one, four, and six of a public school system. On the cognitive test (perception of social differences), age and sex were found to be related to performance. On the behavioral test (recognition of behavioral correlates to cognitive cues), age, social class, and IQ are related to performance. The evaluative test (attachment of evaluations on the basis of cognitive cues), shows no significant results. Possible reasons for this are discussed. With the myriad of literature in social stratification, it is surprising that so little work has been done on the development of class awareness. Lasswell (1958, 1961; Lasswell and Parshall, 1961) suggested that most adults hold to popular conceptions of social classes which are comprised of sets of stereotypes, and he has found evidence to support this. There is also research suggesting that these stereotypes are well developed by the time a person reaches adolescence (Centers, 1950; Congatton, 1952; Himmelweit et al., 1952; Montague, 1954). However, little work has been done investigating the developmental process of acquiring the stereotypes-that is, little work has been done on the development of class awareness. Estvan (1952, 1958, 1965, 1966; Estvan and Estvan, 1959) has done extensive studies of what he calls children's social perception. He is interested in seeing how children view their social world, and in particular what they say about "life situations" presenting contrasts in rural and urban environment; in child and adult activities; and in upperand lowersocioeconomic background. With regard to social-status perception, Estvan finds increasing age to be related to increasing awareness of status symbols. He finds intelligence and rural-urban background of the child unrelated to perception of social status. There seem to be only two studies entirely focused on children's perceptions of social class. The first and most comprehensive is a work by Celia Stendler (1949). Stendler sampled children at four grade levels: first, fourth, sixth, and eighth. She stratified her sample by class position and sex, and tested class awareness in the children by several procedures. The major test was a picture test containing four sets of pictures illustrating homes, recreation, clothes, and jobs at three socioeconomic levels. Stendler found that the first-grade child is "not yet dealing with the problem of social class," but that as the child grows older he becomes increasingly able to rate the pictures as do adult raters, and able to give sophisticated reasons for his ratings. She also found that social class and intelligence are related to the ability to correctly rate the pictures. The only other study in this area was done in 1959 by Gustav Jahoda. Jahoda's testing procedure involved making "socially congruous" pictures by assembling puzzle-like drawings, with lower-middle class drawings fitting together as well as lower-lower or middle-middle drawings. Unlike Stendler, Jahoda found that even at the first-grade level children were perceiving class differences. He also partialed out the effects of intelligence from social class and concluded that the relation between performance scores and social class of the child was largely the result of intelligence. The general problem of class awareness to be investigated in this study involves three dimensions: a cognitive dimension; a behavioral dimension; and an evaluative dimension. The cognitive component of social class awareness involves the simple perception or recognition of social differences. The behavioral dimension involves recognition that behavioral * This research was supported by a National Science Foundation grant, GS2520.

The Organizational Substructure of Disorderly Politics

Abstract: According to the theory of mass society, organizational membership drags people into routine politics, while discouraging participation in direct action movements. The thesis of this paper is that this theory needs severe modification. First, for an important class of organizations-organizations with goals unincorporated by the larger society or containing members with interests or values that are unincorporated-membership increases rather than inhibits participation. This is indicated both by a theoretical analysis of mass theory and by data gathered from participants in the Negro sit-in movement of the early 1960s. Second, this data indicates that the existence of many such organizations is necessary if a direct action movement is to be strong. Thus, not only routine, but disorderly politics too requires an organizational substructure to create and sustain it. Third, the data indicates that the substructures of these two types of politics are not totally distinct, but rather share certain organizations in common. Some organizations seem to generate participation in both types simultaneously, while others may, at one point in their history, inhibit participation, but at another, when a movement has gathered strength, generate it. In short, theory and data suggest the need for a major revision of mass theory which takes the variable of unincorporation into account. Most formal democracies have experienced mass movements that go outside the routine channels of political action to obtain their aims. These movements often engage in illegal or semi-illegal acts and occasionally, intentionally or unintentionally, generate violence. Because they bypass normal channels, they are generally termed direct action movements. Because they frequently give rise to disorder, the kind of politics they involve might best be termed disorderly politics. The cause of such movements has not yet been resolved. One of the more common explanations for their rise argues in terms of "hard" deprivations-deprivations of income, status, power-and facilitating conditions. Thus, for example, the trade union movement, which involved considerable disorder in its early phases, has been explained in terms of such deprivations as long hours, low prestige, and the like, and in terms of such facilitating conditions as the growth of large plants, plants which broke the personal ties between worker and manager, permitted the concentration of activists, etc. This perspective can be termed the class theory of direct action movements. Recently, however, sociologists and other writers have stressed an alternative theory, the theory of massification.1 This theory points to organizational membership as its central explanatory variable. A rich organizational life, it is argued, drags people into routine politics while simultaneously reducing their alienation from the system, leading in this way to restrained "rational" politics. An impoverished organizational life, on the other hand, leads to alienation and a conse* We would like to thank Charles Perrow for comments on an earlier draft of this paper. As he disagreed with parts of our analysis, however, he should not be held responsible for any errors it may contain. 1 The writers and works generally cited as being in this tradition are Arendt (1958), Kornhauser (1959), Nisbet (1953), and Selznick (1960). It is not clear to us that all these writers are saying precisely the same things. We have had to choose among them, therefore, in our exposition of the theory. As Kornhauser's statement is the fullest and as it consists of a synthesis of the writings of the others (insofar as this is possible), our treatment shall rely primarily on his analysis.

