<<

Comparative , 1950-1963

by

ROBERT M. MARSH

1. Introduction

I.i. Definition, Scope and Rationale of the Field Comparative sociology may be defined as that field which is concerned with the systematic and explicit comparison of social phenomena in two or more societies. Cross-societal (cross-cultural, cross-national) comparison is the essential ingredient; intra-societal comparisons may or may not be made concurrently with cross-societal comparisons. Studies which are clearly ex- cluded from comparative sociology as defined are those which make intra- societal comparisons-e.g., between middle and working class voting patterns in one society-without also making cross-societal comparisons. Even with this stipulation comparative sociology is a broadly inclusive field indeed. It contains a great variety of methodological strategies-&dquo;holistic,&dquo; descriptive analyses as well as studies which rigorously test theoretically-derived, or at least theoretically relevant propositions and hypotheses. Comparisons may be of total societies [940-976], though more usually they are of subsystems of different societies, such as socialization practices or polity. Moreover, as

. here conceived, comparative sociology includes both social psychological and social anthropological studies, providing, of course, that they are genuinely cross-societal and not limited to data from one society. The rationale of this field needs to be stated clearly, for there are skeptics who would not dignify cross-societal comparisons as constituting a field in their own right. The skeptic remarks, &dquo;I am interested in , attitude formation, etc., and I am quite indifferent as to the societies from which my data are taken.&dquo; The reply, and the rationale of comparative sociology, is that, as an historical fact, what passes today as &dquo;&dquo;-be it attitude theory, organizational theory, or whatever-has been developed primarily in one, rather small corner of the world and may therefore be highly limited as a general explanatory scheme. Many socio- logical propositions are, of course, stated as universals. That is, they are 5 6

stated as though the relationships and generalizations hold for all societies, for all social systems, even for all social action. But such propositions have rarely been tested outside of modern United States, or Western industrial society. The fundamental rationale of comparative sociology is, then, the need to universalize sociological theory. It is sometimes claimed that complex societies contain within themselves enormous and sufficient variation for most comparative purposes, and that therefore cross-societal comparisons are unnecessary or extraneous. Against this view it may be pointed out that Murdock has shown there is only a small degree of overlap in cultural elements as between European and non-European societies [54]. Almost 50 per cent of the cultural items (con- cerning economy, family, government, etc.) contained in a random sample of non-European societies belong in cultural categories which were com- pletely unrepresented in any of 10 selected European societies [54]. This suggests that while complex societies do contain much internal variation, the degree of variation is nevertheless severely limited. The same conclusion appears justified on the basis of several analyses of variance in comparative studies. In at least four different studies, cross-societal differences in the phenomena being studied-e.g., achievement motivation, value orientations, and other social psychological variables-were shown to be statistically significant relative to intra-societal differences, as between social classes, etc. ~422, 432, 535, 699]. In sum, there is considerable evidence that intra-societal variations, even those within complex societies represent a much narrower range than do cross-societal variations. When sociologists are asked to foretell the likely outcome of submitting existing theory and hypotheses to extensive cross-societal tests, the reply is often either: (1) Modern sociology may be a product of the West, but its theory will be shown to be universally applicable. One needs only to specify the minor, culturally idiosyncratic &dquo;initial conditions&dquo; of different societies in order to apply the theory successfully. Or, (2) Modern sociological theory is so intimately bound to the distinctive characteristics of modern Western industrial society, that it will totally fail to predict and explain social phenomena in other societies; it must be thoroughly reconstituted. On the basis of the hundreds of cross-societal comparative studies published between 1950 and 1963 and reviewed here, both these extreme positions may safely be repudiated. (1) is refuted because some Western-based propositions have had to be sharply modified or have been disconfirmed altogether. (2) is refuted because some Western-based propositions have been shown to &dquo;hold&dquo; for other societies. The rationale of comparative sociology can now be re- stated in a way that gives full recognition to its high calling. Comparative sociology has tlae task of progressively specifying which theories, propositions, etc. hold for all societies, whicla for only certain classes or types of societies, and which for only individual societies. 7

In short, comparative sociology is a legitimate field in its own right. The cross-societal analysis it carries out is a necessary extension of intra-societal analysis. It is a necessary step, but a step which many sociologists, social psychologists, and even some anthropologists (!) abrogate. The purpose of this trend report is to codify the experiences of those who, since 1950, have taken this step. In order that the trends of the period 1950-1963 may be seen in perspective, my report appears under the following three main headings: (1) Comparative Sociology in the Past, (2) Comparative Sociology in the Present 1950-1963, and (3) Comparative Sociology in the Future. In ( 1 ) I shall briefly contrast the &dquo;comparative method&dquo; of the nineteenth-century with present-day com- parative sociology. In (2), the main body of the report, the emphasis will be on developments in the following fields: A. Kinship, family and marriage B. Polity and bureaucracy ’ C. and mobility D. Social Psychology: socialization, personality, national character E. Conformity and F. Cultural patterns and value orientations G. Ecology and - H. Economic development and social change Some of these fields have, of course, been reviewed in earlier issues of Current Sociology. There, cross-societal comparative studies were reviewed along with individual society studies. A consequence of this mode of presen- tation is that, since in all these fields individual society studies far outnumber cross-societal comparisons, the focus of attention could not be centered upon the distinctive problems and contributions of cross-societal analysis. For this reason, the present review adopts a different strategy of presentation. Its focus is exclusively on cross-societal comparative studies; at the same time it does not limit coverage to any one sub-field, such as kinship and the family. It is hoped that by examining cross-societal comparative studies in a number of sub-fields of sociology, social psychology and anthropology, concentration will be upon the distinctive problems and contributions of the field of com- parative sociology. Finallv, after reviewing comparative studies in the period 1950-1963, I shall, in (3), sketch a strategy by which some of the limitations of present-day comparative sociology may be overcome, and the somewhat scattered efforts of comparativists may become more integrated and cumulative.

Lii. Comparative Sociology in the Past

Present-day comparative sociology is anything but a wholesale revival of the late and little-lamented &dquo;comparative method&dquo; of the nineteenth-century 8

[1-5, 8-11]. In the hands of Comte, Bachofen, Maine, McLennan, Spencer, Frazer, Westermarck and others, that method was relatively highly unified in terms of the &dquo;theory&dquo; of unilinear social evolution. Today’s comparative sociology, on the other hand, includes a heterogeneous assortment of partly conflicting theories, frames of reference and assumptions: we have no widely accepted general theory to match nineteenth-century evolutionism. A second difference is that the general level of methodological sophistication is higher in present-day comparative analysis. In the nineteenth-century comparative method, data from primitive, ancient and modern societies were fitted to the procrustean bed of one or another &dquo;stage&dquo; of evolution. The methodology was all too often argument by illustration. Today, on the other hand, com- parativists are more likely to follow accepted canons of sampling, measure- ment, &dquo;deviant case&dquo; analysis, and the like. The reaction against evolutionism in the early part of the twentieth-century tended to throw out the baby with the bath. Cross-societal research into problems of co-variation was superseded by intensive analyses of particular societies and by discussions of general concepts. The revival in recent years of comparative sociology can be seen, therefore, as a counter-reaction in which some of the concerns of nineteenth-century studies are again being taken seriously. White, Service, Parsons, and others have made important contributions to evolutionary comparative theory and analysis [6, 7, 927, 937] Other comparativists have begun to apply scale analysis, factor analysis and other modern statistical and mathematical techniques to the study of evolu- tionary social change [18, 19, 27, 38, 69, 105, 119, 134, 174, 227, 266, 891, 905, 924, 939]. What is of paramount importance is that these and other recent developments in comparative sociology must be evaluated in their own terms, not dismissed out of hand as a wholesale return to the nineteenth- century comparative method.

