Comparative Research across Cultures and Nations [Reprint 2021 ed.] 9783112416389, 9783112416372

Citation preview

COMPARATIVE RESEARCH ACROSS CULTURES AND NATIONS

Publications of the International

8

MOUTON PARIS - THE H A G U E MCMLXVIII

Social Science

Council

Comparative Research across Cultures and Nations edited by

STEIN

ROKKAN

MOUTON PARIS - THE HAGUE MCMLXVIII

© 1968 Mouton and École Pratique des Hautes Études Printed in the Netherlands

Table of contents

INTRODUCTION and KAZIMIERZ SZCZERBA-LIKIERNIK - A programme for the advancement of comparative social science research: the action of the International Social Science Council

STEIN R O K K A N

1

PART I: CROSS-CULTURAL COMPARISONS 1.

J. F . K Ô B B E N - T h e logic of cross-cultural analysis: why exceptions? 17 2. ROBERT B. TEXTOR - Computer summarization of the coded cross-cultural literature 54 ANDRÉ

PART II: COMPARATIVE HISTORIES OF PROCESSES OF DEVELOPMENT 3. REINHARD BENDIX - Concepts in comparative historical analysis 67 4. DANIEL LERNER - Comparative analysis of processes of modernization 82 5. LUCIAN W. PYE - Political systems and political development 93 6. VAL R. LORWIN - Historians and other social scientists: the comparative analysis of nation-building in Western societies 102 7. RICHARD ROSE - Modern nations and the study of political modernization 118 8. LEE BENSON-The empirical and statistical basis for comparative analyses of historical change 129

Table of contents

vi

PART III: QUANTITATIVE APPROACHES TO CROSS-NATIONAL COMPARISONS R. ALKER JR. - Research possibilities using aggregate political and social data 1 0 . G O R A N O H L I N - Aggregate comparisons: problems and prospects of quantitative analysis based on national accounts 1 1 . PHYLLIS DEANE - Aggregate comparisons: the validity and reliability of economic data 1 2 . E R W I N K . SCHEUCH - The cross-cultural use of sample surveys: problems of comparability 13. ROBERT E. MITCHELL - Survey materials collected in the developing countries: obstacles to comparisons 9. HAYWARD

143 163 171 176 210

Introduction

INTRODUCTION

A programme for the advancement of comparative social science research: the action of the International Social Science Council STEIN ROKKAN University of Bergen

KAZIMIERZ SZCZERBA-LIKIERNIK International Social Science Council An International Conference on Comparative Social Science Research Since 1962 the International Social Science Council has been making efforts to advance systematic comparative research across cultural and political boundaries. To evaluate the progress made within this programme and to discuss the possible directions for further expansion, in 1964 the Council decided to convene a Round Table Conference on Comparative Research in direct conjunction with its General Assembly. The Conference was organized under a joint arrangement with the Committee on Comparative Politics of the Social Science Research Council (USA), and was attended by experts from all the disciplines of the social sciences, some of them members of the issc, others invited to contribute papers or to take part in the discussion of the various themes. The Conference was held at UNESCO House in Paris from 22 to 24 April, 1965 and was attended by 40 experts and 12 observers, as follows: GABRIEL ALMOND, Institute of Political Studies, Stanford University, Conference Chairman; REINHARD BENDIX, Department of Sociology, University of California, Berkeley; LEE BENSON, Department of History, University of Pennsylvania; GUSTAVO BEYHAUT, École Pratique des Hautes Études, Paris; KOFI A. BUSIA, St. Antony's College, Oxford; ISAAC CHIVA, École Pratique des Hautes Études, Paris; FRANÇOIS CLOUTIER, World Federation for Mental Health, Geneva; HANS DAALDER, University of Leyden; PHYLLIS DEANE, Newnham College, Cambridge;

2

Stein Rokkan and Kazimierz Szczerba-Likiernik

University of Thessaloniki, Salonica; MATTEI DOGAN, Centre d'Études Sociologiques, CNRS, Paris; FYODOR N. FEDOSEEV, USSR Academy of Sciences, Moscow; DARYLL FORDE, Department of Anthropology, University College, London; FRANÇOIS FURET, École Pratique des Hautes Études, Paris; G I N O GERMANI, Institute of Sociology, University of Buenos Aires; RONALD GRAVESON, King's College, University of London ; SJOERD GROENMAN, Institute of Sociology, University of Utrecht ; President of the ISSC; PENDLETON HERRING, Social Science Research Council, New York; BERTRAND DE JOUVENEL, Société d'Études et de Documentation Économique Industrielle et Sociale, Paris; OTTO KLINEBERG, Faculté des Lettres, University of Paris; ANDRÉ J . F. KÖBBEN, Institute of Cultural Anthropology, University of Amsterdam ; RENÉ KÖNIG, Forschungsinstitut für Soziologie, University of Cologne; GEORGE KURIYAN, Delhi School of Economics, New Delhi; DANIEL LERNER, Center for International Studies, Massachusetts Institute of Technology, Cambridge, Mass.; EMMANUEL L E ROY LADURIE, École Pratique des Hautes Études, Paris; VAL R. LORWIN, Department of History, University of Oregon; LASZLO MAKKAI, Institute of Historical Sciences, Hungarian Academy, Budapest; THOMAS H . M A R S H A L L , Cambridge University; JEAN MEYRIAT, International Committee for Social Sciences Documentation, Paris; JOSEF NUTTIN, University of Louvain; GORAN OHLIN, OECD Development Center, Paris ; LUCIAN PYE, Department of Political Science, Massachusetts Institute of Technology, Cambridge, Mass. ; E . A U S T I N G.ROBINSON, The Marshall Library, Cambridge University; STEIN ROKKAN, The Chr. Michelsen Institute, Bergen; Rapporteur général; RICHARD ROSE, Department of Politics, University of Strathclyde; E R W I N K. SCHEUCH, Institut für vergleichende Sozialforschung, University of Cologne; ROBERT B.TEXTOR, International Department, Education Center and Department of Anthropology, Stanford University; PIERRE U R I , Atlantic Institute, Boulogne-sur-Seine; FRANCESCO Vrro, University of the Sacred Heart, Milan; JERZY WIATR, University of Cracow.

DÉMÉTRIOS DELIVANIS,

UNESCO

was represented by : Director of the Department of Social Sciences ;

ANDRÉ BERTRAND,

Introduction

3

Chief of the Division for the International Development of Social Sciences.

SAMY FRIEDMAN,

A number of international associations were represented by observers : L u c FAUVEL, for the International Economic Association; ROGER GIROD, for the International Sociological Association; E.GREBENIK, for the International Union for the Scientific Study of Population; SERGE H U R T I G , for the International Political Science Association; LAWRENCE KRADER, for the International Union of Anthropological and Ethnological Sciences ; F . H . L A W S O N , for the International Association of Legal Sciences; JEAN D'ORMESSON, for the International Council for Philosophy and Humanistic Studies; HÉLÈNE RIFFAULT, for the World Association for Public Opinion Research; HENRI TAJFEL, for the European Association for the Advancement of Experimental Social Psychology. MASSIMO FICHERA represented the Olivetti Foundation. Themes of discussion The Conference was organized around five themes : I. The Cross-cultural Method. II. Comparative Analysis of Historical Change. III. Comparative Analysis ofProcesses of Modernization. IV. Aggregate Statistical Comparisons of Nations. V. Comparative Sample Surveys. None of these themes could be dealt with in much detail at the Conference: the aim was to discuss recent examples of comparative research in each field and to review alternative strategies in the further advancement of systematic research across cultural and political boundaries. Within the limited time set aside for the Conference only one session could be devoted to each of the five themes and this clearly proved a frustrating experience to the participants: their eagerness to go further in the discussion provided ample proof of the need for broader international confrontations within each of the five fields of comparative research. The comparative studies chosen as texts for the discussion were the following: Theme I : A . B A N K S and R . T E X T O R , A Cross-Polity Survey. Cambridge, Mass., M. I. T. Press, 1965. R . T E X T O R , A Cross-Cultural Summary. New Haven, Conn., Human Relations Area Files Press, 1967. Theme II : R . B E N D I X , Nation-Building and Citizenship. New York, J . Wiley, 1 9 6 4 . K . W . D E U T S C H and W . J . F O L T Z , eds., Nation-Building. New Haven, Conn., Yale University Press, 1963.

4

Stein Rokkan and Kazimierz

Szczerba-Likiernik

S.N. EISENSTADT, The Political Systems of Empires. New York, The Free Press, 1963. S. M. LIPSET, The First New Nation. New York, Basic Books, 1963. Theme III: D. LERNER, The Passing of Traditional Society. Glencoe, III., The Free Press, 1958. G . A . ALMOND and J . S . COLEMAN, eds., The Politics of the Developing Areas. Princeton, N.J., Princeton University Press, 1960. L.W.PYE, ed., Communication and Political Development. Princeton, N.J., 1963. J . LAPALOMBARA, ed.. Bureaucracy and Political Development. Princeton, N . J . , 1963.

eds., The Political Modernization of Japan and Turkey. Princeton, N.J., 1964. Theme IV: N . GINSBURG, Atlas of Economic Development. Chicago, University of Chicago Press, 1961. B . M . RUSSETT et al., World Handbook of Political and Social Indicators. New Haven, Conn., Yale University Press, 1964. L. NEUNDÖRFER, Atlas sozialökonomischer Regionen Europas. BadenBaden, Aug. Lutzeyer, 1964. Theme V: G . A . ALMOND and S. VERBA, The Civic Culture: Political Attitudes and Democracy in Five Nations. Princeton, N.J., Princeton University Press, 1963. R . A L F O R D , Party and Society: The Anglo-American Democracies. Chicago, Rand MacNally, 1963. R.E.WARD andD.RusTOW,

Background This choice of themes must be seen against the background of earlier efforts to further cross-national and cross-cultural comparisons within the framework of UNESCO.

In its first attempts in this direction UNESCO focused on a few substantive fields of direct interest within its over-all programme: studies of the sources of tensions among nations and races, studies of stereotypes and prejudices, studies of opinions on international issues. The principal product of these early efforts of comparative research was the CANTRIL sample survey in nine countries in 1949:, the study reported on in the volume How Nations See Each Other.1 These early efforts did not generate a long-term programme: large-scale survey research across a number of countries demanded a complex administrative apparatus, and the studies themselves came under criticism for their tendency to pursue abstract comparisons without consideration of the historical contexts and the structural conditions of particular response constellations. By the mid-fifties the

Introduction

5

UNESCO Social Sciences Department became more and more heavily involved in the promotion of training and research in the developing countries and found it increasingly difficult to pursue explicitly comparative studies. An attempt was made in 1956 to launch a programme of comparative surveys but this proved very difficult to get under way. A small four-nation study was in fact organized but it was soon realized that this was the wrong tack. 2 In a sense the current programme grew out of this realization of failure. It became more and more obvious that UNESCO could use its limited funds much more effectively if instead of organizing fresh comparative studies from scratch it concentrated its efforts on what might be called the infrastructure of comparative research: if it took on as a long-term task the establishment of better facilities for research workers interested in cross-cultural or cross-national analysis of one type or another. This meant, at least in the beginning, a concentration on methods, on sources of information, on access to data for analysis. This new line found a first expression in the UNESCO programme for 1961-1962; all scholars concerned in the advancement of comparative research have reasons to be grateful to T. H. Marshall, André Bertrand and Samy Friedman for their efforts to make this new departure administratively acceptable within the UNESCO framework. The International Social Science Council was given a central role in the development and execution of the new programme. The Council decided to concentrate the first set of Conferences under the programme on quantitative methods of comparison. The Conference held at La Napoule on the Còte d'Azur in June, 1962, was devoted to comparative survey research,3 and the Conference at Yale University in September, 1963, focused on aggregate national statistics and the possibilities of correlational analyses using nations as units. 4 Both these lines have been pursued in subsequent conferences and publications. The work in the field of comparative survey data has concentrated on problems of data access, data archives, data retrieval. 5 The work on aggregate comparisons has been pursued in two directions. The Yale Conference recommended that the approach developed in the Russett et al., World Handbook of Political and Social Indicators be discussed in detail within each major region of the world in order to ensure broader coverage, better evaluations and more realistic analyses of the data at hand. A first regional conference within the programme was organized at Buenos Aires in September, 19646 and a second in March, 1967, in New Delhi. The Yale Conference also recommended that data programmes of the type developed by Deutsch and Russett should be supplemented by within-nation ecological archives to allow studies of the sources of variations between different types of localities and between advanced and backward areas of each country. 7 A first discussion of the development of such ecological archives took place at the Second Conference on Data Archives in September, 19648 and a technical conference on Quantitative Ecological Analysis took place at Evian, in September, 1966.

6

Stein Rokkan and Kazimierz

Szczerba-Likiernik

The issc has for some time been concerned to go beyond the exploration of such strictly quantitative methods and to study further approaches to systematic comparisons among cultural and political units: at the Paris Conference the two 'old' themes were therefore put last on the agenda and three 'new' themes first. The first of these new themes was the cross-cultural method: the qualitative alternative to the aggregate comparisons espoused by Karl Deutsch and his team. Bruce Russett and his collaborators have chosen as their units of analysis the politically and territorially defined units termed nations and have assembled the available quantitative data on their properties. George Peter Murdock and his co-workers have chosen as their units a sample of culturally defined societies and have developed a system of qualitative codes for characterizing each such unit. Robert Textor has recently tried to organize on the basis of these codings a computer-produced cross-cultural summary for a sample of 400 cultures: this great effort of data processing, at that time still only in the form of direct printouts, formed the basis for the discussion of the first theme of the Conference. The second and the third themes of the Conference were closely related to each other: the discussion of Comparative Analysis of Historical Change focused on the building of nation states in Europe and the West and the discussion of Comparative Analysis of Processes of Modernization concentrated on the possibilities and limitations of generalized developmental models in the study of social and political change in any part of the world, whether in the older nations of Europe and the West or in the emerging units of post-colonial Africa or Asia. Notions of development, directional change, modernization had already been extensively discussed within the issc programme but so far mainly in terms of the availability of codable and quantifiable data for systematic processing: data on levels and rates of growth, on differences between advanced and backward areas, on the spread of material and cultural innovations, on the speed of economic, social and political mobilization. But such data have to be analyzed and interpreted in a broader context of historical knowledge: the social sciences can only become 'developmental' through close co-operation with the students of the time dimensions of social life, the historians. For the first time within the issc programme steps were taken to bring historians and social scientists together to see how useful they could be to each other in comparative studies of nation-building and processes of modernization. It was hoped that these initial discussions would lead to more systematic exchanges in the future, and generate pressures for a new line of activity within the issc programme. These were the three new themes explored at the Conference. There are many others waiting to be investigated but the programme can only be expanded step by step. We are particularly conscious of our failure to give due consideration in the

Introduction

7

planning of the Conference to the very interesting developments under way in cross-cultural psychology: fortunately several of the active leaders in the movement were present at the Conference and helped to formulate suggestions for co-operative arrangements within the over-all framework of the issc programme. Organization of the volume of Conference papers Altogether eleven papers were presented and discussed at the five sessions of the Conference: all of these have, after some revision, been printed in this volume. Three further papers were added shortly after the Conference to round off the presentations, one, by the U.S. mathematician and political scientist Hayward R. Alker, on the potentialities of the 'data programme' for aggregate comparisons built up at Yale University, another, by the English economic historian Phyllis Deane, on the limitations of comparisons of economic data, and a third, by the American sociologist Robert E.Mitchell, on difficulties encountered in the use of data from sample surveys carried out in traditional societies at a low level of development. To simplify the structure of the volume these fourteen papers have been grouped in three Parts: The first concentrates on the efforts of cultural and social anthropologists to develop a basis for controlled comparisons across world samples of elementary societies (Theme I in the original agenda). The second deals with approaches to the study of processes of mobilization, differentiation and integration in complex national polities (Themes II and III). The third takes up a variety of methodological problems in the use of quantitative techniques in the comparative analysis of societies at different levels of economic, cultural or political development (Themes IV and V). Further developments The discussion of the eleven papers and the sixteen background volumes (cf. pp. 3-4) revealed a broad consensus on priorities in the development of cross-cultural and cross-national research. There appears to be general agreement that the International Social Science Council is in a position to exert an important brokerage function in the advancement of comparative research across political, ideological and cultural boundaries. It is equally clear that the Council cannot make much headway on its own: it has to base its action on initiatives taken at the centres of intellectual innovation in the advanced countries and to invest its resources in the development of facilities for cross-communication and co-operation among the most active research groups in the different countries. Great advances, it is true, have been made and can still

8

Stein Rokkart and Kazimierz

Szczerba-Likiernik

be made through single-nation initiatives: the bulk of comparative data-gathering operations have to this day in fact been planned in the United States and executed under some form of contract in the other countries of comparison. The International Social Science Council can contribute decisively to the development of world-wide co-operation among social scientists by opening channels for initiatives from a wider range of research groups and by facilitating the matching of research interests across the national boundaries. This is essentially the operational strategy of the European Coordination Centre for Research and Documentation in Social Sciences9 which was created by the Council in 1963 and has already established itself as a useful facility for a wide range of research groups anxious to initiate crossnational comparisons within Europe. Europe has in fact been ripe for such initiatives for some time: what was lacking was an organizational focus, a concrete institutional basis for concerted action. The Vienna Centre made it possible for a sociologist such as the Hungarian Alexander Szalai to develop, over a remarkably short period of time, a large-scale project of co-operative research across a dozen countries. The decision to set up the European Coordination Centre came just at the right moment: communications between sociologists in the East and the West had reached a point where co-operation on concrete tasks of empirical research was possible, and the regional organizations of the West had so far concentrated their efforts on purely economic studies and failed to offer a minimum of infrastructure for cross-national research in central fields of sociology. The initial successes of the Vienna Centre hold an important lesson for the future: cross-national research requires an institutional framework, an organizational basis. Great plans and important pilot studies can result from haphazard encounters of enthusiasts but a cumulative tradition of cross-national research can only develop within a clear-cut organizational setting. The demographers and the economists have been able to build up broad international professions within the framework of large-scale intergovernmental organizations: the UN, the Regional Commissions, the World Bank, the OECD and the EEC all offer continuous opportunities for experiences in the handling and evaluation of data masses from wide ranges of countries and help to develop genuine cross-national expertise. There is no such firm basis for cross-national endeavours in the other social sciences: in anthropology, in sociology, in political science. There is some movement in the fields closest to demography and economics. It is interesting to observe that the two Research Committees under the International Sociological Association which have moved farthest toward the development of a cumulative programme of cross-national studies are those dealing with the Family and Mobility: both of them focusing on variables close to the concerns of demographers and both relying heavily on data from enumerations or from surveys approximating the model of the census. In other fields of the social sciences it has proved much more difficult to develop

Introduction

9

continuous programmes: there have been no institutional frameworks for longterm commitments to cross-national inquiries and, still worse, hardly anything has been done to evaluate or to standardize the production of data across any two or more nations. This situation sets a series of important tasks for UNESCO and the International Social Science Council. The experts consulted all agreed on the need for active exploration of new strategies in the advancement of cross-cultural and cross-national comparisons. At the final session of the International Conference a number of proposals were discussed for further steps to be taken by UNESCO and the Councils. It proved impossible to reach full agreement on priorities among substantive projects but there was broad consensus on major lines of action to strengthen the institutional foundations for research initiatives across cultural and national boundaries. The following Recommendations were finally adopted: 1. Cross-cultural analysis The Council is urged to organize, if possible during 1966, a first international seminar on the methodology of cross-cultural comparisons on the basis of coded information of the type assembled in the Human Relations Area Files and other collections. This Conference should provide an opportunity for field workers and collection organizers to discuss strategies for the improvement of the basic data for comparison and at the same time allow social scientists from a variety of fields to acquaint themselves with the potentialities of cross-cultural analysis and to discuss the logic of inference procedures. 2. Comparative studies of social and cultural development The Conference welcomes the initiative taken by the Council in bringing historians and social scientists together in joint attempts at comparative analysis and recommends that efforts be concentrated in the first phase on the collation and evaluation of qualitative as well as quantitative data on processes of nation-building, urbanization, industrialization and demographic transformation. An interdisciplinary symposium should be organized in 1967 or 1968 to give detailed consideration to variations in sequences of nation-building and to discuss alternative theoretical approaches to the study of such processes. Further symposia on the comparison of national contrasts in processes of urbanization, industrialization and demographic change should also be planned within the same context. The Conference urges the International Social Science Council to give high priority to such studies of the phases of modernization and emphasizes the importance of such comparisons for an understanding of the prospects of social, cultural and politica development in the new nations of Africa and Asia. 3. Comparative studies of regional disparities of development The International Social Science Council is commended for its initiative in developing plans for a conference on regional disparities within the nations of South Asia and urges similar action in other areas and groups of countries in the years to come. The planning of symposia on regional disparities in Eastern and Western Europe should be given high priority in the programme: these are probably the areas where the data are most abundant

10

Stein Rokkan and Kazimierz

Szczerba-Likiernik

and where the prospects for theoretically significant comparative analyses of types of within-nation inequalities are the most favourable. These symposia should be closely co-ordinated with the work on backward regions within developed countries currently under way at the European Centre at Vienna. Action should also be taken to plan similar work in Africa. It is also recommended that the International Social Science Council take the necessary action to follow up these efforts through the compilation and evaluation of statistical indicators for the principal regions within each country covered in the symposia: this would add an important new dimension to the compilation of aggregate indicators for entire nations attempted in the Russett et al., World Handbook of Social and Political Indicators and also help to pinpoint weaknesses in current international statistics and to identify priority tasks for new data gathering operations. 4. The development of data archives The Conference urges the International Social Science Council and the International Committee for Social Sciences Documentation to take action to accelerate the development in different countries of historical data archives allowing computer processing of statistical time series for different units of the national territory. International cooperation among the developers of such archives is an essential condition for the exploration of the possibilities of detailed comparisons of rates of modernization, mobilization and integration. The Conference also asks the International Social Science Council to pursue its efforts to establish a federation of archives for survey data and archives for ecological data in Europe and urges action to develop close links between this federation and the American Council of Social Science Data Archives. Action should also be taken to bring into this network of co-operating bodies archives under development in other areas of the world, specifically in Latin America. To ensure effective communication among all those concerned to develop and to make use of such archives, the International Social Science Council is urged to establish a Committee on Social Sciences Data Archives during 1965 or 1966. 5. The advancement of the methodology of comparisons The Conference urges increased attention to the problems of analysis and inference in comparative research and emphasizes the importance of systematic confrontations of the procedures used in qualitative as well as in quantitative comparisons and in crossnational and cross-community comparisons. The Conference welcomes the plans developed within the International Social Science Council for a Seminar on Ecological Data and Analysis Techniques in 1966 and recommends that action be taken to organize similar seminars for the confrontation of procedures and techniques within other methodologies, particularly content analysis and experimental studies. This series of activities should be planned as steps toward the organization of a broadly conceived Symposium on the Theory and Logic of Comparative Research. The Conference also emphasizes the need to encourage systematic training in the techniques of comparative data-gathering and analysis and recommends that training seminars of the type already developed within Europe for comparative surveys be planned for other methodologies and other areas as well. The International Social Science Council should encourage national social science councils and national research centres to take initiatives in this direction.

Introduction

11

6. Documentary guides The Conference urges the International Committee for Social Sciences Documentation to co-operate with the International Social Science Council in developing further its plans for a series of Guides to Data on Comparative Research and urges specific attention to the need for handbooks of evaluated historical statistics for developed countries. The Conference urges, the International Social Science Council to encourage developments in this direction. 7. Co-operation with the disciplinary associations and national social science councils The Conference urges the International Social Science Council to take up contact with the disciplinary associations and with national research councils in the social sciences to enlist their co-operation in the programme of comparative research. Specifically each association should be urged to organize detailed reviews of problems and prospects for comparative research in its fields at its Congresses and to invite research workers in neighbouring disciplines to report on corresponding development in theirs. 8. Co-operation with UNESCO The Conference recognizes the decisive importance of the initiatives taken by UNESCO in advancing the theory and methodology of comparative cross-cultural and crossnational research and strongly recommends that the Organization accelerate its efforts in this direction and give full support to the projects planned within the long-term programme currently under development. On all these points, steps have subsequently been taken to ensure follow-up or to plan projects: 1. A Symposium on Cross-Cultural Research Tools in Comparative Social Anthropology got organized in Paris under the direction of Claude Lévi-Strauss and in close co-operation with officers of the Human Relations Area Files and other leaders in the movement of comparative anthropology : this, we trust, will be followed up by further action to ensure international co-operation in the development of retrieval systems for information about primitive societies and by a series of work groups exploring the possibilities of systematic comparisons of societies at different levels of complexity. 2. Plans for the organization of a Conference on Comparative Research on Processes of Nation-Building are far advanced and will be linked with a number of working parties on the establishment of historical data archives. 3. The Symposium on Quantitative Ecological Analysis will be followed by a Conference on Regional Disparities of Development in Eastern and Western Europe: this will provide the basis for a series of cross-national guides to the available timeseries data at different levels of territorial aggregation. 4. The Third Conference on Data Archives held in London in April, 1966, set up a Standing Committee on Social Science Data Archives. This Committee will take charge of further action to ensure detailed exchanges of information and technical know-how among archives and will seek ways of encouraging the buildup of a wide range of data files for computer processing : not only data from samples

12

Stein Rokkan and Kazimierz

Szczerba-Likiernik

surveys, but also from censuses and other official enumerations, from administrative records, and from historical and biographical sources. 5. Plans have been prepared for a series of working parties on the methodology of comparisons, at first discipline by discipline, later across all fields of the social sciences. A Symposium on Cross-cultural Psychology is scheduled to take place during 1969 and will be followed by discussions along similar lines within other disciplines. The UNESCO Department of Social Sciences is taking an increasing interest in the exploration of new procedures for the training of graduate students in techniques of cross-cultural and cross-national analysis. The strategy found most likely to pay off seems to be the 'computer plus data bank plus analysis tearrí procedure: bodies of prima facie comparable data for two or more countries are brought together for computer analysis and joint interpretation by small groups of social scientists familiar with the historical and structural contexts in each of the countries compared. This formula has still to be worked out in some detail but it is clear that this is a procedure much more likely to produce contextually controlled comparisons than the traditional exchanges of statistical tables and papers. 6. The International Committee for Social Sciences Documentation will continue its series of International Guides to Electoral Statistics and has proceeded to launch a parallel guide to Mass Media Statistics. A similar guide to Cultural Statistics will be discussed in 1967-68 and plans for a variety of guides to historical time series statistics will be worked out through further consultations. 7. Sessions on progress in cross-cultural psychology have been organized within the International Congress of Psychology in Moscow and a plenary session on "Cross-national Sociology" has been convened at the Sixth World Congress of Sociology at Evian. The International Social Science Council is in constant contact with the disciplinary associations and will urge continuous efforts to advance comparative studies. 8. UNESCO has increased its support for the programme in comparative research for the biennium 1967-68 and appears to be committed to continued support of the programme for possibly another eight years. The International Social Science Council will continue its efforts to increase the resources for the programme and has entered into close contact with a number of national research councils to explore the possibilities of joint action. On all these points the International Social Science Council and its sister organizations depend heavily on the enthusiasm of the devoted research workers in the national centres and on the good will of the officers of national councils and foundations. There is no cheap short-cut to the goal. The social sciences cannot be internationalized by fiat or from above: we must encourage the 'grass-roots' of the research community to take an active part in international work and this can only be done through the development of specialized institutions and facilities.

Introduction

13

NOTES

1. W.Buchanan and H.Cantril, How Nations See Each Other (Urbana, 111., University of Dlinois Press, 1953). 2. See S.Rokkan, 'Sample Surveys of Common Ideas About Foreign Countries', International Social Science Journal, 9 (1), (1957), pp. 121-128; E.Reigrotski and N.Anderson, 'National Stereotypes and Foreign Contacts', Public Opinion Quarterly, 234 (Winter 1959-60), pp. 515-528; and M.Brouwer, 'International Contacts and Integration- Mindedness : a Secondary Analysis of a Study in Western Europe', Polls, 1 (2), (Summer, 1965), pp. 1-11. 3. A first brief report on this Conference was published in Social Science Information, 1 (3), (1962), pp.32-38. A fuller report and a selection of the papers appeared in a special issue on 'Data in Comparative Research', International Social Science Journal, 16(1), (1964), pp. 2-97. 4. The papers of this Conference have been published in R.Merritt and S.Rokkan, eds., Comparing Nations (New Haven, Conn., Yale University Press, 1966). 5. A report of the First Conference on Social Science Data Archives (Cologne, June, 1963) appeared in Social Science Information, 2 (4), (1963), pp. 109-114. The papers of the Second Conference (Paris, September, 1964) have been published in S.Rokkan, ed., Data Archives for the Social Sciences (Paris, Mouton, 1966). A report on the Third Conference (London, April, 1966) appeared in Social Science Information 5 (1), (1967) pp. 63-68. 6. See reports in Social Science Information, 4 (2), (1965), pp. 156-172; and in Revista latino-americana de Sociologia, 65 (1), pp.39-151. 7. See report in Social Science Information, 2 (4), (1963), pp. 98-103; and the chapters by E.Allardt, S.Rokkan and H.Valen in Comparing Nations (New Haven, Conn., Yale University Press, 1966). 8. See Introduction to S. Rokkan, ed., Data Archives for the Social Sciences. 9. See reports in Social Science Information, 2 (3), (1963), pp. 90-123; 3 (3), (1964), pp. 70-76; and 4 (2), (1965), pp. 128-155.

CHAPTER

I

The logic of cross-cultural analysis: why exceptions?* ANDRÉ J. F. KÔBBEN University of Amsterdam

1.

Introduction

In the course of time hundreds of statements have been made in cultural anthropology of the type 'where A, there B'. Or again, 'where A, there B, there C...'. Statements, therefore, concerning the functional relationship between two, three or a whole series of social phenomena. A statement of this type may be made merely in passing or even remain implicit in the argument, or it may be explained at length and more or less carefully tested. The remarkable thing is, now, that such statements never hold good in all cases (for all societies). There are always exceptions - deviant cases - to the rule, sometimes even quite a few. An extreme example is found in Murdock (1949), p. 152. The hypothesis put forward there is 'where bilocal residence, there kinship terminology of the generation type'. In order to test this, he examines 20 bilocal societies, nine of which prove in fact to have generation-terms. Eleven of them, however, have a different type of terminology and are, therefore, exceptions. In addition, Murdock notes 33 societies which are not bilocal, yet do have generation-terms (and are thus also exceptions). Even so, Murdock need not discard his hypothesis: Kinship Terminology

Residence rules Bilocal Other

Generation terms 9 33 Other terms 11 187 The large number of societies in the fourth quadrant (187) compels us to assume the existence of a significant correlation between bilocality and terminology of the generation type.1 * Translation: Mrs. M. J. van de Vathorst. A variant version of this chapter was printed in Current Anthropology, 8 (1-2), (1967), pp. 3-19 with comments by a dozen anthropologists and a reply by the author, pp. 19-34.

18

Andre J. F. Kobben

Given this situation we have accustomed ourselves to saying that cultural anthropology may discover regularities but not scientific laws. Some anthropologists even make this observation in a somewhat self-complacent fashion, as if it were slightly vulgar to be looking for, or to find, 'laws'. Anthropologists who have succeeded in establishing the existence of a statistically significant relationship between two social phenomena are usually quite satisfied with their achievement and consider their task completed. In this belief, however, they are wrong. Exceptions are there to be explained. For reasons we shall discuss more fully later on (see in particular sections 9 and 10), I do not believe we shall ever achieve laws that are totally without exceptions. Not even if we had nothing but perfect data and nothing but perfect anthropologists. Many exceptions, however, may be eliminated, either by proving them to be spurious or else by including them in the rule. The latter method, besides making the 'law' statistically more significant, has the advantage of giving it a greater subtlety, thus adding to its value. Here, as elsewhere in this paper, I propose to use the phenomenon of slavery as a paradigm to illustrate my point. One of the earliest works in which a serious attempt is made to discover intercultural regularities is H. J.Nieboer's book about slavery.2 One of his hypotheses is 'where hunting, food-gathering and fishing are exclusive means of subsistence, there no slavery'. Testing of this hypothesis does in fact show a clearly significant correlation. But there is one important exception, namely, the North Pacific Coast Indians, who do have slaves. It is all to Nieboer's credit that he attempts to give an explanation for this exception. For one thing, he points out that these people primarily live on fishing and have a plentiful supply of food. This makes it possible for them to dwell together in fairly large groups and have permanent settlements (whereas in general hunters cannot afford to do this). For the same reason, fairly big differences in wealth and status may arise.3 Now he may correct his thesis as follows: 'where hunting, foodgathering and fishing, there no slavery, unless there is a plentiful supply of food, and for that reason... etc.' Consequently, our understanding of the phenomenon of slavery is increased.

But is it quite correct to say that there are no laws without exceptions in cultural anthropology? This is not true of statements on the highest level of abstraction and therefore, of the highest triviality, as for instance: 'where rapid and sudden change, there violent reaction'. But we may with impunity disregard such laws. Some authors, however, make statements on a lower level of abstraction, yet do claim them to be without exceptions. For different reasons, however, their claims are not, or at least not completely, tenable. Let me give four examples: S.N.Eisenstadt (1956); K.A.Wittfogel (1957); G.Sjoberg (1960); R.B.Textor (1967). Eisenstadt discusses age-groups, and particularly the question of why these occur in one type of society but not in another. One of his hypotheses is that they

The logic of cross-cultural analysis: why exceptions?