The Lost Letter Technique as a Measure of Social Variables: Some Exploratory Findings

Abstract: Milgram's lost letter technique was used to test the hypothesis that large-city dwellers are less responsive to the needs of their fellow man than are small-town dwellers. Two experiments are reported in which lost letters were dropped in major cities, medium size communities and very small towns. In the first experiment the letters were stamped, but in the second experiment they were unstamped. Significant differences were found but they did not support the hypothesis. The results suggested the presence of regional differences in return rates and a possible source of bias was identified. Milgram, Mann, and Harter (1965) introduced an imaginative research tool which they labeled the lost letter technique. The lost letter technique consists of placing a large number of addressed, stamped, sealed, but unmailed letters in public places. When a person discovers one of these letters it appears to have been lost. When confronted with a lost letter, the finder may chose to ignore it, mail it, or destroy it. In the initial use of this technique, Milgram et al. varied the name of the organization to which the letter was addressed and were thereby able to obtain specific return rates for various organizations. Presumably, the individual return rates reflected the popularity or public acceptance of the organizations. Milgram et al. demonstrated that community attitudes toward various organizations could be measured with the lost letter technique. Similarly, Milgram (1969), and Shotland, Berger, and Forsythe (1970) demonstrated that the outcome of an election could be predicted by the return rates of lost letters addressed to campaign organizations. It would appear that the usefulness of the lost letter technique is not limited to the assessment of community attitudes. The present paper reports two attempts to expand the lost letter technique into other areas of social research. THE RESEARCH PROBLEM The mass communications media tend to characterize the urban dweller as isolated, selfcentered, and unconcerned with the needs of his fellow man. The widely publicized murder of Kitty Genovese on the streets of New York is often used as an illustration of the callousness and indifference of the urban dweller. In marked contrast, the small-town dweller is often characterized as friendly, helpful and socially responsible. The mass media appear to enjoy reporting stories of the humanitarian efforts of small towns in helping their members in need. If these characterizations are correct, if smalltown dwellers are indeed more concerned about their fellow man, then one could hypothesize that more lost letters would be returned from small communities than from large urban areas.

On the Logic of Post-Factum Explanations: The Hypothesis of Lower-Class Frustration As the Cause of Leftist Radicalism

Abstract: The common theme underlying most theories on lower-class leftist extremism views this orientation as dependent on frustration with life situation. Employing data collected in lower-class slums of Santiago, Chile, five theories bearing on this notion are tested. None is supported. Leftist radicalism, however, is associated with imputation of responsibilty for frustrations to the social structure. Results support a definition of radicalism as a complex orientation requiring antecedent cognitive variables for its emergence. The popularity of the frustration-radicalism hypothesis is interpreted as a partial result of the post-factum self-legitimation of successful revolutionary movements. The study of working-class politics has been permeated by the idea, expressed in a thousand different forms, that radicalism of the left arises from these sectors as a result of unbearable frustration with their position in the socioeconomic structure. A brief survey of currently available theories for the prediction of leftist radicalism reveals frustration with life conditions as the dominant common theme underlying most hypotheses. Though other variables may be posited as contributory or intervening, the process of working-class radicalization is made ultimately dependent on increasing discontent by its members with their social and economic situation. In classic Marxist theory, revolutionary activism hinges, in an immediate sense, on the acquisition of "consciousness" by the proletariat. Yet, the crucial process underlying the increasing politicization and increasing intensity of proletarian struggles against the bourgeoisie is the inability of the capitalistic order to prevent pauperization of the masses and evermore exploitative arrangements of production, which inevitably generate frustration and discontent (Marx, 1939, 1963, 1967; Marx and Engels, 1955). Intra-bourgeois competition and the process of concentration of capital lead to an ever-widening gap between the economic overabundance of the diminishing few and the abysmal misery among the growing many. The frustration of the latter, the mounting rage among those who have nothing and less to lose constitutes the basic force of revolutionary change in a capitalist order. At the most general level, this view has been embodied in the hypothesis, unanimously accepted by political sociologists since Marx, of an inverse correlation between reception of socioeconomic rewards from the existing social order and tendencies toward revolutionary extremism. Such notion is certainly present in Max Weber's (1958; 1965) sociology of power, as well as in successive Marxian and neoMarxian formulations of a theory of classes (Dahrendorf, 1965; Mills, 1970; Moore, 1969). In contemporary sociology, the notion is usually rephrased as predicting a general negative association between socioeconomic status and leftist radicalism. Status is directly an indicator of differential socioeconomic rewards and indirectly, so the hypothesis assumes, of different levels of satisfaction and, hence, commitment to the existing social order. Lipset (1963:129) * Revised version of a paper presented at the annual meeting of the Rural Sociology Society, Washington, D. C., 1970. The data on which this paper is based were collected under the auspices of the Midwestern Universities Consortium for International Activities (MUCIA). Partial support for statistical analysis was provided by the Graduate Research Board of the University of Illinois. The author is grateful to Professors Archibald 0. Haller, Donald J. Treiman, and William Sewell for most valuable criticisms and suggestions. Responsibility for the shortcomings of the paper rests with the author.