II. Comparative Sociology in the Present, 1950-1963

ILi. Kinship, Family and Marriage

The continuity of research interest from nineteenth-century to contemporary comparative sociology is well indicated in the problem of the relative primitiveness of matrilineal and patrilineal institutions. Aberle has shown [105] on the basis of Murdock’s 1957 World Ethnographic sample of 565 societies [101] that matrilineal descent is clearly not a characteristic of one general stage of cultural evolution. Instead, its incidence is explained in terms of specific evolutionary adaptations. Driver and Massey (119] and Blalock [16] have dealt with one aspect of this same problem, namely, is 9

there a parallel evolutionary sequence in societies with matricentered insti- tutions, those with patricentered and those with bicentered institutions ? The hypothesized sequence is : (1) change in sexual division of labor in subsistence pursuits, (2) postnuptial residence, (3) land tenure, (4) descent and (5) kinship terminology. Methodologically, Driver and Massey show how synchronic data from over 200 North American Indian tribes can be interpreted as though they were diachronic, thereby testing hypotheses of evolutionary change. The data support the hypothesized multilinear evolutionary sequence. Driver and Massey [119] and Driver [27] have also made cross-societal tests of the relative predictive power of evolutionary, functional and historical theories. The conditions under which matrilineal, patrilineal and other forms of descent systems occur and vary have been specified by other comparativists [142, 184]. The influence of economic variables on marital residence, authority and interaction patterns within extended families has been highlighted by Richards [181] and by Heath [150]. A second broad range of problems in the kinship area concerns the variable kinship solidarity, especially when the degree of solidarity is measured among extended kin, rather than restricted to solidarity within the nuclear family. When degree of kinship cohesiveness or solidarity is related cross-societally to the overall degree of societal complexity, there is at least some evidence which supports the hypothesis of a curvilinear relationship [128, 159, 174]. That is, clan organization, indicative of highly developed unilineal kinship systems and high degrees of kinship solidarity, tends to be absent in both the simplest and the most complex (industrialized) societies. In these two types of societies kinship ties between related nuclear families tend to be weaker. It seems clear, furthermore, that the findings of recent studies in the United States and Britain [145, 197], to the effect that extended kinship ties are still viable, do not contradict this curvilinear relationship. The degree of kinship solidarity in these modern urban areas is considerably lower than that observed in &dquo;middle range&dquo; societies with unilineal clans, etc. Variations in demographic aspects of &dquo;family life cycles,&dquo; as between India and the United States, also explain why industrialized societies have less kinship solidarity [115]. The measurement of kinship solidarity is a needed area of research, and Bardis’s 16-item Familism scale is a step in the right direction [109]. A third focus in comparative studies of kinship and the family has to do with world changes in family patterns during the last century or the last few decades. Goode [140] has brought together a considerable body of data from the modern West, Arabic Islam, Sub-Saharan Africa, India, China and Japan. He argues that recent changes in the family, in contrast to earlier changes, are not only more momentous, more universal and perhaps more rapid, but, most importantly, are moving all societies toward a single, similar family type. Regardless of the family type from which change takes place, 10

it is the conjugal type toward which family systems everywhere are now tending. There is an important corollary of this proposition. When societies undergo modern changes in family patterns, the base lines from which change takes place may vary greatly from society to society. Therefore, the societies may appear to be moving in different directions, even though in fact they are converging toward the conjugal type. For example, in both Arabic Islam and Tokugawa Japan, divorce rates were already high (by United States and Western standards) prior to industrialization and modern social changes. With these changes, divorce rates in Arabic Islam and Japan have declined, rather than increased, as in the West at a similar &dquo;stage.&dquo; Some analysts have taken this fact as disconfirmation of Western-based theory. To the extent, however, that divorce rates in the West, in Arabic Islam and in Japan &dquo;level off&dquo; at similar (and relatively high) levels-as is predicted by Goode- the overall trends will have been trends of convergence, and at least a modified version of Western-based theory will have been confirmed. In a fifth area of comparative family analysis-marital dissolution and its causes-traditional assumptions based on individual Western industrial socie- ties have been questioned. Comparative evidence suggests that divorce rates in the United States and other industrial societies are not uniquely high in world history, and that in no society can high divorce rates necessarily be taken as indicators of the disorganization of that society’s family system [140, 172]. For example, in many societies where divorce has been institu- tionalized for centuries, the family system qua system has remained intact. Anthropological and sociological studies have both dealt with causes of differentials in divorce rates, the former stressing some aspect of kinship structure, the latter the homogeneous versus heterogeneous social attributes of those who marry. The analytical parallels between these sets of causes were inadequately noticed, since anthropologically-cited causes tended to be specific to simpler societies, and sociologically-cited causes appeared to be specific to more complex, differentiated societies. It remained for Ackerman [106] to integrate these explanations of divorce rates in terms of hypotheses concerning the conjunctive versus disjunctive structural affiliations of spouses. Ackerman’s theory of affiliations as a structural determinant of differential divorce rates is, indeed, a model of the kind of leverage cross-societal com- parative analysis can provide. Another major proposition in this area is Goode’s : divorce rates are inversely related to social class. Goode has not only confirmed this proposition in a number of nations, but has also developed it from a synchronic into a diachronic statement in which the relation between social class and divorce rate is specified according to the industrialization phase of the society [138, 140]. Radcliffe-Brown had stated that the tensions which arise between parents and children, given the unequal distribution of authority during early so- cialization, tend to draw closer together the grandparent and grandchild. In 11

recent years, by means of some elegant comparative analysis, this proposition has been modified by Nadel and Apple [5, 107]. In a final body of comparative family research, problems of the Oedipal situation and of incest have received attention. Somewhat competing expla- nations of these phenomena-Freudian theory, learning theory, social struc- tural theory and cultural theory-have been sifted and better integrated [17, 130, 143].

Il.ii. Polity and Bureaucracy

In their concern with the political organization of primitive societies and the evolution of polity in more differentiated societies, modern comparativists have kept alive some of the interests of the nineteenth-century evolutionists [206, 209, 210, 215, 216, 217, 226, 237-240, 250, 261, 272, 275, 278, 284, 294, 297]. Yet here, as in kinship studies, new problems, methods and conclusions have come to the fore. One of these is the shift from formal to functional definitions of &dquo;government&dquo; [272]. Instead of asking &dquo;What ought a government to look like ?&dquo; and concluding that primitive societies lack government, since they have no collectivities which resemble Western governmental collectivities (legislatures, formal councils, etc.), the question has become, &dquo;What does the polity or ’government’ do in any society ?&dquo; On this basis, all societies can be said to have a polity, i.e., protection of members of the community &dquo;against lawlessness within and enemies without,&dquo; and decision-making on behalf of the community in matters which concern all [272, p. 16). Much has been written, by British social anthropologists in Africa and others, about &dquo;stateless, segmentary lineage&dquo; societies [187, 275]. These are societies which lack centralized authority and whose major social groups are segmentary unilineal lineages. The question raised recently by LeVine [264] is whether this class of societies is sufficiently similar so that one can predict their contemporary political behavior, both under colonial rule and after independence. By following the history of two African segmentary societies, the Nuer and the Gusii, LeVine asserts a negative answer to the question. Specifically, &dquo;stateless, segmentary&dquo; societies may vary significantly as to authority values, and socialization practices. These variations give rise to different responses to colonial rule and political independence. Moving more in the direction of modern, complex societies, there have been a number of attempts by comparativists to formulate general models of political systems and to analyze the governments of modern nations [202- 204, 208, 214, 222, 223, 230, 238, 244, 252, 257, 271, 295, 298, 308]. The structure of political parties, their membership and partisan voters have also been analyzed [211, 219, 221, 236, 241, 280, 281, 301]. The problem of the social, economic and other requisites of political 12