19

are found wherever the family unit impedes the attainment of full social status by its junior members (p. 248). He is so sure of this being a general rule that he does not even consider it necessary to test it systematically (pp. 123-124). One of the examples he gives is that of the Nuer. Now what causes him to think that their family units so impede their junior members? Mainly the fact that they have age-groups! In other words, there is an element of circularity in his argument. Small wonder, then, that he finds no exceptions.4 Wittfogel believes there is a sufficient and necessary relation between despotic societies on the one hand and hydraulic works (on a large scale) on the other. Therefore: 'where hydraulic works, there despotism' and: 'where despotism, there hydraulic works'.5 No doubt this rule holds good in many cases, but there are certain societies which do have hydraulic works, but no despotism (Maya, Cambodia, Sinhala), and on the other hand some which have no hydraulic works, but do have despotism (Ruanda, Urundi, and the other so-called Hima-states). Wittfogel himself names a number of other cases - ranging from the Roman Empire to the Soviet Union - but these, he says, arrived at despotism through diffusion. They have borrowed this cultural trait from societies with hydraulic works, so these are the ultimate causes after all! Wittfogel is thus able to maintain his thesis intact in the first place because he does not present all relevant data, in the second place because he attributes too much to the factor of diffusion, and in the third place - a factor which I have not yet mentioned - because he is not sufficiently explicit in defining his terms.6 Sjoberg opens his book about the pre-industrial city with the statement that 'the city and civilization are inseparable'. Thus we find, here too, the claim of a necessary and sufficient relation. Much depends, however on the content of the terms employed. Sjoberg's condition for 'civilization' is the presence of writing. In that case, how are we to regard the Yoruba towns, which are indeed non-literate but otherwise fulfill quite well our usual notions of 'urban'? They may be - and this was equally true in traditional times - very populous, some of them having more than 50,000 inhabitants. They have an extremely high density of population, a fair amount of division of labour and functional diversification. And, finally, their inhabitants have a typically urban mentality.7 Sjoberg gets out of the difficulty by calling these towns 'quasi-urban', for the very reason that they are non-literate. He uses this same device for the Inca, Ashanti and Dahomey towns. In itself this procedure is not inadmissible, but it does detract from the heuristic value of his thesis. Textor's book, which is entitled A cross-cultural summary, is of quite a different order. In its 2700 pages more than 20,000 statistically significant correlations are presented, calculated by the computer. Of these about twenty are exception-less. So it is here not so much the author himself who offers rules without exceptions but rather the machine. These are, however, of a special character, in so far that in

20

Andre J. F. Kobben

all of these cases only an extremely small number of societies are involved in the comparison. For instance: Fear of ghosts, spirits, etc. High Low

Structural complexity High Low 2 0

0 4

Consequently the predictive value of this outcome is particularly low (X 2 = 2.34; P = 0.066). The addition of a handful of new cases may radically change the picture. No, for the time being I will let my own thesis stand unchanged, worded in the following somewhat Cretan formula: 'where an anthropological rule, there exceptions'. And then I ask the question: why these exceptions? How should they be explained? This paper is intended to provide the beginning of an answer to such questions. I would point out, however, that this limited answer has a wide application: It is equally or almost equally valid for the other social sciences, including history. Ten, or perhaps eleven, factors, which cause exceptions to occur, are to be distinguished. These are: 1. Defective classification: defective procedures of ethnologists 2. Defective classification: defective procedures of ethnographers 3. Multicausality 4. Pluricausality 5. Functional equivalents 6. Intervening variables 7. Diffusion; external contacts 8. Cultural and social lag 9. Coincidence 10. Personality 11. Combinations of two, or more, or all of these factors. We shall now take a closer look at each of these factors in turn. 1. Defective classification by ethnologists (comparative anthropologists) Assuming the hypothesis 'where A, there B', we get the following situation: B Not-B

A I. A, B III. A, Not-B

Not-A n . Not-A, B IV. Not-A, Not-B

The cases of (A, B) and (Not-A, Not-B) are in accordance with the hypothesis; (A, Not-B) and (Not-A, B) are in conflict with it. Now exceptions result if what should have been called 'A' is mistakenly called 'Not-A'. Thus, cases which be-

The logic of cross-cultural analysis: why exceptions?

21

long in the first quadrant (A, B) are placed in the second (Not-A, B). Similar mistakes may be made with respect to the other quadrants. Example. One of Nieboer's hypotheses concerning slavery is: 'where all land in use (where there is a shortage of land), there no slavery'.8 He reasons that where there is a land shortage, there will be people who have no land and will thus be obliged to offer their services to the land-owners. In such cases, slavery would simply be unnecessary. He then proceeds to test this hypothesis by examining data concerning Micronesia, where in most cases all available land is in use.9 Great is his disappointment: many ethnographers report that 'slavery' does occur there. On closer inspection, however, it appears that those ethnographers and Nieboer differ in their interpretations of the terms 'slave' and 'slavery'. Nieboer (p.9, p. 30) calls a slave a person who is another person's property and who is therefore forced to stay with his master. The so-called 'slaves' of Micronesia, however, prove to be landless people, desperately poor and in a position of subjection with respect to the landowners. But such a 'slave' in the service of a master may yet, as it proves, leave the latter and place himself in someone else's service, providing the other person will take him. Far from conflicting with Nieboer's hypothesis, therefore, these cases prove to lend it strong support. In this case, the spurious 'exceptions' were exposed as such, but in many cases the ethnologist relies on the ethnographer's terms - often no other procedure is possible and his classification will, in consequence, be incorrect. A natural objection that might be raised is that in this way, though 'right' cases may be classed as 'wrong', 'wrong' cases may equally be classed as 'right', so that with large numbers the net effect will be near enough to zero. This, however, is not true, at least not in cases where there is a real correlation between the phenomena A and B. In order to understand how this can be, let us assume a correlation between A and B in 90 % of the cases, so that a correct classification of all cases would yield the following picture:

B Not-B

A

Not-A

I. 90 III. 10

II. 10 IV. 90

But mistakes will of course occur and some cases will be placed in the wrong quadrant. Now it is clear that there is a 90 % chance of a 'right' case being classed as 'wrong' - that is, of being put in quadrants II or III instead of I or IV - while there is only a 10% chance of a mistake being made the other way round. 10 It is important to place some emphasis on this fact. So often we hear doubts expressed as to the value of significance tests: 'what is the use of them, if all sorts of mistakes are made in classifying the cases?' We must reply that if the result is significant in spite of this, it pleads very strongly for the correctness of the hypothesis, since incorrect classification detracts from the result. Unless - and this is an important restriction - unless the ethnologist is biased, so much hoping that his hypothesis will hold good that his classification of the cases he examines, is affected. This may be explained as follows:

22

André J. F. Köbben

Assume that there is in reality no correlation at all between A and B, so that: A 8 8

B Not-B

Not-A 8 8

Assume further that the ethnologist driven by bias puts 8 of his 32 cases in the wrong categories, thus:

B Not-B

A 8+4 8-4

Not-A 8-4 8+4

Then, by doing so, he has created a statistically significant correlation between the phenomena A and B, which does not exist in reality. Example 1. Horton (1943, p. 277) assumes a relation between insobriety and premarital sexual freedom. He tests his hypothesis with the following results: Premarital sexual freedom

Strong

Relatively unrestricted

20

Relatively restricted

4

Insobriety Moderate or slight 4 After correction 7 (or 8) 8 After correction 5 (or 4)

These results conform to his expectations (X 2 = 9, which is significant at the 1.0% but not at the 0.1 % level)... but has he classified his cases correctly? On careful re-testing of the 8 cases in the fourth quadrant, I found that at least 3 and perhaps 4 of these belong in the second one, that is, with the exceptions! When this correction is carried out there is no longer any question of a statistically significant relation. 11 One cannot avoid having the impression that prejudice in favour of his hypothesis has led the author astray. Example 2. Beatrice Blythe Whiting (1950) puts forward the following hypothesis : 'where superordinate punishment, there sorcery unimportant; where no such punishment, there sorcery important'. A test amply confirms her hypothesis: 1 2 Sorcery Unimportant Important

Superordinate punishment Existant Non-existant 12 5

3 30

But did she classify her cases correctly? Pilling (1962, pp. 1057-59) tells us how, distrusting these results, he checked 4 of the 30 cases in the fourth quadrant and that in all four of these he found the classification to be incorrect. They really belonged in the third quadrant (superordinate punishment - sorcery important), that is, with the

The logic of cross-cultural analysis: why exceptions?

23

exceptions! Pilling does no further checking, considering it unnecessary to do so, and merely suggests that her approach would seem to be worthless. One of the members of my staff went to the trouble of going over the same ground once more, rechecking 10 of the 30 cases in the fourth quadrant, including 3 of the cases checked by Pilling. His conclusion, strangely enough, was that Whiting's classification was correct in all of these cases (Bax, 1964b)! The bias in this matter, therefore, would appear to be on Pilling's part. On the basis of these and similar experiences it is possible to set up a number of practical rules for intercultural comparisons: Use only those ethnographies which give incontrovertible information concerning the phenomena in question. Try to find a clear criterium (an operational definition) of that phenomenon. Let the classification be carried out by two neutral judges (preferably two who are unaware of the hypothesis that is to be tested) and use only those cases which both these judges, separately, place in the same category. Definitions. The examples that have been discussed so far have shown that much depends on the definitions of the terms we use. What exactly do we mean by 'slavery', by 'insobriety strong' or 'moderate', and what do we mean by 'superordinate punishment'?13 Of course, this is a matter of convention. Radcliffe-Brown (1946, p. 39) expresses it in this way: 'Words and the concepts expressed in them are the tools of scientific reasoning. When we are defining abstract terms, there is no question of one definition being right and another wrong, but only of which definition gives us the most useful tool for scientific analysis and generalization.'

This is so true that it is almost a truism. Nevertheless, it may be profitable to repeat it, since only too often anthropologists apparently assume that a particular term must irrevocably represent a particular thing - and thus are guilty of what in logical terms is called realism or substantialism.14 This is a continual source of misunderstanding and of 'exceptions'. Take, for instance, the concept of law. Murdock(1955, pp. 4-5) declares that there are many 'universals of culture': he names no less than 73 elements which, according to him, are common to all human cultures. These include, for example, property, government, trade, and law. Is it true that all societies have law? Yes and no: it all depends on our definition. To put it as Murdock does is a piece of 'realistic' thinking. Radcliffe-Brown (1952, p.212) defines law as 'social control through the systematic application of the force of politically organized society'. From this point of view there are hundreds of societies which have no law. An example of this are the Nuer. Their ethnographer (Evans-Pritchard, 1940, p. 162) does, in fact, tell us that they have 'no law in a strict sense', since there is 'no authority with power... to enforce a verdict'.

24

Andre J. F. Kobben

Hoebel's (1954) conception of law is somewhat less restrictive, although this is perhaps not immediately clear from his definition. He would also call 'law' those cases where the hurt individual himself takes action to redress the wrong against him, at least if public opinion upholds him and sanctions his actions. 15 Looked at in this way, the Nuer do have law,16 and so do many other groups. But this still doesn't make 'law' a universal phenomenon, for Hoebel (p. 28) will use the term 'law' only where there is, 'in threat or fact, application of physical force by an individual or group'. Now there are many groups which do not fulfill this requirement. This is true, for instance, of the Bete (Ivory Coast). An old man was surprised in the act of washing the mud off his feet in the village well. In doing so he committed an offence. In the court (?)-session devoted to this case, the man's guilt was established and after some hours of deliberation he was fined a goat. The culprit, however, declared in forceful terms that he was not going to pay up. He was heard in silence, after which the meeting dispersed. When I asked what was going to happen now, they replied: 'Nothing! But if he ever appeals to us in future to try a case for him we shall refuse'. There was, however, no question of physical coercion at all. Only when we employ the definition given by Pospisil (1958, p.267) do all human societies have law, including the Bete. Pospisil, in fact, would speak of law if any kind of coercion at all is used. Even if it is psychological coercion: ridicule, contempt, boycott. And, also, if the pressure brought to bear on the individual consists of the denial of rights or privileges. Now, let us assume the following hypothesis: 'only there law, where more complex and more extensive political structure'. This would be true according to Radcliffe-Brown's definition; not true according to Hoebel's ('many exceptions'); absolutely untrue according to Pospisil's. Conversely the thesis: 'where human society, there law' (Murdock) is true according to Pospisil, not true ('many exceptions') according to Hoebel and absolutely untrue according to RadcliffeBrown. Still, we haven't yet discussed the main cause of exceptions-through-classification, at least we have not explicitly done so. This is: representing as discontinuous what is really a continuum. 17 The anthropologist, when working on intercultural comparisons, generally classes his cases on an all-or-none basis. A simple example of this is found in Tylor (1889, p. 247), who advances the thesis 'where mother-in-law taboo for the man, there matrilocal residence'. In order to prove this, he divides societies into: A = such a taboo present

Not-A = no such taboo present

While actually these societies should be placed on a continuum:

The logic of cross-cultural analysis: why exceptions? 1 p

1 Q

1 R A

1 S

1 U

1 V Not-A

1 W

25 1 X

One extreme is'P': absolute avoidance of mother-in-law by ego for as long as the marriage lasts. The other extreme is 'X': informal behaviour on a basis of equality between mother and son-in-law. Q.: initial strict avoidance, but becoming less strict after the birth of the first child or after a number of years. Or else: strict avoidance observed by persons of the upper stratum, less strict avoidance in the lower strata. R.: mother and son-in-law may meet, but not without a third person present. They may not address one another directly nor eat in one another's presence. This continuing with ever-diminishing intensity (S:...;T...; U:.; V:...). Untilfinally: W.: mother and son-in-law must avoid using obscene language in one another's presence. What we do now, in intercultural comparison, is to place a caesura in a rather arbitrary fashion, say a t ' T ' . The whole field to the left o f T we then call A and that to the right we call not-A. But it is quite possible that in the cases under'S' the mother-in-law taboo is of so little importance that it has scarcely any social effect or none at all: these are going to be exceptions. Conversely, it is possible that in the cases under 'U' or ' V' the mother-in-law taboo is still so strong that it does have a social effect: these are exceptions the other way. It is therefore advisable to make it a rule in intercultural comparison only to use extreme cases.18 The fact that cultural anthropology has to use this type of dichotomy clearly demonstrates how great a distance separates us from the natural sciences! Imagine a science of physics in which Boyle's law went like this: 'where pressure great, there volume small, where pressure small, there volume great'. Instead of 'P x V = C \ We can also put the same thing into different words: the natural sciences and economics usually employ the method of concomitant variations, while we use the so much poorer tool of the method of agreement and difference.19 But we have the excuse - a convincing enough one! - of having no choice. Given the nature of our data, the method of concomitant variations is practically never applicable20 and it looks as if for the time being it will stay like that. But we can try and use more than the two categories of 'A' and 'Not-A' and thus get a little closer to reality. In our previous example, for instance, this might be done as follows: Strict and permanent (prolonged) avoidance

Moderately strict and temporary avoidance

No avoidance or only to a slight degree

Such a division into three categories is used notably by Murdock (1949, p.277). A point is, however, that this remedy in its turn produces the disease, because by using two caesurae instead of one the chances of creating exceptions are doubled.

26

André J. F. Köbben

In rather the same fashion Textor, in his earlier-mentioned computer book, divides societies into five categories as to political structure.21 Large state

Little state

Minimal state

Autonomous Independent local community families Unfortunately data are only put into the computer in dichotomized form, and results are exclusively in two-by-two tables. Now in order to do justice to the continuous character of his subject, Textor runs his data through the machine four times. The first time he divides his societies into two groups: those having the 'large state' versus all the other ones together. And these two categories are then correlated by the machine to a large number of social phenomena. The second time he takes 'large state' plus 'little state' as one category, and 'minimal state' 'autonomous community', 'independent families' as the other one. Etcetera for the third and fourth runs. In this way he does not yet represent reality in his tables - naturally that will never be possible - but he approximates it not too unsatisfactorily. It may be good to point out in this connection that even a continuum never does do full justice to reality. Let us take the example of matrilinearity. We may divide all societies into two categories: Matrilineal A better model is a continuum:

versus

Non-matrilineal

Extremely

Matrilineality

matrilineal

absolutely

of no

significance

I

I

P Q R S T U V W X But still more adequate is a rendering as in the diagram shown below. It is based on six features of matrilinearity. The six lines radiating from the centre each represent a continuum with this point indicating the total absence of the feature in question and the other end of the line its presence to an extreme degree. The diagram has been filled in for two matrilineal societies with which the author is acquainted. Inheritance of offices completely matrilineal

Inheritance of goods completely matrilineal

Absolute uxorilocal residence

Matri-group absolutely exogamous

Descent only reckoned along matrilineal lines

Matrilineal kin highly corporate group Djuka

Agni

The logic of cross-cultural analysis: why exceptions?

27

Operational definitions. To make sure that things which are equivalent are actually classified in the same category, it is important to use operational definitions. That is, descriptions which offer a clear criterion of classification. Such criteria are often numerical ones. Murdock (1949, p. 28), for instance, wishes to designate a society as characterized by polygynous households if 20 % or more of the households in that society are in fact polygynous. But the criterion used need not necessarily be of a numerical nature. Take the thesis proposed by M. Fortes: where simple society, there no unilineal groups, or, in his own words: 'Unilineal descent groups are not of significance among peoples who live in small groups, depend on a rudimentary technology, and have little durable property'. (Fortes, 1953, P. 24). It is impossible to test the hypothesis in this form. What do we mean by 'small groups'? and what is 'rudimentary technology'? or 'little durable property'? We can only get somewhere if we translate this into operational terms. In this case a workable criterion for 'simple society' is 'society of hunters and collectors'. The thesis then becomes: where hunters and collectors, there no unilineal descent groups. This is easily tested with the help of Murdock's (1961) World Ethnographic Sample:

Non-unilineal kinship structure . kinship structure

Hunting and collecting Dominant Non-dominant ^ j ^

23

311

Although there are many exceptions (23 +178 = 201), there is a clearly significant correlation and consequently the thesis may be regarded as tenable. But of course, the nominal and operational definitions do not completely coincide - in other words, the validity of the operational definition is not perfect. 22 For there are societies of hunters and collectors which are not simple in Fortes' sense, and on the other hand there are societies of cultivators which are very simple indeed. This lack of coincidence of the nominal and operational definitions is, in its turn, a source of exceptions! Take the Tsimshian (North West Indians): they are hunters and collectors; but not 'simple': unilineal structure. Take on the other hand the Kiman (New Guinea) who do have intensive cultivation; nevertheless a 'simple' society: no unilineal structure. 23 An elegant application of the operational definition - though they do not actually use the term - is offered by Homans and Schneider in their well-known and much criticized study on unilateral cross-cousin marriage. In which cases, they ask themselves, will ego be willing to marry mother's brother's daughter and not father's sister's daughter? In those cases where mother's brother does not hold jural authority over him but father's sister does. Such a situation normally exists in a patrilvasaX society. And conversely: in which cases will ego not be willing to marry his mother's brother's daughter

28

Andre J. F. Köbben

while having 110 objections against father's sister's daughter? In those cases where mother's brother does hold jural authority over him, but father's sister does not. This second situation is normal for a matrilineal society. They test their hypothesis as follows: Matrilateral cross-cousin marriage Patrilateral cross-cousin marriage

Patrilineal 22 2

Matrilineal 4 5

So there are 2 + 4 = 6 exceptions. Homans and Schneider differ from most authors in that they attempt to explain these exceptions. They note that 'patrilineal' is merely a translation into operational terms of: jural authority held by father's sister, not by mother's brother. And 'matrilineal': jural authority held by mother's brother, not by father's sister. Practically all their exceptions prove to be due to the fact that nominal and operational definitions do not coincide. In the four matrilineal cases, for instance, the locus of authority proves not to lie with mother's brother, although these societies are matrilineal, but with the father and his relatives. As Homans and Schneider say themselves: 'we were wrong for the right reasons'. A similar situation exists in one of the two patrilinealpatrilateral societies. An interesting circumstance is that Homans and Schneider proceed also to check their positive cases, i.e. to see whether the nominal and operational definitions coincide. In one case this proves not to be so and this they immediately relegate to the exceptions: the society concerned is that of the Yir Yoront (Australia) who, according to the sources consulted by Homans and Schneider are patrilineal, have matrilateral cross-cousin marriage while nevertheless the locus of authority lies with mother's brother. Ego is treated as an equal by father and father's sister, who will stand by him and protect him whenever necessary. As Homans and Schneider have it: 'in this case we were right for the wrong reasons' (p. 41). 2. Defective classification by ethnographers Under this category we may class those cases where the ethnologist (comparative anthropologist) classifies incorrectly because the information supplied by the ethnographer is insufficiently precise, or incorrect. Take the example of those two legendary anthropologists who studied the same African society. One stated that divorce was 'relatively unimportant', the other said divorce was 'relatively important'... the former being an American, the latter an Englishman. Gluckman (1950, p. 190) has advanced a proposition on this subject. As he says: 'divorce is rare... in tribes organized on a system of marked fatherright, and frequent and easy to obtain in other types'. Actually he was obliged to test this hypothesis with data which were to a large extent no less impressionistic than in the example just given. What, then, does it mean if a certain case is classed with the category 'patrilineal/divorce frequent', that is, with the exceptions? Perhaps it is a 'true' exception, but it may just as well be the ethnographer who is to blame. A second example is found in Murdock's Social Structure. In this book residence is given a key position, for Murdock sets out to show that changes in kinship structure take place through changes in residence. Matrilinearity, for instance, does

The logic of cross-cultural analysis: why exceptions?

29

not change directly into patrilinearity; a change in the economic or political conditions of a matrilineal society causes patrilocality, which in turn causes patrilinearity. Everything therefore depends on the reliability of his data concerning residence. Among his 'sample' of 250 peoples are the Ashanti, whom he classifies as being patrilocal. And rightly so, as long as he relies on that great ethnographer of the previous generation, R.S.Rattray (1929, pp.2-3,22). According to the latter an Ashanti household consisted of ego plus his wife (wives) and his unmarried children; his married sons plus their wives and children; with, in addition, as the case may be, ego's mother and/or younger brothers, slaves, etc. Rattray states this quite positively and makes no reference to any other possibilities. Only in quite another connection and on a different page altogether (p. 49) he says in a note: 'We seem... to have an indication that matrilocal unions were not unknown.' Shortly after Murdock wrote his Social Structure, however, new data concerning residence among the Ashanti were published. Fortes (1949, pp.54ff), who collected very precise data concerning two Ashanti villages, devotes fourteen pages to residence. In more than half the number of marriages, husband and wife prove not to be living together at all. Especially in the beginning, when the marriage is new, they both stay at home (in their own compound) and pay each other visits both ways. Sometimes it stays like that, in other cases either the wife goes to live with her husband (virilocal), or the husband with his wife (uxorilocal). Altogether six types of households may be distinguished. The strictly virilocal and patrilocal type described by Rattray is one of these six and is of relatively rare occurrence. We might now console ourselves by reflecting that this is only one of the 250 cases in Murdock's sample. Unfortunately, however, most of Murdock's sources are equally impressionistic. Only in this case we happen to have proof of it! Of course, this is not to be held against Murdock, since those sources were all he had to go by. It should be added that Rattray is regarded, and deservedly so, as one of the best among the older ethnographers. Moreover, in his World Ethnographic Sample (1961, p.205) Murdock has most ingeniously managed to include the newer and far more complicated data furnished by Fortes. Let us continue with Rattray and Murdock a little longer. Rattray (1929, p. 23) informs us that the Ashanti have cross-cousin marriage and he makes it appear as if this is the usual type of marriage. The only restriction he makes is that 'biological and other reasons must sometimes [my italics] have rendered this ideal impossible of fulfilment'. Fortes (1950, pp. 281-282) however, found that of 525 Ashanti women only 8% were married to their cross-cousins! It is true that this percentage was undoubtedly higher formerly. Fortes can even prove that this was so: of the women under 30 years of age, about 2 % were married to cross-cousins, of those over 50 years, about 14%. But even the latter percentage is surprisingly low when seen in the light of Rattray's account.

30

Andre J. F. Kobben

Something similar is reported of the Tswana (South Africa). Their ideal is marriage to a cross-cousin, but Schapera (1950, pp. 156ff) found that of the 2500 marriages he checked, only between 4 and 5 % were really between cross-cousins! In other societies, however, this percentage is much higher: for the Polynesian Tonga it is probably about 40%, for the Muria (India) around 90 %.24 Murdock (pp. 173-174), however, lumps all these societies together and puts them indiscriminately into the category of 'cross-cousin marriage present'. Once more we can hardly blame him, for his data don't permit any other course of action. But no wonder, then, that he meets with exceptions when he tries to establish a correlation between this phenomenon ('cross-cousin marriage present') and another ('same term for father's sister and wife's mother'). Evidently, therefore, it is important to know the degree (frequency) of occurrence of the phenomena in question in the various societies. What we also need to know is their intensity. Take the rule proposed by G.M.Foster (1962, Ch.3): 'where peasant society, there interpersonal relations strained: suspicion, enmity, envy, lack of cooperation'. Foster gives a number of examples to illustrate his thesis. But naturally these phenomena (suspicion, enmity, etc.) occur in all human societies, so that Foster may find his thesis confirmed in every case. (It should be added that Foster uses only the Method of Agreement and does not inquire into the possible occurrence of the same phenomena in non-peasant societies.) The question is how frequently this suspicion and enmity occur in a certain society and, particularly, with what degree of intensity. Only if the intensity is greater than in any other type of society, does his proposition really have some value. Unfortunately, no one gives an exact indication of this intensity and I must admit that it would be extremely difficult to find a way of measuring it. The conclusion is inevitable that, for exact intercultural comparisons, exact ethnographic data are needed, preferably of a numerical nature. Happily we may note a growing tendency in this direction in cultural anthropology. In his study on divorce mentioned earlier, Gluckman (1950, p. 190) complains that 'the literature is not clear on most points', and that he has no numerical data at his disposal. At present, that is fifteen years later, it would almost be possible to retest his thesis exclusively with societies for which the degree of divorce is known exactly.25 This is in no small part due to Gluckman's own work and that of his collaborators. Now, by a caprice of fate, it looks as if such a retest will not leave much of his thesis intact, and I do not doubt but this will be a source of satisfaction to Professor Gluckman! Statistical data are not a necessity for all the phenomena we study. We must know the frequency or degree of occurrence, for instance, of cross-cousin marriage, polygyny, or divorce in societies where these are legally permitted. No such data are necessary when it is a question of whether or not such things are legally permitted in a certain society. In Dutch society it needs no statistical research to know that polygynous marriages are prohibited!

The logic of cross-cultural analysis: why exceptions?

31

And compare linearity and residence. Of the Tonga, Elisabeth Colson (1958, p. 23) says: 'though descent and inheritance are matrilineal, residence is a matter of personal preference'. This is true of many societies, and for this reason an anthropologist should always scrupulously collect statistical data concerning residence, particularly in a matrilineal society. For the reckoning of descent such data are usually not necessary. But here again there are exceptions. In some societies (notably Papuan ones) the individual is free to choose whether he will live with his mother's or with his father's relatives and to reckon his descent, correspondingly, either with his mother or with his father. For such cases the term ambilineal society is used.26 In other respects too, the individual in these Papuan societies is allowed a considerable amount of variations and alternatives in his behaviour,2' more so than in many African societies. Hence the greater need, in general, for statistical data where Papuan societies are concerned.

Since, then, the data supplied by ethnographers are often defective, let us try to determine what are the causes of this, apart from the more ordinary ones such as superficial observation, errors, and allowing oneself to be hoodwinked. In this connection, we should distinguish between three things: - social structure itself; - the informant's image (model) of it; - the ethnographer's image (model), influenced as it is by his personality and by the 'school' in which he is educated, and by the culture, or subculture, in which he has lived. Image (model) of the informant. Many ethnographers rely rather heavily on one or a few informants only, a dangerous procedure even - or: particularly - if these are honest, interested and intelligent. What is obtained in such cases is not the society itself, but the model of this society as it exists in the mind of the informant. This model may, and often will, deviate greatly from reality, since such an informant is necessarily biased in all sorts of ways. He may, for instance, be polygynously married, be a member of the upper class28 and a landowner... all these factors influence his view of the social situation. Even irrespective of this, the informant will tend to represent the ideal existing in his society as if it were reality although he knows quite well that this ideal is often departed from. Either because he is ashamed of such departures or else because he does not consider them worth mentioning. A few examples (to which I am sure every fieldworker may add others from his own experience). Djuka (Surinam): All the members of a kinship group ought to form one solidary group. So the informant says: 'they form one solidary group'. Agni (West Africa): during one particular week the villagers ought not to work on the land. So the informant says: 'They do not work during this week'.29 An Indian village: Brahmins should not pay marriage-money; so the informants say: 'Brahmins do not pay marriagemoney'. 30 Winnebago: Boys should not visit girls during their menstruation period; so the informant says: 'they don't visit them...' etc.81

32

André J. F. Kdbben

Even in the case of phenomena involving no strong emotions misrepresentations may occur. Both the Bete (West Africa) and the Djuka (Surinam) have a considerable measure of polygyny. When these people are asked how that is possible (whether there is no shortage of women) the reply is, in both cases, that there are considerably more women than men. In actual fact the opposite is true, there being a surplus of men.32 An Agni (West Africa) informant assured me that sons always grow up in their father's house. When he found out that I was checking this statement of his, he took offense: 'don't you trust me, maybe?' This in spite of the fact that his sister's son was living with him and not with his own father.33 For this reason, Lévi-Strauss (1953, p. 527) even talks about the model hiding social structure like a screen. Of course, this model is interesting enough in itself, and as such worthy of our attention. The model, moreover, which exists among the members of a certain society should certainly constitute a factor in our appreciation of that society; but it is imperative to know the extent to which it is merely a model and to what extent it corresponds to the actual social structure. And that is something about which we often lack sufficient knowledge. The image (model) of the ethnographer. Every ethnographer returns from the field with a chaotic and confused mass of data. From these he must extract a coherent report; he must somehow organize them. What inevitably follows is what Den Hollander (1955, pp. 235-236, 255) calls the rounding-off process. He organizes his data according to certain guiding principles, supplementing, extrapolating and putting in accents not compellingly prescribed by these data. This is by no means a reprehensible way of proceeding unless data are suppressed or invented. On the contrary, an anthropological monograph that has no well thought-out plan behind it, that has no themes, is a bad monograph. Pure description doesn't exist. But the comparative anthropologist should keep in mind that he only sees societies through the optics of the ethnographers. For the comparison of different cultures this is a very real handicap.34 A famous example is Malinowski who, in his Crime and Custom in Savage Society (1926), described the whole of Trobriand society in terms of reciprocity. Chapter 8 is entitled 'The principle of give-and-take pervading tribal life'. Chapter 9: 'Reciprocity as the basis of social structure'. Under these headings he deals with obvious matters such as the exchange of fish and tubers which takes place between the people of the coast and those of the interior. But he also regards as an instance of reciprocity, for example, the tears a woman sheds for her husband at the latter's burial, - these tears being shed in return for the gifts she has received from her husband's relatives. Similarly: the gifts a husband gives to his wife and children in return for her being married to him. It is interesting to note that Malinowski, as he says himself, did not arrive at this conception until many years after his fieldwork.35

The logic of cross-cultural analysis: why exceptions?