democracy in the modern world is one with which sociologists have been more concerned than anthropologists. In recent comparative studies the same substantive findings have tended to recur: political democracy (Almond and Coleman’s &dquo;modern political system&dquo;) is associated with a relatively high degree of economic development; democracy is likely to be present only in the more economically developed societies [203, 227, 256, 266, 268]. In these studies one finds an increasingly sophisticated grasp of methodology. The methodological trend is from descriptive studies within one region of the world, e.g., the Middle East, to correlation and regression analysis, based upon a world-wide sample of societies. The best study to date in this area is Cutright’s [227]. Cutright develops more objective measures of democratic political development than Lipset. Lipset had concluded that economic and political development are positively correlated. Cutright re-analyzes some of Lipset’s own data and shows that this conclusion does not hold up as well on a world-wide basis as it does within regions. Societies Lipset classifies as &dquo;European dictatorships&dquo; have greater wealth, industrialization, and education than have Lipset’s &dquo;Latin American democracies.&dquo; This is counter to Lipset’s proposition. In other words, a &dquo;regional&dquo; factor is important, as well as level of economic develop- lent, in determining degree of political democracy. (The same point is made by Coleman [203] in his comparison of Latin American and Asian-African nations. On five of eleven indicators of economic development, Latin American countries are higher in economic development than Asian-African countries, regardless of degree of democratic-competitiveness in their political system.) Cutright finds that a communications development index (based on newspaper consumption, telephones, and domestic mail per capita) is a better predictor of political development than is economic development, though, of course, communications and economic development are themselves highly correlated. With data on 77 nations, Cutright constructs a scattergram and a prediction equation for the regression of political development on com- munications development. This enables him to emphasize what Lipset did not emphasize, namely, the tendency for many countries to be either over- developed politically relative to level of communications development, or underdeveloped politically relative to level of communications development. This moves Cutright to raise such questions as, Do both these kinds of &dquo;deviations&dquo; make for disequilibrium and strain within the nation ? When a given nation is either more of less democratic than its level of communica- tions development &dquo;warrants,&dquo; which is more likely : the &dquo;regression&dquo; of its political institutions toward less democratic forms, the advance of its com- munications level, or some other type of &dquo;strain to consistency&dquo; ?y Given the large number of voting studies done within individual societies, it is not too surprising that a number of cross-societal analyses of electoral behavior have also been published in recent years. Most of these deal with 13 voting and other forms of political participation among all strata in the population [199, 212, 213, 228, 246, 267, 268, 286-288, 290-292, 032, 306]. Some studies, however, focus on voting patterns among specific groups, such as the working class [234, 242, 268, 270], or women ~233, 283]. A number of closely related comparative studies analyze attitudes on political issues, apart from voting behavior itself [218, 224, 225, 229, 232, 263, 289]. The political attitudes and behavior of those in various elites-legislators, ruling classes, intellectuals-have also been explored [249, 262, 274, 296]. Since comparative voting studies have been reviewed in an earlier Trend Report, in this Journal, I shall conclude my remarks on studies of polity by citing a few other areas of research which have received some attention in recent years. Among these would be comparative studies of the [254, 255, 282, 293, 305, 494], communism and communists [200, 207, 220, 259, 260, 310], community power structure [247, 248, 276, 277] and interest groups [201]. Finally, another hopeful sign is the extension from intra-societal studies to cross-societal comparisons of the applications of &dquo;boundary&dquo; fields to the study of the polity. I refer here to laboratory small group studies of political processes [279, 307], studies of cultural values concerning authority and obedience [299], and studies which relate per- sonality variables to comparative political behavior [243]. At least one comparative study has much potential relevance for the field of bureaucracy (large scale organization, formal organization) as for polity. March [273] tests and confirms with data from 15 societies the following two hypotheses : (1) The greater the autonomy of a subsystem with respect to a larger system of which it is a part, the greater the range within which the subsystem can manipulate the orientations of the individual members of the subsystem to behavior situations. (2) The more autonomous the subsystem with respect to the larger system, the more effective will be the subsystem’s control over its members and the less frequent the deviation from the norms of the subsystem. March’s analysis is highly suggestive. Explicitly comparative empirical studies of bureaucratic organization have been much less numerous than studies of the polity. There are some hopeful signs, however. One is that the few comparativists in studies of bureaucracy exhibit much concern with Weberian theory and have attempted to test . further such aspects as authority and obedience patterns [312, 313, 317], charisma ~316, 328], diffuseness versus specificity in work organizations [311], particularism, performance and professionalism [314, 315, 320], advancement of executives [326], and role conflicts, as between &dquo;of~cial&dquo; roles on the one hand and kinship or political roles on the other hand [318, 319]. One of the most important developments is the attempt analytically and empirically to segregate the influences of (a) the external socio-cultural environment, (b) the technology used by the organization, and (c) organizational variables per se : here the work of Udy [333-337], Richardson [329] and Harris and Kearney 14

[324] merits serious attention. Another methodological development is the attainment of Guttman scales for bureaucratic variables [315, 334].

II.iii. Social Stratification and Mobility. Comparative studies in this field fall under at least four major headings. First, nineteenth-century evolutionary questions have again claimed atten- tion. For a number of Polynesian societies, Sahlins [438], working in the evolutionary tradition of Leslie White, has demonstrated that the form and degree of stratification is a positive function of the economic productivity of the society, which is in turn dependent on physical environment and technology. Goldman [455] has also analyzed data from Polynesian societies to show the changes in stratification and mobility concurrent with increasing societal complexity and cultural evolution. It is to be regretted that so little comparative work has been done on the evolution of systems of stratification. A second body of comparative analysis has been concerned with the replication of United States occupational prestige studies in other societies. At first, as might be expected, these replicational studies were restricted to other highly industrialized Western societies, the Soviet Union and Japan [478-482]. Because all these studies revealed a relatively invariant structure of occupational ranking among industrial societies, despite differences in culture and history, Inkeles and Rossi [477] advanced a &dquo;structural,&dquo; as opposed to cultural explanation, an explanation in terms of the &dquo;requirements of any industrial society.&dquo; One of the most significant break-throughs of cross-societal research since 1950 has been the repeated discovery, unexpected from the point of view of the Inkeles-Rossi theory, that occupational prestige ranking in relatively non-industrialized societies-India, Brazil, Indonesia, and the Philippines-also correlates highly witla that of the industrial nations [475, 476, 483, 484]. These findings make clear that the Inkeles-Rossi in- dustrial society explanation is inadequate. All these findings, on the other hand, can be accounted for by the functionalist theory of stratification, which asserts that in any society, industrial or non-industrial, major roles, e.g., occupational roles, are ranked on the basis of the degree of knowledge and/or responsibility called for in that occupation [416, 430a]. A third group of comparative studies focuses on the amount of vertical social mobility in different societies. The proposition that the rate of mobility is high and basically similar in all industrial societies [460, 461] has been subjected to close scrutiny; lB1iller’s analysis [465] makes clear that when the strata between which mobility occurs are analyzed according to finer distinctions than the manual-nonmanual one, there are considerable differ- ences among industrial nations’ mobility rates. Another approach accepts the finding that the amount of mobility is generally higher in industrial than in non-industrial societies, but goes on to separate, analytically and empirically, 15

two sources of mobility in any society. One source is changes in the labor force structure, e.g., an increase in the proportion of middle- and high-status occupations relative to lower-status occupations. A second source, often hypothesized as &dquo;explaining&dquo; differences in mobility rates, is the degree to which universalistic-achievement values are institutionalized in mobility channels. Both Rogoff [437] and Marsh [463] have sought to test these competing explanations by comparing societal mobility rates when intersocietal differences in occupational demand are held constant. The conclusion suggested by these studies is that the greater rate of social mobility in industrial societies is due almost wholly to sheer quantitative occupational demand differences, rather than to values and norms of a universalistic-achievement type. That is, values have not been shown to have a significant relationship to mobility rates, independent of occupational demand structure. Another development in comparative mobility studies has been the applica- tion of mathematical models, e.g., Markov chain models, to intergenerational mobility data [454, 466-468]. If the third body of research is concerned with the question of how much mobility there is, the fourth directs attention toward the processes of mobility, ie., the ways in which mobility takes place, whatever its amount. The formal educational system of a society, insofar as it recruits people on the basis of &dquo;intelligence-related&dquo; criteria, is often believed to be a major mobility channel. It is no accident, therefore, that a number of recent comparative studies have dealt with mobility in the educational system [449, 453, 456, 457, 470]. Two interesting developments in this context have been Anderson’s &dquo;skeptical note&dquo; on the alleged positive relationship between education and vertical mobility [448] and Turner’s concepts of &dquo;sponsored&dquo; and &dquo;contest&dquo; types of mobility processes [473]. Finally, several propositions concerning mobility processes, originally tested in the United States, have been replicated in other societies. Propositions about the mobility-relevant values and aspira- tions of lower class people have been partly confirmed, and partly discon- firmed, in Japan [415]; Janowitz’s hypotheses about the recruitment processes in the military elite in the United States have had the same mixed fate in a British study [469].

ILiv. Social Psychology : Socialization, Personality, National Character.

Of the relatively numerous cross-societal studies in this area in the 1950-1963 period, a number share common orientations to theory and method to such an extent that they can be spoken of as a &dquo;school.&dquo; The &dquo;Whiting-Child- Lambert&dquo; school has in common (a) the concern with systematic empirical- statistical tests of propositions from neo-Freudian and learning theories, (b) the use of extensive, world-wide samples of societies, chiefly those drawn from Human Relations Area Files (HRAF) data, (c) the development of 16

sophisticated techniques for the coding, rating, etc. of data taken from secondary ethnographic sources, as in the HRAF. The particular causal theory which has come to be associated with this &dquo;school&dquo; can be schematized as follows :

Not all these studies have dealt with the entire chain, of course; several focus on only one or two links in it. The studies of the Whiting-Child- Lambert school have analyzed a large number of child socialization variables: indulgence, nurturance, obedience, independence, self-reliance, responsi- bility, achievement, frustration-aggression, identity, anxiety, etc. [495-497, 499, 500, 505, 519, 529, 540, 541, 544-548]. Cultural variables are seen as a &dquo;projection&dquo; of personality and child-socialization practices. Among the cultural variables that have received attention from this school are: folktales [505, 528, 600], religious and magical beliefs [545, 547, 744, 751], beliefs about the causes of sickness and curing practices [546], and games [534, 724]. Perhaps the major study to emerge from this tradition is Whiting and Child’s Child Training and Personality [546]. The argument presented was as follows: in the socialization process in any society, each of five drives- oral, anal, sexual, aggressive and dependence-typically may be frustrated and treated with underindulgence, or gratified and indulged or overindulged. The former treatment gives rise to socialization anxiety, the latter to socializa- tion satisfaction. Popular &dquo;explanations&dquo; of illness in a given society are adult symptoms (&dquo;projections&dquo;) of socialization anxiety. Notions concerning how to cure illness in a given society are adult &dquo;projections&dquo; of socialization satisfaction. Thus, the following hypotheses were tested: (1) the severity of weaning (oral socialization anxiety) is associated with &dquo;oral explanations of illness,&dquo; e.g., when sickness is said to be caused by eating or drinking magically poisoned food, or by the verbal spells and incantations of sorcerers. (2) The severity of aggression training (aggression socialization anxiety) is associated with explanations of illness which involve aggression, e.g., hostility toward or disobedience of the spirits, use of magical weapons by a sorcerer, etc. (3) The severity of independence training is associated with dependence explanations of illness, e.g., the belief that illness results from &dquo;spirit pos- session,&dquo; and so forth. The rise of the Whiting-Child-Lambert school with its explanations of cross-societal socialization practices, personality and cultural patterns, has 17

not gone unchallenged. Young in particular has mounted a frontal attack, in which he recodes Whiting and Child’s own data from 54 societies and attempts to demonstrate the explanatory superiority of a very different hypothesis : namely, that initiation ceremonies are the dramatization of the sex-role characteristics of societies with a high degree of male solidarity [549]. Some of the same problems studied by Whiting, Child and others have also been studied by investigators who collected their own original data instead of using the HRAF. These investigators have necessarily therefore been restricted to much smaller-scale cross-societal tests of their hypotheses, typically to two or three societies per study [498, 501, 503, 513, 517, 518, 523, 526, 530, 533, 535, 538, 539]. Among researchers which have collected their own primary data, some of the more interesting have involved syste- matic observations of child play situations [503, and Whiting, Child and Lambert, forthcoming]. Studies of &dquo;national character&dquo; have also been conducted on a cross- societal basis [508, 512, 537]. But it appears that this type of analysis has declined somewhat in the 1950-1963 period, in contrast to its vogue in the ’thirties’ abd ’forties’. If this downward trend is real, it may be due in part to the well-directed attacks on the older traditions of &dquo;national character&dquo; and culture and personality studies. A closely related body of explicitly comparative analysis published between 1950 and 1963 deals with national stereotypes; the images the members of one society have of other societies [506, 508, 532, 610]. A number of new developments in comparative studies of socialization and personality should be noted. One is the comparison of responses to the California F-scale for authoritarian personality by subjects in a relatively non-authoritarian society with those of subjects from more normatively authoritarian societies. Data on authoritarianism have been explicitly com- pared for the United States, Britain and several European countries, Latin America, Lebanon, Egypt and other Arab states, and India [617-617]. Not only is there some evidence that people in authoritarian societies have higher mean F-scores than those in more democratic societies, but children in the two types of societies have also been shown to differ in several other behavioral manifestations [618]. A second development has been a movement away from studies in which the culture is treated as a unitary thing which produces a model personality type or national character, and toward studies which present a more differ- entiated analysis. Thus, socio-economic status differences within societies have been related to personality variables [422, 432, 520]. The extension of research on &dquo;the Achievement motive&dquo; (Need-Achieve- ment, nAch) from clinical studies within the United States to extensive cross-societal tests is a third important development [578-580]. In the work of McClelland and his associates data have been gathered from ancient

2 18

Greece, England between 1400 and 1800, and from 40 contemporary nations, in order to demonstrate the bearing of levels of achievement motivation upon successful entrepreneurship and economic development. Two other pieces of comparative analysis deserve mention for their origi- nality and theoretical relevance. The first of these is that the historian Elkins has heightened our understanding of the &dquo;Sambo&dquo; personality syndrome, documented in historical writings on the ante-Bellum American plantation Negro slave [509, 510]. He has done so by comparing the United States Deep South with data on the history of slavery in Latin America, and with data on Nazi concentration camps. Elkins shows that the &dquo;Sambo&dquo; personality syndrome appeared both among Negro plantation slaves, and among prisoners of Nazi concentration, but not among Latin American slaves. He explains these variations in personality in terms of a common set of variables, including the diversity of the symbols of authority, and the range of permissible adjustment to them. The second piece of work has direct bearing on one major aspect of Parsonian theory. In that theory, social, cultural and per- sonality systems are said to have boundaries, and boundary-maintaining mechanisms which handle the inputs of functional exigencies from other systems, &dquo;external&dquo; to the system being analyzed. Siegel has shown that in the social systems of the Hopi Indians, the European Ghetto Jews and the Hutterites, &dquo;The conscious maintenance of relatively high anxiety plateaus as an adaptive pattern occurs among those infrequently encountered groups whose cultures are tightly integrated and faced by serious threats by hostile environmental forces&dquo; [538, p. 48]. Here, then, is a comparative study which explicitly demonstrates the role of a personality variable (high anxiety plateaus, restrictions on the release of anxiety) as an adaptive, boundary- maintaining mechanism, in the face of the functional exigencies of the social and physical environment.

II.v. Conformity and Deviance.

Comparative studies in this field may be grouped under five headings : (1) general studies of norms and social control, (2) homicide and suicide, (3) illegitimacy rates and sexual mores, (4) drinking behavior, alcoholism and drug addiction, and (5) delinquency and crime. The more or less explicit common problem in all these studies is the identification of the norms of given social systems and the analysis of the conditions and consequences of conformity and deviance with respect to these norms. A few words about these studies are in order. 1. Among the general studies of norms and social control, the more interesting deal with group autonomy in relation to internal deviance [273], conflicts between universalistic and particularistic norms, in which Philippine students are found, unexpectedly, to be more universalistic than are American 19 students [667], patterns of permissiveness and internal inconsistencies within primitive societies [664 and 670, respectively], and sorcery as a form of social control [669]. At least one study attempts to compare tendencies to con- formity behavior by Norwegian and French students in a laboratory experi- mental situation [663]. 2. Given the early work of Durkheim and others, it can be said that the theory of suicide is much better developed than the theory of homicide. Moreover, until recently, suicide and homicide behavior were studied in relative isolation of each other. Henry and Short’s Suicide and Homicide, though limited in its data to the United States, made a theoretical advance by suggesting that suicide and homicide could be viewed as two forms of aggressive behavior. Suicide represents aggression directed toward the self; homicide, aggression toward others. Some of the elaborations of this theory were tested in Ceylon, where some findings on the relationship between economic cycles and the male suicide rate were contrary to United States findings, thereby necessitating some revision of theory [671, 672]. Another offshoot from Durkheim’s theory of suicide has also been replicated in Ceylon [657]. In a more anthropological vein, suicide and homicide patterns in tribes in Nigeria, Uganda and Kenya have been analyzed [646]. 3. Comparative studies in the area of sexual behavior have dealt with the causes of illegitimacy rates in Latin America and the Caribbean [658, 659], general patterns of deviation from sexual mores, e.g., incest, abduction and rape, and punishments for these deviations [647], and, finally, with premarital sexual intimacy [651-653, 655]. The latter studies present an interesting research design, and considerable real variation in the phenomena to be explained (from the conservative sexual norms of the Utah Mormons, and the &dquo;typical&dquo; United States midwestern pattern, to the more permissive premarital sex mores in the Scandinavian countries). These studies are noteworthy because they chart some of the systematic relationships among sexual norms, personality-level consequences of deviation from these norms (e.g., guilt, satisfaction), and social consequences of deviance, i.e., illegitimacy rates and marriage. 4. A pioneering comparative study by Horton, &dquo;The functions of alcohol in primitive societies,&dquo; appeared in 1943, prior to the period covered here. More recently, Horton’s analysis has come under attack in Washburne’s study of drinking in primitive societies [668]. Another study of alcohol in primitive societies [661] and one of alcohol consumption rates in Scandinavia [648, 649] also deserve note. In one of these studies [648] a Guttman scale of drinking permissiveness was shown to correlate highly with drinking habits. In the one comparative study of drug addiction found in the literature, Clausen’s proposition was tested with United States and British data. The hypothesis states that &dquo;the prevalence and consequences of addiction in any society depend as much upon the social and legal definitions placed upon 20