33

This viewpoint has not failed to produce its effect. Thus we find Hoebel (1958, p. 346) saying: 'The principle of reciprocity is operative in all societies. However, some societies are content to leave it implicit in their cultures without placing much formal emphasis upon it. Others, such as the Trobrianders and many other Melanesians, go to great length in institutionalizing reciprocity and throwing it into the spotlight of social awareness.' We may wonder: is this really so or does it only seem that way because of the emphasis Malinowski has placed on reciprocity? Personally, I believe the latter to be true. The extent to which the ethnographer's own personality and private values may influence his picture of the social situation is illustrated clearly by the RedfieldLewis case. The facts are known to many of us. During the twenties Redfield carried out fieldwork in the Mexican town of Tepoztlan. This same town was the object of fieldwork once more during the forties, this time carried out by Lewis.36 Both anthropologists reported their findings in a book. And, as Redfield (1955, p. 134) himself says: 'the two books describe what might almost seem two different peoples occupying the same town'. Redfield's own book gives an impression of a fairly homogeneous, smoothly functioning and well integrated society, composed of contented people who get on well with one another. Lewis, on the other hand, notes the presence of tensions, fear, envy, distrust and lack of cooperation and considers these to be dominating elements in this society. Redfield (1955, p. 136) makes the following comment: 'the hidden question behind my book is: what do these people enjoy. The hidden question behind Dr. Lewis's book is: what do these people suffer from'. Redfield realizes quite well that it is the personal factor which had made the two pictures so very different. The cause is to be found in a romantic trait of Redfield's personality. To quote this author one last time (1955, p. 135): 'I looked at certain aspects of Tepoztecan life because they both interested me and pleased me. I saw the almost ritual meaningfulness to the Tepoztecan of his daily work; I saw the pride the people had in their little mountain-walled country, so long inhabited, so deeply grown into their thoughts and feelings.' Redfield and Lewis, therefore, give totally different pictures. Which of these two is correct? Redfield declares they both are and that, in fact, the two pictures are complementary. I am not quite so sure that this is true. But even if it is, it is also true that Redfield's description is so one-sided as to give outsiders a decidedly wrong impression. In this connection we may recall the hypothesis mentioned before, the one Foster proposes: 'where peasant society, there interpersonal relations strained; suspicion, enmity, envy, lack of cooperation'. If Foster, when testing this hypothesis, had only had Redfield's book at his disposal in the case of Tepoztlan, he would have been forced to class this society as an exception ('peasant society,

34

Andre J. F. Kobben

nevertheless harmonious interpersonal relations'). Now, however, he relies heavily on Lewis, and Tepoztlan becomes one of the eloquent illustrations in support of his thesis! If all existing ethnographies were to be revised, must we not assume that many of them would prove to be rather one-sided in this way? Once more: in this case we happen to know... as for the rest we are blissfully ignorant. 37 It is not only the ethnographer's own personality which influences his description, but also the school in which he has been educated. Malinowski's influence in anthropology, for instance, has been disproportionately great: everyone saw reciprocity and set out to describe 'his own' society in terms of reciprocity. This was possible in every case because, naturally, every culture contains some elements of reciprocity. But it is quite certain that far too much emphasis has been placed on this phenomenon, while one-sided relations, where one party has the privileges and the other the obligations, have been neglected for many years. The Malinowskians have exaggerated quite as much on the one side as the Marxists on the other. Another example may be found in the writings of a group of Leyden anthropologists. In 1935 Van Wouden (p. 96) pointed out - and this was quite a discovery at that time that patrilineal and matrilineal structures might exist side by side in one society. Enchanted by this discovery, a number of Leyden anthropologists described or attempted to describe various societies in terms of double descent, often founding their arguments on what, to an outsider, seem to be very feeble indications.38

And finally, there is the culture (or sub-culture) from which the ethnographer stems himself. Den Hollander (1965) remarks: 'he will never succeed in freeing himself completely from the dominant attitudes, presuppositions, presumptions, opinions, values and prejudices of his own culture'. We have already given the example of the American ethnographer who said that divorce was 'relatively unimportant' in a certain society while an Englishman considered it to be 'relatively important'. Unconsciously each applied the norm of his own culture. A description by an Englishman of land tenure in an African community may be hard to follow for a Dutchman on account of terms such as 'fee simple', 'lease-hold', 'free-hold'. 39 By using such terms, the English writer is imposing a juridical construction that is typical of his own culture, on that other culture. Although we cannot, for all these reasons, expect absolute objectivity on the part of the ethnographer, the latter should nevertheless guard against too much subjectivism. Unless this is kept in mind, intercultural comparisons would indeed be meaningless - and anthropology, for that matter, would degenerate into mere novel-writing. The ethnographer too, would therefore be wise to make it a rule to look for operational definitions and numerical criteria. We should not too hastily assume this to be impossible. When comparing the Tapanahony-Djuka with the Cottica-Djuka the fieldworkers had the impression that the Cottica people experienced their religion less intensely than the other group. One would think that such an impression does not readily lend itself to translation into operational terms. It

The logic of cross-cultural analysis: why exceptions?

35

proved to be possible all the same. For in this religion an important part is played by mediums: a person may be the medium of a deity, who from time to time speaks through this person's mouth. In the Tapanahony region 37% of the adult population proved to be mediums in this way while in the Cottica region this was only 20 %. The ethnographers accepted this as a criterium and considered it a confirmation of their impression. Note that the percentages were compared after they had already stated their impression; so it really was a conclusive test.

Itis.of course,notalwayspossible to set up an operational definition—perhaps it is in theory, but not in practice. Through field-work carried out among these same Djuka, for instance, the present author grew convinced that backbiting in this society has an important function as a means of social control. 'Important' can here only mean: more important than in at least a number of other societies. The cause of this seems to me to lie in the fact that in Cottica-Djuka society only minimal differences in power exist between its members. This pronouncement amounts to a general hypothesis somewhat like this: 'where minimal differences in power, there backbiting important'. But this is a hypothesis which I cannot prove in any way. It is merely an impression and will have to continue as such for the time being. For how am I to find a way of measuring the degree of backbiting? How can I be sure that the phenomenon occurs less frequently or that it is less intense elsewhere? Purists might be of the opinion that it would be better to drop the hypothesis altogether. I believe, however, that we should leave room for speculation, providing it is a disciplined sort of speculation and is presented as such. If we concern ourselves only with 'safe' questions, those which are most relevant theoretically might remain undiscussed. 3. Multicausality Every social phenomenon is due to a large, in principle even unlimited, number of factors. Most of these factors may be left out of account in intercultural comparisons for the simple reason that they are common to all cases. But if this is not true of a certain factor, neglect of it may give rise to exceptions. Suppose:

C, that is: B

factor A and factor B are both necessary to bring about the phenomenon C. Suppose further, that someone arrives at the hypothesis : 'where A, there C', while he is unconscious of the fact that B is also a necessary condition. Then we would get the following picture : A

C

I

Not-C

III

Not-A II x

IV

36

Andre J. F. Kobben

In the first quadrant would be classed those cases where A is present, B also happens to be present, and therefore also C. The third quadrant ('exceptions') would contain those cases where A is present but where B happens to be absent, so that C is absent as well. Let us take the example of Nieboer's main thesis (1910, Part II) concerning slavery: 'where slavery, there open resources'. By 'open resources' in an agricultural society he means a situation where there is land in plenty, or where, at least, not all land is being used and where, besides, few capital goods are required to till the soil (for instance only a hoe or a digging-stick). Baks et al. (1961, pp. 191-196) carefully re-tested this hypothesis with data from Africa, Oceania, and Indonesia. The result was as follows: Slavery No slavery

Open resources 15 20!

Closed resources 2 7

This seems to be disastrous for Nieboer's hypothesis. There is no question at all of a significant correlation. So the hypothesis, as it stands, must be rejected. But let us approach the phenomenon from a different angle. We may then note that slavery is a form of coercion: one person forces another to work for him and prevents the other from leaving him. Is coercion of this type 'technically' possible in all societies? The answer is no; only in those where, even apart from slavery, social stratification is clearly present. In otherwise egalitarian societies slavery seems not to be easily practicable. Nieboer neglected to take this factor into consideration, but let us make allowance for it. Of the 44 societies examined by Baks et al., 19 have social stratification and 25 are egalitarian. When Nieboer's hypothesis is separately tested for the stratified societies and for the egalitarian ones, the results are as follows:

Slavery No slavery

Stratified Open resources Closed resources 10 2 2 5

Egalitarian Open resources Closed resources 5 18! 2

It we take only stratified societies, Nieboer's hypothesis is tenable, at least in the sense that a statistically significant correlation exists between the two phenomena in question. His hypothesis should therefore be amended as follows: 'where slavery, there stratified society with open resources'. When only egalitarian societies are taken, no less than 18 of the 25 cases examined prove to be in conflict with Nieboer's hypothesis. We might almost say: those are the societies which could use slavery but are not capable of introducing it. This example then takes us to an important conclusion: if I have proved that there is no statistically significant correlation between A and B, that does not prove that there is no functional relation between the two phenomena. 40 Conversely, however, it is true that if there is a demonstrable correlation between these phe-

The logic of cross-cultural analysis: why exceptions?

37

nomena, then there must also be functional (or historical) relation. The nature of this relation is then still to be decided.41 4. Pluricausality I would use the term pluricausality when a phenomenon C may be brought about by A but also, quite independently, by B. This might be symbolized as follows A B

>C >C

Let us assume that somebody proposes the hypothesis: 'where A there C', forgetting that B may, quite independently from A, also bring about phenomenon C. We then get the following picture: A

Not-A

c

I.

II.

Not-C

III.

IV.

X

In this case the second quadrant will contain a whole series of exceptions, namely those cases where A is absent, but B happens to be present, and therefore C as well. Example 1. Westernization. A question of current interest is: why do some nonWestern peoples offer an almost insuperable resistance to modernization and westernization while others show much less such resistance? A factor often named in this connection is the existence of a rigid and strongly hierarchical social structure.42 This factor, however, is at most a sufficient but certainly not a necessary condition. A 'rival' factor, for instance, is cattle nomadism. Egalitarian societies of cattle nomads - typified for instance by the Nilotic peoples as described by EvansPritchard (1940), Lienhardt (1958) and Schneider (1959)-even show maximal resistance to development in the western sense. Assume now that someone were to test the first-named factor without taking the other into account, that is, test the hypothesis: 'where strongly hierarchical structure, there great resistance to westernization*. He would then have to cope with all cases of cattle nomadism as exceptions. Example 2. Bilateral kinship structure. A bilateral structure is found on the one hand in very simple societies (small egalitarian groups with little functional diversification) and on the other hand in modern, Western society. This circumstance even caused Murdock (1949, p. 187) to say: 'The forms of social organization appear to show a striking lack of correlation with levels or types of technology, economy, property rights, class structure or political integration'.

The twofold statement given above may be tested with the help of Murdock's World Ethnographic Sample. In the first place the relation between Western society and bilateral kinship:

38

Andre J. F. Kobben Modern (Western) society

Other societies

21 1

206 336

Bilateral kinship Other

There is clearly a significant correlation, 43 but the number of exceptions in the second quadrant is uncomfortably large (206!). This number may be reduced by taking into account the 'rival' hypothesis: 'where simple society, there bilateral structure'. If we translate 'simple' into operational terms in the same way as on page 27, this becomes: 'where hunters and collectors, there bilateral kinship structure'. The World Ethnographic Sample contains 74 societies of hunters and collectors, 49 of which have a bilateral structure while 26 are not bilateral. This gives us: Modern (Western) society

Other societies (hunters and collectors left out)

21 1

206-49 = 157 336-26 = 310

Bilateral kinship Other

In this way the number of exceptions is reduced and the significance of our correlation is considerably increased. The same operation may of course be carried out by starting from the other side: Hunting and collecting Dominant Not-dominant Bilateral kinship Other

49 26

178 311

The number of 178 exceptions in the second quadrant may be reduced by leaving out modern (Western) societies: Hunting and collecting Dominant Not-dominant (Western societies left out) Bilateral kinship Other

49 26

178-21 = 157 311-1 = 5 / 0

Now there is one objection that might be raised against this way of looking at the facts. For if factor A and factor B are really different and operate quite independently of one another, then their respective effects are not the same. These effects may resemble each other, but they are not identical. Old-fashioned logicians, therefore, reject pluricausality in this sense. 44 Take for instance bilateral kinship structure: without going into details it is clear that our Western bilateral system is not exactly the same, and has not exactly the same functions, as that of the Bergdama or the Eskimo. Agreed! But this objection is one of more theoretical than practical importance. Usually our categories are large enough to include in one category phenomena

The logic of cross-cultural analysis: why exceptions?

39

which are the effect of different causes and which are therefore not quite identical. I know that some fellow anthropologists will object that this is just what we shouldn't do, that it leads to superficiality and perhaps even to 'tabulated nonsense'.45 But unless we want to lapse into solipsism we shall all have to use these larger categories. In practice everyone does so, in our branch of science as much as in any other. Biologists put man, mouse and mammoth into one category, and does anyone blame them for it? 5. Functional equivalents Once more we start with the hypothesis : 'where A there B', while assuming that factor A is the cause of factor B. In a number of cases, however, we will find that A does not give rise to the phenomenon B but to C, which is related to B, in the sense that it fulfills the same functions. Such cases are thus exceptions to the rule 'where A there B'. I have taken the term 'functional equivalents' from Merton (1956, pp. 33-35). He gives the example of religion. Several authors have remarked that religion fulfills certain necessary functions : those of social control and of integration in terms of sentiments and beliefs. That is why these authors have concluded that religion is a necessary condition of human life: 'where human society, there religion'. But Merton points out that it is not so much religion itself which is necessary but rather those functions. And these same functions may be fulfilled by equivalents: nationalism, political ideals, a humanist system of ethics or anything similar. Our second example is concerned with a closely related phenomenon, that of prophetic movements. We may ask ourselves when such movements will occur. Some factors may be named: oppression; internal tensions; sudden changes; catastrophes ; diseases ; poverty. It is true that, where all or some of these factors occur, prophetic movements will frequently arise. Sometimes, however, a secular revolutionary (nationalistic) movement may arise instead, or else an organized attempt at economic betterment, for instance through the founding of cooperatives (cf.Kòbben, 1964, pp. 144-145). These are functional equivalents which weaken the rule: 'where such and such conditions are fulfilled, a prophetic movement will arise'. For a final example I turn once more to the phenomenon of slavery. Nieboer says : where open resources, there people will try to make others work for them, therefore slavery. But here again alternative solutions are possible : for instance in the form of heavy tributes or corvée (cf.Bakseia/., 1961,p. 193), indentured labour under penal sanctions, debt relationships (cf. Fahrenfort, 1943, Kloosterboer, 1960). With regard to this factor we must make the same observation as was made with regard to pluricausality, namely that old-fashioned logicians cannot accept it. For them, one cause produces one effect. In the strict sense this is of course true. If 'factor A' in one case brings about effect B and in the other effect C, then those

40

André J. F. Kobben

two 'factors A' are not exactly identical. That this is so may be demonstrated: during the 19th century prophetic movements were in Western Europe for the most part replaced by revolutionary ones. All or almost all of the conditions we mentioned were still present, but in such a special social context that it need not surprise us if the effect was different. Generally, however, we use such rough-andready categories that functional equivalents do have a practical significance in explaining exceptions.

6. Intervening variables On the subject of exceptions which are caused by intervening variables I can be quite brief. We discussed them already when we were dealing with operational definitions (p. 27). Let us take once more the hypothesis: 'where hunters and collectors, there bilateral kinship structure'. This bilateral structure is not a direct result of the fact that the members of such a society live by hunting and food-gathering. The causal nexus here is, rather: hunting and food-gathering >• small group > bilateral structure. So we have a twofold source of exceptions, arising in the first place from the fact that not every society of hunters has extremely small groups, and in the second place that people who live together in extremely small groups do not always have a bilateral kinship structure. But even where there is no explicit use of operational definitions there may be one or more intervening variables. Take slavery, for example. Nieboer tests the relation between open resources and slavery. But the causal nexus is not: open resources >• slavery. Rather: open resources > need for people to work the landowner's land > slavery. Again a twofold source of exceptions. The significance of intervening variables as a source of exceptions is very neatly illustrated by an example from Driver and Massey's (1957) book on North American Indians. They test Murdock's (1949, p. 113-115) hypothesis which holds that there is a relation between the economic position of woman and matri-terminology, which relation would be an indirect one; it is not: women principal producers-*- matri-terminology. But rather: women principal producers-*-matrilocal residence-»matrilinearity-> matri-terminology. When these factors are correlated, the correlation-coefficient proves to be positive in every case, but the more intervening variables the weaker the correlation:

1. 2. 3. 4.

Women principal producers Matrilocal residence Matrilinearity Matri-terminology

1.Women principal producers

2. Matrilocal residence

3. Matrilinearity

4. Matriterminology

x

high x

moderate high x

low moderate high

x

The logic of cross-cultural analysis: why exceptions?

41

7. Diffusion, external influences Let us assume that some one were to embark on a study of Surinam society around 1875. He would find a stratified society, of a plural type.46 As in other plantation colonies of the Caribbean, the top stratum consisted of a small group of whites, the middle layer of coloureds and the lowest of the great mass of the negro population. At the same time the situation was very clearly one of 'open resources'. There was a shortage of Creole labour on the plantations, so much so that it became necessary to import labourers from the other end ot the world: from India and later also from Indonesia. Nevertheless there was no slavery! For which reason Surinam was clearly an exception to Nieboer's rule (in the amended form I gave in section 3): 'where agricultural society of a hierarchical type with open resources, there slavery'. The explanation is simple. Some years earlier (1863), slavery had been abolished through external influences. These consisted of a parliamentary resolution in the mothercountry which was passed despite protests on the part of the slave-holders.47 The opposite also occurs. The Kaska Indians (British Columbia) are hunters and collectors; they live in small groups and have few hierarchical differences. All factors which would lead us to expect no slavery. Nevertheless they have slavery! They borrowed the institution with name and all from the much more highly developed Tlinkit Indians. 48 A numerical indication of the factor of diffusion is given by Driver and Massey (1957). These authors test the hypothesis already mentioned: 'where women principal producers, there matrilocal residence, matrilinearity and matri-terminology'. The hypothesis proves to be tenable in the sense that there is a statistically significant correlation. But, as they say themselves, there are a host of exceptions. In order to explain these they give up trying to find functional explanations and instead they use, preferably, diffusionist ones. Otherwise, as they show, almost half their facts would remain unexplained. 49 This is all the more poignant when seen in the light of Murdock's remark that 'the forms of social organization seem singularly impervious to diffusion'. 50 In passing, I would point out that diffusion, as we have seen, is a disturbing factor when we are looking for functional relations. But that, conversely, it is equally true that functional relations are a disturbing factor when we are looking for diffusionist relations.51

A closely allied problem, which has been given much attention by theorists, is the question of what exactly should, for purposes of intercultural comparisons, be regarded as one society, that is, as a statistical unit. 52 This problem is already quite old. In 1889 Tylor delivered his famous lecture to the Royal Anthropological Institute, in which he showed there to be an 'adhesion' - we would say a correlation - between avoidance of parents-in-law and residence. During the subsequent discussion, Galton, the well-known statistician, remarked:

42

André J. F. Kôbben

'It would be extremely desirable for the sake of those who wish to study the evidence for Dr. Tylor's conclusions that full information should be given as to the degree in which the customs of the tribes and races which are compared together are independent. It might be that some of the tribes had derived them from a common source, so that they were duplicate copies of the same original.' (Tylor, 1889, p.270) Here, the problem is very neatly expressed! All I wish to show at present is that this difficulty again may be a source of exceptions. Say I am going to re-test Tylor's hypothesis : Wife's avoidance Residence

Patrilocal Other forms

Avoids I

ni

ofparents-in-law

Does not avoid

n

IV

Now the Bete (West Africa) for instance, would be classed in the second quadrant, that is, with the exceptions. After her marriage a Bete woman goes to live in her husband's village, but she does not avoid her parents-in-law. Apart from the Bete the same is true of a handful of neighbouring tribes who are closely related to the Bete. If we were to include all these tribes in our sample that would give us as many exceptions! I must, however, emphatically add that the same factor may also 'embellish' the results. Lowie (1947, pp. 94-97) gives an instance of this: among Siberian tribes there are a dozen or so cases of the daughter-in-law taboo, that conform to Tylor's rule. If we ignore the evident relationship of these tribes, counting them all as positive, then the case for a causal nexus between these phenomena is tremendously strengthened. It might appear to be quite easy to offer a prescription for avoiding this source of error: 'for purposes of intercultural comparisons, from a set of historically related groups only one should be taken'. But this is no effective solution to our problem. Consider for example the Southern Bantu tribes. Their historical relationship is beyond dispute. Yet if one tribe is to be selected out of this group for the subject in question, which one are we to take? The trouble is that in this respect, notwithstanding their historical relationship, there are essential differences between them! In some we find both husband and wife avoiding their parents-in-law. In others the husband does avoid his parents-in-law but the wife does not - she only has to show them marked respect. 53 And, concerning the Kgatla (also a Southern Bantu tribe) Schapera (1940, pp. 113-114) says: 'There is little of that formal restraint between a man and his parents-in-law reported of other Bantu tribes, and when I told my Kgatla friends of the famous mother-in-law taboo they were invariably amused to find how stupid other people can be!' Now these Kgatla are a sub-tribe of the Tswana. There are about fifty of these Tswana sub-tribes and presumably none of them have the parents-in-law taboo. 64 For the purpose of testing the rule in question - the relation between parents-in-

The logic of cross-cultural analysis: why exceptions?

43

law taboo and a particular form of residence - it would thus be necessary and sufficient to take one of the Tswana sub-tribes as a 'case' and to disregard the rest. Perhaps this solution might be raised to the status of a general rule? So that we could say: 'in testing an intercultural rule, no more than one of the various sub-tribes of a tribe should be taken'? Unfortunately, no, and this for two reasons. In the first place those sub-tribes are in some respects not uniform - obviously not, or else they would not be perceived as separate units. A person for example, who is investigating the occurrence of parallel-cousin marriage cannot afford to lump all the Tswana sub-tribes together. In five of these sub-tribes such marriages are permitted; in two they are only allowed if the prospective bride is ego's father's /¿«//-brother's daughter; in two others again they are on no account permitted.55 Besides, it is a question of definition. What do we mean by 'tribe' and what by 'sub-tribe'? There is no generally accepted operational definition for these concepts. What one author calls this, the other calls that. As much as thirty-five years ago Hobhouse et al. (1930, pp. 8-12) pointed out the inconsistencies in the use of these terms, but without offering a solution: they listed as 'tribes' what the respective ethnographers called 'tribes', in the hope that the law of averages would rectify their mistakes! We can't say that we have advanced much since then. The latest proposal on this subject comes from Naroll(1964, pp. 283-291). To begin with, he introduces a new word, 'cult-unit'. Apparently he is a victim of the old delusion that a new concept is created with a new word. A 'cult-unit', then or to use the old words: tribe, society or ethnic unit - is defined by Naroll as follows: 'a group of people who are domestic speakers of a common distinct language and who belong either to the same state or the same contact group. 'Distinct language' he defines as: 'all those dialects mutually intelligible to speakers of a stated dialect'. And 'contact-groups': people who are all interconnected by successive contact links. Two families may constitute a contact link 'if their dwellings are no more than two hundred air kilometers apart'.56 In the comments which followed Naroll's paper, quite a few pertinent critical remarks were made. I shall not attempt to reproduce them here, since the article in question is of such recent date and appeared in such a widely read journal as Current Anthropology. In any case, Naroll's definition does not appear to be generally applicable as yet in its present form, nor, should I add, does the author have this pretension. 8. Cultural and social lag Let us assume that phenomenon A is the factor (or one of the factors) which gives rise to phenomenon B. If phenomenon A disappears, the effect B will not, usually, immediately cease to exist. Assume now that some one has formulated the rule: 'where A, there B'. In this case he will not find A, but B, therefore an exception.

44

Andre J. F. Kobben

Take for example the so-called agro-towns such as they are found in Southern Italy or in Yoruba-land (S.W.Nigeria). 57 These are settlements with an urban character and with a population of 20,000 to 100,000 living mainly by agriculture. Now we may readily assume that insecurity of some sort was one of the causes of their origin. So: 'where agro-towns, there insecurity'. This insecurity, however, has largely ceased to exist in the present day both in Southern Italy and in Nigeria. Nevertheless the agro-towns continue to exist. No dispersion, or at least no immediate or total dispersion takes place. The question which unavoidably arises is: why does no dispersion take place in these cases?58 It is a queer situation, after all, peasants living in town while the land they have to work is situated at a great distance, - in the case of the Yoruba, sometimes as much as 30 to 80 miles away.59 Human inertia is not necessarily the only factor here. There are other factors, such as sociability (Southern Italy) or the status that goes with living in a town (S. Italy, Yoruba) which deter those people from leaving. Additional factors may be the large markets of the town (Yoruba), the schools, transport facilities. In this connection it is interesting to note that with a different people, the Mambwe, the larger population concentrations did disappear quite soon after the arrival of the Pax Britannica... but these had few functions apart from the one, precisely, of offering security.60 When looked at in this way, this source of exceptions - social and cultural lag might seem to be merely another form of our fourth factor (pluricausality), but on account of its special character we would rather continue to regard it as a factor in its own right.

9. Coincidence In accordance with the generally accepted view, I use the term coincidence when two independent chains of cause and effect happen to meet. In what way 'exceptions' can come about by coincidence I will illustrate with an example. Suppose: in a certain region circumstances are on general grounds judged favourable for community development. A team of community development experts arrives, but as they start work an outbreak of epidemic disease occurs. The people in question believe there is a causal connection between the epidemic and the strangers' presence and refuse to cooperate. The coincidence of the epidemic has in this case rendered the general pronouncement invalid. Such a situation occurred among the Bushnegroes of Surinam in the 1920's, when an attempt was undertaken at 'community development', as we would call it now. The official who was in charge was fetched in the capital by two kinsmen of the Paramount Chief. Both got influenza. While on their way back, one died, the other one recovered after many months only. At the same time, in the villages, epidemic diseases took a high toll. The animals which had been taken along to be introduced into the Bushnegro economy, died as well. All this was seen as a bad omen. After that, no Bushnegro

The logic of cross-cultural analysis: why exceptions?

45

was prepared to accompany the unfortunate official to where the Bushnegroes live. It was a clear case of cum hoc ergo propter hoc. As the person in question himself wrote: 'With my arrival, death and misery fell upon them. I was shunned, and as a proof of how frightened they were of me, I might add that when 1 appeared in a village, people literally fled.'" I am convinced that in this way coincidences often disturb our beautiful rulesOftener, I think, than the anthropological literature shows: in many instances coincidence is not diagnosed as such. For the mind seeking order in the apparent chaos of data, coincidences are uncomfortable things, and often we seek unconsciously to explain them away.

10. Personality (the personal factor) An uncommon personality may very well bring about an exception to a general rule. Take as an example prophetic movements. Earlier (p. 39) I have indicated under what social conditions we may expect them to occur: foreign domination; internal tensions; rapid social change; poverty; diseases; calamities. Thus we may formulate the rule: 'where such conditions present, there prophetic movements.' Prophetic movements may, however, very well arise where such conditions are only weakly developed, simply by there being a strongly charismatic personality who feels called upon to act as a prophet. Conversely, where such factors are present in a high degree, there may be no prophetic movement, because there is no person who sets himself up as a prophet. I know there are people who do not agree. Some sociologists and anthropologists believe that when the situation is ripe a prophet will appear.82 It seems to me doubtful, however, and to testify to a sort of professional deformation, to think that circumstances will, as it were, create the prophet. In this way many exceptions may be attributed to the personal factor. We must beware, however, of too soon having recourse to this factor. Take, for instance, the matrilineal conflict. On the basis of data from numerous societies the following general pronouncement might be made: 'where matrilineal society, there specific conflicts', namely between a man and his sons on the one hand and his sister's sons on the other, the issue at stake being the inheritance. 63 Among the matrilineal Bushnegroes (Surinam), however, such conflicts do not occur, or very rarely. Can this be attributed to the personal factor? No, for although we are here considering the phenomenon as occurring in one tribe only, it is still sub specie repetitionis that we see it: there are hundreds of matri-segments consisting of a man, his sisters and his sisters' sons, within which this conflict might potentially arise. If it does not, there must be a general factor (factors) which prevents it from occurring. In the society of the Agni (W. Africa), on the other hand, conflicts of this sort abound. Here too, however, there are, by way of exception, certain individual matri-seg-

46

André J. F. Kòbben

ments where the matrilineal conflict does not occur. These again may be explained by the personal factor : mother's brother and sister's son may both happen to be easy-going, tolerant persons, or have a particular affection for one another. 11. Combination of factors Most of our general rules have exceptions not just of one type but of a whole series of types, or even of all ten of them in combination. This may be illustrated by Nieboer's thesis concerning slavery. We have already seen that factors 1, 3, 5, 6, and 7 may give rise to exceptions in this case, and I shall now show that this is true also of the remaining five factors on our list. Factor 2. The proportion of slavery may range from a few percent of the population (Tahiti)64 to more than 90% (18th century Surinam).65 Similarly, there are great differences in intensity. Slavery with the Ashanti was 'a position of servitude surrounded by rights',66 while in Surinam a slave was for all practical purposes without rights.67 Nevertheless, we usually place all these cases in the one category of slavery. Factor 4: pluricausality. Apart from open resources, another quite separate factor may lead to slavery, for example conspicuous leisure.68 Factor 8: social lag. Nieboer (1910, p. 366) himself gives an example of the occurrence of this factor in his data: at the beginning of the 14th century, England had closed resources; in conformity with this situation there was no slavery. In 1349 came the Black Death. The ensuing mortality was so great that a shortage of labour arose (open resources). But this did not result in an immediate introduction of slavery. Factor 9: coincidence, and Factor 10: personality. We may think, for example, of a particularly well-disposed master, having a particularly faithful slave (personal factor), who is manumitted on account of his having saved the life of his master (coincidence). Manumission occurred even in such an extremely hard slave-society as Surinam.69 12. Conclusion Looking back and reviewing the whole field again, we realize very clearly indeed that in anthropology we are concerned with a refractory type of data which do not easily lend themselves to tabulation. It is not surprising that we have exceptions, it is rather surprising that we have regularities. The fact that we are able to formulate these in spite of everything tells greatly in favour of their intrinsic significance. It is inadmissible for Lévi-Strauss (1958, p. 344) to dismiss in a footnote - though a long one - the theory developed by Homans and Schneider on unilateral cross-cousin mariage without challenging their numerical data. (Realize that Homans and Schneider wrote their book as a reaction to Lévi-Strauss' own Structures élémentaires de la parenté.)

The logic of cross-cultural analysis: why exceptions?

47

He might have done one of three things: given a theory which accounted for Homans and Schneider's results at least as adequately as their own does; or shown that Homans and Schneider classified their cases incorrectly on a large scale; or again, with the help of a number of new 'cases', shown that the results reached by Homans and Schneider were really due to coincidence. We may note that such a re-test with more societies has in fact been carried out independently of Homans and Schneider, even twice, and in both cases has produced a statistically significant result.70 Ever since Murdock's Social Structure was published in 1949, many fellow anthropologists have made it a sport to attack his results on either logical or empirical grounds.71 From time to time the present author has willingly joined in this. But Murdock (1949, p. 178-179) quite rightly points out that all 120 coefficients calculated in his book are positive in sign. This cannot possibly be due to coincidence or to any slip or error. In spite of everything his results are substantially valid. As a matter of fact Murdock's results too have been re-tested with a much larger sample, and again the result was positive (Murdock, 1961, p. 193; Textor, 1965a, p. 5). So, let us not be discouraged. Let us work harder and think harder and sometime the day will come when we have more complex and more subtle 'laws' than we have at present. That will be a rather glorious day for anthropology.

NOTES

1. The numbers given concern only the terms for FaSi, MoSi and Mo. 2. Nieboer (1910); for a critical appraisal and further elaboration of his theses see Fahrenfort (1943); Siegel (1945); Kobben (1952); Kloosterboer (1960); Baks el al. (1961). 3. Nieboer (1910), pp. 201-255. 4. Cf. also Blok (1964a). 5. Note, however, the restrictions Wittfogel (p. 12) himself gives: '...beyond the influence of strong centres of rainfall agriculture, and below the level of a property-based industrial civilization...' 6. Cf. also Bax (1964). 7. Bascom (1955); Mitchel (1962); Hoffmann-Burchardi (1964); Mabogunje (1964). 8. Nieboer (1910), pp.97-110; 314-324; 328-346. 9. Nieboer (1910), pp. 104-110; 328-346. 10. Cf.Naroll (1962), pp. 18-19, and Driver and Massey (1957), p. 430. 11. See for a more detailed discussion Kobben (1952), pp. 142-143. 12. Whiting (1950), p. 87. 13. Cf. Udy (1964), p. 178. 14. Cf. Adler (1964). 15. For his definition see p. 28. For examples see Hoebel (1954), Ch.V. 16. Evans-Pritchard (1940), pp. 165,168,171. 17. Hempel and Oppenheim (1936), Ch.IV, par. 5. 18. One notes with satisfaction that this is regularly done in Banks and Textor (1963), and in Textor (1967). 19. Cf. Berliner (1962), p. 54. 20. Only Naroll (1956; pp. 700-701) gives concomitant variations in the form of the following 'laws': P = (T/2)" = C4, in which P means Population of largest settlement in

48

21. 22. 23.

24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.

44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56.

André J. F. Kôbben thousands; T means Team types (power groups); C means Craft specialization. These 'laws' are, as Naroll himself points out, no more than rough approximations. This classification is largely derived from Murdock (1961), p. 203. Cf. Zetterberg (1954), pp. 36-43. Garfield (1939), pp. 177-178 resp. Serpenti (1965) Chaps. Ill and IV. I suggest here that there is a causal connection between the fact of Kimam society being simple and its not having a unilinear kinship structure; this, of course, I cannot prove. Gifford (1922), p. 16 resp. Elwin (1947), p. 621. Gilford's sample is too small, however, to permit his data to be regarded as conclusive. Cf. Garbett (1960), p. 42. Pouwer (1960), pp. 115 if. Pouwer (1962), passim. See for a striking example Berreman (1962), p. 10. Kôbben (1964), Ch. VIII. Oral information. Radin(1920), pp.387-388; 393-394. Kôbben (1964), p. 181. Kôbben (1956), p. 27. Cf. for examples and a useful discussion of this point: Gladwin and Sarasin (1953), pp. 442-445. Malinowski (1926), pp. 37,41. Cf. Redfield (1930) and Lewis (1951). Cf. for further examples Leach (1964a) pp.427-429; Pouwer (1962), p.28. For examples see De Josselin de Jong (1951); Nooteboom (1935); VanderLeeden (1950). For a critical appraisal see Kloos (1963, 1964), cf. De Josselin de Jong (1964). Rattray (1929), pp. 365 ff. Cf. Brown (1963), pp. 130-131. For another striking example see Otterbein and Swanson (1965). For a further discussion of this problem I refer to my forthcoming book on intercultural comparisons. Examples in Kôbben (1964), pp. 70-72. I do not suggest that the fact of so many Western societies having a bilateral kinship structure has to be accounted for only functionally; common origin is undoubtedly an important factor. CohenandNagel(1955),pp.270. Cf. Leach (1964b), p. 299. Cf. Van Lier (1949), Chaps. VIII and IX. For the notion of 'plural society' see idem, pp.9ff.; Furnivall (1948),pp.304ff;Hoetink (1962),p. 148;Despres(1964), pp.1052 ff. Van Lier (1949), Ch. VH. Honigmann (1954), p.86; for a second example see Baks et al. (1961), pp. 160-161. Driver and Massey (1957), pp.438; 434-435. For a refutation of this point of view see Wilson (1952). Haekel (1956), p. 32. Cf. Kôbben(1952), pp. 132-133 ; Schapera (1953); Naroll (1961) ; Naroll and d'Andrade (1963); McEwen(1963), p. 162; Naroll (1964). Kloos (1961),passim. Schapera (1950), pp. 145-146. Schapera (1950), p. 151. Naroll (1964), pp. 286-287. The author has painstakingly built up a set of interconnected

The logic of cross-cultural analysis: why exceptions?

57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71.

49

definitions. I do not reproduce these here in full. Blok (1964), resp. Bascom (1955) and Hoffmann-Burchardi (1964). Cf. Blok (1964),passim. Oluwasanmi (1964), p. 3. Watson (1958), Ch.Vn. de Groot (1965), pp. 41-44. E. g. White (1949), Chaps. VI and VII. For a historical survey see Opler (1964), pp. 527ff. Cf. Kobben (1964), Ch. II. Jongmans (1955), pp. 33-35. Van Lier (1949), p. 31. Rattray (1929). Van Lier (1949), Ch. VI. Baks etal. (1961), pp. 107,110,118,193. Van Lier (1949), p. 100. Berting and Philipsen (1960), p. 66. Cf. comment by Homans (1962), pp. 252-256. The latest example is Evans-Pritchard (1965), pp. 25-27.

BIBLIOGRAPHY

Adler,F., 'Social Realism' and 'Social Nominalism', in J.Gould and W.L.Kolb, eds., A Dictionary of the Social Sciences (Glencoe, 111., The Free Press, 1964). Bäks, C., Breman, J. C., and Nooy, A. T. J., Slavernij. MS., 1961; cf. summary in Bijdragen Taal-, Land- en Volkenkunde, 122, (1966), pp. 90-109. Baks,C., et al., 'De betrouwbaarheid van informatie in een kaste-samenleving'. Sociologische Gids, 12 (3), (1965), pp. 167-174. Banks, A. S. and Textor, R.B., A Cross-Polity Survey. Cambridge, Mass., M.I.T. Press, 1963. Bascom, W., 'Urbanization Among the Yoruba'. American Journal of Sociology, 60 (5), (1955), pp. 446-545. Bax.M.M.G., Commentaarop Whiting en Pilling. MS., 1964. —, Despotie en irrigatie: een critische evaluatie van Wittfogel's 'Oriental Despotism'. MS., 1964. Berliner, J.S., 'The Feet of the Natives are Large; an Essay on Anthropology by an Economist'. Current Anthropology, 3 (1), (1962), pp. 47-61. Berreman, G.D., Behind Many Masks; Ethnography and Impression Management in a Himalayan Village. Ithaca, N.Y., Society for Applied Anthropology, Monograph No 4,1962. Berting, J. and Philipsen, H., 'Solidarity, Stratification and Sentiments'. Bijdragen TaalLand- en Volkenkunde, 116 (1), (1960), pp. 54-80. Blok, A., Commentaar op 'From Generation to Generation'. MS., 1964. —, Enige aspecten van het geconcentreerde nederzettingspatroon in Zuid-Italie. MS., 1964. Brown, R., Explanation in Social Science. London, Routledge and Kegan Paul, 1963. Cohen, M.R. and Nagel, E., An Introduction to Logic and Scientific Method. London, Routledge and Kegan Paul, 1955. Colson, E., Marriage and the Family Among the Plateau Tonga of N. Rhodesia. Manchester, Manchester University Press, 1958. De Groot,S.W., An Early Attempt at Community Development. MS., 1965. De Josselin de Jong,P.E., Minangkabau and Negri Sembilan. 's-Gravenhage, M.Nijhoff, 1952. —, 'Circulerend connubium en het dubbelunilineale principe'. Bijdragen Taal-, Land- en

50

Andre J. F. Köbben

Volkenkunde, 120 (1), (1964), pp. 181-194. Den Hollander, A.N. J., 'Sociale beheersing in een Bengaalse stad', in De wereld der mensen (Groningen, Wolters, 1955). —, 'Soziale Beschreibung als Problem'. Kölner Zeitschrift für Soziologie, 17(2), (1965), pp. 201-233. Despres, L. A., 'The Implications of Nationalist Politics in British Guiana for the Development of Cultural Theory'. American Anthropologist, 66 (5), (1964), pp. 1051-1077. Driver,H.E. and Massey.W.C., Comparative Studies of North American Indians. Philadelphia, Transactions of the American Philosophical Society, 47 (2), (1957). Dube.S.C., Indian Village. London, Routledge and Kegan Paul, 1955. Eisenstadt,S.N., From Generation to Generation; Age-Groups and Social Structure. London, Routledge and Kegan Paul, 1956. Elwin.V., The Muria and their Ghotul. London, Oxford University Press, 1947. Evans-Pritchard,E.E., The Nuer; A Description of the Modes of Livelihood and Political Institutions of a Nilotic People. Oxford, The Clarendon Press, 1947. —, 'The Comparative Method in Social Anthropology', in E.E.Evans-Pritchard, The Position of Women in Primitive Societies, and Other Essays in Social Anthropology (London, Faber and Faber, 1965). Fahrenfort, J. J., 'Over vrije en onvrije arbeid'. Mens en Maatschappij, 19 (1), (1943), pp. 29-51. Fortes,J., 'Time and Social Structure: An Ashanti Case-Study', in Social Structure. Studies Presented to A.R.Radcliffe-Brown (Oxford, The Clarendon Press, 1949). Fortes,M., 'Kinship and Marriage Among the Ashanti', in A.R.Radcliffe-Brown and D. Forde, eds., African Systems of Kinship and Marriages (Oxford University Press, 1950). —, 'The Structure of Unilineal Descent Groups'. American Anthropologist, 55 (1), (1953), pp. 17-41. Foster, G.M., Traditional Cultures and the Impact of Technological Change. New York, Harper & Row, 1962. Furnivall.J.S., Colonial Policy and Practice. London, Macmillan, 1948. Garbett,G.K., Growth and Change in a Shorn Ward, Occasional Paper No 1. Salisbury, University College of Rhodesia and Nyasaland, Dept. of African Studies, 1960. Garfield,I.E., 'Tsimshian Clan and Society'. University of Washington Publications in Anthropology, 7 (3), (1939). GifFord.E.W., Tongan Society, Bernice P. Bishop Museum Bulletin, 61. Honolulu, 1929. Gladwin, T. and Sarason,S.B., Truk: Man in Paradise, Viking Fund Publications in Anthropology, No 20. New York, Wenner-Gren Foundation, 1953. Gluckman,M., 'Kinship and Marriage Among the Lozi of N.Rhodesia and the Zulu of Natal', in A.R.Radcliffe-Brown and D.Forde, eds., African Systems of Kinship and Marriage (Oxford University Press, 1950). Haekel, J., 'Zum heutigen Forschungsstand der historischen Ethnologie', in J. Haekel, et at., eds., Die Wiener Schule der Völkerkunde (Horn-Wien, E.Berger, 1956), pp. 17-90. Hempel,O.G. and Oppenheim, P., Der Typusbegriff im Lichte der neuen Logik. Leiden, A.W. Sijthoff, 1936. Hobhouse, L. T., Wheeler, G. C. and Ginsberg, M., The Material Culture and Social Institutions of the Simpler Peoples. London, Chapman & Hall, 1930. Hoebel.E. A., The Law of Primitive Man. Cambridge, Mass., Harvard University Press, 1954. —, Man in the Primitive World. New York, McGraw-Hill, 1958.

The logic of cross-cultural analysis: why exceptions?

51

Hoetink,H., De gespleten samenleving in het Caribisch gebied. Assen, Van Gorcum, 1962. Hoffmann-Burchardi.H., 'Die Yoruba Städte in Südwest-Nigeria'. Erdkunde, 18 (3), (1964), pp. 206-234. Homans, G. C., Sentiments and Activities. Essays in Social Sience. London, Routledge and Kegan Paul, 1962. —, and Schneider,D.M., Marriage, Authority and Final Causes. Glencoe, 111., The Free Press, 1955. Honigmann, J. J., The Kaska Indians; A Historical Reconstruction. New Haven, Conn., Yale University Press, 1954. Horton, D., 'The Functions of Alcohol in Primitive Societies'. Quarterly Journal of Studies on Alcohol, 4 (1943), pp. 199-321. Jongmans.D.G., Politiek in Polynesia. Beverwijk, J.deSwart, 1955. Kloos,P., De relatie ego (man) versus schoonmoeder, en ego (vrouw) versus schoonvader bij enkele Zuid-Bantoes. MS, 1961. —, 'Duolineaire afstamming en het M.C.C.-huwelijk'. Bijdragen Taal-, Land- en Volkenkunde, 119 (3), (1963), pp. 287-295. —, 'Duolineaire afstamming en het M.C.C.-huwelijk; eenrepliek'. Bijdragen Taal-, Landen Volkenkunde, 120 (3), (1964), pp. 368-375. Kloosterboer, W., Involuntary Labour Since the Abolition of Slavery. Leiden, E.J.Brill, 1960. Köbben, A. J.F., 'New Ways of Presenting an Old Idea; the Statistical Method in Social Anthropology'. Journal of the Royal Anthropological Institute, 82(2), (1952), pp. 129-146. —, 'Le planteur noir'. Etudes Eburnéennes, 5, (1956). —, Vanprimitieven tot medeburgers. Assen, Van Gorcum, 1964. Leach, E., 'Comment on Schiffler's Note on the Mangaian Kopu'. American Anthropologist,66(2), (1964). —, 'Comment'. Current Anthropology, 5 (4), (1964), p. 299. Lévi-Strauss, C., 'Social Structure', in A.L.Kroeber, ed., Anthropology Today (Chicago, University of Chicago Press, 1953). —, Anthropologie structural. Paris, Plön, 1958. Lewis, O., Life in a Mexican Village: Tepoztldn Restudied. Urbana, University of Illinois Press, 1951. Lienhardt,G., 'The Western Dinka', in J.Middleton and D.Tait, eds., Tribes Without Rulers (London, Routledge and Kegan Paul, 1958). Lo wie, R. H., Primitive Society. New York, Liveright, 1947. Mabogunje, A., 'Urban Landscape of Ibadan', in Ibadan in the Changing Nigerian Scene (Ibadan, 1964). Mimeo. McEwen, W. J., 'Forms and Problems of Validation in Social Anthropology'. Current Anthropology, 4(2), (1963), pp. 155-183. Malinowski,B., Crime and Custom in Savage Society. London, Kegan Paul, Trench, Trubner and Co., 1926. Merton, R.K., Social Theory and Social Structure. Glencoe, 111., The Free Press, 1956. Mitchel.N.C.,'Yoruba Towns', inK.M. Barbour and R. M.Prothero,eds., Essayson African Population (London, Routledge and Kegan Paul, 1961). Murdock.G.P., Social Structure. New York, Macmillan, 1949. —-, 'Universals of Culture', in A.Hoebel, J.D.Jennings, and E.R.Smith, eds., Readings in Anthropology (New York, McGraw-Hill, 1955). —, 'World Ethnographic Sample', in F.W. Moore, ed., Readings in Cross-Cultural Methodology (New Haven, Conn., HRAF Press, 1961).

52

Andre J. F. Köbben

Naroll.R.,'A Preliminary Index of Social Development'. American Anthropologist, 58(4). —, 'Two Solutions to Galton's Problem', in F.W.Moore, ed., Readings in Cross-Cultural Problems (New Haven, HRAF Press). —, Data Quality Control', a New Research Technique. Glencoe, 111., The Free Press, 1962. —, 'On Ethnic Unit Classification'. Current Anthropology, 5 (4), (1964), pp. 283-291. —, and D'Andrade, R.G., 'Two Further Solutions to Galton's Problem'. American Anthropologist, 65 (5), (1963), pp. 1053-1067. Nieboer, H. J., Slavery as an Industrial System. 2nd ed., 's-Gravenhage, M. Nijhoff, 1910. Nooteboom, C., Oost-Soemba, een volkenkundige studie. 's-Gravenhage, M. Nijhoff, 1940. Oluwasanmi.H. A., 'Agricultural Environment of Ibadan', in Ibadan in the Changing Nigerian Scene (Ibadan, 1964). Mimeo. Opler.M., 'The Human Being in Culture Theory'. American Anthropologist, 66 (3), (1964), Part 2. Otterbein, K. F. and Swanson, C. L., 'An Eye for an Eye, a Tooth for a Tooth; a Cross-Cultural study of Feuding', American Anthropologist, 67 (6), (1965), pp. 1470-1482. Pilling,A.R., 'Statistics, Sorcery and Justice'. American Anthropologist, 64 (5), (1962), pp. 1057-1059. Pospisil, L., Kapauku Papuans and their Law, Yale University Publications in Anthropology, No 44. New Haven, Conn., 1956. Pouwer, J., 'Loosely Structured Societies in Netherlands New Guinea'. Bijdragen Taal-, Land- en Volkenkunde, 116 (1), (1960), pp. 109-118. —, Het individu in samenleving en cultuur. Groningen, Wolters, 1962. Radcliffe-Brown, A. R., 'A Note on Functionalist Anthropology'. Man, 46 (29-30), (1946), pp. 38^1. —, 'Primitive Law', in Structure and Function in Primitive Society (London, Cohen and West, 1956). Radin.P., 'The Autobiography of a Winnebago Indian'. University of California Publications in American Archaeology and Ethnology, 16 (7), (1920). Rattray, R.S., Ashanti Law and Constitution. Oxford, The Clarendon Press, 1929. Redfield,R.S., Tepoztlan, a Mexican Village, Chicago, University of Chicago Press, 1930. —, The Little Community. Chicago, University of Chicago Press, 1955. Schapera,I., Married Life in an African Tribe. New York, Sheridan House, 1941. —, 'Kinship and Marriage Among the Tswana', in A. R. Radcliffe-Brown and D. Forde, eds., African Systems of Kinship and Marriage (Oxford University Press, 1950). —, 'Comparative Method in Social Anthropology'. American Anthropologist, 55 (3), (1953), pp. 353-362. Schneider,H.K., 'Pakot Resistance to Change', in W.R.Bascom and M.J.Herskovits, eds., Continuity and Change in African Cultures (Chicago, University of Chicago Press, 1959). Serpenti, L., Cultivators in the Swamp; Social Structure and Horticulture on Fred.Hendrikeiland( West New Guinea). Assen, Van Gorcum, 1965. Siegel,B. J., 'Some Methodological Considerations for a Comparative Study of Slavery'. American Anthropologist, 47 (3), (1945), p. 357. Sjoberg,G., The Pre-Industrial City. Glencoe, 111., The Free Press, 1960. Textor.R.B., Computer Summarization of the Coded Cross-Cultural Literature. MS, 1965. —, A Cross-Cultural Summary. New Haven, Conn., HRAF Press, 1967. Tylor.E.B., 'On a Method of Investigating the Development of Institutions'. Journal of the Royal Anthropological Institute, 18, (1889), pp. 245-280. Udy,S.H., 'Cross-Cultural Analysis; a Case Study', in P.E.Hammond, ed., Sociologists

The logic of cross-cultural analysis: why exceptions?

53

at Work (New York, J.Wiley, 1964). Van der Leeden, A.C., Hoofdtrekken der sociale structuur in het Westelijk binnenland van Sarmi. Leiden, Udo, 1956. Van Lier,R.A.J., Samenleving in een grensgebied. 's-Gravenhage, M.Nijhoff, 1949. Van Wouden,F.A.E., Sociale structuurtypen in degrote Oost. Leiden, Ginsberg, 1935. Watson, W., Tribal Cohesion in a Money Economy. Manchester, Manchester University Press, 1958. White, L. A., The Science of Culture. New York, Farrar, Straus & Co., 1949. Whiting, B. Blythe, Paiute Sorcery, Viking Fund Publications in Anthropology No 15. New York, 1950. Wilson, T. R., 'Randomness of the Distribution of Social Organization Forms. A Note on Murdock's Social Structure'. American Anthropologist, 54 (1), (1952), pp. 134-138. Wittfogel.K. A., Oriental Despotism. New Haven, Conn., Yale University Press, 1957. Zetterberg, H.L., On Theory and Verification in Sociology. Stockholm, Almqvist and Wiksell, 1954; New York, Tressler Press, 1954; Totowa, N.J., Bedminster Press, 3rd enl. ed., 1965.

C H A P T E R II

Computer summarization of the coded cross-cultural literature ROBERT B. TEXTOR Stanford University

1. Introduction This paper will discuss the application of a computer technique, the Pattern Search and Table Translation Technique, as it has been used to summarize the entire body of available coded cross-cultural literature. The development of this technique, and its application to cross-cultural analysis, took place between 1959 and 1965 at Yale, Harvard, and Stanford Universities. Results will be published in 1967 by the Human Relations Area Files Press of New Haven, Conn., USA, in a book to be entitled A Cross-Cultural Summary. The primary hoped-for value of the book will lie in the selection, organization, and presentation of information in a form that will stimulate the reader to construct more general and powerful hypotheses as to: 1) the interlinkages among various social, cultural, and psychological phenomena in the cross-cultural universe; and 2) the mode of causation, or functional significance, of such interlinkages. A secondary value of the book will be to serve, in some instances, as a hypothesis-checking device. 2. Computer procedure The materials used in the book are drawn selectively from all available sources of coded cross-cultural data, both published sources and such unpublished sources as I was authorized to use. These sources, 38 in number, cover the entire range of interests of cross-cultural scholars to date. (Details appear in Section 9 at the end of this chapter.) Input to the computer consists of cro ss-cultural data in dichotomous, or dichotomized, form. (In deciding which dichotomies to include, I

Computer summarization of the coded cross-cultural literature

55

was advised by Professor George Peter Murdock of the University of Pittsburgh, and others.) The computer crossed each dichotomous variable against all the others. A very large number of two-by-two tables is thus generated. Next, the computer automatically winnows out all 'nonsignificant' or 'uninteresting' tables — that is, all those whose 'P' values are higher than the (arbitrary) cut-off level of 10 per cent. The computer then prints out the remaining 'significant' or 'interesting' tables. For each table, it also prints out the chi-square value, the phi coefficient, and the two-tailed 'P' value. (Where N is equal to or less than 40, the 'P' value is determined by the Fisher Exact Test.) A departure from previous practice is that the computer also 'translates' all significant tables, printing out the translations in the form of grammatical Englishlanguage sentences. These sentences are logically organized into paragraphs, and the paragraphs into sections. Such translations hopefully make the book a more convenient reference work than would otherwise be the case, and facilitate rapid cross-reference inquiry. It is also hoped that these translations will enhance the heuristic value of the book, since most social scientists seem to think more creatively in linguistic than in numeric terms. Details of an earlier, and slightly different, version of the Pattern Search and Table Translation Technique appear in Chapter II of A Cross-Polity Survey (CPS), by Arthur S.Banks and RobertB.Textor (Cambridge, M.I.T. Press, 1963). The finished book is the result of four successive computer runs. After each run, I examined the printout and decided which dichotomous variables could advantageously be dropped and which, in the interest of brevity, should be run only one way instead of two. Additional brevity and readability were achieved by selective 'hand-winnowing' (see CPS, pp. 40—41). Professor Murdock served as adviser throughout this process of selective reduction of output. A copy of the most relevant of the preliminary runs will be deposited in the library of the Human Relations Area Files headquarters in New Haven, for use by scholars who might wish to examine portions of it not included in the final volume. 3. Scope and organization The finished book will be approximately 2,794 pages long. The first 208 pages will be an introduction and commentary, set in letterpress. The remaining 2,586 pages will be computer printout, printed by photo-offset. By the use of unusually thin, opaque paper it will be possible to bind the entirety in one volume. The introduction will contain the following elements: A. List of the 400 cultures used, and the principal enthnographic sources for each, as cited by the Ethnographic Atlas. (Details in Section 4 below.) B. List of alternative names, if any, for each of the 400 cultures. C. Citation and description of each source of codings.

56

Robert B. Textor

D. Full listing and commentary on the sample of each of the 38 studies from which codings were taken. E. Detailed explanation and evaluation of the statistics and the methodology. F. Full description of the coding rules used in formulating each input variable. G. General and particular caveats against misuse of the printout. H. Suggestions for possible further analytical operations to be performed by the reader. I. Bibliography of cross-cultural sources, both substantive and methodological. In short, the introductory 208 pages will be similar in intent and nature to the introductory chapters of CPS. The computer printout portion is organized as follows: Each dichotomous variable provides a pair of contrasting subjects. Each pair of subjects, taken together with all associated pairs of significant predicates, is a 'paragraph'. The arrangement of predicates within a paragraph follows a uniform order. This order is the same as that of the paragraphs themselves, and accords as closely as possible with the order typically followed in an ethnographic monograph. In other words, information is presented in an order that is congenial to the accustomed thinking habits of most readers. Thus, the paragraphs are grouped into sections dealing with the following subjects: location of the culture in question, linguistic affiliation, natural environment, settlement pattern, diet, subsistence base, technology, writing system, demography, political integration, societal complexity, social stratification, work organizations, occupational specialization, economics, law, jurisprudence and medicine, community organization, largest noncognatic kin group, lineality of kin group, lineality of inheritance, authority within the family, avoidance, mode of marriage, divorce, status of women, fertility, pregnancy and childbirth, infancy and childhood, adolescence, sex, illness and therapy, aggression and warfare, religion-magic-eschatology, games, culture contact and culture exchange, miscellaneous methodological controls. The Appendices, which follow the introductory portion, will be similar to those appearing in CPS, and will include a listing of all codings and all handwinnowings. 4. Sample The sample used in the study is the 400-culture sample employed by Professor Murdock in the Ethnographic Atlas, which appears serially in the journal, Ethnology. This is a judgmental sample and reflects the experience of Professor Murdock over a period of thirty-odd years of cross-cultural research. Cultures included are largely nonliterate and pre-industrial. It can be convincingly argued that these 400 cultures comprise a relatively satisfactory sample for many purposes of crosscultural research. In my view, however, the sampling question remains an open

Computer summarization of the coded cross-cultural literature

57

one. Both the Human Relations Area Files and I will offer every cooperation to researchers desirous of devising and using other sampling designs. Hopefully, as results from a variety of sampling designs begin to accumulate, many issues concerning sampling will gradually be resolved. 5. Built-in safeguards To serve as a check on sampling and related types of error, a number of safeguards have been built into the book. Dummy 'whiskers' paragraphs have been included. (See CPS, page 51.) Also included is a paragraph for each of the 38 sources, in which are contrasted the cultures included in the particular coder's sample (out of the Murdock 400), versus the cultures not so included. Thus the reader is enabled to inspect the 'biases', or attributes, of each coder's sample, and evaluate these biases for himself. He is also enabled to see how closely one coder's sample overlaps with another coder's sample. It is reassuring to note that the degree of such overlap is fairly high in most cases. There are also several paragraphs that permit inspection of the peculiar biases or attributes of cultures whose principal ethnographer(s), as listed by Murdock, come from a particular country. For example, it turns out that there are 52 cultures out of the 400 where the principal ethnographer(s) were reared and trained in the British area (Britain, Australia, New Zealand, South Africa). There are 113 cultures where the principal ethnographers were reared and trained in the American area (U.S. and Canada). When this British-American contrast is crossed with the contrast between 'patrilineal or double descent' versus 'matrilineal', it turns out that British ethnographers have been inclined to choose patrilineal or double-descent cultures for study, while American ethnographers have been inclined toward matrilineal societies, and this difference is significant beyond the three per cent level. (This appears in the printout as Statement 483/190.) In similar fashion, paragraphs are included which contrast different periods of ethnography. Thus, for example, it is possible to examine the attributes of cultures for which the principal ethnographies were written prior to 1950 as against 1950 and afterward. 6. Summary of possible advantages As is now clear, the Summary makes no innovations of a statistical nature. The innovations that it does make are ones simply of data presentation, and do not mark a very sharp break with the past. Nonetheless, it is felt that this new method of presenting information does offer some advantages to some users. These possible advantages may be summarized as follows: A. The entire coded cross-cultural literature is covered. Codings from a particu-

58

Robert B. Textor

lar scholar's work (always identified by his initials) are systematically crossed with codings from each other scholar's work. B. While the scope of the book is widely inclusive, it is also selective. Those analytical categories have been selected which seem to promise the greatest yield in terms of theory-building. In making the selections, I was guided by an inspection of three preliminary runs, and by the advice of experts who had also had the opportunity to inspect preliminary runs. C. All tabular information is printed out, plus all types of statistical information customarily preferred by scholars interested in cross-cultural work, namely, the chi-square, the phi, and the 'P'. D. Experience to date indicates that the heuristic value of the tabular material is enhanced when it is 'translated' into linguistic form. These 'translations' will hopef ully help attract to the cross-cultural field a number of creative social scientists and other thinkers who find untranslated tabular material uncongenial or even forbidding. E. Numerous techniques have been employed to minimize reading burden while maximizing information. (See CPS, pp. 34—36, 39—41.) F. By alphabetically listing each culture in its appropriate coding category at the head of each paragraph the printout makes it easy to check coding assignments, and stimulates the discovery of errors. This listing feature might also prove heuristically valuable in that it provides illustrative examples of cultures belonging in each coding category. (See CPS, p. 31.) G. Experience indicates that the logical-conventional ordering of variables stimulates the reader to broaden the scope of his theory-building activities, and to perceive previously unsuspected relationships between different domains of data, for example between: level of political integration and child-rearing practices; subsistence base and sexual behavior; or family structure and religious belief. H. The structure and ordering of the variables is planned to facilitate explorations by researchers concerned with various social evolutionary approaches. I. By use of composite variables and residues, the book controls for a variety of intervening or interpretive variables and focuses on particular contrasts believed to be of crucial interest. (See CPS, pp. 24—25.) J. All tables and associated sentence-pairs that pass the winnowing test are automatically printed out. The reader is thus forced to confront all the significant information, perhaps including some which is contrary to his biases, hypotheses, or expectations. This discourages the kind of unconscious intellectual 'cheating' that can creep into the work of any scholar, even the most conscientious. K. The 53 methodological paragraphs are aimed at making maximally visible to the trained reader whatever biases and inadequacies might inhere in the samples used by various scholars, as well as in the 400-culture sample of the Ethnographic Atlas itself.

Computer summarization of the coded cross-cultural literature 59 L. The paragraphs on the nationality of the ethnographer might stimulate insights of a 'sociology of knowledge' nature. M. The paragraphs on date of ethnography will perhaps give the reader a feel for overall patterns in the development of ethnology as a science, and in the development of ethnographic techniques. N. Since all significant sentence-pairs print out, and since they appear in a uniform order, the reader readily develops a sense of holistic context and pattern. In other words, the traditional anthropological emphasis on context and pattern within a particular culture may now be brought to bear in crow-cultural research as well. O. Context is also achieved in the additional sense that cross-referencing is made easy. When the reader encounters a theoretically interesting predicate-pair, he can instantly turn to the paragraph where that predicate-pair is the subject-pair of its own paragraph, and examine the significant association that it has with other variables. Take a hypothetical example. In reading through Paragraph 187, the reader discovers in Statement 187/390 that cultures coded as matrilineal never severely punish premarital sex relations. He then turns to Paragraph 390 to see what are the other significant attributes of cultures coded as severely punishing premarital sex relations. Here he finds Statements 390/84—87, which reveal patterned, persistent positive associations between strong premarital sex controls and complexity of political organization. P. By studying the entirety of a particular paragraph, the reader might discover some other explanatory variable, rather than the one under consideration, which seems to explain a particular phenomenon more convincingly or adequately. Q. In other instances, the failure of a potential predicate-pair to print out can in effect mean, or suggest, that a previously suspected relationship does not hold. A hypothetical example of this latter case is as follows: the reader has for a long time suspected that strength of desire for children is more importantly determined by subsistence base than by form of social organization. He happens to be reading Paragraph 190 and his attention is arrested by Statements 190/282,295. These tell him that cultures which are patrilineal (or double-descent) are characterized by a high desire for children and a high punishment for abortion — while matrilineal cultures, by contrast, are characterized by a low or absent coding on both counts. He also notes Statements 190/51, 53, 54, 55, 62,63, which reassure him by revealing positive associations between patrilineal (and double-descent) cultures, and more complex forms of subsistence. To be strictly honest, however, he turns to Paragraphs 282 and 295. Here he discovers that, in both paragraphs, potential statements 51 through 70 — those dealing with subsistence base — completely fail to print out. On the face of this, perhaps the only permissible tentative conclusion is that the patrilineality-matrilineality distinction has a more important bearing on strength of desire for children, than does type of subsistence base. (For an interest-

60

Robert B. Textor

ing minor exception, see Statements 58/282,295, which are fairly well explained by Statements 58/186,190,198.) R. The above two items suggest what might in the long run prove to be the most important advantage of the book, namely that it provides a relatively objective basis upon which social scientists who are wary of cross-cultural statistical analysis immediately can and conveniently enter into meaningful dialogues with social scientists who are committed to such methods. S. If the reader takes the stance of testing a directional a priori hypothesis, then the book will serve as a device for 'instant hypothesis testing'. T. The book might sometimes serve as an instant argument-settler. For example, suppose that a hypothetical political scientist contends that presence of some kind of state organization normally implies social stratification based on wealth (rather than stratification based on something else). He turns to Paragraph 86 and sees, in Statement 86/106, that there is indeed a significant correlation. However, it is in the reverse direction from that which he predicted. The burden is now quite heavily upon him either to explain this away or to alter his position. In other cases, presumably more numerous, there wille be no significant association either way. U. The book can be used to avoid 'false starts'. It facilitates preliminary investigation of a hypothesis to see if 'there is something there' which would warrant the expenditure of substantial time and money on a further, full-fledged investigation calling for new theoretical formulations and new coding operations. V. Experience to date indicates that the format of the printout often readily suggests to a trained reader what further analytical techniques are needed. These might involve such techniques as deviant case analysis, the construction of multivariable contingency tables for secondary analysis, factor analysis, cluster analysis, or scaling. W. As a researcher employs the book on a focused research problem, he will often discover that new coding categories are needed in order to attack problems central to his inquiry. He will thus hopefully be stimulated to make creative use of existing codings by refining and combining them in new ways, and/or to establish new coding categories in accordance with his theoretical interests, and then proceed to do his own coding. In any case, the field of cross-cultural research will be the gainer. X. Experience to date indicates that the printout is useful to a teacher preparing a lecture. Y. As a tool for the field worker to carry with him into the field, the book will hopefully serve as a stimulus to more creative inquiry, or at least as a prod upon the field worker to get more complete descriptive coverage of the variables that have been used in previous cross-cultural studies. Z. The Pattern Search and Table Translation Technique is not bound to the English language, but presumably can be used for any language. It has already been

Computer summarization of the coded cross-cultural literature

61

used experimentally for French, German, Russian, and romanized Siamese. Production of this book in another language would hence be possible if demand were sufficient. Translation costs would be minimal, because equivalents need be found for only five verbs, two adverbs, one subject noun, and a limited number of clauses.

7. Follow-up activities planned by Human Relations Area Files The publisher plans a sustained follow-up of this book. H.R.A.F. will be the central distributing point of all the data decks. Any scholar, anywhere in the world, may write to H.R.A.F. and receive, at cost plus postage, a complete copy of these decks. In cooperation with Professor Murdock and the Ethnographic Atlas, H.R.A.F. will be constantly incorporating into these decks addenda and corrigenda submitted by experts on particular cultures. H.R.A.F. will also systematically add new codings from published sources or from scholars who wish to share their codings in advance of publication. At appropriate intervals, it might be advisable to bring out a revised edition of A Cross-Cultural Summary incorporating addenda, corrigenda, and codings from new studies that have appeared in the interim.

8. The farther future While the Human Relations Area Files promises maximum cooperation to scholars wishing to pursue their own follow-up inquiry, it is to be hoped that advances in computer technology will also soon come to our assistance. One may look forward to the day when any scholar, wherever located, can sit down at a special typewriter and type out instructions to a computer to process cross-cultural materials in a certain way. He should be able to lump, refine, and composite variables from A Cross-Cultural Summary at will. The tables and matrices he desires (together with associated statistical information) would then instantaneously print out. After studying the results, the scholar reformulates his hypotheses in accordance with his theoretical orientation, and then types out new instructions to the computer. In such manner, computer technology can take the drudgery out of inquiry, and free us for creative thinking.

9. Sources Variables selected for inclusion in the book will utilize codings from works by the following authors: Ackerman,C., 'Affiliations: Structural Determinants of Differential Divorce Rates'. American Journal of Sociology, 69(1), (1964), 13-20. Anthony, A. S., Male Initiation Rites. Unpublished Doctoral dissertation, Harvard University.