the non-medical use of narcotics as upon the nature and effects of narcotics or the nature of the persons who become addicted&dquo; [666]. This and another hypothesis were supported: Britain had less prevalence of narcotic addiction than the United States, less crime connected with it, and a less developed narcotics subculture than in the United States. These facts are explained in terms of the differing social and legal definitions of addiction in the United States and Britain. 5. Comparative studies of juvenile delinquency have gained some leverage by viewing gang delinquency as an adaptive mechanism in societies in which, unlike many simpler societies, there is an inadequate legitimate patterning of the transition from adolescence to adult status [645]. Other studies in this area describe patterns in several modern nations [665] and analyze anti-social and delinquent behavior among Mexican and United States children as a product of &dquo;value disturbance&dquo; [662]. A few studies deal with adult crime and penology [650, 654, 669, 671, 672]. While the studies of conformity and deviance just alluded to cover at least four specific problem-areas-homicide and suicide, illegitimacy and sexual deviance, drinking behavior and drug addiction, and delinquency and crime -as well as a few more general areas of deviance, they do so rather thinly. There has simply been too little comparative research in this area, and a number of problems have been almost totally neglected, e.g., the role of deviant behavior in conflict and in social and cultural change, the analysis of conformity and deviance in the context of work organizations and other large-scale organizations-is the &dquo;Organization Man&dquo; a distinctively Ameri- can phenomenon, or does the concept betray the lack of comparative per- spective on the part of its creator ?

II.vi. Cultural Patterns and Value Orientations.

Under the influence of Boas and the Historical school of American anthro- pology, the prevailing conception of &dquo;configurations of cultural patterns and value orientations&dquo; was radically relativistic. Values and configurations of culture were described and analyzed in ways which were &dquo;too particularized to single cultures to permit systematic comparisons betzveen cultures&dquo; [699, p. 3). The emphasis was on the uniqueness of each society’s values and culture. One of the most important trends in comparative analysis since 1950 has been the movement away from the older relativism and toward the delineation and measurement of values and configurations of culture, seen as systems of analytical variables on which all societies can be compared. The older tradition, typified best perhaps by Ruth Benedict’s Patterns of Culture, still of course has its adherents. But the contrast with this tradition can best be seen in recent comparative work by Cattell, Schuessler and Driver, Morris, Rettig and Passamanick, Rodd, Kluckhohn and Strodtbeck, 21 and others [20, 22, 23, 65, 75, 685, 693, 696, 699, 703, 705, 706]. I shall try briefly to point up the sharp contrast in method and theory, yet basic similarity in problem-focus, between the Boas-Benedict school and these recent comparative studies. The first advance made by the newer comparative analysis has been to show that if Benedict was right in her insistence on the uniqueness of the values and cultural configurations of a given society, she was not justified in her inference, &dquo;Because uniqueness, therefore incommensurability.&dquo; In Benedict’s own words, the ends and the means of one society &dquo;cannot be judged in terms of those of another society, because essentially they are incommensurable&dquo; (Patterns of Culture, 1946 ed., p. 206). Modern com- parative analysis here follows the lead of the field of linguistics: the uniqueness of natural languages is not denied; rather, both the unique and the common elements of different languages are given more precise statement in terms of key linguistic variables, and treated analytically and quantitatively. Second, modern analysis has shown that the analysis which Benedict and her school carried out intuitively, qualitatively, through &dquo;total immersion&dquo; in a given culture, can also be done in more formal, systematic, quantitative and replicable ways. For example, a number of recent studies have used factor analysis to get at three of Benedict’s major concerns: (1) the problem of the major factors which define a given culture or value system, (2) the problem of distinguishing types of cultures and culture areas from one another, and (3) the problem of identifying unique cultural configurations and values. Let us consider these three problems in turn. 1. With respect to the major factors which define a given culture, Cattell’s factor analysis of over 70 cultural variables in samples of 69 and 40 modern nations has revealed &dquo;factors&dquo; such as &dquo;enlightened affluence versus narrow poverty,&dquo; &dquo;vigorous order versus unadapted rigidity,&dquo; &dquo;size,&dquo; etc. [22, 23]. These factors have stable meanings and help to define the particular con- figurations of given cultures. Schuessler and Driver have factor analyzed 2,500 cultural items in a sample of 16 California Indian tribes; two major factors emerged: a &dquo;Northwest California culture pattern&dquo; and a &dquo;Central California culture pattern.&dquo; The former is defined by wealth-orientation, competitiveness, and individualism; the latter is more cooperative, outgoing, and nomadic than the former, but also is less accomplished [75]. The philosopher Morris factor analyzed his comparative data on 13 &dquo;Ways of Live&dquo; and uncovered such underlying factors as &dquo;social restraint and self control,&dquo; &dquo;enjoyment and progress in action,&dquo; &dquo;with-drawal and self- sufficiency,&dquo; &dquo;receptivity and sympathetic concern,&dquo; and &dquo;self-indulgence&dquo; [703]. In these and other studies, the &dquo;factors&dquo; have the same logical status as Benedict’s &dquo;appolonianism&dquo; and &dquo;dionysianism,&dquo; or her characterizations of Kwakiutl culture as dominant in rivalry and Zuni as dominant in group activity and orderliness. But because the methods are more precise and 22

because different societies can be compared in terms of a common set of factors, real gains can bc realized in the newer approach. 2. The greater use of quantitative methods in recent studies also lends itself to the distinguishing of types of cultures and culture arcas from one another. Thus, (.’attell’s Index of Pattern Similarity measures the degree of similarity between the &dquo;cultural syntality&dquo; profiles of two or more societies [22, 23]. An index score of -~-1.00 indicates perfect similarity, .00 indicates complete lack of similarity, and -1.00 indicates that the cultural profile of one society is the inverse of that of the other society. From this Index of Similarity, Cattell defines clusters and &dquo;families&dquo; of societies, each having similar culturel profiles. With the same objective in mind, Morris calculates a statistic, D, to measure the difference between any two &dquo;Ways to Live&dquo; factors. There are 13 variant Ways to Live, e.g., &dquo;preserve the best that man has attained,&dquo; &dquo;control the self stoically,&dquo; etc. By means of the D statistic every society can be represented as a point in 13-dimensional space, the location of each society being determined by its scale values on the 13 Ways to Live. In another recent study, the similarities and differences in values among societies have been measured in terms of invariance coefficients and &dquo;transformation analysis&dquo; [65]. In this study, American and Korean college students’ evaluation of 50 morally prohibited types of behavior was factor analyzed. Two of the factors which emerged together accounted for over 64 per cent of the explained item variance in each sample. These factors, &dquo;general morality&dquo; and &dquo;religious morality,&dquo; had very high invariance co- efficients, indicating that the evaluations by American and Korean students are highly similar with respect to these factors. Still another study of values used the Allport-Vernon-Lindzey Study of Values test cross-societally in order to identify similarities and differences among American, Japanese, Taiwanese and Mainland Chinese subjects [706]. 3. It is but a short step from the above attainments of recent comparative studies to a third advance, the more precise delineation of unique cultural configurations and values. Each of the measures cited above can do this. In Morris, unique cultures can be defined in terms of extreme scores on the &dquo;Ways to Live&dquo; factors. For example, in comparing United States, Indian, Chinese, Japanese, Canadian, and Norwegians students’ responses, Americans were shown to be lowest on the Restraint factor, highest in Self-Indulgence, highest in Self-Orientation and lowest in Society-Orientation ; Chinese ap- peared as the most action- and socially-oriented of the cultural groups, etc. Cattell found that about 10 percent of the 69 societies in his sample are isolates in the sense that their individual syntality pattern profiles have little resemblance to any other societies. The isolates include both leading coun- tries, Germany, the United States, the Soviet Union, the United Kingdom and Japan, and some very small countries, Nepal, Salvador, Luxembourg, and Eire. Uniqueness here, then, is defined by being an isolate, having a 23