62

Robert B. Textor

Apple,D., 'The Social Structure of Grandparenthood'. American Anthropologist, 58 (1956), pp. 656-663. Ayres, B. C., A Cross-Cultural Study of Factors Relating to Pregnancy Taboos. Unpublished Doctoral dissertation, Harvard University. Bacon,M.K., Barry,H.M., and Child,I.L., Child Rearing. Unpublished codings, Yale University, 1955. Barry, H. Art. Unpublished Bachelor's thesis, Harvard University. Brown, J. K., 'A Cross-Cultural Study of Female Initiation Rites'. American Anthropologist, 65(1963), pp. 837-853. D'Andrade.R.G., 'Anthropological Studies of Dreams', in F.L.K. Hsu, ed., Psychological Anthropology (Homewood, 111., 1961). Evan, W. M., Law. Unpublished codings, 1963. Ford.C.S., A Comparative Study of Human Reproduction, Yale University Publications in Anthropology, No 32. New Haven, Conn., Yale University, 1945. — and Beach, F. A., Patterns of Sexual Behavior. New York, Harper & Row, 1951. Freeman,L.D. and Winch,R.F., 'Societal Complexity: An Empirical Test of a Typology of Societies'. American Journal of Sociology, 62 (1957), pp. 461-466. Goodman,J.F., Mourning Behavior. Unpublished Bachelor's thesis, Harvard University. Harley,J.K., Adolescent Peer Groups. Unpublished Doctoral dissertation, Harvard University. Hickman,J.M., 'Dimensions of a Complex Concept: A Method Exemplified'. Human Organization, 21 (3), pp.214-218. Horton,D., 'The Functions of Alcohol in Primitive Societies, a Cross-Cultural Study'. Quarterly Journal of Studies on Alcohol, 4 (1943), pp. 199-320. Jackson, M., A Study of the Evolution of Social Control: The Organization, Theory and Practice of Jurisprudence and Medicine. Ann Arbor, Mich., Mental Health Research Institute, University of Michigan, 1962. Lambert, W. W., Triandus, L. M., and Wolf, M., 'Some Correlates of Beliefs in the Malevolence and Benevolence of Supernatural Beings: A Cross-Cultural Study'. Journal of Abnormaland Social Psychology, 58, pp. 162-169. Leary.J.R., Food Taboos and Level of Culture: A Cross-Cultural Study. New Haven, Conn., Human Relations Area Files, 1961. McClelland,D.C., The Achieving Society. Princeton, N.J., Van Nostrand, 1961. Moore, F.M., Unpublished codings on type of natural environment, New Haven, Conn., Human Relations Area Files, 1963. Murdock,G.P., 'World Ethnographic Sample'. American Anthropologist, 59 (1957), pp. 664-687. Nag,M., Factors Affecting Human Fertility in Non-Industrial Society: A Cross-Cultural Study, Yale University Publications in Anthropology. New Haven, Conn., 1962. Naroll,R., 'A Preliminary Index of Social Development'. American Anthropologist, 58 (1956), pp. 687-715. Roberts,J.M., Arth,M.J., and Bush,R.R., 'Games in Culture'. American Anthropologist, 61 (1959), pp. 597-605. Shirley,R.W. and Rommey.A.K., 'Love Magic and Socialization Anxiety: A CrossCultural Study'. American Anthropologist, 64 (1962), pp. 1028-1031. Simmons, L.W., 'Statistical Correlations in the Science of Society', in G.P.Murdock, ed., Studies in the Science of Society (New Haven, Conn., Yale University Press, 1937). Slater,P.E. Unpublished coding guide for the cross-cultural study of 'narcissism', Brandeis University, 1963.

Computer summarization

of the coded cross-cultural literature

63

Stephen,W.N., The Oedipus Complex: Cross-Cultural Evidence. New York, The Free Press of Glencoe, 1962. Swanson.G.E., The Birth of the Gods: The Origins of Primitive Beliefs. Ann Arbor, Mich., University of Michigan Press, 1960. Udy,S.H., Organization of Work: A Comparative Analysis of Production Among NonIndustrial People. New Haven, Conn., HRAF Press, 1959. Veroff.J., Achievement Motivation as Revealed in Folktales. Codings contributed to D. C. McClelland, op. cit. Whiting,B.B., Paiute Sorcery. New York, The Viking Press, 1950. Whiting, G.M., Diet. Unpublished Doctoral dissertation, Harvard University. Whiting, J. M. and Child, I. L., Child Training and Personality: A Cross-Cultural Study. New Haven, Conn., Yale University Press, 1953. Whiting,J.W.M., Kluckholm.R., and Anthony,A., 'The Function of Male Initiation Ceremonies at Puberty', in E.Maccoby, T.Newcomb, and E.Hartley, eds., Readings in Social Psychology (New York, Henry Holt, 1958).

CHAPTER III

Concepts in comparative historical analysis* REINHARD BENDIX University of California, Berkeley

The renewed interest in comparative studies of social change dates from World War II. This intellectual repercussion of the war and its aftermath is most apparent in the discontinuities of interest which have marked the work of American social scientists in recent decades. Before World War II American scholars devoted their primary attention to the study of American society. Even if one considers the tremendous popularity of theories of social evolution in the United States before the 1920's, one is struck by the fact that these theories were largely applied in Social Darwinist fashion to an interpretation of the competitive economic struggle. The predominant American concern was domestic in contrast with Europe, where these theories originated and where they were used to interpret the encounter between the advanced industrial societies of Europe and the peoples and cultures of colonial and dependent areas. With the notable exception of anthropologists, this intellectual 'insularity' of American social scientists may be related to America's anti-colonial heritage, just as the renewed interest in comparative studies may be related to America's worldwide political involvements since World War II. As a result the earlier parochial orientation has declined, as economists, political scientists, and sociologists attempt to assess the relativity as well as the characteristics of the American experience. A steadily increasing number of social scientists are concerned with non-Western areas, especially with regard to problems of 'modernization'. This concern has benefited markedly from the wartime experience * Revision of a paper originally presented to the Round Table on Comparative Research, International Social Science Council, Paris, April 22-24,1965. The paper develops themes, presented in more empirical detail in my book Nation-Building and Citizenship (New York, J.Wiley, 1964).

68

Reinhard Bendix

of many scholars, including extensive training in foreign languages. It has also been affected by the revolution in research methods which modern computers make possible with regard to the storage and evaluation of data. We seem to find ourselves in a period of intellectual transition, and the participants in this reorientation naturally tend to emphasize its novelty. The shift of emphasis is evident if one compares the preoccupation of scholars with the conventional teaching of history. Typically, the student learns the history of his own country in considerable detail, while the histories of other countries are presented to him much more selectively, or not at all. At one time professional historians defended this conventional method on the ground that the development of each country is unique so that the obvious concern with education for citizenship coincided with a plausible, intellectual conviction. Yet today historians no longer adhere to this position as firmly as they did some two generations ago. In notable instances they have presented comparative studies of their own. A series of publications has been concerned with problems of generalization in historical studies. And while preoccupation with national history remains predominant, many scholars so preoccupied are nevertheless concerned with the questions of conceptualization of central interest to social scientists.1 Yet the change in intellectual orientation may be more apparent than real. The greater receptivity towards a conceptual and comparative approach to the study of history is not matched by much agreement on what such a program of study implies positively. There is little agreement on what is to be understood by such recurrent terms as analysis, change, social structure, and comparison. Taking each in turn, I shall try to indicate the issues involved in our use of these terms. My purpose is to propose an approach to the comparative analysis of historical change at an intermediate level of abstraction.2 Analysis. At least three divergent approaches to the study of historical change may be distinguished for purposes of orientation. The older, evolutionist approach tended to be classificatory. It assumed that the less developed countries will follow the 'steps and sequences of change' through which the more developed have passed already. Analysis becomes a matter of assigning culture traits or even a whole country at a given time to a specific stage of development. Once this is done, it is possible to assess the progressive or regressive significance of ideas and actions, either because the future or next stage is 'known' in advance, or because it seems plausible to examine the past of the 'developed' countries for purposes of such retrospective evaluation. To be sure, evolutionist theory is no longer expounded in such simplistic terms. Scholars have become more cautious than their predecessors: concepts of differentiation or increasing complexity are substituted for the idea of progress, and allowance is made for multilinear developments and the reversal or omission of 'stages'.3 But while these modifications go far, it is not clear that the original theory has been abandoned. The proliferation of synonyms of

Concepts in comparative historical analysis

69

change like development or modernization with their several adjectives warns us that this is an area of uncertainty and confusion; the new vocabulary often employs older theories of evolution uncritically. Related to this older approach, but more modern in its nomenclature, is the view that societies should be analyzed as 'natural systems'. In this perspective a social structure appears as an interrelated, functioning whole with systemic prerequisites, properties, and consequences, which may be identified as a 'stage of development'. Typically, such analysis runs the danger of reification, which occurs whenever a society is identified as a unit that maintains or changes itself, in order to 'survive' as such. I shall comment on this view below, but wish to refer here to one modern tendency which is related to, but not identical with, this holistic or systemic approach. I refer to the social engineering approach, which is oriented towards planned social change. In this view analysis should aim at the discovery of critical independent variables, since control of these will entail predictable changes in the dependent variables. Indebted to images derived from controlled experiments or from medical practice, this approach is less classificatory than the older, evolutionist theory and less organicist than system theory proper. But like these theories, its simplifying assumptions and tests of truth depend upon a ceteris paribus treatment of historical constellations. For example, the record of economic growth in the developed countries is employed as a model, however provisionally, so that historical preconditions reappear as logical prerequisites, without which growth cannot occur. In this way the engineering approach comes close to the 'natural systems' approach in that both operate with the concept of 'indispensable prerequisites', though the engineering approach is perhaps more candid in generalizing from the Western experience.4 Comparative analysis of historical change attempts a closer approximation to the historical evidence than is possible either on the assumptions of evolutionism, or of systems-theory, or of social engineering. As a result it promises less in the way of prediction and of guiding social actions towards defined goals. Whether this sacrifice is permanent or temporary remains to be seen. Studies of social change in complex societies may hold in abeyance the tasks of causal analysis and prediction while concentrating on the preliminary task of ordering the phenomena of social change to be analyzed further. This task can be characterized by reference to the meaning of 'change' and of 'social structure'. Change. At the risk of oversimplification, I shall assume that at a minimum considerations of change involve two terminal conditions, so that the word 'change' refers to the differences observed before and after a given interval of time. Since the future is uncertain, studies of historical change deal in the first place with past changes, the better to understand what the contrasts between 'before and after' are and how they have come about. Naturally it is hoped that a better understanding of historical changes will contribute to a fuller exploration of developmental

70

Reinhard Bendix

possibilities, perhaps even to constructive action, but the relations between knowledge and action are complex and should not be prejudged. Studies of change then depend upon contrasts between social structures before and after change has occurred. Without knowledge of the respects in which a later social structure differs from an earlier one, we do not know what changes to look for and explain. This is one reason why such studies use familiar concepts like feudalism, democracy, totalitarianism, etc., despite the many justified criticisms levelled against these terms. Such concepts express something we want to express, namely that in some overall and important, but rather general sense, an old social structure has passed away and a new one has taken its place. Dissatisfaction with such conventional terms is understandable, but it is no solution to substitute universal terms for these concepts of limited applicability. In a recent contribution, Gabriel Almond has suggested, for example, that 'interest aggregation' is a term cutting across all the conventional distinctions between political systems and hence can be applied universally. Such a term has the utility of prompting us to look for 'interest aggregation' in unfamiliar social structures to which our conventional terms do not apply, but it does not dispense with the utility of terms like 'class' or 'estate' which already differentiate — however approximately — the more familiar types of 'interest aggregation'. I suspect that we will invent new terms to fit the unfamiliar types of 'interest aggregation' once we have analyzed them sufficiently, for concepts are the result of inquiry as much as they are its precondition. What then is meant by 'social structures' and how do we study them comparatively? Social structures retain certain of their characteristics while individuals come and go. The specification of such enduring characteristics involves abstractions from observations of behavior and from historical evidence. On this basis studies of social change should be able to state that one type of social structure has ceased to prevail and another has taken its place. Yet to make such an assertion involves the hazards made clear by the debates concerning Max Weber's ideal type. Definitions of structures like feudalism, bureaucracy, etc., usually take the form of enumerating several, distinguishing characteristics. Such enumerations necessarily 'freeze' the fluidity of social life, as Weber himself emphasized. They say nothing about the strength or generality with which a given characteristic must be present, nor do they say anything about structures in which one or another element of the definition is missing. The result has been uncertainty. Abstractions are needed to define the characteristics of a structure and thus they remove the definition from the evidence. On the other hand, when we approach the evidence 'definition at hand', we often find its analytic utility diminished, because the characteristics to which it refers are in fact neither unequivocal nor general.5 Concretely: impersonal definition of rights and duties is one of the distinguishing criteria of bureaucracy. But 'impersonal definition' has meant many things: the

Concepts in comparative historical analysis

71

rights and duties of the classic Chinese bureaucrat and of an English official in the administrative class are worlds apart, even if both are impersonally defined. Nonetheless, the criterion is indispensable if we are to find all instances of 'bureaucracy' or properly identify those instances we do find. Then we will want to know how general and important the phenomenon of 'impersonal definition' is in a given case. Typically, this involves us in the task of analyzing the methods by which the rights and duties of officials are defined, and the degree to which these definitions correspond to behavior. That analysis will reveal the characteristic discrepancy between formally stipulated methods and actual implementation, and that discrepancy will raise questions about the utility of the criterion ('impersonal definition') with which we started. Thus, the criterion employed simplifies the instances to which it applies, and hence its analytic application poses difficulties. The dilemma is genuine, but there are proximate solutions. Examination of comparative studies suggests, it seems to me, that definitions of social structures are contrast-conceptions. Implicitly or explicitly, we define such terms as feudalism, capitalism, absolutism, caste-system, bureaucracy, and others by contrast with what each of these structures is not. For example, fealty-ties are contrasted with contractual, absolutist centralized with feudal decentralized authority, caste with tribe or estate, impersonal with personalized administration, the unity of household and business with their separation, etc. My suggestion is that contrast-conceptions are indispensable as a first orientation (they serve a function as benchmarks), which introduces analysis, but should not be mistaken for analysis. Since social structures are defined by several characteristics, more than one contrast conception may be found analytically useful. The choice depends in good measure on the purpose of inquiry and the historical context. In the emergency of modern bureaucracy, as Weber defined it, the recruitment of officials and their exercise of authority were emancipated from the direct intrusion of kinship relations and property interests. This aspect was in the foreground of attention, as long as hereditary privileges prevailed, but has declined in importance along with the rise of equalitarianism in all spheres of modern life. The exclusion of 'every purely personal feeling' remains an important desideratum and a proximate characteristic of official conduct, but this condition may not do as much today to ensure administrative impartiality as it did, as long as government by a social elite encouraged the intrusion of family loyalties and property interests upon the conduct of public business. That is, recruitment to official positions on the basis of impersonal criteria and the separation of office and incumbent remain characteristics of bureaucracy, but the changed structure of modern politics has altered their significance. For certain purposes it would be useful, therefore, to formulate an early and a later type of bureaucracy which would take account of this altered environment of government administration.

72

Reinhard Bendix

In this view of the matter the definition of a social structure in terms of a cluster of traits can only serve as a first approximation. On closer inspection every such trait proves to be an abstraction from the contentions among groups of men. The fealty relation between king and vassal is one of the defining characteristics of feudalism. But the contentions over their reciprocal rights and obligations between these classes of men are resolved in a variety of ways without thereby divesting that relation of the quality of 'fealty'. In this way social structures are defined by a set of issues which comprise the characteristic areas of contention among the constituent groups of a society. If we then say that one social structure has ceased to exist and another has taken its place, we mean that the terms of reference have changed by which issues are defined, relationships maintained, or contentions resolved. This is the meaning, it seems to me, of Tocqueville's classic specification of the contrast between a feudal and a democratic society.6 One corollary of these considerations is that concepts of social structure should be used in two forms. By bureaucracy we mean a depersonalized form of governmental administration, but we know that depersonalization is a matter of degree. Hence we use 'bureaucracy' when we wish to contrast one type of administration with another, and 'bureaucratization' when we wish to emphasize that the new terms of reference like 'depersonalized personnel selection' continues to be problematic, an issue whose every resolution creates new problems as well. Similarly one can distinguish between democracy and democratization, nation and nationbuilding, centralized authority and the centralization of authority, etc. Such usage will create linguistic problems from time to time. For example, Max Weber's usage of Vergesellschaftung instead of Gesellschaft had much the same purpose that I suggest here, but there is no proper English equivalent of this word-form, and it is not exactly usual in German either. Whatever the linguistic difficulties, we should keep the substantive distinction in mind. By defining social structures in terms of a set of issues, we not only avoid the reification of concepts, but make them 'operational'. If, in this way, we reformulate Max Weber's definition of bureaucracy, we obtain a specification of the issues over which individuals and groups contend in their effort to realize their ideas and maximize their chances, however they define these. The consequence of such contentions is a development in the direction of bureaucratization or debureaucratization as the case may be. Analysis of such contentions can account for the changing strength of the 'traits' which characterize a social structure, but are 'never twice the same'. 7 A third corollary is a reformulation on the concept of equilibrium. Having been taken over from feedback mechanisms like the thermostat or from biological analysis, the term is widely used by social scientists who employ the concept of 'social system'. Such 'systems' are believed to 'survive' as long as 'they' are in a condition of equilibrium or return to it. The idea has merit in the very general

Concepts in comparative historical analysis

73

sense that we combine with the concept 'social structure' the notion of some stability and identity over time. We must account for such stability as exists. However, I do not consider the concept 'equilibrium' useful for this purpose, because it is not the social structure or 'the system' which maintains itself in 'equilibrium', whatever that means, but men who by their actions (however conditioned) achieve a certain degree of stability, or fail to do so. Here the definition of social structure in terms of a set of issues helps, because it points to the contentions through which individuals and groups achieve a measure of accommodation or compromise between conflicting imperatives. By way of illustration, I shall reformulate Max Weber's types of domination in keeping with this perspective. The charismatic quality of a personality proves itself by its supernatural attributes (ultimately by miracles), and thereby gains recognition from the ruled. A leader will claim unconditional acceptance of his authority, but, as Weber says, if the test of this claim remains forever wanting, he will appear forsaken by his God or bereft of his heroic powers. Between the leader's unconditional claims and the followers' secret longing for visible signs of his 'gift of grace' this authority-relationship will fluctuate one way or another, but it will also endure as long as that tension exists. Similarly, under traditional domination, authority is exercised by the ruler in conformity with established precedent. Tradition also confers on him a certain latitude, so that the ruler acts arbitrarily in keeping with tradition. But when he regularly infringes upon the limits set by tradition, he runs the risk of jeopardizing the legitimacy of his own position. Guardians of traditional limits and guardians of the king's prerogatives are, therefore, typical groupings under this type of domination. Finally, Weber distinguishes between formal and substantial rationality of law. The legal order exists as long as neither principle is allowed an absolute ascendance. It is a continuous political and legal task to maintain enough balance between these antagonistic tendencies, for insistence on some principle of material justice can destroy the legal framework just as exaggerated formalism can undermine confidence in the legal system. In this view, stability of a social structure is not an 'equilibrium' that can be attributed to a 'system', but the end-product of always proximate efforts to maintain stability. Here may be the place also to comment briefly on a problem raised by a German and an English historian, both of whom warn us against the dangers of substituting inevitably arbitrary categories for the terms in which the historical participants themselves think about the questions at issue.8 The point is well taken, I believe, and the definition of social structure suggested here allows us to take account of this subjective dimension. But it is also necessary to go beyond that dimension and define the social structure which eventually results from all these contentions, and that cannot be done in subjective terms alone. Indeed, some abstraction and arbitrariness will be unavoidable in order to 'freeze' the fluidity of historical change for purposes of obtaining benchmarks, as suggestedearlier. It may be that the

74

Reinhard Bendix

deliberate employment of static and dynamic terms like bureaucracy and bureaucratization, democracy and democratization, etc., provides a way of conceptualizing both, the group contentions that are an essential part of change, and the altered social structures which from time to time result from that change. Comparison. The points discussed may now be considered in relation to the comparative analysis of historical change, and specifically of the 'steps and sequences of change in the processes of nation-building and national integration'. Ideally, we should be able to consider all such changes in the same terms, and there is a powerful intellectual legacy which invites us to do so. That legacy goes back to the contrast between tradition and modernity which was first formulated in the romantic period and has been reformulated ever since. Familiar dichotomies like status and contract, Gemeinschaft and Gesellschaft, folk and urban society, and others have been given their most systematic formulation in Talcott Parsons' scheme of pattern variables. The utility of these distinctions has been diminished, in my opinion at least, by a tendency toward reification. Nineteenth century evolutionary theory, for example, imputed to the different aspects of a society 'a strain of consistency, with each other, because they all answer their several purposes with less frictions and antagonism when they cooperate and support each other' (W.B. Sumner). Modern reformulations of this idea in terms of system theory and equilibrium are more sophisticated no doubt and have an impressive array of analogies to draw on, yet they continue to attribute a 'strain of consistency' to social structures such that the 'frictions and antagonisms' between the several traits will diminish—in the famous long run. 9 On these assumptions it is certainly possible to consider all societies in comparative terms, irrespective of time and space. This approach has been most fully developed with regard to the social and psychological consequences of industrialization, and in the field of national integration with regard to the study of the central value system. For the scholar interested in comparative studies, the alternative to this approach is to think in terms of concepts applicable to some (rather than all) societies. This strategy of analysis proceeds in the belief that concepts of universals — even if useful for certain orientating purposes - are so emptied of content that they require specifications in order to be applied to some body of evidence and these specifications are concepts of more limited applicability. Examples: interest aggregation is a universal concept while class, estate, political party, etc. are more limited; administration (or should I say: goal-attainment?) is universal but administration by disciples or bureaucrats or patrimonial servants are limited, and so on. By concepts of limited applicability, I mean concepts that are usefully applied to more than one society for a period whose approximate beginning and end are themselves an object of research. Such delimitation is always debatable. But however difficult in detail, I doubt that it is useful, for example, to speak of class in the absence of

Concepts in comparative historical analysis

75

a formal legal equality and the freedom of movement and expression that go with it, or of political parties in the era of politics among cliques of notables, or of the nation-state in the absence of a monopoly of legitimate coercion in the hands of government - although in these and other cases the qualifying criterion is a variable as well as a contrast-conception, as discussed earlier. These considerations are equally applicable to 'nation-building'. The concept 'nation' requires delimitation against a period and condition to which it does not apply, and it may be that with variations in such preconditions different types of 'nationhood' will have to be formulated. A nation is always in the process of change, for example in the extent to which consensus prevails or different sections of the people have formally equal rights - hence the phrase 'nation-building'. With regard to this process, the specification of an early and a late condition is necessary, the 'before-and-after model' to which I referred. This is in fact a crucial step in the procedure since comparative studies of nation-building depend on the success with which the different dimensions of nation-building can be conceptualized and then compared with one another. In this respect Stein Rokkan and I have experimented with categories derived from T. H . Marshall's analysis of Citizenship and Social Class, while Karl Deutsch has marshalled a great body of evidence on the physical indexes of nation-building. In some instances it may even be possible to combine quantitative indexes with more qualitative criteria, for example in the study of the franchise which has the unique advantage of involvingdichotomous choices like eligibility vs. nonelegibility and voting vs. nonvoting. The following table is suggestive in this respect: Extentions of the Franchise and Changes in Participation in Presidential Elections in the United States, for Selected Years Year 1860 1880 1900 1920 1940 1952 1956 1960

Population (1,000,000's) 27.6 50.3 76.1 106.5 132.0 157.0 168.9 180.7

Per cent of population eligible to vote 17 23 25 51 61 62 61 60

Per cent of eligible population that voted 84 78 74 49 62 64 60 64

Source: Murray Gendell and Hans L.Zetterberg, A Sociological Almanac for the United States (New York, Charles Scribner's Sons, 1964), p. 54. The table is based on figures originally assembled by Robert Lane, Political Life (1959) but supplemented for the later years. With the exception of 1920, the first year adult women had the right to vote throughout the U.S., the table gives a graphic picture of the transition from a politics of notables to the mass-politics under a universal franchise. Voting participa-

76

Reinhard Bendix

tion is high as long as voting eligibility exists only for the few, but declines as the franchise is extended. It would be interesting to assemble comparable data for other countries and inquire into the reasons for the differences which this would reveal - a task that would supplement Almond's and Verba's suggestive study of Civic Culture by the addition of an historical dimension. But in what way would it enable us to speak of 'steps and sequences' in the process of nation-building? In this respect comparative studies are obliged to develop a typology of nationbuilding processes before proceeding further, or so it seems to me. This approach is implied in the work of Max Weber and Otto Hintze, whose comparative studies aim at delineating and distinguishing features of the Western-European development. In an effort to account for the initial development of capitalism, Weber is concerned with the larger complex of Occidental rationalism, whereas Hintze restricts himself to the relation between social structure and political institutions and seeks to account for the emergence of modern administration and representation. Comparison for both writers means in the first the use of contrasts with other civilization in order to define more precisely what they wish to explain. Both writers rejected the evolutionism of the nineteenth century not only because they criticized its biological analogies, but because they were interested in developments that were true of more than one but less than all societies.10 I want to suggest in what ways I think this perspective to be especially useful for the comparative study of recent historical changes, those broadly suggested by the twin terms of industrialization and democratization. In the introduction t o Capital, Marx points out that he had chosen England as his model, because it exemplified the 'laws of capitalist development', which would govern by and large the future development of other capitalist countries. Thus, he felt that he could say to his German readers: de te fabula narratur. This position is, of course, based on the assumption of necessities emanating from the economic structure of societies, which - in the long run - determine political change including international relations. We can now say, I believe, that the facts do not bear this out. Once industrialization had been initiated, no country would go through the same process in similar fashion. Not only were English mechanics used in the early industrialization of Germany, for example: English institutions were used by German intellectuals as points of reference for the development of German institutions. The point is a general one: industrialization itself has intensified the international communication of techniques and ideas, which are taken out of their original context and adopted or adapted to satisfy desires and achieve ends in one's own country. What is here said with reference to the international repercussions of English industrialization, applies mutatus mutandis to the international repercussions of the ideas of the French Revolution. English industrialization and the French Revolution altered the terms of reference by which 'issues are defined, relationships maintained, and contentions resolved'.

Concepts in comparative historical analysis

77

Looking backwards from the vantage-point of the eighteenth century, one is justified in emphasizing the continuity of changes in Western Europe which culminated in these events, and which Weber and Hintze analyzed as distinguishing characteristics of occidental civilization. That industrialization and democratization emerged from a long and distinctly European development may help to account for the strong tendency of social scientists to consider change a phenomenon that is internal to the societies changing. It is equally legitimate to consider industrialization and democratization as having been 'initiated' at a particular time and place and as constituting a 'breakthrough' to a new historical era. If one considers the great transformations which followed, one is inclined to highlight the contrast between pre-revolutionary traditions and post-revolutionary modernity. This contrast has been a dominant theme of social theory from the eighteenth century to the present, and it underlies many of the generalizations that have been derived from the Western experience. However, these same perspectives may be used in a different manner. Emphasis may be placed on the persistent distinctiveness of the Western experience which is as notable in its feudal traditions as it is in its modern industrialism and democracy. Although long in the making in Western Europe as a whole, the twin revolutions of the eighteenth century came to a head in England and France, and since then this impetus to change has had repercussions in other Western European countries and in European settlements overseas. These repercussions may be considered an extension of the internal continuities of the prerevolutionary development analyzed by Weber and Hintze. Let me try to characterize these repercussions by restructuring what we know about certain pervasive differences between intellectual and working-class alienation and agitation during the nineteenth century. The alienation of intellectuals is a by-product of industrialization itself, for industrialization creates a mass public and a market for intellectual products and thus accentuates the elitism of some, the populism of others, and the ambivalence of all intellectuals, especially through their awareness of the discrepancies between high culture and popular culture. This general alienation was overshadowed as well as greatly intensified in the countries which witnessed from afar the rapid economic advance of England and the stirring events of the French Revolution, so that their own economic backwardness and autocratic institutions appeared still more backward and autocratic by comparison. Under these conditions a polarization of cultural life has typically occurred between those who would see their own country progress by imitating the 'more advanced countries', and those who denounce that advance as alien and evil and emphasize instead the wellsprings of strength that exist among the people and within the native culture. This reaction was typified by the difference between Westernizers and Slavophils in Tsarist Russia; recently its convolutions have been analyzed with great subtlety by my colleague, Joseph Levenson, in his book Confucian China and its Modern Fate. The general pattern has occurred again

78

Reinhard Bendix

and again; it has been a mainspring not only of intellectual alienation but also of movements for national independence where these have occurred. Working-class alienation and agitation during the nineteenth century involved very different processes, as I have argued in Nation-Building and Citizenship. Intellectuals could experience industrialization as an emancipation from their previous subservience to the Church and to private patrons. But workers experienced it initially as economic destitution exacerbated by legal and political changes which - under the slogan of individual freedom - made a mockery of their position as members of the community. For workers alienation meant simply second-class citizenship reinforced by police measures and the ideological indignities heaped upon them through sermons and public debates. As a result, radical agitation among workers represented a protest against this type of discrimination, a point largely obscured by Marx whose theory of alienation attributes to workers types of dissatisfaction more often found among intellectuals. There is a strong civic component in much working-class agitation that is missing from the radical agitation of intellectuals, since their citizenship was never in question. And conversely: the alienation of intellectuals also occurs in countries like the United States in which workers have a strong sense of citizenship and reject radical appeals. However, the two kinds of movement have joined in various blends of socialism and nationalism, where the workers' protest against second-class citizenship and the intellectuals' ambivalence about the comparative backwardness of their country and their own role in it, are not sufficiently assuaged. In Europe these nineteenth-century movements occurred in the context of an emerging nation-state. They were preceded by developments which furthered an absolutist concentration of power, on the one hand, and a more or less individualized citizenry, on the other. At the risk of putting a very complex matter too simply, I would say that the desire of the workers for full citizenship and the search of the intellectuals for a power capable of removing the backwardness of their country, had a common precondition in the prior decline of kinship ties, religious belief, linguistic affiliation, territorial and racial communalism. None of these ties or associations disappeared, but some of them had been weakened by the ascendance of Christianity, others by the Renaissance and the Reformation, and others still in the course of struggles between enlightened absolutism and the estates. It will be recalled that Max Weber's lifework documents the proposition that Christian doctrine and the revival of Roman law militated against familial and communal ties as foci of loyalty that could compete effectively the universal claims of legal procedure and the Christian faith. By these prior developments men were freed very gradually for such alternative solidarities as those of social class and national citizenship, though the relative decline of 'traditional' and the relative ascendance of 'modern' solidarities remain an issue to this day and hence a major subject of comparative study in Europe as well.

Concepts in comparative

historical analysis

79

Here is the place, it seems to me, to refer to that loss of community which has been a recurrent theme of social theory ever since the eighteenth century. The reference can be made without nostalgia or false romanticism, if it is accompanied by an appreciation of these alternative solidarities, and hence by a proper balancing of the assets and liabilities of such a development. The 'new nations' of today are in a fundamentally different position from the 'new nations' of Western Europe during the nineteenth century. Indeed, the applicability of the term 'nation' itself is in question. The long, European pre-history in the course of which familial and communal ties gradually weakened, is notably absent from the societies which have gained their independence after World War II.11 In these societies industrialization has become an almost universal ambition, and the populism of the franchise (if not democracy) an equally universal reality in good part as a reaction against Western colonialism but still also as a result of influences emanating from Europe. One effect of the right to vote has been that just those familial and communal ties are mobilized politically, which militate against the emergence of civic loyalties and hence at least against one of the preconditions of the nation-state that is familiar to us from the Western experience. The consequence of this new historical pattern may be as great eventually as the consequences of the twin revolutions in eighteenth century Europe. At any rate, there is no precedent in our experience for the emergence of 'nations' in the context of three competing world systems which can quickly transform every tension of a social structure into an issue of international relations under the threat of nuclear war.12

NOTES

1. Note the two Bulletins of the Social Science Research Council, dealing with the relations between history and the social sciences. See also Louis Gottschalk, ed., Generalizations in the Writing of History (Chicago, University of Chicago Press, 1963). The statement in the text is especially well illustrated, however, by the many contributions of historians published in the pages of the collection Comparative Studies in Society and History, edited by Sylvia Thrupp (The Hague, Mouton). 2. Though more directly concerned with the study of historical change, the intention here is similar to that of Robert K. Merton in his discussion of theories of the middle range. See his Social Theory and Social Structure, 2nd ed. (Glencoe, 111., The Free Press, 1957), pp. 9-10 and passim. For certain purposes higher levels of abstraction may well be useful, and logically one cannot speak of intermediate levels of abstraction without acknowledging the existence and possible utility of higher levels as well. However, considerable differences of judgment and emphasis remain with regard to the direct, analytic utility of such higher levels, as the following discussion indicates. 3. Cf. the recent contributions of Talcott Parsons, Robert Bellah, and S.N.Eisenstadt in the June, 1964 issue of the American Sociological Review. Regrettably, these writers

80

4.

5.

6. 7.

8.

9.

10.