cultural syntality pattern that is &dquo;off on some unique track of its own&dquo; [22, 23]. The contrast between the Benedict tradition and modcrn comparative analysis of value-orientations is most fully epitomized in the Harvard Values Project [695, 696, 697, 699, 709-711, 714]. The summary volume of this project, Kluckhohn and Strodtbeck’s Variations in Value Orientations [699], weaves together the several dimensions of this developing field of comparative value analysis. First, it presents an explicitly comparative conceptual scheme, based on the assumptions that there are five universal &dquo;existential problems&dquo; which all cultures must come to terms with, and three logically exhaustive alternative &dquo;solutions&dquo; to each of the five problems. Second, it presents a number of testable theoretical propositions concerning the relationship between given value orientations and other aspects of society and culture. For example, certain types of value-preference ordering are more likely than others to be found in societies undergoing rapid social change. Third, instead of characterizing societies only in terms of dominant values, assumed to characterize virtually all people and all situations in the society, value systems are seen as &dquo;interlocking networks of dominant and variant value positions which differ only in that there is a variable ordering of the same value- orientation alternatives&dquo; [699, p. 366]. Fourth, a formal, structured interview schedule is the instrument with which the basic data on value orientations in different societies are collected. Fifth, predictions about the ordering of value orientations are made on the basis of standard ethnographic data and then compared with the findings based on the interview schedule of 22 items. There was a rather high degree of correspondence between these two in- dependent estimates. On the basis of their comparison of five cultural groups living in the Rimrock area of the American southwest, Kluckhohn and Strodtbeck con- clude that there are &dquo;significant within culture regularities and significant between-culture differences&dquo; [699, p. 139]. In elaborating these findings, Kluckhohn and Strodtbeck do all the things Benedict would have done. They characterize the value orientations of each cultural group. They test whether pairs or groups of cultures are significantly different in their mean values on given value orientations, and accordingly cluster together similar cultures. They identify cultural groups which are &dquo;unique&dquo; in comparison with other groups on given value dimensions. In short, there are healthy signs of both continuity in problem-focus and increasing theoretical and methodological sophistication when recent com- parative analysis of values and cultural configurations is judged against the background of the Boas-Benedict Historical tradition.

’ II.vii. and Urban Ecology Sociology. , In the three decades prior to 1950 a body of theory and a number of generalizations concerning &dquo;the City&dquo; were formulated in the field of ecology 24

and urban sociology. This research was largely restricted to North America, but at first little attention was paid to the question whether &dquo;the City&dquo; could be analyzed independently of degree of industrialization, culture, and other elements. Then, as more sociologists began to conduct cross-societal studies of cities, particularly pre-industrial cities, criticisms of the &dquo;Chicago school&dquo; of ecology and urban sociology began to mount [791, 797, 798, 800, 802, 806, 816, 835, 836, 838, 840]. On the basis of studies of cities in latin America, Europe, Asia and Africa, these recent comparativists have noted the following ecological features which deviate sharply from what would be expected on the basis if American generalizations. ( 1 ) The center of the city is more often the hub of administrative, governmental and religious activity than of business, (2) Where there is an approximation to the central business district, it lacks the dominance found in the United States. (3) There is a minimum of specialization in land use; less separation of place of residence from place of work: residence and shops occupy the same buildings and both are scattered throughout the city. Many sites concurrently serve religious, educational, business and residential uses. (4) The elite and higher socio-economic status groups tend to live in or near the center of the city, while the poorer strata live in the periphery of the city, the reverse of the United States pattern. (5) Given the above, slums are found at the periphery of the city and there is no &dquo;zone of transition&dquo; next to the business district (if indeed there is a central business district at all). (6) Processes of invasion and succession are much less in evidence, since traditional land use patterns are maintained. (7) There is less centralization, and no marked increase in centralization as city size increases. (8) There is no uniform tendency for population density to decrease toward the periphery of the city. (9) Suburban growth from residential decentralization is limited. (10) Land values are not all-important determinants of urban patterns-historic parks, palaces, etc. are able to resist commercial encroachments. Assuming that these are, with variations, the facts of non-Western ecology, we can justly acknowledge the significance of the empirical contributions of the recent comparativists. On the other hand, the theoretical conclusions drawn by some of the comparativists must be regarded much more skeptically. Sjoberg, for example, notes that there were great limitations on economic expansion, credit and capital formation in the pre-industrial city, that the upper classes often denigrated business values and entrepreneurial activity [836]. He is aware that these differences subject the pre-industrial city to very different parameters than those assumed in the &dquo;Chicago school.&dquo; But Sjoberg and some other comparativists do not draw the correct conclusion from all this. Given that industrial and pre-industrial cities differ in their parameters and in the values of the major independent variables specified in &dquo;Chicago&dquo; ecology, the crucial comparative theoretical question is not, &dquo;Does the structure of industrial and pre-industrial cities differ ?&dquo; Of course 25

the structure differs ! Classic &dquo;Chicago&dquo; theory could have predicted that (pre-industrial) cities without a dominant and expanding central business district, etc. would differ from (industrial) cities which had an expanding central business district. The fault with classical ecological theory is not that it is wrong, but that it is oriented to a narrowly conceived range of variation in its major variables, namely, the variation comprehended by cities in only the most highly industrialized societies. Fortunately, a few of the recent comparativists realize this, and have qualified the theoretical import of their findings accordingly [798, 816, 838]. Caplow, for example, wisely pointed out that &dquo;... the trend in Guatemala has been toward increased centralization, suburbanization, outward dis- placement, and commercial dominance. If we arrange Mexico City, Guate- mala City, Merida, Quezaltenango and Oaxaca ... in order of size, it is at once apparent that the larger the community the further it has departed from the traditional colonial pattern, and this rough relationship appears to hold in some detail&dquo; [838, p. 347]. In summary, the proper conclusion which should be drawn from ecological analysis is that the ecological patterns of industrial, pre-industrial and &dquo;developing&dquo; cities are determined by the same set of variables. Since the values of these variables differ among these different types of cities, the ecological patterns observed also differ. But if and insofar as the values of the independent variables in pre-industrial and &dquo;developing&dquo; cities change in the direction of the industrial city, the other, dependent variables and patterns should also come to resemble those of the industrial city. A closely related body of &dquo;Chicago&dquo; theory which developed between the two world wars asserted that the heterogeneity, mobility and centralization of large cities resulted in a new form of social organization : the weakening of primary group ties and atomization of social relationships, secularization, social disorganization-in a word, &dquo;mass society.&dquo; At least part of this theory has recently come under attack in studies of American and British cities, which have shown that primary groups survive, even flourish, among kin, and friends, in work groups and in large organizations, despite their metro- politan settings. If we think of the ratio of primary group relationships to secondary group relationships as a continuum, or of degrees of &dquo;primary-ness&dquo; versus &dquo;second- ary-ness,&dquo; then the crucial question is: where on this continuum do these recent United States and British findings belong ? They are stated often as though they belong well over toward the &dquo;primary group dominance&dquo; end of the continuum. Before accepting this conclusion, we need much more cross-societal comparative evidence. And it is precisely here that we are worst off. The deplorable situation is far from unfamiliar in sociology: the recent American and British studies were not explicitly designed to collect and measure data in ways that would allow ready comparison with the studies 26

of primary and secondary groups in Timbuctoo [829], in Cairo [791] and in other pre-industrial cities [835, 836]. Nor do the latter, non-Western urban studies reveal much awareness of the findings of the recent American and British studies. (1B1iner’s book, of course, appeared in 1953, before the recent British and American studies were published, and this criticism cannot apply to his work on Timbuctoo.) In short, the theoretical relevance of these two bodies of research on primary versus secondary group ties in cities is compromised by the fact that each body of research, in effect, talks past the other one. Until this defect is remedied, the comparative analysis of &dquo;mass society&dquo; in industrial, pre-industrial and &dquo;developing&dquo; cities can only continue to flounder. Students of this problem in different societies must work out cooperatively a common set of variables, measures and analyses, so that the comparative problems in this field can have a happier resolution.