Reinhard

Bendix

take no note of the important contribution by Ian Watt and Jack Goody, 'The Consequences of Literacy', Comparative Studies in Society and History, V (1963), pp. 304345. Note also the closely related analysis by Bruno Snell, Scenes from Greek Drama (Berkeley, University of California Press, 1964), passim. For scholars interested in evolution these studies have the great advantage of focussing attention on a more or less documented record of transition from a preliterate to a literate society. In a recent contribution, Daniel Lemer suggests that since rising output per head depends especially upon a people's willingness to change, politicians are well advised to promise economic benefits only after people have changed their ways in the requisite direction. He is silent, however, on how politicians can be induced to act in this manner, or how people are likely to change in the absence of promises, or why in the movements for independence the value of independence has priority over the value of economic growth. See below: Daniel Lerner, 'Comparative Analysis of Processes of Modernization', pp. 82-92. These and related issues are discussed in Arthur Schweitzer, 'Vom Idealtypus zum Prototyp', Zeitschrift für die gesamte Staatswissenschaft, Heft 120 (Jan. 1964), pp. 13-55. The article will be useful even to those who do not follow all of the author's stimulating suggestions. I have used this and other suggestions of the literature in my elaboration of this point in Nation-Building and Citizenship. I use the example of bureaucracy, since in Nation-Building and Citizenship I have formulated the implications of the general points made in the text on pp. 107-115. Similar points are suggested elsewhere in the volume with regard to the contrast between patrimonialism and feudalism, the plebiscitarían and the representative principle in a democracy, the double hierarchy of government in totalitarian regimes, the relation between central and local authority in Indian history. None of these other concepts is as clearly worked out as the concept of bureaucracy. See Otto Brunner, Neue Wege der Sozialgeschichte (Göttingen, Vandenhoeck & Ruprecht, 1956) and E.P.Thompson, The Making of the English Working Class (London, V. Gollancz, 1963). In his major work Land und Herrschaft (Baden bei Wien, R. M. Rohrer, 1939), and Adeliges Landleben und Europäischer Geist (Salzburg, O. Müller, 1949), Brunner reanalyzes feudalism in terms of the legal, economic, and ethical categories employed by those directly involved in feudal relationships, but the volume of essays puts this perspective in the larger context of European social history. Thompson for his part wishes to restore the meaning of the term 'class' and accordingly he rejects abstract definitions. Class he says is a historical phenomenon 'which happens when some men, as a result of common experiences, feel and articulate the indentity of their interests as between themselves and as against other men whose interests are different from theirs'. (E. P. Thompson, op cit., p. 9). Note incidentally that the same point is made despite the rather marked difference in political orientation of the two authors. It is a short step from this thought to a metaphoric language which attributes actions of various kinds to society, the famous fallacy of misplaced concreteness against which Whitehead warned. To me at least it has always seemed odd that a theory which began by placing human action at the center of its attention ends up by referring to the actions of systems, though that consequence is probably related to the way action was defined in the first place. Max Weber's critiques of evolutionism are contained in his Wissenschaftslehre, Hintze's in several critical essays contained in Soziologie und Geschichte.

Concepts in comparative historical analysis

81

11. That, it seems to me, is the questionable part of my colleague S. M. Lipset's analysis in The First New Nation (New York, Basic Books, 1963), since the achievement of political independence at the end of the eighteenth century is comparable with a similar achievement in the middle of the twentieth century only on the assumption that all achievements of independence by former colonies are comparable - irrespective of time and place. I do not consider the utility of that assumption very great, but whatever it may be is diminished, it seems to me, by the neglect of the obvious differences between independence movements then and now. 12. With the permission of the editor this paper was published also as a contribution to Tom Burns and S. B. Saul, eds., Social Theory and Economic Change (London, Tavistock Publications, 1967).

CHAPTER

IV

Comparative analysis of processes of modernization DANIEL LERNER Massachusetts Institute of Technology

Modernization is, in my lexicon, the social process of which development is the economic component. If economic development produces 'rising output per head', then modernization produces the societal environment in which rising productivity is effectively incorporated. As an acceptable first approximation of a definition, we would consider as modernized a society that is capable of 'self-sustained growth' over the long run. This definition, while oriented toward the economic process, is not so restrictive as it may seem at first glance. To begin with, economic development is in fact a high-priority objective of every modernizing society - the prime mover, when it is not indeed the only motivation, for modernization. Moreover, and this is the crux of the matter, the attainment of 'self-sustained growth' involves far more than the economic processes of production and consumption. It involves the institutional disposition of the full resources of a society, in particular its human resources. For an economy to sustain its growth by its own autonomous operation, it must be effectively geared into the main components of the skill infrastructure and the value suprastructure of its societal framework - i.e., the skills and values of the people who make it work. On this view, a society capable of operating an economy of 'self-sustaining growth' is ipso facto a modernized society. We orient our definition in this sense in order to focus attention upon the proposition that is central to the analysis presented in this paper: namely, that there is a single process of modernization which operates in all developing societies - regardless of their colour, creed, or climate and regardless of their history, geography, or culture. This is the process of economic development, and since development cannot be sustained without modernization, we consider it appropriate to stress this common mechanism underlying the various faces of modernization.

Comparative analysis of processes of modernization

83

Having stated this proposition so baldly, let me promptly add the necessary disclaimers. The proposition does not assert that colour, creed, climate are of no consequence in modernization; nor does it assert that history, geography, culture are irrelevant to development. These characteristics of a society clearly modify the development process, by varying its mode and adjusting its tempo. But they cannot change the basic mechanism underlying the particularities of mode and tempo. If a society is to develop, it must achieve 'rising output per head'. And if a society is to modernize, it must bring development to the level of 'self-sustaining growth'. A society that did not give high priority to these objectives would have little motivation to undergo the trials and tribulations, the pangs and pains that accompany modernization everywhere. For these objectives impose their price namely, that people make themselves behave in ways required to operate the only mechanism whereby these objectives can be attained. Note that nothing normative in favor of modernization has been said. In fact, I do not believe that modernization, as here conceived, is either desirable or indeed feasible in every society of the contemporary world - a judgment based on the estimate that, a generation or two hence, some types of society will be able to modernize more effectively at lower human cost. What we have said is that most traditional societies - and all that are new states - appear to have in fact opted for modernization (without my advice or consent). Havingdone so, they have committed themselves to a process they may not comprehend - a process that imposes upon them demands they may neither understand nor accept - but a process whose demands cannot be ignored, unless the objectives of modernization are abandoned. Consider one set of demands imposed by the objective of 'rising output per head'. Economists are agreed that to achieve this objective a new division of labor must be created that will shift working people from the primary agricultural sector (which will continue to feed the society as well or better) into the secondary and tertiary sectors. This sounds simple enough in doctrine, but what a transformation of human lifeways - skills and values - is required to bring it off in practice! Most developed societies feed their entire population (at the rate of 3,000 or more calories per head per day) with 20 % or less of their labor force employed in agriculture. Most underdeveloped societies employ 80 % or more of their labor force in agriculture and still do not supply their entire population with the minimum caloric requirements stipulated by FAO and WHO - often, indeed, supplying only half or less of the quantities produced by the advanced nations. To achieve modernization, the developing societies would have to double or triple their output with onethird or one-fourth of their present agricultural workers. This requires 'rising output per head' in orders of magnitude ranging from 600% to 1200% - simply to bring their agricultural sector into an acceptable relationship with the rest of the economy. To this add the extra order of magnitude imposed by the 'population explosion' in just these developing countries.

84

Daniel Lerner

What is to be done with the 60 % or more of the labor force displaced from agriculture? They are to be re-employed in the industrial and service sectors. Again, if one says this fast, it is possible to overlook the enormous transformation of human lives that this process requires. Transport must be built to move them, homes to house them, schools to train them, factories and offices to employ them, and the array of urban facilities to service them. These are the processes of urbanizationindustrialization that, in the western world, evolved over the course of several centuries. Even so, they entailed suffering and misery for many millions of people as revealed in the 19th century studies of the urban poor by Frédéric Le Play in Europe, by Charles Booth in England, by Jane Hull in America. But the modernizing lands today are societies-in-a-hurry. Emulating what the advanced western societies have become today, they want to get there faster. Accordingly, they force the tempo of western development. Even more serious, as a result of their hurried pace, they often disorder the sequence of western development. The most conspicuous symptom of the contemporary disorder is what happened to urbanization in the developing areas. Every student of development is aware of the global spread of urban slums - from the ranchos of Caracas and favellas of Rio, to the gecekôndu of Ankara, to the bidonvilles and 'tin can cities' that infest the metropolitan centers of every developing country from Cairo to Manila. The point we wish to stress, in referring to this suffering mass of humanity displaced from the rural areas to the filthy peripheries of the great cities, is that few of them experience the 'transition' from agricultural to urban-industrial labor called for by the mechanism of development and the model of modernization. They are neither housed, nor trained, nor employed, nor serviced. They languish on the urban periphery without entering into any productive relationship with its industrial operations. These are the 'displaced persons' of the development process as it now typically occurs in most of the world - a human flotsam and jetsam that has been displaced from traditional agricultural life without being incorporated into modern industrial life. The hapless condition of the D.P. enters into the lives of their children as well, thus prolonging displacement into the next generation. Typically, their children do not go to school, do not find work (other than the most menial and unrewarding), do not become urbane in any significant sense (other than the urbanity of big-city delinquency and crime). If they do go to school, rarely is attendance sustained enough for them to acquire truly functional literacy. What they learn they forget; what they remember is insufficient to serve any useful purpose. Thus, the D.P. status, once acquired, is prolonged and even perpetuated. The postwar generation of urban D.P.'s, around most of the world, provides the nucleus of 'urban poor' that will multiply well into the next century. Their existence, dreadful in personal terms, is at the same time one of the major

Comparative analysis of processes of modernization

85

drags on the development process wherever they are. They are not merely unemployed and impoverished. They are counter-productive in the sense that considerable resources must be allocated to maintain them in even this miserable condition of life, since the social conscience of a modernizing society will not allow - and its self-protection cannot afford - the sorts of death bred by squalor. While I was in Caracas, during January 1962, an 11-month old infant was eaten alive by a rat in one of the ranchos. If this sort of thing were to 'go too far' in the ranchos, what guarantee would there be for the safety of babies in the better neighbourhoods down the road a piece? Indeed, what protection is there for decent folk against those D.P.'s who survive infancy in the ranchos, and their functional equivalent around the world, in order to grow into juvenile delinquents, hoodlums, criminals? Every such new urban agglomeration makes substantial demands upon the resources of its environment more police and firemen, more hospitals and schools, more housing and related facilities. Much of this outlay is wasted, but there is no way in which it can be saved. Not so long as the environment cannot provide the one productive outlet from the D.P. condition - namely a job. The foregoing is intended to dramatize the need for comparative analysis of the processes of modernization by focussing on one such process that has gone seriously awry in most developing countries. This is the flooding of great urban centers by persons who have no work there. What this augurs is the contemporary decoupling of the twin processes of urbanization and industrialization whereby most of the developed countries of the world attained their present condition. In this sense, the rat in Caracas is not a cautionary tale but a piece of evidence for functional analysis. It is a principal thesis of this paper that the modernization of most of the world is going badly - is costing too much for too little benefit - owing partly to the lack of comparative analysis on which rational control of the modernization processes could be based. In support of this thesis we propose to identify some of the major processes and their interaction - then to suggest some priorities for the comparative analysis that is yet to be done. Major processes and their interaction We have already identified some of the major processes that animate modernization: urbanization, industrialization, education, sanitation, transportation. The first two are 'basic' variables in the sense that they have to be generated first, before the rest can come into productive operation. The last two are 'sequential' variables, in the sense that they invariably come into operation once the first two are operative. Whether the last two operate more or less efficiently in cost: benefit terms is an important, but technical, question subordinate to the great policy decisions activating the first two. Historically, where people displaced from agriculture went to the

86

Daniel Lerner

towns and found work, there industrialization-urbanization (they must be hyphenated) entrained in due course the sequence of urban services that are symbolized by sanitation and transportation. I say 'symbolized' because there are literally hundreds of technical and administrative sequences that must be worked out in order for sanitation and transportation to operate effectively in the modernizing environment. Education is rather different from the others. It is an 'intervening' variable. It cannot be fully independent of the urban-industrial processes, for it depends upon the human motivations and material facilities supplied by these. Yet it cannot be left to come along as part of the 'sequential' aftermath. For while the urbanindustrial processes may be initiated independently of education, they cannot long be sustained without equivalent growth of education as the main supplier of those skilled human resources upon which urban-industrial growth utterly depends. This point is illustrated in a variety of ways throughout the 8-volume U.N. report on Science and Technology for Development. Perhaps the most simple and vivid statement of the case is that by Professor R.V. Garcia of Argentina: 'The very basis of the problems of development, the core of everything related to it, is Man. Technology cannot canalize his potential with the same speed as it can canalize, for example, a river. Today we know how to deal with a desert, to turn it into an orchard in relatively few years; but far more time is necessary to train men who are capable of growing oranges in a desert'. To grow oranges in a desert requires men to do something other than what comes naturally. A man used to orange-growing will be strange to deserts; one used to deserts will be strange to the culture of oranges. Either way, something new must be learned in order to bring into efficient working relationship sets of lifeways the desert and the orchard - never before integrated in the experience of a single person. The skills required for the planting, tending and harvesting of citrus crops are considerable, but far more than technical skills must be learned. Living in orchard oases means blending two disparate 'cultures' with different, often divergent, values. At a rudimentary level it requires a recasting of one's perception and evaluation of the natural elements - sun and wind, sand and water. At a somewhat higher level, it reshapes one's use and valuation of animals and people. In the orchard oasis, one is less likely to give absolute priority to camels, and rather more likely to find uses for women and children, than elsewhere in the desert. Yet, the men and families living in an orchard oasis cannot wholly abandon the lore and law of the desert either, for it remains their proximate environment. We take the case of the oasis orchard as only a more vivid instance of the process that is activated whenever people move from a less to a more developed environment. Something new, which transforms old values while reshaping old skills, must be learned. This 'something new' is nothing less than how to live productively

Comparative analysis of processes of modernization

87

in a new and strange and ever-changing environment. When people move from the familiar routines of village life (where each person's role and status is known to all) to the unrelated sets of 'contacts' imposed by urban living, their style of life abruptly undergoes a deep transformation. To cope with the consequences of their physical mobility, they must acquire psychic mobility. For mobility is the primary process whereby modernization is activated - the process of which urbanization-industrialization is the major mode. But the world's educational facilities are vastly inadequate to provide the quantity of psychic mobility required by the amount of physical mobility that is now occurring over most of the world. This is why we pointed first to the millions of D.P.'s who now populate the periphery of the world's great urban centers. These are the most numerous class of people victimized by mobility without education. But people of the same class can be found as well in smaller towns and even in orchard oases. These millions of people form the hard core of the 'revolution of rising frustrations' that confronts the modernizing world over the next two generations,i.e., the rest of our century. This is the heavy price the developing countries must pay for their failure to understand and adapt the urban-industrial sequences revealed by the Western Model of modernization. The West spentmostof the 19th century dealing inadequately with its unprecedented problems of the urban poor. From the French revolution of 1789 to the British General Strike of 1926, the most 'advanced' countries of Europe failed to find satisfactory solutions for urban-industrial poverty. These failures traced their effects in the revolutionary civil wars throughout Europe and the international wars that spread beyond the European continent. The agonizing civil wars imposed communism on Russia in 1918, fascism on Italy in 1924, nazism on Germany in 1933, falangism on Spain in 1937. Its wider effects appeared in the 'popular front' that invaded all of the Western world during those critical 1930's. These conflicting movements of protest ignited the global war of the 1940's and shaped the global Cold War of the 1950's. During these decades the failures of the Western Model were put on display for all the world to see. The persistence of urban poverty was the main failure. Its projection into civil and global wars certified the incapacity of modern society to maintain external peace without resolving its internal conflict. It was Hitler who divided Europe into 'have' and 'have-not' nations and declared that he would move Germany into the 'have' category by force. His effort laid waste to Germany and much of Europe. Paradoxically, if Germany thirty years later has become a 'have' nation, it was by losing Hitler's war, abandoning Hitler's aim of European hegemony, and entering into cooperative economic development with its European partners. We are spanning great historical sequences in apparently cavalier fashion. To articulate the propositions underlying our summary sketch of Western history over the past 200 years would go far beyond the scope of this paper. Accordingly, we limit ourselves to pointing out that these two centuries of Western history project

88

Daniel Lerner

a cautionary tale of major import for the modernizing societies of the rest of the world. Those new nations that permit their attention and resources to be focussed on external 'confrontations' - as between Egypt-Israel, India-Pakistan, ChinaIndia, Indonesia-Malaysia - are likely to reenact the worst follies of Western history. To follow Hitler's schema of dealing by military force with an international map of haves and have-nots is to follow a devious, and deviant, course that cannot bring other than grievous effects in the form of civil and international wars. This sort of strategy is the more senseless in that none of the parties to these 'confrontations' between developing countries is very 'have'. The idea of enriching oneself by the plunder and booty of a defeated adversary is, in this context, a capital error. Getting a bigger slice of the other fellow's small pie will impoverish him without enriching you. Beggar thy neighbour, in the neighbourhoods of poverty, is a game without payoffs to the 'winner'. The present level of aggressivity among poor developing countries enjoins us, if we are to deal realistically with their severe problems of internal growth, to take cautionary note of their tendency toward fruitless external adventure. For the great strategy of development is not to snip a piece from a neighbour's little pie, but to learn how to bake bigger pies for oneself. For this, one needs the contructive guidance of political economy - what has been called the 'policy sciences of democratic development' - rather than the destructive operation of military force. Since the policy sciences of democratic development require nothing less than a comprehensive understanding of the development process in operational terms, it is to this requirement that we turn in the concluding section of this paper.

On the agenda of comparative analysis The policy-science approach to comprehensive understanding of societal processes begins with a formulation of goals: the value-objectives that we want developing societies to achieve. Next, it describes the trends of recent and current events, in terms of movement toward (or away from) the postulated goals. It then undertakes to analyze the conditions under which these trends are occurring - this, given the instability of most operational indexes and the inadequacy of most statistical time-series, being the most difficult, but scientifically the most essential step in policy-science projections from a comprehensive map of the past and present to a comprehensive forecast of the probable future. Our definition of goals for modernizing societies is reasonably clear from our earlier references to the 'policy sciences of democratic development'. The notorious ambiguity of all such key symbols as 'democracy' need not obscure our meaning in this context, since the institutional procedures that best serve democratic goals may vary in different societies at different times. Thus, France was not cast out of the democratic world when irresponsible parliamentary supremacy under the

Comparative analysis of processes of modernization

89

Fourth Republic brought it to Timmobilisme'; nor will France be cast out because the Fifth Republic is governed by an autocratic president with a strong 'cult of personality' tendency. The French municipal elections in progress at this moment (14 March 1965) are an acceptable indicator that France remains democratic, however its institutional procedures may change from one decade to another. We could not say this much for the Indonesia of Soekarno, the Ghana of Nkrumah, or even the Egypt of Nasser (who is trying harder than most 'charismatic' leaders to involve his people). For, in those countries, the autocratic cult of personality remains unchecked by public opinion and uncontrolled by popular vote. Without some institutional procedure to activate these popular controls, there is no democracy and no democratic development. For the essential elements of a democracy are the mobility and participation of its individual citizens. Mobility is essential because it liberates the individual that was bound to his inherited place in traditional society. Liberated from his native soil, he gains physical mobility by changing his position in space. Liberated from his native status, he gains social mobility by changing his position in society. Liberated from his native self, he gains psychic mobility by changing his personality to suit his new place and status in the world. This powerful transformation liberates the individual from the constraints that bound him to his place and kin. But liberation from place and kin also implies isolation from the communitas whereby individuals were related to each other by traditional custom. This passage from community to society has been sufficiently articulated in the sociological literature of the past century that we need only allude to it here. We stress only that as a man transforms himself - from being defined as his father's son to being defined as the citizen of a nation - he acquires a need for participation in public life that was not felt in the traditional society. As a citizen, each man must make and remake his own relations with the larger social environment in which he lives. This is why the participant citizen rapidly becomes the cash customer, the radio listener, and the voter. For a citizen of a modern society can function only by participating actively in its market, its forum, and its voxpopuli. These distinguishing marks of citizenhip in ancient Athens and Rome are still the principal traits of the urban man today. In the modern urban-industrial context, however, these traits take on significant new forms. The modern forms are associated largely with the rapid tempo af modernization in the contemporary world. In previous centuries, two full generations might elapse before the transition from rural to urban living would slowly enter into the lifeways of a village or clan. Today, around much of the world, this transition is occurring within the lifetime of millions of individuals. Clearly, swifter means are needed to teach and train modernizing individuals in the lifeways of urbanity. The chosen instruments for this process are the mass media, which have come to function, only in the past two decades, as a 'mobility multiplier'.

90

Daniel Lerner

The multiplier power of the mass media derives from their capacity to reach a vast number of individuals rapidly and repeatedly. The messages they transmit operate directly, continuously and simultaneously upon people of diverse place and status, in isolated rural hamlets as in crowded city slums. They are thus capable of diffusing an 'apperceptive mass' that transforms individual personalities while weaving threads of interaction among them. Because the mass media have this power of affecting millions of individuals severally and collectively, they are uniquely empowered to carry forward the two processes of 'interest articulation' and 'interest aggregation' which Gabriel Almond has identified as principal components of political development - or, in the more comprehensive term of this conference, of modernization. It is our conviction, based on a decade of data-collection and evaluation, that the wartime and postwar spread of the mass media around most of the world has been the most important single factor in producing the global 'revolution of rising expectations' that animated development activity in the 1950's. It is our further conviction that the activity of the 1950's produced a severe setback to rising expectations in many developing areas, owing to a faulty communication strategy for the mass media - or the absence of an informed communication strategy - on the part of the charismatic leaders of the emerging nations and their technical advisers from the advanced societies. Largely as a result of the strategic failure to use the mass media effectively for development purposes, much of the underdeveloped world in the 1960's faces a 'revolution of rising frustrations'. The source of frustration is the large - and growing - disparity between the bright new world 'promised' by the mass media (at least by implication) and the miserable world in which the audience for these promises actually live. The mass media, for two decades or more, have led people to believe that with the achievement of independent nationhood would come the beginnings of paradise on earth. In the event, however, independence brought only the beginning of new and harder problems than the peoples of the emerging nations had ever faced while they were in a condition of dependence. As they rapidly learned, independence did not automatically bring prosperity. It brought only the opportunity to enter the long hard road of raising 'output per head' whereby prosperity may, in some distant future, be attained. Small wonder that people who thought income would rise along with the raising of their new national flag have found themselves frustrated by the facts of independent political life. The fault lies with the communication strategy of the new leaders because they, eager to mobilize popular support and maintain public morale, either promised good things they could not deliver - or, if they were more prudent (not necessarily more wise), they permitted people to believe that these good things had been promised to them. In short, they allowed their peoples to learn to want more than they could hope to get. This created the unbalanced want: get ratio which is the psy-

Comparative analysis of processes of modernization

91

chosomatic malady affecting most of the underdeveloped world today. As I have described this malady elsewhere at considerable length, I need say here only that I consider this an essential starting point for comparative analysis of the process of modernization in the years ahead. Some recent studies provide excellent guidance for the types of research that are needed. The work of the Almond-Pye Social Science Reseach Council Committee on Comparative Politics is outstanding and well-known to all participants in this conference. I would stress only that the work sponsored by this committee has not been confined to any doctrinaire conceptions of methodology. The initial AlmondColeman volume made its great contribution by dealing comparatively with all the areas of the world according to an objective list of operational indexes. The WardRustow volume focussed a similar approach on the comparison of two particular countries, Japan and Turkey, that are entirely different in geography, history, religion, culture, but have in common that they are considerably more advanced in the modernization process than their neighbours. The Pye volume dealt with a variety of developing countries, but with a focus on the comparative analysis of their handling of the communication process. Each of these, as well as the approach through survey research in the Almond-Verba Civic Culture, provides a useful model for future research. To close on the theme announced at the beginning of this paper, I would recommend that future research start from the recognition that, within the comprehensive process of modernization, there is a relatively autonomous component usually called 'economic development'. Economic development is autonomous in the sense that its basic rules are applicable to all developing peoples regardless of colour, creed, or culture. The rule of 'rising output per head' governs economic development regardless of the socio-political environment or the psycho-cultural context. The latter components may, in any given society, facilitate or hinder the achievement of'rising output per head'. But the rule remains that without 'rising output per head' there is no economic development. The larger form of this proposition is that a truly modern society is capable of 'self-sustaining growth'. This means that an input-output level has been achieved which guarantees that overall economic growth will continue without requiring inputs external to the system. Future researchers will do well to start from the conception that only a modern society is capable of self-sustaining growth, and ipso facto, that any society capable of self-sustaining growth is modern. The reason, as I have indicated, is my data-based conviction that no society can achieve selfsustaining growth in its economy without having attained a reasonably full measure of modernity in its polity and in the matrix of its social relations. In conclusion, I would point out that this is no brief for turning the study of modernization over to the economists. Quite the contrary. Economic development is too important to be left to the economists. What economists have to teach us

92

Daniel Lerner

about these matters must now be incorporated into the research designs of those who study societal development from a more comprehensive perspective. An example of some relevance to the concerns of this volume is the work produced by the M.I.T. Center for International Studies. There, with the close collaboration and indeed with the initiative of outstanding economists, fruitful efforts have been made to sketch a model of the development process precise enough to include economic behaviour and comprehensive enough to encompass the intricate web of human relations.

CHAPTER V

Political systems and political development LUCIAN W. PYE Massachusetts Institute of Technology

It is with great pleasure that I respond to the invitation to report on the work of the Committee on Comparative Politics of the Social Science Research Council in the field of Political Development. In doing so, I shall be giving some of the intellectual background of the series of Studies in Political Development now in the process of publication by the Princeton University Press and which are listed in the programme of the Conference. (See Introduction p. 4.) It is not easy to present a full and accurate intellectual history of a group experience which has been as dynamic as that of the Committee on Comparative Politics. There is first, and above all, the impossible task of recording our inestimable debt to the intellectual and human leadership of Gabriel A. Almond. Whatever great merit this enterprise has had can be traced directly to his imaginative conceptualizations and sound judgments. It is also difficult to recapture the flow of intellectual currents among the vigorous and independent minds of all the committee members. It must finally be recognized that the work of the Committee took place during a period of rapid change in the general state of comparative politics and thus our work has been greatly enriched by the contributions of many people. When the Committee was first established it took as its task a need to broaden both in geography and in theory the scope of comparative politics. Our initial reaction was perhaps typical of a group of American political scientists: we felt it might be appropriate to apply to foreign political systems the types of empirical analysis which have been so fruitful in developing the study of the American political system. In short, in looking beneath the level of formal institutions of government we tended naturally to turn to the phenomenon of parties and interest groups and to examine the ways in which social forces and public attitudes become political phenomena. We were further encouraged in doing this because in our initial survey

94

Lucían W. Pye

of the newly emerging polities we were struck with the degree to which their political life revolved around specific elite groups, such as armies, bureaucracies, nationalist leaders, intellectuals, and not constitutional structures and processes. One of our early activities was thus to sponsor a program of assisting research into the political orientations and dynamics of the critical groups in various political systems. After this start it soon became apparent that group analysis could never provide the basis for systematic comparative analysis without a broader theoretical framework. The need for some common theoretical focus was further intensified by our desire to encourage a more inclusive comparative politic which might encompass political systems at all stages of development throughout the world and throughout history. We began our work at a time when developments in the technology of data handling were taking place which would make it possible to manage in a far more orderly fashion greatly increased quantities of data. These advances in technology, which have now made it possible to establish various forms of 'data banks', do make it possible to rely upon raw empiricism far more than ever before. Huge quantities of data can be collected, collated, and then freely sifted about in order to uncover various patterns of relationships without analysis being as firmly committed to as narrow schemes of organization as had been necessary in the past. It has, however, been our judgment that in spite of the promise of significant breakthroughs in the handling of quantitative data the current situation in fact called for a more serious effort than ever before at developing general theories for comparing total political systems. Our feeling was that it might be all too tempting to rely excessively upon available socio-economic data, which are customarily collected not with an eye to their relevance for comparative political analysis, and thus not to establish the essential political categories for which new data should be gathered. We were from an early stage acutely aware of the potential for using sample survey methods for measuring political attitudes, but again we felt that this relatively costly method could only be justified if it was based on an equally comprehensive and sophisticated theoretical foundation. For these and other reasons we found that when we moved beyond the study of groups we needed a framework which stressed functions and related structures and actors to such analytically defined categories. Our concern was with identifying functions common to all political systems and in terms of which we could not only define the political system but also establish significant typologies for differentiating types of systems. It was at this stage that Gabriel Almond led us to the formation of the functional theory which he described in and which served as the organizing principle of The Politics of the Developing Areas which he and Coleman edited. It is certainly not necessary for me to elaborate on the particular functional model in the Almond-Coleman volume. With respect to the continuing work of the Committee the model served as a substantial guide in planning the seven volume

Political systems and political development

95

series of Studies in Political Development. The first volume in the series represented an effort by people from a variety of disciplines to examine different dimensions of the communications function. Bureaucracy and Political Development, edited by Joseph LaPalombara, was our effort to make a more detailed analysis of some aspects of the authoritative functions and particularly the rule application function. Our concern with the functions of interest articulation and interest aggregation led us to organize the conference on political parties and development, which was also noteworthy because it explicitly sought to deal with the processes of political development in Europe and America as well as in the underdeveloped areas of the world. In seeking to investigate further the political socialization function, we felt it appropriate to limit sharply our focus on an agent of socialization which can be relatively easily controlled by public authority, and thus we organized the study on Education and Political Development which has been edited by James S. Coleman. From the beginning of the Committee's work we have had a strong interest in the cultural and attitudinal dimensions of comparative analysis, and thus when we reached the point of planning our series, we recognized that we could not limit ourselves to just a single study of political socialization. Indeed, at this stage we were becoming increasingly concerned with the role that political culture plays in affecting political development. It was a recognition of this problem which first prompted us to engage in the comparative analysis of two modernizing systems in two quite different cultural areas, Japan and Turkey, which was planned by Robert E.Ward and Dankwart Rustow. Concurrently we also planned a more general comparative study of political cultures in which area specialists on ten different countries have analyzed the relationship between political cultures and modernization in their countries of specialization. Gabriel Almond took the lead in planning this enterprise which is culminating in a study edited by Sidney Verba and myself. The Committee's interest in the study of political culture stemmed initially from our functional view of the political system and also our early concern with the orientations of elite groups. Indeed some of our conceptualizations about political culture were directly tied to an interest in the styles and modes of behavior and calculation of power holders in transitional systems and the ways in which traditional concepts of authority and the uses of power have been affecting the possibilities of building more modern systems. The Committee's interest in political culture was really given its great impetus, however, by Gabriel Almond's and Sidney Verba's impressive demonstration in The Civic Culture of the potentialities of using survey methods to identify the critical dimensions of types of political cultures. Several of the studies in Political Culture and Political Development build on both the concepts and the data of The Civic Culture. The aspect of work on political culture ties back to the tradition in political science of concern over the role of citizenship training and of the attitudes of the public in determining the type and the viability of different forms of politics and particularly of democracy.

96

Ludan W. Pye

In relating political culture to a functional model of the political system, we became increasingly sensitive to several problems in theory. First, there is the question of measuring the performance of the political system as a whole. The very manifest problems of performance of many of the new states would certainly have called attention to this general question even if our concerns in theory development had not led us in that direction and caused us to employ such terms as 'capabilities' and 'crises of development'. On this advance in theory we were again greatly indebted to Gabriel Almond and his creative formulation of the problem, a statement of which is to be found in his extremely inclusive and disciplined article, 'A Developmental Approach to Political Systems' which appeared in the January 1965 issue of World Politics. The problem of relating political culture to the functional political system also made us more aware of the problems of adequately defining the concepts of 'modernization' and 'development'. From the beginning of our work on the functional model we have relied very heavily upon the Parsonian distinctions in the 'pattern variables' to differentiate traditional and modern modes of behavior. We found it necessary to go beyond these considerations, first, with respect to evaluating the performances of particular sanctions, as for example in the Communications study; and then, second, in characterizing total political systems, as in the comparative Turkey and Japan volume. Concern with this problem has led us to the characterization of a development syndrome which encompasses the essential ingredients of a generic concept of development. Very briefly, we see this development syndrome as having three dimensions. First, with respect primarily to the spirit of the political culture we note the theme of Equality. This includes the transition in mass attitudes from people being subjects to becoming active citizens; the greater reliance upon achievement rather than ascription in the recruitment to political roles; and the belief that legitimacy should be associated with universalistic and impersonal rules and codified laws rather than particularistic customs. Political modernization thus involves mass participation. A second dimension of the developmental syndrome relates more to the authoritative function of government and can be summarized under the heading of Capacity. Capacity deals with the ability of the system to mobilize resources and to evoke psychic commitment from the population. It thus is directly related to the scope, magnitude and effectiveness of the bureaucratic and formal legitimate structures of the government in reaching equally throughout the territory of the state. Modernization thus involves secular rationality in the organization of administration and an expansion in the capabilities of the political system. The third theme in the development syndrome is that of Differentiation which relates primarily to the areas of political structures and social organization. Development then involves greater functional specificity, structural specialization and at the same time effective integration of roles and structures.

Political systems and political development

97

In thinking of development according to these three themes of Equality, Capacity and Differentiation, and at the same time seeking to relate the concepts of political culture to the capacities of a political system, it seems helpful to modify slightly the list of basic functions which constitute the political system and to designate a particular set of crises or problems which are typical of the developmental process. Thus we may designate those functions most closely associated with the political culture and which involve the establishment, maintenance, and strengthening of norms about proper authority and conduct as the Legitimizing Functions, which specifically consist of (1) Political Socialization, (2) System Legitimization, (3) Authority Allocation, and (4) Commitment Evocation. The function of Political Socialization is well recognized and requires no further comment here. The System Legitimization function involves the creation of fundamental law, constitutionmaking, the establishment of basic institutions, and the commitment to such ideals as democracy, socialism, development, etc. The Authority Allocation function involves the decisions as to what structures and processes are to be authoritative and what should be the recognized authority of particular individuals, offices, parties, cliques and other such political actors. The Commitment Evocation function relates to much of the affective dimension of politics, and involves both the loyalties and enthusiasms of the people for their system and the attitudinal prerequisites which make it possible for leaders to appeal for sacrifices that go beyond immediate material rewards. The second category of functions are involved in the relationship between the political system and the social environment in which it is set. These may be called the Process Functions for they describe the political process which consists of the selection of personnel, the advancement and clash of issues, and the dissemination of information about this process. Specifically, the process functions include: (1) Political Recruitment, (2) Interest Articulation, (3) Interest Aggregation, and (4) Political Communication. Finally, since the political system is responsive to the conscious efforts of its actors to devise programs and policies in which power and authority are applied to the maximization of values, there are the functions related to decision-making which can be called the Policy Functions, and which consist of (1) Resource Mobilization, (2) Resource Allocation, (3) Order Maintenance, and (4) External Relations Management. It should be noted that these functions are most frequently but not exclusively performed by authoritative structures, and thus are most closely associated with government in the narrower sense of the word. As we have suggested, the development syndrome of Equality, Capacity and Differentiation tends to affect the performance of the functions, producing tensions or crises. Specifically, we have singled out the following crises as being of typical importance in the developmental process: (1) Identity and (2) Legitimacy, which are both closely related to problems in the political culture and in the performance

98

Lucían W. Pye

of the legitimizing functions; (3) Participation and (4) Integration which relate mainly to the process functions; and (5) Penetration and (6) Distribution which relate mainly to the policy functions. At present one of the activities of the Committee is a joint effort to define and describe in historical detail the six crises and relate them to various sequences of national political development. I can only here give a rather preliminary description of the six crises. I trust, however, from this brief characterization of the crises I can make clear what are the central problems we are concerned with in viewing the historical process of political development. The Identity Crisis. The first and most fundamental crisis is that of achieving a common sense of identity. The people in a new state must come to recognize their national territory as being their true homeland, and they must feel as individuals that their own personal identities are in part defined by their identification with their territorially delimited country. In most of the new states traditional forms of identity ranging from tribe or caste to ethnic and linguistic groups compete with the sense of larger national identity. The identity crisis also involves the resolution of the problem of traditional heritage and modern practices, the dilemma of parochial sentiments and cosmopolitan practices we have emphasized. As long as people feel pulled between two worlds and without roots in any society they cannot have the firm sense of identity necessary for building a stable, modern nation-state. The Legitimacy Crisis. Closely related to the identity crisis is the problem of achieving agreement about the legitimate nature of authority and the proper responsibilities of government. In many new states the crisis of legitimacy is a straightforward constitutional problem: What should be the relationship between central and local authorities? What are the proper limits of the bureaucracy, or of the army, in the nation's political life? Or possibly the conflict is over how much of the colonial structure of government should be preserved in an independent state. In other new states the question of legitimacy is more diffuse, and it involves sentiments about what should be the underlying spirit of government and the primary goals of national effort. For example, in some Moslem lands there is a deep desire that the state should in some fashion reflect the spirit of Islam. In other societies the issue of legitimacy involves questions about how far the governmental authorities should directly push economic development as compared with other possible goals. Above all, in transitional societies there can be a deep crisis of authority because all attempts at ruling are challenged by different people for different reasons, and no leaders are able to gain a full command of legitimate authority. The Penetration Crisis. The critical problems of administration in the new states give rise to the penetration crisis, which involves the problems of government in reaching down into the society and effecting basic policies. As we have noted, in

Political systems and political development

99

traditional societies government had limited demands to make on the society, and in most transitional systems the governments are far more ambitious. This is particularly true if the rulers seek to accelerate the pace of economic development and social change. To carry out significant developmental policies a government must be able to reach down to the village level and touch the daily lives of people. Yet, as we have observed, a dominant characteristic of transitional societies is the gap between the world of the ruling elite and that of the masses of the people who are still oriented toward their parochial ways. The penetration problem is that of building up the effectiveness of the formal institutions by government and of establishing confidence and rapport between rulers and subjects. Initially governments often find it difficult to motivate the population or to change its values and habits in order to bring support to programs of national development. On the other hand, at times the effectiveness of the government in breaking down old patterns of control can unleash widespread demands for a greater influence on governmental policies. When this occurs the result is another crisis, that of participation. The Participation Crisis. As we noted in seeking to define political development, one dimension of the concept involves an expansion of popular participation. The participation crisis occurs when there is uncertainty over the appropriate rate of expansion and when the influx of new participants creates serious strains on the existing institutions. As new segments of the population are brought into the political process, new interests and new issues begin to arise so that the continuity of the old polity is broken and there is the need to reestablish the entire structure of political relations. In a sense the participation crisis arises out of the emergence of interest groups and the formation of a party system. The question in many new states is whether the expansion in participation is likely to be effectively organized into specific interest groups or whether the pressures will lead only to mass demands and widespread feelings of anomie. It should also be noted that the appearance of a participation crisis does not necessarily signal pressures for democratic processes. The participation crisis can be organized as in totalitarian states to provide the basis for manipulated mass organizations and demonstrational politics. Integration Crisis. This crisis covers the problems of relating popular politics to governmental performance and thus it represents the effective and compatible solution of both the penetration and the participation crises. The problem of integration therefore deals with the extent to which the entire polity is organized as a system of interacting relationships, first among the offices and agencies of government, and then among the various groups and interests seeking to make demands upon the system, and finally in the relationships between officials and articulating citizens. In many of the transitional systems there may be many different groupings of

100

Lucían W. Pye

interests, but they hardly interact with each other, and at best each seeks to make its separate demands upon the government. The government must seek to cope with all these demands simultaneously. Yet at the same time the government itself maynotbe well integrated. The result is a low level of general performance throughout the political system. The Distribution Crisis. The final crisis in the development process involves questions about how governmental powers are to be used to influence the distribution of goods, services, and values throughout the society. Who is to benefit from government, and what should the government be doing to bring greater benefits to different segments of the society? Much of the stress on economic development and the popularity of socialist slogans in the new states is a reflection of this basic crisis. In some cases governments seek to meet the problem by directly intervening in the distribution of wealth; in other cases the approach is to strengthen the opportunities and potentialities of the disadvantaged groups. For the purposes of discussion I would suggest that attention might be focused on conceiving of political modernization in terms of these six crises and the ways in which they arise in the development of different polities. In most of the new states these crises have not arisen in the order in which I have just listed them. Indeed, in some cases, for example, we find leaders seeking to use the Distribution crises, which only appeared late in the history of most European systems, as a means of resolving the basic Identity crises. We find also that often the Participation crises can destroy whatever prior resolution may have been achieved with respect to the Legitimacy problem. Space unfortunately does not permit me to elaborate any further on these crises and their relationship to the modernization syndrome, but I trust that the outline of our analysis is now apparent. I am afraid that I have already been excessively schematic and dwelt too much upon only a limited aspect of the Committee's work. This is in part a problem of space, but also because I have now brought the account down to the present and far from completed phase of our endeavors. Indeed, I must make it absolutely clear that in speaking of what is still very much an ongoing intellectual process, I can only give a tentative and personal statement; other members would certainly describe our current stage of concept building in different terms and with different stresses. This state of affairs does, however, give me an opportunity to say that the Committee on Comparative Politics has never sought to create a fixed common theory or analytical approach beyond anything which would be of value in the work of the individual members of the Committee. We have on the other hand felt that if we are going to be of any help in stimulating the field of comparative politics we would have to clarify in our own minds many questions about the theory of political systems and the dynamics of political development.

Political systems and political development

101

It is our hope that in seeking to establish some greater clarity in the problems of conceptualizing about total political systems we may be able to contribute some theoretical considerations which will make more intellectually rewarding and significant our greater technological capacity to collect data on the world's political systems. We also feel that a concern for the theoretical basis for comparative politics will facilitate a new linkage between empirical social science and the concerns of classical political theory. Likewise, we would hope that any effort to find theoretical concepts which can help to classify the entire universe of known political systems will help to narrow the gap between contemporary concerns with problems of political development in the new states and historical perspectives on man's total experiences with building polities.

C H A P T E R VI

Historians and other social scientists: the comparative analysis of nation-building in Western societies VAL R. LORWIN University of Oregon

This paper* attempts to do four simple things. First, it recalls the utility of more historical studies as the basis for continuing comparative research and theory concerning the processes of nation-building and, more specifically, the utility of work within one cultural area, that of Western Europe. Second, it discusses various kinds of 'one-nation studies' which are needed for further international comparisons. Third, it looks at a few characteristics of the work of professional historians in relation to the possibilities of more, and more effective, cooperation between historians and other social scientists. Finally, the paper describes a series of comparatively oriented interdisciplinary studies, now under way, dealing with the smaller European democracies. The term 'nation-building' is now being used to cover the historical processes of both nation-state formation and civic integration within the national life. The two processes are of course related and in some cases overlapping, notably in many new states that would be nations. For the Western nations, mostly older creations, we can usually separate the two processes. It is with national integration, the extension of civic, political, and socio-economic rights and participation, in the sense of T.H. Marshall's classic essay 'Citizenship and Social Class',1 that I shall be chiefly concerned, rather than with the nation-state's territorial formation or consolidation. * This statement is an entirely rewritten version of the paper presented to the Round Table, April 22-24, 1965. An earlier text appeared in Vol. XVII, N o 4 (1965) of the International Social Science Journal. Mycolleagues and I thank the Ford Foundationformaking possible the cooperative research described in Part 4 of this paper. We also wish to record our great debt to the Center for Advanced Study in the Behavioral Sciences at Stanford, Calif., for the hospitality and the services offered.

Historians and other social scientists

103

1. Studies of Western society In the study of national integration, we should not linger too long over the dichotomies between developed and underdeveloped societies. These dichotomies blur the vital differences among nations in the group labeled 'developed' (sometimes quaintly called 'fully developed') as well as among those labeled 'developing'. They deny us some of the most significant dimensions of comparative analysis. It is essential, as Lipset has urged, 'to move beyond the pointing out of gross variations at different levels of technical development and to specify the key sources of differences among nations at comparable levels' of development.2 Even among those advanced societies we have cheerfully assumed to be well known, systematic empirical knowledge of nation-building and theory on the subject are still underdeveloped, even if developing. Western political science, as Almond and Verba have pointed out, 'has only begun to codify the operating characteristics of the democratic polity itself'. 3 No doubt historians have already produced more materials than have yet been absorbed by those engaged in the tasks of 'codification'. Yet for Western society, we are still in the early stages of comparative research on the processes of nation-building. Western society has the richest history of national achievement and of pathological nationalism. It is also richest in trans-national associations and in attempts at supra-national political organization. The vocabulary and concepts of the discussions ofnation-building and national integration-even in the developing nations are derived from Western experience, chiefly by way of Western social science. Although we are still redefining the concepts of nation-building and national integration, research and communication on these problems in Western society suffer less from terminological ambiguity and conceptual uncertainty than do discussions which combine the experiences of developed and developing societies. There is enough common historical experience and common cultural context in Western society to make comparisons realistic and fruitful. On the other hand, there are differences enough, among the Western nations, to make comparisons exciting and significant. 2. One-nation studies International studies alone will not sustain international comparisons. To compare the nations, we need more knowledge of processes and institutions and social forces within the nations. One would be happy to learn that it is breaking through an open door to plead the case of one-nation studies. But it often looks as if, in our anxiety to compare sequences of historic change, we are trying to skip sequences of needed historical research. There are reasons for dissatisfaction with much of the existing single-country

104

Val R. Lorwin

specialization in Western societies. Ideally, single-country studies should beinformed by a lively sense both of the varieties of historical experience within the nation and of some of the relevant experience beyond present national frontiers. They should separate the workings of factors specific to the nation and those which are characteristic of many nations. Ideally, no doubt, historical researches on nationbuilding should be related to explicit hypotheses about nation-building in general, with the hypotheses in turn being reformulated in the light of research findings. So much for the ideal. The mills of comparison, however, must rely for grist on many sources, including some which will seem parochial indeed. Most of the information for international comparison is likely, for some time, to come from single-nation studies, unrelated to any trans-national hypotheses. We can hope and we can work for a comparative framework in the basic national studies; we cannot expect it. Now we must be at once grateful for materials produced out of concerns remote from our own and critical in our handling of such materials in comparative research. Hopeful of getting more materials for comparison, one may list some of the major themes on which, for so many Western nations, we still need historical studies: the formation, attitudes, and behavior of political parties, social movements, interest groups, regional associations, and patriotic societies; crises in national identity and in civic cohesion; the interaction of local and national politics; the growth of the public administration at various levels; the role of national and local police forces and popular images of them; the recruitment of elites; the adjustment of old elites to the increasing importance of the masses; the absorption of immigrants. Nobody doubts the existence of national character who knows anything about schools and armies, said Lord Acton.4 But we need to know a great deal more about the role of schools and armies—and churches and other voluntary organizations - in political socialization. We need to know a great deal more about popular literature and folk art, the nature of folk heroes, national symbols and the language of national identification (positive and negative), popular mythology and popular conceptions of fundamental law. Such studies call upon the materials and methods of intellectual history and cultural anthropology, as well as upon more traditional historical sources and styles of research. There is nothing new in this sort of an agenda, but most of the work remains to be done. The study of nation-building in Western society is not a set of historical variations on a swelling theme of widening and deepening community. So eminent a historian as Halvdan Koht could speak of 'a constant progress toward national unity through the rise of successive classes' toward a 'fuller and richer national solidarity'.5 But that was over half a century ago, and we know now, if we did not then, that the progress is not 'constant' and the solidarity not always 'fuller and richer'. Although Koht's vision has not been nation-bound, what a Norwegian might see around him would be more reassuring than what a German or a Spaniard saw.

Historians and other social scientists

105

Pathology is a well-known aid to the study of physiology, as Durkheim reminded us. We must give more study to the discontinuities and the retrogressions of civic integration, to xenophobia and nativism as well as assimilation; the loss of spontaneity and liberating purpose in social movements; the apathy in the exercise of rights which often follows the struggle for equality; the vapidity as well as the virtues of high-consensus societies. One pauses, lest his desiderata begin to resemble a sort of shopping list prolonged to the last scholar's last years - or ultimate research grant. And I would not be so unappreciative of much work I know, and more I do not know, as to imply that everything remains to be done. There are valuable studies hardly tapped by comparative analysis, in the less accessible European languages and even in the international languages. There are many old studies of great substance and insight: to give only a single instance, the University of Chicago series of 1929-1933 called 'Studies in the Making of Citizens', 6 of which I might mention only two volumes, Carlton J.H.Hayes's France: a Nation of Patriots' and Oscar Jaszi's The Dissolution of the Habsburg Monarchy,8 In a sense, however, the existence of such works underlines the needs which have not been filled. I am mindful of a distinguished economist's wry introduction to his own call for more research in certain areas: 'It is difficult to support allegations about the absence of a given type of scientific work; often the allegation illuminates only the reading habits of its author'. 9 Out of my own limitations, my comments here are restricted chiefly to American, English, French and Belgian literature. To understand the present, we must - to quote Durkheim again - from time to time stand back from it. To understand the nation, we must from time to time stand back from its center and study its components and its peripheries. If the nation is more than the sum of its parts, the parts are more than fractions of the whole. Sometimes national aggregates cancel out, rather than sum up, important local and regional, class or occupational, ethnic and cultural developments. The history of political parties and of trade unions, for example - in fact, the history of the nation itself - has too often been written and rewritten only on a national basis. Even for national history and international comparison, we must disaggregate the the national data by studies of the nation's geographic or occupational or cultural and other components. We need studies of limiting as well as more or less typical cases. Local studies, moreover, permit the testing of hypotheses of national behavior, for example, phenomena of mass behavior which, asa historian of Jacksonian democracy has said, are 'better assayed by increasing the precision of research than by widening its geographic scope'. 10 Some medieval communes are better known than their nineteenth-century successors. 'Social history is local history', Dorothy George has remarked. 11 So is a good deal of political and economic and cultural history, if the local is also seen in its larger context. The processes of political socialization, for example, often

106

Val R. Lorwin

revolve about the local political scene. So begins, and sometimes continues, the cursus honorum for many a national politician : a Joseph Chamberlain or a Theodore Roosevelt or a Camille Huysmans (to cite an internationalist who remained a local potentate while becoming a national figure). We must be grateful for the chance to utilize what local history we have, even when it offers erroneous local causations. If it is written in the broader context of comparison, within the nation or beyond it, it is of course likely to be more interesting, sounder and more meaningful. Asa Briggs has shown how one may write modern urban history, and compare community with local community, within the same and in another country, to illuminate the relations between politics and social structure at a crucial period of class relations and political ferment. 12 Regional studies furnish essential information and bases of comparisons. One may study the region, or another administrative or political or cultural or economic unit below the nation, for its own sake, as the scholars of French electoral geography and electoral sociology have been studying the political history and public opinion of département after département. Or one may compare periphery and center and their relations to each other, as Rokkan and Valen have done for Norway, 13 Erik Allardt for Finland, 14 and Juan Linz for Spain. 15 Charles Tilly has recently shown how a regional historical study may use social science techniques for systematic comparisons within the region and thereby clarify its long mythical relationship to the national regime in a crisis period. 16 One may compare different kinds of urban and industrial environments in which members of a social class have played their life roles, as Georges Duveau has done with sensitive attention and sympathy for the workers of the Second Empire in France. 17

3. Historians and other social scientists Some of the most valuable of the historical work I have been discussing is by men and women whose formal attachment is to another discipline. 'History, it is clear, has no monopoly on the study of the past'. 18 Now I should like to turn to historical work by members of the guild of historians, qualified to publish, as it were, under the appellation contrôlée. How can we help make available more of the historical data and evoke more of the basic historical studies which are still needed, even for the comparatively well studied nations of Western Europe, as we go forward to the fullness of systematic comparisons within nations and among nations? Clearly a closer and freer cooperation between historians and other social scientists is needed. Theorists of nation-building must rely chiefly on secondary authorities. 19 'Life is too short to do anything else when using the comparative method', as T.H. Marshall reminds us. 20 And, he adds with an irony authorized by a rare mastery of both disciplines, surely historians 'will not rebuke the sociologist for putting his faith in what historians write'.

Historians and other social scientists

107

There is no space here for tribute to the merits of historians, nor do they need my tribute. No doubt, as Felix Gilbert was recently moved to write, 'What historians say about the theory of history is mostly feeble and their prescriptions for what other historians ought to do are usually impossible to realize. Non-historians writing about history are often irrelevant and don't seem to know what the writing of history involves.' 21 1 must avoid the rarefied atmosphere of the theory of history and I hope to avoid the stuffy atmosphere of preachment, for I wish to write about what may be barriers to cooperation between historians and other social scientists. The term 'cooperation' I shall use sometimes literally to mean collaboration in joint endeavors, whether of research or discussion; 22 sometimes loosely to mean all sorts of mutual stimulus - reading others' works, talking about them, and profiting from them, even if at times in opposition to methods or findings.23 The value of such cooperation I shall not argue here but take for granted. Nor shall I enlist the reader for the indoor pastime still so amazingly popular in American academic circles: the debate over whether history 'belongs' among the humanities or among the social sciences. The entertainment value of this game is somewhere between that of 'twenty questions' and that of charades, yet it is often played by as few as two persons, even two on the same 'side'. 24 Of course to say that history is of both the humanities and the social sciences is to say nothing of us relationship to the other social sciences. For the moment, however, let us merely say of social science and history what Sainte-Beuve said of poetry and history: 'A little poetry takes one away from history and the reality of things; much poetry brings one back to them.' 25 If the writing of history, as Sir Lewis Namier once pointed out, calls for a 'lively sense of how things do not happen', cooperation of other social scientists with historians calls for a lively sense of how most historians do not work. Most of us do not work an international vein of material. I am speaking of modern or - in Continental terminology - contemporary historians of Europe, not of historians of earlier periods or of most other areas of the world, who simply cannot be national in focus, nor (on the whole) of Marxist historians, for whom further discussion would be needed. Finally, I am speaking not of the great masters of the craft, nor of philosophers of history, but of the mass of historians or a sort of Weberian ideal type. If I discuss limitations rather than achievements, I am painfully aware that de te fabula narratur. Most of us do not work in a comparative perspective. Historians, it is true, have been making analogies and comparisons, first across space and then across time, since Thucydides compared the politics of Athens and Sparta. They have embodied implicit comparisons in their terminology and in the moral lessons they have drawn from man's experience. But comparisons are not necessarily comparative history. As a whole, the historical fraternity has not paid much attention to the methods and opportunities of explicit and more or less systematic comparative

108

Val R. Lorwin

study. The comparative method is an integral part of linguistic studies, art and religious history, law, sociology, and political science. But history, as Bert Hoselitz gloomily observed, has been 'outstanding among the social sciences in rejecting longest the application of this method'. 26 This continues to be true despite the interesting discussions of philosophy and method, and applications of method, not only in books and old-established journals, such as the Annates and the Journal of the History of Ideas, but in such new journals as Comparative Studies in Society and History, History and Theory, and the Archives Européennes de Sociologie?7 For these discussions only very slowly and indirectly affect the writing of most historians. If Pirenne, in a famous address, 28 praised what he called the comparative method, it was to diminish national, racial, and political prejudices among historians. He did not even identify the comparative method, beyond suggesting that it was a matter of studying other countries than one's own and taking 'a point of view which would be none other than that of universal history'. The excellent recent English symposium volume on Approaches to History29 has no chapter on the comparative approach. Barraclough's chapter entitled 'Universal History' is a plea for 'ecumenical' history, for universal coverage, but it does not mention comparative methods in the world history for which it argues. Nor does the valuable collective French work on L'Histoire et ses méthodes30 go into comparative methods. It refers 14 times to Marc Bloch, but not once to his remarkable 'Pour une histoire comparée des sociétés européennes', 31 a locus classicus for discussion of the subject. To be sure, lack of general recognition has not prevented some scholars from developing comparative approaches, notably in intellectual history and economic history. In recent work in economic history, comparative methods are stimulated by historians' confrontations with model-building social scientists in the study of economic growth and by the fever of national concern with differences in rates of growth. 32 'Historical research does not permit autarchy', said Marc Bloch.33 But most historians study a problem or an institution, a personality or an event, within a single country, usually their own. The reasons for a national focus are many: to begin with, the roots of national history in national culture, and then the facts of organization of school curricula and documentary administration. The nation-state produces vast quantities of materials out of its internal political processes, its dealings with its administres, and its external relations. It also shapes the scope and direction - hence the documents - of voluntary organizations, interest groups, and social movements, as well as cultural activities of the modern era. Authority is responsible even for many of the materials of protest against authority: the police archives contain essential materials for the history of labor and radical movements. The modern historian who is a teacher is likely to be teaching the history of his

Historians and other social scientists

109

own country; if not, the history of another, a leading, nation. If he is an archivist, he is likely to be organizing or administering archives of his own nation, at the national or local level. He meets his colleagues in national organizations for the promotion of scholarship, teaching, and professional interests. Language facility and access to materials in the homeland direct the historian to the study of his own country. He may expect to find more of a public for works on his own country, whether among the general reading public or his professional colleagues or in those local history societies so congenial in their blend of disinterested scholarship and antiquarianism, local and family piety. More or less consciously the guardians of their nation's traditions, historians, in their finest hours, revivify those traditions and bring the best in them to bear upon the problems of the present. But most, said Pirenne, 'behave toward the nation like the architect toward his client'; they 'seek above all to furnish him a history in conformity with his tastes and his customs, in short, a habitable history'. 34 Pirenne was well placed to speak, for he was one of the few men to have become a 'national historian' in the sense of a national institution or symbol. Yet historians are not perforce any more ethnocentric than the many economists, political scientists and sociologists, who - despite the canons of their crafts - do not hesitate to generalize about all human behavior from their study of some of their own countrymen's behavior. 'Economic teaching and research in the United States have suffered from a myopic preoccupation with Anglo-American ideas, institutions, and problems', a leading American economist recently found. He called for a 'geophysical stretching' of the scope of economics, which is 'at least as necessary as the "historical stretching" that is rightly urged by historians and others'. 38 Obviously, economics is not the only discipline, nor are Americans the only people, to fall short in this regard. Individual labor comes more easily to historians than collective enterprise, except for such work as the editing of national series of documents or war histories. Small-scale enterprise is more natural than large; 'they reject frescoes in favor of portraits'. 36 International projects are likely to be more collective and larger in scale than the work most historians envisage. Moreover, most of us are more accustomed to looking down the corridors of time than across space. We are modest about our capacity to comprehend many cultures, and loath to compare without at least the illusion of such comprehension. Our literary canons do not induce tolerance among historians for the specialized vocabularies in which some behavioral scientists seek to give greater precision to their ideas. Reluctant to study languages used by only a few million people, historians are even less eager to make an effort for vocabularies which they (perhaps hastily) conclude are destined to remain the vehicles of a few dozen. Another source of difficulty in cooperation with other social scientists is the reluctance of historians to accommodate themselves to explicit models. Abstraction

110

Val R. Lorwin

from the complexities they know or would know is a wrench of the spirit. They tend therefore to look upon model-builders with a skepticism even deeper than that of the social psychologist who spoke of 'the authors who seem to entertain their models because the models entertain them'.37 Yet this skepticism cannot prevent us from seeing our evidence in relation to our own implicit models, usually as cherished as unexamined. An English historian, in sardonic reference to philosophers of history, has remarked that, like the character in Moliere's comedy, 'one may be writing History without formally knowing what it is'.38 We are even more likely to be writing social science without formally knowing, or properly caring, what our assumptions about man's nature or society are. As the Committee on Historical Analysis of the Social Science Research Council observed 'Historians borrow ready-made generalizations, whether they know it or not. If they were to borrow them knowingly, they might be in a stronger intellectual position'.39 'The art of theorizing, like that of fiction, largely consists in knowing what to leave out', as Michael Balfour says.40 Historians usually want to find out more about what needs to be put in, before they rigorously leave out. For their colleagues in the social sciences as for themselves, they have a wholesome fear of what T.H.Huxley called the greatest tragedy, a beautiful theory killed by a simple fact. The less sophisticated historians have not acquired a proper respect for the processes by which stimulating theoretical formulations may arise out of summary and imperfect perception of the historical evidence.41 Nor do they always appreciate the virtue of the 'outrageous hypothesis' in generating useful inquiry. About the use of the material they have amassed and analyzed and qualified, historians often feel as psychologists and anthropologists do about sociologists' use of their data. They have an innocent capacity for being shocked by inattention to, or careless handling of, facts in works of greater theoretical range than their own. They may not be charmed when a well-known sociologist gracefully asks 'the reader's indulgence if the next few pages leave much to be desired in terms of historical accuracy and detail'. They are likely to worry, not about absence of factual detail, but about carelessness of broad statements of presumed fact. They will not be reassured, even if they are grateful, when they read that 'All too often the societies that appear in the work of sociologists are merely historical constructions borrowed from earlier works or even invented in order to provide an impressive contrast for contemporary data.'42 Short-run time sequences may concern the historian more than other social scientists. But those concerned with grand sequences of social change cannot disregard the short-run sequences in periods of crisis and decision. John Plamenatz has stated this concern with his usual cogency, in warning us against the reification of the concept of 'revolution'.43 (In the study of nation-building the traps of reification are many.)

Historians and other social scientists

111

There is a constant tension in the work of most historians between the claim that they alone are serious in the handling of the facts of the recorded past and — on the other hand - the disclaimer of the aims of science for their work because theirs is not a discipline of regularities and laws and predictions. The claim is exaggerated; the disclaimer out of place. As Max Weber observed, 'The professional historians, unfortunately, have contributed not a little to the strengthening of the prejudice that "historical work" is something qualitatively different from "scientific work" because "concepts" and "rules" are of "no concern" to history.'44 Since Weber wrote, moreover, we have learned much more about indeterminism in the natural and physical sciences, about the role of description in the generalizing disciplines, and about 'the circumstantial, the irrational, and nonrational, as well as the logical and systematic nature of social research'.45 Historians may come to lose some of the feeling of uniqueness in their intellectual processes and in the findings of their discipline.46 Meanwhile, the social sciences not being very scientific, Bertrand de Jouvenel has suggested, they ought to be very social. A contemporary economist distinguished for his work in economic history once deplored the separation of the two disciplines in that field: the economists seemed to have all the problems and the historians all the data. Surely, by now, in our work on nation-building, as on economic growth and other problems of social change, enough people within the relevant disciplines have enough realization of mutual interest and interdependence to call more upon the historians in the formulation of the problems and upon other social scientists in the pursuit of the historical data? At a conference on the study of comparative politics,47 an American political scientist demanded, with asperity, 'Must we all become sociologists, or whatever the most general field of study is?' In the dozen years or so since that conference, an answer has come from some of the most effective contributions, empirical and theoretical, to comparative politics. Now members of the more theoretically and behaviorally oriented disciplines may ask, in their work on nation-building: Must we all become historians? Clearly, history is too important to be left to the professional historians alone. We all mine the endlessly complex seams of a fascinating and frequently tragic human experience. We have everything to gain by making the confrontations of historians and other social scientists in that work more frequent and - to use a word dear to Marc Bloch - fraternal.48 4. An example of comparative studies: The politics of the smaller European democracies. This communication may be of most practical use (its writer's modest aim) if it now turns to a fraternal research undertaking of an international and interdisciplinary nature. The series of studies recently launched on the Politics of the Smaller

112

ValR.Lorwin

European Democracies will, we hope, indicate something of the possibilities of a cooperative project which combines the national specializations of individual authors with comparative and theoretical concerns from the outset. It also combines the autonomy of the individual scholar with the stimuli of continuing confrontation and mutual criticism. The collaborating authors are people whose training and interests include history and behaviorally oriented disciplines. Steeped in the development and problems of the countries they are working on, many of them also have a 'built-in' comparative focus because of their research on other countries than their own. The initiators of the project 49 have had an experience of candid collaboration on a volume of country studies and comparative analysis, launched by Robert A. Dahl, on Political Oppositions in Western Democracies,50 which includes both large and small democracies. In working conferences, the authors have hammered out a common outline and a set of priorities for the types of information they hope to bring together on most of the countries, at various historic periods, for the individual country studies and for comparisons. One result of these meetings has been to show us how serious are some of the gaps in the availability of basic historic information for countries of advanced political, cultural, and administrative development. Departures from the common outline will of course be necessary because of the contrasts in national experiences. These contrasts give us some of the dimensions ofcomparison against the background of common characteristics. The small democracies include old states and newly independent ones; old monarchies and republics old and new; unitary states and federal; unilingual nations and multilingual; religious homogeneity and religious pluralism; nations with experiences of ruling other peoples and nations with memories of being ruled. Economic development came early to some countries and only very recently to others. The similarities and the differences among these states - the five Scandinavian nations, the three Benelux countries, Ireland, Austria, and Switzerland - indicate something of the historical and comparative interest of the individual country studies and the later more theoretical analysis. The smaller European democracies show a number of different styles - perhaps later we can speak of 'types' - of civic participation. One need only contemplate the differences between The Netherlands and Belgium: states of the same degree of economic and cultural development, similar in size and exposed international position, and of similar cultural values. The Dutch show a striking civic cohesion despite all the rigidities of religious, social, political, and cultural particularisms to which they have given their own name: verzuiling. The Belgians have achieved a high degree of social organization but a less orderly relation of particularistic groupings to national political life; their civic culture has more of the frondeur tone of France and less of the rational tone of the Northern countries. Within Belgium the differences among Flemings and Walloons and Bruxellois are at least as striking as those between Belgium as a whole and

Historians and other social scientists

113

The Netherlands. Both countries show the importance in the processes of national integration of the pre-national and even medieval collectivities, institutions, and traditions: church, 'estates', universities, and - in Belgium - linguistic groupings. These make a rich field for the study of continuities and of change in societies whose complexity is so much greater than their size. Some of our concerns are basic to the history and analysis of the consolidation, governing, and civic integration of all national states in a fairly advanced cultural setting. Others are questions specific to small nations. How has smallness of population-and-area affected the substance and intensity of national life? Are there for example - national political myths and ideologies linked to smallness? What has smallness meant for the character of relationships among elites? For relationships between leaders and led in government and in social movements and interest groups? How has it affected the degree of dependence on power resources outside the national system? Accessibility to international economic penetration and vulnerability to international economic changes? What have been the consequences of smallness for governments and voluntary organizations in relation to regional, European, and global international cooperation? Following an understandable bias in research and teaching, Western theorists of democratic politics have tended to take account only of the experiences of the main 'Anglo-Saxon' states, France, and Germany, with occasional reference to Italy or Sweden or Switzerland. The rich diversity of experiences in the smaller European democracies has been almost lost to comparative theory, or confused in its occasional exploitation, for lack of available information. Yet each of the smaller democracies offers fascinating historical and current evidence, on nationbuilding among other subjects, which needs assembly and modern analysis, comparison with the larger democracies and with existing theories of democratic political behavior. If human history is permitted to continue, the West will not forever furnish the models (whether in an analytical or normative sense) of nation-building. As long as it does, however, the models need not be based only on England and France and the United States. Through the smaller European democracies, we can see experiences significantly different from those of the large nations. A distinguished Swedish economic historian remarked that no citizen of a small country should expect other people to study his country's history merely because it happened. May we not turn his modest statement around and say that no scholar of a larger country should ignore the history of another country simply because it is a small country? Let us not yield to the snobbishness of size, which is unlovely and, like all snobbishness, stifling to the intellect as well as to the heart. The value of human experience is not in the size of the political community. Ancient Athens knew cultural greatness among a body of citizens no more numerous than that of

114

Val R. Lorwin

Iceland today. Shakespeare's England had a population no greater than that of Switzerland now. One speaks softly in suggesting that any academic studies have utility for the tasks of statesmanship. But in the life of many of today's new nations, smallness of population-and-power is a crucial characteristic. Many of their nation-builders find much of the experience of the large nations apparently unrelated to their own problems simply because of the differences in numbers of electorate and of elites, in weight of national resources and power. Some of the people in the new nations and their friends have therefore turned with interest to the social, political, and administrative experiences of the smaller European democracies. However differently we may evaluate the elements of human freedom, we must remain sensitive to its forms and uses in our discussions of modernization and nation-building. Nowadays other objectives, n o doubt worthy and pressing, generally take pride of place, and push individual freedom out of sight or hearing, in discussions of the 'political alternatives of development'. 61 Is it unreasonable to hope that study and comparison of the development of the smaller European democracies will throw light on the conditions under which peoples may - at least for a historic moment - harmonize the often contradictory goals of freedom and discipline, spontaneity and order, compassion and justice, growth and stability, national integrity and international cooperation? 'Between the individual and humanity, stands the nation', said Friedrich List over a century ago. 52 It has been true, since then, in ways he could not envisage. But it is among the attractions of research (wertfrei though it be) in most circles of the smaller European democracies that there the nation does not so much stand between the individual and humanity as lead him toward humanity.