ILviii. Economic Developnrent and Social Change. A relatively large number of comparative studies of economic development and industrialization were published between 1950 and 1963 [338-375]. Of these, several explicitly confronted the problem of whether contemporary underdeveloped societies can and will recapitulate the developmental histories of the highly industrialized nations during the last century or two. That this is unlikely, at least in some respects, is the conclusions of Kuznets [356], Hoselitz [352, 353] and Habakkuk [346]. The level of economic performance in today’s underdeveloped countries is lower than that of the already de- veloped countries prior to their industrialization. Agricultural densities are three or more times as high in Asia today as in Europe at a corresponding time. Urbanization is running ahead of industrialization to a greater extent than in the history of the European cases. If underdeveloped countries do make a successful transition to industrialism, it is probable that it will be through models other than that of European [195, 912]. For example, societies with high densities in their agricultural population may drain off their labor surpluses by small-scale, labor-intensive industry [350, 352]. Sociological and social psychological variables have been shown to influence economic development. Of particular interest here is the work of McClelland and his associates on need-achievement [578, 579], Hagen on personality variables [908], Goldschmidt’s test of the Weber &dquo;Protestant Ethic&dquo; thesis among California Indian tribes [687], and Levy’s analysis of the internal social structural elements which account for Japan’s conspicuously greater success in industrialization than China’s [917]. Among the more dramatic processes of social change are, of course, revolutions, rebellions and disasters, and we have had a number of com- 27 parative analyses of these [251, 676, 896, 897, 899, 903, 910, 918, 928b]. One of the most important problems in this area is to determine the types of, and conditions under which, revolutions, rebellions, and disasters actually give rise to basic social change, as opposed to merely elaborating or restoring previous social arrangements. Among other processes of social change which have been investigated comparatively are ideology and mass movements [892, 929], and the role of the military in developing nations [898]. On a different plane from all these relatively specialized studies are the macroscopic analyses by Sorokin [931] and White [937]. The studies of economic development and social change referred to above have for the most part cut across two or more regions of the world, and emphasized generic comparative problems. Comparative studies couched in regionally-specific terms should also be noted as a continuing trend. Thus, we have studies of change and development in Central and South America [351, 362, 374, 890, 904, 918, 935], Africa [892, 894, 901, 932], the Middle East [909, 915], Asia [344, 350, 352-355, 358, 363, 917, 919, 920], Oceania (388, 933~ and, of course, Europe and North America [346, 347, 349, 368-370, 375, 377, 378, 381-383, 385]. In concluding my review of substantive studies of social change in the period 1950-1963 I wish to emphasize the revival of interest in the analytical variable societal differentiation. This variable had a central place in 19th cen- tury comparative analysis, especially in the work of Spencer, and though comparative analysis from 1950 to 1963 has only begun to return it to its earlier central position, its potential significance is great indeed. Not only has social change itself been conceived as a process of differentiation (or de-differentiation), but the variable, societal differentiation, has begun to appear even in synchronic comparisons in such sub-fields as kinship, con- formity and deviance, etc. In the final pages of this Report I want to argue the case for making the variable of societal differentiation the major organizing principle for the future integration and progress of the field of comparative sociology. The re-emerging concept of differentiation refers to social-structural level variables. It is defined as the multiplication of one structure of a society (e.g., a role, a collectivity) into two or more new structures, each structurally distinct. Each new structure typically becomes more functionally specialized than the earlier single structure out of which it emerged. The functionally specialized and differentiated structures make complimentary contributions to the larger system of which they are a part. But taken together, they are functionally equivalent to the original unit. Differentiation must not be confused with segmentation, in which two or more structurally distinct units perform essentially the same function in the system, rather than comple- mentary functions. Thus, populous agrarian societies may be highly segmented -e.g., thousands of village communities, millions of peasant households, all 28

performing essentially the same functions-but they are typically much less differentiated than are modern industrial societies [4, 6, 7, 10, 362, 895]. The measurement of structural differentiation at the societal level is in its infancy, but there have been some advances in recent comparative analysis. Naroll [924] has proposed a &dquo;preliminary index of evolutionary social develop- ment,&dquo; one of whose indicators is occupational craft specialization. Naroll counts the number of differentiated occupational crafts in each of 30 societies and thereby provides us with an ordering of societies, ranging from the Yahgan of Tierra del Fuego to the Inca and Aztec, on what we should call degree of structural differentiation. In the same manner, more complex societies could be ordered by the number of differentiated occupations in their labor force. Freeman and Winch [905] have shown that the variable of societal complexity, which underlies the classical typologies of Tönnies and others, has the property of a unidimensional Guttman scale, and that a sample of 48 societies can be ordered in terms of the six scale types. In their Guttman scale of community development, Young and Young [939] have been able to order some 54 societies on a scale of which at least some items tap the variable of structural differentiation. The same is true in Carneiro’s applica- tion of scale analysis to evolution [18, 19]. Carneiro is particularly lucid in distinguishing between the kinds of evolutionism implied in Guttman scalo- gram analysis and nineteenth-century evolutionism. Carneiro and Tobias [18] impressively scale 100 societies, ranging from the least differentiated to Han dynasty China, according to pools of 354 and 50 cultural traits. The list of traits would have to be pruned in order to restrict it to only social structural differentiation type variables. Another approach to the measurement of societal differentiation has been in terms of factor analysis. Here, instead of compiling an ever larger list of items which presumably measure societal differentiation, the attempt is to reduce such lists to a few basic factors. To date factor analyses have, not unnaturally, incorporated mostly technological, economic and other kinds of relatively extensively available data. Until such time as more direct measures of social structural differentiation can be obtained for large samples of societies, one can only assume that these data are valid. Among factor analytic studies in this area, those by Berry [340] and Schnore [371] are outstanding. Berry factor analyzes 95 nations on 43 indices of development, and finds five factors which account for 94 per cent of the total sum of squares. Factor one-a technological factor-alone accounts for more than 84 per cent of the total sum of squares. This technological factor results from the similar ranking of societies on such variables as energy production and consumption, Gross National Product, etc. Whereas the classical typologies of Gemein- schaft-Gesellshaft, Mechanical-Organic, Folk-Urban, etc. emphasized binary or ordinal scale categorizations, the findings of Berry’s factor analysis suggest a different conclusion. On all of Berry’s five factors, the distribution of 29

societies is continuous and linear; there are no sharp breaks of discontinuities between richer and poorer countries. The implication of this is that any typing of societies, at least on this technological factor, can be done only by arbitrarily introducing cutting-points in the continuous array of societies.

III. Comparative Sociology in the Future

There is a serious limitation in most studies which have attempted to measure societal differentiation, complexity, development and the like and then scale or order societies accordingly. Each study tends to deal with societies only at one end of the scale: either the relatively more primitive societies [18, 19, 905, 924] or the relatively more complex societies [340, 371]. One might have no objection to this division of intellectual labor were it not for the fact that different indicators are used in measuring degree of differentiation, and that while societies at one end of the scale may be nicely distinguished, societies at the other end are too grossly lumped together. The Freeman and Winch scale [905] for example would group all of the one hundred-odd modern nations, from Gabon and Laos to the United States, in the same scale type. Obviously, if the field of comparative sociology is ever to have a valid measurable variable of societal differentiation, its measurement must be in terms of indicators for which there are at least convertible data in all societies, from the simplest hunting and gathering bands to the most differentiated modern industrial nations. This requirement greatly restricts the possible indicators that might be used. It is difficult to predict how this problem will eventually be resolved. Among the indicators of societal differentiation more likely to be widely adopted are: ( 1 ) direct indicators, such as craft and occupational role differen- tiation [813, 924], and degree of social stratification and of political integration [101]; (2) indirect indicators, such as amount of energy consumption per capita from all sources [340, 371]. However these problems of measurement are resolved, it is possible now to sketch briefly how the variable societal differentiation may be utilized as an ordering principle in terms of which both the results of comparative analysis in recent years may be codified and the integration and cumula- tiveness of future work in comparative sociology may be maximized. I shall first outline a logic of codification for comparative sociology, and then illustrate how existing substantive findings may be integrated in terms of this codification scheme. It should be clear that from this point on in this Report, my primary task changes from being a reporter of what has been done in the field to being an advocate of what I should like to see done. I claim that the codification scheme to be presented can make a great and 30

much-needed contribution to the somewhat disconnected efforts of com- parativists, now and in the future.

IILi. The Logic of Codification for Comparative Sociology.

When one assesses the findings of a given study in which two or more societies or subsystems thereof have been explicitly compared, there are three major questions that can be asked: 1. What is the range of the societies compared in the study, in terms of degree of societal differentiation ? Obviously, other things equal, we have more confidence in a study which tests some general proposition in societies of low, middle and high differentiation than where the proposition is tested only among primitive societies, or only among English-speaking de- mocracies, etc. 2. In any given comparative study there should be some &dquo;phenomena to be explained,&dquo; or dependent variables. The first question one asks of these phenomena or variables is: do they vary among t7te societies compared? Thus, if the phenomena to be explained are &dquo;storm and stress in the adolescent phase of the life cycle,&dquo; we want to know whether all societies have this feature in essentially the same degree, or whether the societies vary in this respect. 3. Finally, if the phenomena to be explained vary significantly among the societies compared, we then ask, do the phenomena to be explained vary with degree of structural differentiation (as defined in (1) above), or do they vary independently of degree of differentiation ? That is, a given com- parative study may test the relationship between an independent variable, x, and a dependent variable, y, in a sample of N societies. After ordering these societies according to their degree of differentiation, we can then hold constaiit degree of differentiation in order to see whether (a) the dependent variable y, or (b) the relationship between x and y, do or do not vary with degree of differentiation. These are the only logically necessary questions to ask in codifying any piece of comparative analysis. Before generating some categories which arise from the various combinations of these three questions or criteria, it should be noted that there are important further steps in comparative analysis, steps which go beyond the requirements of codification per se. Specifically, if, in (3) above, one has shown that &dquo;the phenomena to be explained&dquo; do vary with degree of societal differentiation, the next step-a theoretical- methodological step-is to attempt to explain the phenomena in terms of degree of differentiation. That is, one should not stop when a correlation has been established between degree of differentiation and the phenomena to be explained. One should go on to show what it is about differentiation 31 in this particular study that influences the phenomena to be explained. This link in the chain of a full-blown comparative analysis (of which, alas, we have all too few examples in the published literature) cannot be gone into further here (See 923). The three criteria give rise to the following codification categories, by which comparative studies may be classified.

* But phenomena to be explained may vary systematically with other socio-cultural variables, not correlated with degree of structural differentiation.

z

What remains to be done is to illustrate the use of this codification schema with reference to a few of the hundreds of comparative studies discussed earlier in this Report.

III.ii. Illustrations of how Comparative Studies are Codified.

Space considerations prohibit any but the briefest illustration of the applica- tion of the codification schema to comparative studies. We begin with Goode’s test of the hypothesis that there is an inverse relationship between social class position and divorce rates [140]. The societies compared are the United States, New Zealand, Australia, Sweden, France and England. All these societies have a roughly equal (high) degree of differentiation. The hypothesis is confirmed on the basis of the societies compared. The phenomena to be explained (the relationship between class and divorce rates) do not vary among these societies. That is, the relationship is inverse in each society, not positive in some, but inverse or independent in others. Given these characteristics, Goode’s study must be codified as an example of replication; it cannot be any of the other three codification types. Now consider the findings of a number of comparative studies concerning the ranking of occupational prestige [475-484]. Considered together, these 32 studies show that functional roles (e.g., occupations) whose expected activities are relatively similar in different societies tend to be evaluated hierarchically in a similar rank order. Here, the range of societies on the differentiation variable is greater than in Goode’s study: the occupational prestige data come from societies at the highly differentiated extreme (the United States, Western Europe, Australia and New Zealand) and also from less differentiated societies (Japan, Brazil, Indonesia, the Philippines and India). The range is sufficiently great to warrant treating these societies as representing more than only one type on the differentiation variable. The hypothesis of an invariant occupa- tional prestige hierarchy is supported, though of course only a very small number of occupations has thus far been ranked in all these societies. The phenomena to be explained (occupational prestige ranking) do not vary significantly among the societies compared. Therefore, these comparative studies are codified as an example of a universal generalizations, at least ten- tatively, until larger samples of occupations can be ranked across societies. As an illustration of the third codification category, contingency generaliza- lion, lB1arch’s study of group autonomy and social control [273] may be cited. Communities in 15 societies were compared; the communities were deliber- ately varied in the extent of their integration versus autonomy in the larger society of which they were a part. The 15 societies themselves varied signifi- cantly in degree of overall differentiation, from the relatively undifferentiated Jivaro, Andamanese and Siriono to the much more highly differentiated Ashanti, Kazak and Yapese. The two hypotheses tested were both confirmed: &dquo;The range within which a group can manipulate the orientations of the individual members to behavior situations increases monotonically with increase in the autonomy of the group&dquo; [273, p. 325], and the more auto- nomous the group, the less frequent the deviation from group norms. The phenomena to be explained-extent of manipulation of group members’ orientations and frequency of their deviance from group norms-vary among the societies compared; moreover, these phenomena vary with degree of differentiation of society. The more differentiated the society, the less the group can manipulate its members and the more deviation there is from group norms. March’s hypotheses, then, are codified as contingency gener- alizations. Finally, a study of cleavages in voting behavior by Alford [199] may be used to illustrate the codification category of specification. Alford compared four English-speaking democracies (the United States, Britain, Canada and Australia), which are so highly similar in degree of differentiation as to be one &dquo;type&dquo; of society, yet despite this similarity, there were marked inter- societal differences in the degree of political cleavage along class lines. Canada and the United States have less cleavage along class lines than do Britain or Australia. The phenomena to be explained-class cleavage in voting- vary among the four societies; since the societies do not vary commensurately 33

in differentiation, it is clear that the phenomena to be explained vary in- dependently of differentiation. This discovery that societies similar in degree of differentiation have &dquo;unexpected&dquo; variations in the extent of class cleavage in voting is the first step in the decision to codify the finding as specification. That is, we begin with a negative finding: class cleavages in voting behavior are not due to societal differentiation. Unfortunately all too many com- parative studies stop at this point, when the analytic task is only half com- pleted. To complete the task of specification, one should ideally be able to state the variable(s) with which the phenomena to be explained do vary. This variable or variables necessarily will be independent of degree of societal differentiation. In the Alford study, voting cleavages in the United States and Canada are much more along religious, ethnic and regional lines than is the case in Britain or Australia. Thus, the relative significancc of religious- ethnic-regional cleavages versus class cleavages in voting, which at least in these four societies varies independently of societal differentiation, appears to account for the phenomena to be explained. In Alford’s study, the societies compared arc similar in degree of differen- tiation. Research in which the societies compared are dissimilar in degree of differentiation may also be codified as specification, provided, of course, that the phenomena to be explained vary among societies but do not vary systematically with degree of differentiation. An example is Aberle’s study of matrilineal descent [105]. The sample, the 565 societies in Burdock’s World Ethnographic Sample, includes a range of societies from the simplest to the most differentiated. Aberle’s analysis shows that the matrilineate is associeted less with degree of differentiation of society than with factors such as type of ecological adaptation. Even when given types of ecological adaptation are associated with a given degree of societal differentiation, the matrilineate is not the only kinship type found; patriliny and bilaterality are found under the same conditions. Aberle’s analysis is a full-blown case of specification: he not only shows that the matrilineate does not vary signifi- cantly with degree of differentiation, he also demonstrates which variables are associated with the marilineate. In summary, insofar as propositions tested cross-societally are confirmed only in societies of the same degree of differentiation, the propositions can only be codified as replication. When the societies compared are dissimilar in degree of differentiation, if the phenomena to be explained. do not vary among societies, the propositions are codified as universal generalizations; if the phenomena to be explained do vary among societies and also vary with degree of differentiation, the propositions are codified as contingency gener- alizations. Finally, no matter whether the societies compared are of similar or dissimilar degrees of differentiation, if the phenomena to be explained vary independently of degree of differentiation, the propositions are codified as specifications.

3 34

IV. Conclusion

In the period of 1950-1963, cross-societal comparative analysis began to gain momentum. There is every indication that the field of comparative sociology-the systematic and explicit comparison of social phenomena in two or more societies-will in the future flourish on an even greater scale. Yet certain tendencies exhibited between 1950 and 1963 must be checked. Chief among these is the failure to ask the right questions when judging the results of comparative studies, the tendency toward a looseness of criteria, in which one comparative study is regarded as essentially as good as another comparative study. I have proposed a codification schema for comparative sociology which, if adopted, will correct this tendency. It compells us to sort out the findings of different comparative analyses according to such major criteria as the range of differentiation of the societies compared, the sampling base of the study, and whether the phenomena to be explained do or do not vary systematically with degree of societal differentiation. The variable of degree of societal differentiation is beginning to re-emerge as a major ordering principle in the comparative analysis of societies. A general theory is developing in which many of the processes and relationships of interest to sociologists, social anthropologists and social psychologists are being shown to be a function of the degree of societal differentiation. On the other hand, it would be folly to insist that all these processes and relation- ships are a function of differentiation: some are clearly a function of other culturally and historically variable elements. The codification schema pro- posed has as its objective the sorting out of these different outcomes of comparative studies. The schema is a vehicle by which we may approach the goal of comparative sociology, which is to distinguish which theories, propositions, etc. hold for all societies, which for only certain classes or types of societies, and which for only individual societies.