NOTES

1. Reprinted in T. H. Marshall, Class, Citizenship, and Social Development (Garden City, N. Y., Doubleday, Anchor Books, 1965), pp. 71-134. 2. S. M.Lipset, 'Democracy and the Social System', in Harry Eckstein, ed., Internal War, Problems and Approaches (Glencoe, 111., The Free Press, 1964), p. 317. For another recognition of the need for further study of the politics of Western societies, notably in their historical dimension, see Gabriel A. Almond, 'A Developmental Approach to Political Systems', World Politics, XVII (Jan. 1965), p. 183. It is needless to multiply the references; the need exists, and is being met partially. 3. Gabriel A. Almond and Sidney Verba, The Civic Culture: Political Attitudes and Democracy in Five Nations (Princeton, N. J., Princeton University Press, 1963), p. 5. 4. Cited by Gertrude Himmelfarb, Lord Acton: A Study in Conscience and Politics (Chicago, University of Chicago Press, 1952), p. 183. 5. Cited by Halvdan Koht in his Driving Forces in History, trans, by Einar Haugen (Cambridge, Mass., Belknap Press, 1964), pp. 120-121.

Historians and other social scientists

115

6. See Charles E. Merriam, The Making of Citizens: A Comparative Study of Methods of Civic Training (Chicago, University of Chicago Press, 1931). 7. Carlton J.H.Hayes, France: A Nation of Patriots (New York, Columbia University Press, 1930). 8. Oscar Jâszi, The Dissolution of the Habsburg Monarchy (Chicago, University of Chicago Press, 1929). 9. George Stigler, 'The Economist and the State', American Economic Review, LV (1965), p. 11. 10. Lee Benson, The Concept of Jacksonian Democracy (Princeton, N. J., Princeton University Press, 1961), p. vii. 11. Dorothy George, England in Transition (Harmsworth, Pelican Books, 1953), p.41. 12. Asa Briggs, 'The Background of the Parliamentary Reform Movement in Three English Cities, 1830-1832', Cambridge Historical Journal, X (1952), pp. 293-317, and 'Social Structure and Politics in Birmingham and Lyons (1825-1848)', British Journal of Sociology, I, No 1 (1950), pp. 67-80. 13. Stein Rokkan and Henry Valen, 'Regional Contrasts in Norwegian Politics : A Review of Data from Official Statistics and from Sample Surveys', in Erik Allardt and Yrjô Littunen, eds., Cleavages, Ideologies and Party Systems: Contributions to Comparative Political Sociology. Transactions of the Westermarck Society, Vol. X, Helsinki, 1964, pp. 162-238, and a number of other studies cited therein. 14. Erik Allardt, 'Patterns of Class Conflict and Working Class Consciousness in Finnish Polities', in Allardt and Littunen, eds., op. cit. (cited in note 13 above), pp. 97-131. 15. See especially Juan Linz and Amando de Miguel's chapter in Richard Merritt and Stein Rokkan, eds., Comparing Nations (New Haven, Conn., Yale University Press, 1966) and a number of articles cited therein. 16. Charles Tilly, The Vendée (Cambridge, Mass., Harvard University Press, 1964). Note the comment of the reviewer in The Economist (March 6, 1965), p. 1026: '...as the book proceeds... Mr. Tilly has evidently begun to wonder if historical study should not be prescribed for sociologists rather than sociology for historians.' 17. Georges Duveau, La Vie ouvrière en France sous le Second Empire (Paris, Gallimard, 1946), pp. 225-230 and passim. These pages are translated in Val R. Lorwin, ed., Labor and Working Conditions in Modern Europe (New York, Macmillan, 1967). 18. Carey B.Joynt and Nicholas Rescher, 'The Problem of Uniqueness in History', History and Theory, I (1961), p. 151. 19. See for example the excellent work of Reinhard Bendix, with one chapter in collaboration with Stein Rokkan, Nation-Building and Citizenship (New York, J.Wiley, 1964). It shows the utilization of, and reflection on, a vast body of historical material, as well as Bendix's usual shrewd theoretical analysis. But, at least by implication, it also calls attention to the lack of information on many problems in many countries. 20. T. H. Marshall, op. cit. (cited in note 1, above), p. 38. 21. Felix Gilbert, American Historical Association Newsletter, III (1965), p. 8. 22. For a fascinating example of such exchanges, formal and informal (although perhaps with an under-representation of professional historians), see the week's papers and discussions in Georges Friedmann, éd., Villes et campagnes: civilisation urbaine et civilisation rurale en France, Deuxième Semaine Sociologique (Paris, Colin, 1953). 23. As Harold D.Lasswell has put it, 'Challenging ideas make their contribution to the intellectual life by mobilizing the energies of those who reject and oppose as well as those who accept and adopt. The result is the sharpening of issues and the scaling down of exaggeration by the consideration of opposites.' 'Impact of Psychoanalytical

116

24.

25. 26. 27.

28.

29. 30. 31.

32.

Val

R.Lorwin

Thinking on the Social Sciences', in Hendrik M.Ruitenbeek, ed., Psychoanalysis and Social Science (New York, E.P.Dutton, 1962), p.8. On American historians' relations to the social sciences, on which this paper can say little, see John Higham (with Leonard Krieger and Felix Gilbert), History (Englewood Cliffs,N. J., Prentice-Hall,1965), esp. pp.132-144, and Louis Gottschalk, ed., Generalization in the Writing of History, A Report oj the Committee on Historical Analysis of the Social Science Research Council (Chicago, University of Chicago Press, 1963). Both books have extensive bibliographies. Higham is more optimistic than I am about the open-mindedness of American historians to the behavioral sciences, but he is discussing outstanding historians, not - as I am in this paper - the mass of practitioners of the craft. Less hopeful on this score than Higham are his collaborator in the volume History, p. 381, and Thomas C. Cochran, The Inner Revolution. Essays on the Social Sciences in History (New York, Harper Torchbooks, 1964). The large number of gifted scholars who took refuge from Nazism in the United States inevitably brought a certain comparative dimension to their historical studies and a historical dimension to their behavioral work. Charles-Augustin Sainte-Beuve, Chateaubriand et son groupe, vol. II, p. 55, cited by Maxime Leroy, La Pensée de Sainte-Beuve (Paris, Gallimard, 1940), p. 183. 'On Comparative History', review of Rushton Coulborn, éd., Feudalism in History (Princeton, N.J., Princeton University Press, 1956), in World Politics (San., 1957), p. 267. See the valuable Bibliography of Works in the Philosophy of History 1945-1957, compiled by John C.Rule, History and Theory, Beiheft 1 (1961) and 'Bibliography of Writings on Historiography and the Philosophy of History', by Martin Klein, part III of Louis Gottschalk, ed., Generalization (cited above in note 24). Henri Pirenne,'De la méthode comparative en histoire', Compte Rendu du Cinquième Congrès International des Sciences Historiques (Bruxelles, Weissenbruch, 1923), pp. 19-32. H.P.R.Finberg, éd., Approaches to History: A Symposium (Toronto, University of Toronto Press, 1962). Charles Samaran, éd., L'Histoire et ses méthodes, Encyclopédie de la Pléiade (Paris, Gallimard, 1961). Marc Bloch, 'Pour une histoire comparée des sociétés européennes', Revue de synthèse historique, XLVI (1928), pp. 15-50: reprinted in his Mélanges historiques, 2 vols., (Paris, S.E. V.P.E.N., 1963), vol. I, pp. 16-40. There is an English translation in Frederic C.Lane, J.C.Riemersma, eds., Enterprise and Secular Change: Readings in Economic History (Homewood, 111., Richard Irwin, 1953), pp. 494-521; it omits the footnotes, which have interesting remarks on methodology. This is not the place to discuss the pitfalls of comparative studies in economic growth. For a stimulating reminder of some of those pitfalls, as well as some of the possibilities of such research, see Witold Kula, 'Histoire et économie: la longue durée', in Annales: Économies, Sociétés, Civilisations, XV (1960), pp. 294-313. For a lively dissent re economists' interest in comparative economic history, see W. Leontieff : 'Comparativism as a method of scientific inquiry is greatly overrated. In economic research, particularly of a quantitative kind, it offers convenient refuge to unimaginative minds. If one is at a loss in finding an effective analytical interpretation of a given set of facts, it is always possible to compare, particularly if one is ready to disregard destinations. But after the comparison is completed, what next? Too often one turns to the comparison of something else.' 'When should History be written backwards?', Economic History Review, 2nd series, XVI (1963), p. 6.

Historians and other social scientists

117

33. Marc Bloch, Apologie pour l'histoire ou Métier d'historien, Cahiers des Annales, 3 (Paris, Colin, 1952), p. 15. Bloch is there speaking against 'autarchy' of disciplines, as well as of periods and nations. 34. Pirenne, op. cit. (cited in note 28), pp. 23-29. Italics in original. 35. Lloyd Reynolds, in National Bureau of Economic Research, The Comparative Study of Economic Growth and Structure: Suggestions on Research Objectives and Organization (New York, NBER, 1959), p. 177. 36. Maxime Leroy, op. cit. (cited in note 25), p. 175. 37. Gordon W. Allport, Personality and Social Encounter: Selected Essays (Boston, Beacon, 1960), p. 67, n. 10. 38. M.D.Knowles, 'Academic History', History, XLVII (1962), p.223. 39. Louis Gottschalk, ed., op. cit. (cited in note 24), p. 209. 40. Michael Balfour, States and Mind (London, Cresset, 1953), p. 49. 41. Cf., however, an economist (and economic historian) in the same vein of protest: 'Intellectual history supplies many examples that suggest that premature generalization stultifies rather than promotes theoretical progress.' Mark Blaug, 'A Case of Emperor's Clothes', Kyklos, xvii (1964), p. 563. 42. Ralf Dahrendorf, Class and Class Conflict in Industrial Society (Stanford, Calif., Stanford University Press, 1959), pp.241 and 242. 43. John Plamenatz, The Revolutionary Movement in France, 1815-1871 (London, Longmans, Green and Co., 1952), pp.xii-xiii. 44. Max Weber, The Methodology of the Social Sciences, E.A.Shils and H.A.Finch, transis, and eds., (Glencoe, 111., Free Press, 1949), p. 115. 45. Phillip E.Hammond, ed., Sociologists at Work (New York, Basic Books, 1964), Introduction, especially p. 2. 46. For one discussion, Maurice Mandelbaum, 'Historical Explanation: the Problem of Covering Laws', History and Theory, I (1961), pp. 229-242. 47. Roy C. Macridis, 'Research in Comparative Politics: A Seminar Report', American Political Science Review, Vol.47 (1953), pp. 641-675. 48. Marc Bloch, Apologie pour l'histoire (cited in note 33), p. 72. Bloch uses the word there in a sense somewhat different from, and much broader than, mine here. 49. Hans Daalder of the University of Leiden, Stein Rokkan of the Christian Michelsen Institute and the University of Bergen, Norway, Robert A. Dahl of Yale University, and the author of this paper. 50. Robert A.Dahl, ed., Political Oppositions in Western Democracies (New Haven, Conn., Yale University Press, 1966). The volume contains chapters on Austria, Belgium, France, Germany, Italy, The Netherlands, Norway, Sweden, Great Britain, and the United States. 51. K.H.Silvert, ed., Discussion at Bellagio: The Political Alternatives of Development (New York, American Universities Field Staff, 1964), illustrates the problem well. Cf. however, the statement of the President of Nigeria: 'If independence means the substitution of indigenous tyranny for alien rule, then those who struggled for the independence of former colonial territories have not only desecrated the cause of freedom, they have betrayed their peoples.' 'Nigerians Angered by a Death', New York Times (Feb. 9,1965). 52. Friedrich List, Das nationale System der politischen Ökonomie (Basel, Kyklos-Verlag, and Tübingen, J. C.B. Mohr, 1959), p. 174.

C H A P T E R VII

Modern nations and the study of political modernization RICHARD ROSE University of Strathclyde

In theory, we would all like to see all kinds of societies studied very intensively from all kinds of perspectives. In practice, the limits of scholarly manpower and resources lead to the uneven accumulation of data and theories about different societies. In the past 15 years, special attention has been given to the study of societies beginning to modernize or in a very intermediate stage of development. The shift of attention from only a few modern Western states to many nations in the course of development has been enormously stimulating to social scientists. Even though theories of modernization presuppose connections between modernized and transitional states, at times we may be in danger of overlooking several advantages of intensively studying already modernized societies.1 From such study, we can learn not only what new concepts can tell us about modernized states, but also what the conditions of already modernized states can tell us about new concepts and newly developing societies. The purpose of this paper is to consider advantages which might accrue to students of political modernization in all parts of the world if more attention were given to comparative analysis involving already modernized societies. The paper is presented in confidence that the need for studying developing nations is already well known, and widely accepted. Comparative studies in modernization may be discussed under two broad headings: studies which involve explicit inter-national comparison, and studies which utilize comparative concepts to elucidate conclusions of comparative significance from the examination of processes within a single nation. Both approaches are valuable; to date, the second appears to be more customary, for reasons of scholarly tradition and convenience.2 Within the field of explicitly inter-national comparative studies, at least five different approaches might be distinguished.

Modern nations and the study of political modernization

119

1. Comparisons of many nation-states in terms of aggregate national characteristics have demonstrated that some major clusters of attributes are characteristic of a number of modern societies. They also provide typologies within which to locate individual societies. Gabriel Almond and Edward Shils have pioneered in the development of realistic typologies of political systems; Phillips Cutright has shown how data on socio-economic and political development can be related with statistical precision and Banks and Textor have recently advanced analysis by exploring techniques for assigning and analyzing nominal as well as ordinal characteristics of nation-states. 3 Among other things, global surveys emphasize the 'extreme' and 'atypical' position of already modernized societies in the world of today. 2. Comparisons of attitudes cross-nationally to date have usually taken place in modernized societies, became only such societies have afforded good facilities for survey research. The Civic Culture, by Almond and Verba, is a landmark in creating attitude data by survey methods in order to test hypotheses comparatively; the authors promise to develop analysis further in less advanced societies. S. M. Lipset's work is a comparable achievement in the field of secondary analysis of previously collected attitude data from Western nations. In transitional societies, the lead given by Daniel Lerner's The Passing of Traditional Society (1958) is now being followed up by survey studies of attitudes among elites and other populations. 4 In The Achieving Society David McClelland ingeniously demonstrates the potentialities of studying attitudes towards components of modernization by content analysis. 3. The comparative analysis of particular structures, functions and processes common to all political systems readily lends itself to simultaneous consideration of modernized and transitional states. In the Social Science Research Council Studies in Political Development series, five volumes are of this type, focussing upon communications, bureaucracy, education, culture and parties. In the past, such studies were often centered on modernized societies; today, the emphasis if anything is in the opposite direction. While this approach rejects the identification of a single structure and a single function, nonetheless, for eminently practical reasons, researchers often use formal institutions as points of departure, e.g., attacking the comparative study of political socialization by analyzing educational institutions. 5 Reciprocally, certain major processes, such as legitimation, are not easy to study comparatively because the function is performed by so many different formal and informal institutions from nation to nation. 4. To date, studies of modernization concentrating upon internal differences of nations have given more attention to urban/rural contrasts rather than to interregional differences, notwithstanding the considerable lags in development which even recently have existed within regions of modernized societies, such as the American South, Quebec in Canada, and the Italian Mezzogiorno. These differences are

120

Richard Rose

particularly important insofar as they indicate that some 'sub'-systems within the boundaries of a single state are better considered systems in their own right. V.O. Key's Southern Politics and Rokkan and Valen's 'Regional Contrasts in Norwegian Politics'6 exemplify intra-regional and inter-regional comparison respectively; Robert Alford's Party and Society combines inter-regional and inter-national comparisons. 5. The comparative analysis of the general process of modernization in traditional, transitional and modernized societies can at least be dated back to Marx and Weber. Among contemporary studies, one might distinguish between the comprehensive theoretical analyses of Gabriel Almond, and of S.N.Eisenstadt's The Political Systems of Empires; theoretical studies involving tests in a number of countries, e.g., E.E.Hagen's On the Theory of Social Change; and symposia on aspects of modernization whether concerning a broad range of countries, as in the S.S.R.C's Bureaucracy and Political Development, or narrow in scope, as in the Social Science Research Council symposium, Political Modernization in Japan and Turkey, edited by Ward and Rustow. An alternative approach to comparative studies concentrates attention upon a single nation, or one aspect of a nation, in order to elucidate or illustrate concepts and relationships of general significance to modernized and transitional societies. 6. Because of difficulties in conducting field work in a number of countries, generalizations are often based upon an intensive analysis of a single developing nation. A book such as Leonard Binder's Iran covers a wide range of phenomena about one society in order to illustrate the author's comprehensive analytic schema. Alternatively, a single concept or process may be studied intensively within a single national context; Lucian Pye's Politics, Personality and NationBuilding is about Burma, but it is not only about Burma. 7. Less customarily, an approach developed in the study of transitional societies can be applied to the contemporary functioning of an already modernized system; for example, Richard Rose's Politics in England is an attempt to indicate the gains in applying Almond's approach to a country already intensively studied from other perspectives. The scope for such studies is considerable; for instance, one might use concepts borrowed from the study of modernization to analyze the process of transition in which American Negroes have been involved in recent years. 8. A growing and potentially very promising type of single-nation analysis involves the study of a stage of transition in the past history of a modernized society, in order to illuminate the present or future problems of societies currently in transition. It should be recalled that W.W.Rostow's work on economic development followed intensive study of the Industrial Revolution in 18th and 19th century England; Neil Smelser's Social Change in the Industrial Revolution, an intensive study of industrial organization and family life in Lancashire, 1770-1840, combines precise historical scholarship with a broad gauge Parsonian framework for studying

Modern nations and the study of political modernization

121

change. Recently, American scholars such as S.M.Lipset, W.N.Chambers and Richard Merritt have begun to re-analyze late 18th century American history in order to understand modernization processes in what Lipset calls 'the first new nation'.7 While the relationships between modernized and transitional societies are treated in a variety of ways, two points require special emphasis. First, in the light of knowledge of processes of political change gained from studying new nations, we need to rethink the conventional wisdom and assumptions about the processes by which already modernized societies have reached this desired condition. It is necessary to do this because the characteristics of modernized societies form the background to much theorizing about development by scholars and politicians. In principle, the concentration of historical scholarship in developed societies should make this a relatively simple task. Yet much historical writing is of little use to social scientists, and that which is relevant is often buried or ignored. To take only one example: several years before the Social Science Research Council's volume on bureaucracy was published, several English historians conducted an extremely suggestive re-analysis of the role of bureaucracy in the modernization of early and mid-19th century England, a topic of particular significance for that volume.8 In introducing this symposium, the editor, Joseph LaPalombara, called attention to the equally important misunderstanding of contemporary processes in modernized societies: 'Principles we try to export do not even operate in the United States'. The editor sought to increase the precision of his analysis of bureaucracy by fixing attention upon an 'Anglo-American' model of politics. Yet historically and today the bureaucratic systems of the two countries differ in a number of fundamental structural features and in role cultures. Englishmen also differ with each other about the merits of their own style of bureaucracy. These conflicts are particularly relevant when nationals from the two countries are involved in administration in developing countries.9 Secondly, there is a danger that the parochial concentration of study upon a few Western nations may be replaced by neo-parochial concentration upon a limited number of developing countries. The point is raised explicitly by David Apter's description of the self-imposed terms of reference of the Committee for the Comparative Study of New Nations. Apter argues that its admittedly 'arbitrary' decision to exclude from consideration nations that achieved independence before 1945 arises from a desire to maintain a 'comparative focus'.10 Ironically, the statement is contained in the preface to a volume entitled Old Societies and New States, a title that might as easily refer to a study of European countries in the past 150 years as to the study of Afro-Asian nations. In reviewing the prospects for increasing our knowledge of the processes of modernization, it is debatable whether more attention should be given to promising intellectual growth points (e.g., the study of political socialization) or to intellectual

122

Richard Rose

bottlenecks. Since systems theory presupposes the interdependence of parts, it might seem appropriate to concentrate considerable attention upon the bottlenecks to greater understanding of modernization and thus, in this paper, to consider ways in which the study of already modernized societies may contribute to a better understanding of bottleneck problems. The first obstacle to understanding the processes of political modernization arises from considerable confusion about the definition of the term. Ten different uses of the concept have been catalogued by Pye. 11 Even the titles of the volumes in the major Social Science Research Council study show slight inconsistencies in usage, for a volume on Political Modernization in Japan and Turkey is followed by one on Education and Political Development. The terms 'change', 'development' and 'modernization' may be used interchangeably by scholars to refer to a family of hypotheses. Within this family of ideas, several distinguishable emphases may be noted. In the broadest sense, the reference may be to all processes of political change; certainly, emphasis upon dynamic rather than static characteristics of political systems involves an important and challenging reorientation of thinking. No specific direction of movement need be implied in writingabout political change. Scholars who employ the term development usually discuss the extent to which a political system moves in the direction of acquiring more or improved capabilities. Modernization, insofar as it is used differently from development, seems to be regarded as a more precise concept (cf. industrialization). Attempts have been made to define modernization in terms of a variety of quantitative and/or structuralfunctional characteristics. Because political modernization involves concomitant fundamental changes in a range of political characteristics, specialists in the field rightly show an interest in a number of important problems which, while conceptually distinct, are more or less related to modernization. The concern with political stability in the context of dynamic change is important to political scientists. Stability may be studied in an ideologically neutral manner or it may be used to support an essentially conservative analysis of international trends. More openly ideological is a linking of modernization with democratization; nonetheless, a good empirical case can be made for the argument that the two are closely and positively related. Yetanother concern associated with modernization is that with 'civility' or 'civic' attitudes in the political culture of a society. Combining the study of modernization with the study of other basic features of a political system leads to a more comprehensive appreciation of politics; at the same time, however, it blurs any specific and restrictive meaning that can be given to modernization. The problem is further complicated by the fact that while analytically distinct, the political system is only one sub-system operating within a society. The growth of inter-disciplinary awareness has rightly made political scientists interested in relating changes in the political sub-system to prior, contemporaneous or conse-

Modern nations and the study of political modernization

123

quential changes in the economic and social sub-systems. There are very great differences in the extent to which self-described studies in comparative modernization concentrate upon political or non-political factors. Phillips Cutright has demonstrated the significance of relating non-political indices of socio-economic development to separate indices of political development. But only 20 of the 75 variables considered in the booklength catalogue of the Yale Political Data Program are labelled as 'political' by the authors. Furthermore, analysis of stages of development gives chief attention to an economic variable, per capita Gross National Product as the basis of a tentative classification of societies.12 The value of the Cross-Polity Survey by Banks and Textor lies in its concentration upon the intrinsically more difficult task of tabulating ordinal and nominal political characteristics. It is important to do this because political modernization cannot be reduced simply to industrialization or urbanization, for the political systems of industrialized and urbanized societies vary considerably in the character of their fundamental development - cf. the 20th century histories of America and Germany. Because the political, economic and social sub-systems of a society may change at different rates and in different ways, the interaction and lags between these sub-systems are themselves of major theoretical importance. Yet another complication in analysis arises from basic differences in criteria used for measuring different types of modernization. Economic and social criteria are usually quantitative or readily quantifiable. Political criteria more often involve qualitative judgments. The increased reliance upon structural-functional analysis in contemporary political science reinforces difficulties in measurement. It is not always easy to generalize about the performance of functions, or the functional significance of particular institutions, in a form that would be acceptable to the great majority of scholars. Attention to the study of the political culture (i.e., a complex of attitudes treated separately from institutions) introduces yet another complication for the study of political modernization: the need to reconstruct the past attitudes of various populations without hope of recourse to survey data. Furthermore, scholars must cope with the importance of the mixture of traditional and modern outlooks even in a modern system. In short, while reliable and valid criteria for modernization are extremely important, these criteria are much more difficult to obtain for the political sub-system than for the economic and social sub-systems.13 Technical problems in the measurement of modernization can only readily be tackled where the dependent variable exists, i.e., in already modernized societies. As Russett notes, 'There is a clear tendency for the quality and availability of data to rise with the level of economic development in a country.'14 The Cross-Polity Survey shows that considerable advances are possible in assessing nominal characteristics of political systems. Yet the development of precise measurements of modernization, important as it is for an explicitly dynamic field of study, is bound

124

Richard Rose

to be a slow process. Banks and Textor comment: 'Dynamic analyses are most easily carried out when the analyst has restricted himself to 1) relatively few variables for which highly accurate data are available, or 2) relatively few polities... for which highly accurate data are likewise available.'15 Concentrating attention upon refining the classification and analysis of modernized political systems is likely to give rise to the awkward question: To what extent are all modernized societies exceptional cases? Whatever definition of modernization is employed, the number of nations included in this category will be a small proportion of the total in the world, and they will be in an extreme category. The U.S., Australia, New Zealand and Canada are exceptional insofar as the absence of a traditional society and the presence of virtually virgin land for settlement are of crucial importance.16 Western European nations such as England, France and Germany differ fundamentally from transitional societies of today because the former had an Imperial rather than a colonial status, beginning their transition before the revolution of rising expectations had become important. Even within Europe, the modernized nations are exceptional in that their achievements differ from those of countries such as Spain, Portugal and Greece, countries with more in common with the majority of the world's nations. The most modern nonWestern nation, Japan, is described in a comparative study as the product of an exceptional configuration of factors enabling it to modernize; Turkey, by comparison, seems to have more in common with the majority of societies, and it has also achieved less modernization.17 The Soviet Union and now Communist China demonstrate that strides toward modernization can be made from circumstances which may be less different from those of most transitional societies than are the backgrounds of Western nations. The most easily defended assumption is that the process of political modernization in a single society is the result of a combination of factors general to many countries and factors that are exceptional or unique. Unfortunately, many historians have repeatedly demonstrated that while the study of unique sequences of events can be extremely suggestive, nonetheless the research does not 'add up', that is, findings are not presented with the intention of accumulating generalizations. In addition, historians concerned with unique events often betray sociological naivete, either ignoring the significance of general social trends or, alternatively, misusing whatever concepts they borrow from social science. The complementary danger in social science is that the necessary process of abstraction leads individuals to employ easily manipulated and internally consistent abstractions which are no longer relevant to the empirical phenomena which were the original objects for analysis. In such a situation, a sensible strategy for social scientists to employ is one which concentrates attention upon middle-range theories and processes where problems of comparison are less complicated by the difficulties discussed above. For instance,

Modern nations and the study of political modernization

125

studies of processes of political socialization and of nation-building in a variety of nations at different levels of development promise to increase our understanding of some of the dynamic aspects of political systems, even while major bottlenecks remain in whole-system comparative analysis. More attention to the processes by which governments plan and allocate resources and control outputs might also be immediately profitable. This is particularly important because governments in transitional societies offer intentions to regulate affairs in such a way that their outputs make a maximum contribution to development. Given the long time periods for which data are available in already modernized societies, political scientists might reasonably hope that the intensive analysis of governmental policy (most simply expressed in budgets, though also contained in other types of legislation) would improve understanding of the extent to which governments can make a positive contribution to national development. Another means of advancing knowledge might come from a more self-conscious choice of deviant cases to illustrate problems in development. The volume edited by Robert Ward and Dankwart Rustow on Japan and Turkey is an excellent example of a careful choice of countries for intensive study. Among more-or-less modernized nations, the political records of France and Germany provide illustrations of pathological problems, whereas the relatively successful development of legitimacy in Eire and Northern Ireland, notwithstanding a major guerilla war in the 1920's of a type common in transitional societies, is an atypical 'success' story. The distinction between typical and atypical need not be a distinction between what is generally significant and what is not. A solution to the basic problem of defining the process of modernization is not likely to come soon, if at all. Investigations into the problem, particularly by political scientists, are still in a preliminary stage. At present, sophistication is represented by recognition that political change does not conform to a unilinear movement through three ideal-type stages labelled 'traditional', 'transitional' and 'modern', convenient as these labels are. Modernization is a multi-origin, multi-track, multi-outcome process. Unfortunately we lack a conceptual vocabulary which readily recognizes the complexity of this process. Theoretically, it would not be difficult to develop a conceptual schema that would place each nation in a separate category - and perhaps leave dozens of categories empty. In practical terms, what is needed is a framework for analysis which allows for variety, yet keeps the total number of categories sufficiently few (say, no more than twelve) so that it is still possible to generalize about historical data. Because the process of development in political systems starts from several origins and involves a variety of alternative modes of transition, it would hardly seem reasonable to expect that all these differences in origins can be obscured in a process of rapid homogenization culminating in 'modernity'. It may be possible to argue that in the economic and social sub-systems of modern societies roles are inter-

126

Richard Rose

changeable from one nation to another. But to argue this about the political sub-system is another matter altogether. To put the case in concrete terms, it suggests that there is nothing 'fundamentally' different in the politics of America, England, Canada, Germany, France and, soon enough if not already, Russia and Japan. While not denying the stimulation that can come from emphasizing similarities across a wide range of political systems, it is nonetheless paying a very high price for abstractions if all major differences are ignored. The differentiation of characteristics of countries which are, or will soon be, treated as modern is particularly important for the delineation of future developments. Implicit in all studies of modernization is the assumption that political change, while gradual, is continuous. The achievement of a particular set of 'modern' characteristics is not the eschatological consummation of politics. While theories of political development remain theoretically open to the discussion of 'post-modern' stages, the point is not, to my knowledge, pursued. Yet for a few societies, the achievement of political modernization is a thing of the past. Unless one argues that countries such as England and America have not changed 'fundamentally' since they crossed the threshold of political modernization and will not change fundamentally in the future, then sooner or later it will be necessary to sketch out some of the characteristics of a 'post-modern' stage, or stages, of political development. Attempting to spell out the character and operational indices by which one can measure when a society moves out of the modern into a 'post-modern' stage can have considerable heuristic advantages. The choice of criteria, particularly if some are qualitatively different from criteria of modernity, calls attention to dimensions of change very different from those currently receiving emphasis. Inevitably, discussion of criteria involves shading by value judgments; accepting and recognizing this can be a virtue. A traditional politician might emphasize military strength as the chief criterion for post-modern development. Socialists might argue for criteria concerning the redistribution of political power, wealth and status. Economists might simply declare that economic criteria should continue to be basic in defining 'post-modern' development. Sociologists' criteria might involve greater differentiation and role-specific behavior. Humanists might argue that postmodern stages of development would be characterized by government showing greater concern for humane treatment of the handicapped, or a sense of'community' among its citizens. Thinking through the value assumptions implicit in assessing changes within modernized societies, particularly our own, may help clarify value assumptions involved in studying transitional societies at various distances from the threshold of modernization. Such an exercise can also usefully call attention to the extent to which societies which take different paths in transition to modernity may also take different paths out of this category to other stages of development. Conceivably,

Modem nations and the study of political modernization

127

some transitional societies might attain non-materialistic features characterizing 'post-modern' societies by leapfrogging past some of the less desirable or more ephemeral characteristics of modern societies today.

NOTES

1. An indication of the imbalance is the need for the programme of Rokkan, Lorwin, Daalder, et al., for collecting basic data about the numerous smaller democracies of Western Europe. See The Politics of the Smaller European Democracies (SED/G/64/6, 1964), mimeographed. 2. See, e.g., the bibliography of H.Eckstein and D.Apter, eds., Comparative Politics (New York, The Free Press of Glencoe, 1963), pp. 741-746. 3. See, e.g., G. A. Almond, 'Comparative Political Systems', Journal of Politics, XVIH, 3 (1956); Edward Shils, Political Development in the New States (The Hague, Mouton, 1962); P.Cutright, 'National Political Development', American Sociological Review, XXVIII, 2 (1963); and A.S. Banks and R.B. Textor, A Cross-Polity Survey (Cambridge, M.I.T. Press, 1963). 4. See, e.g., the contributions by F.Bonilla and H.Hyman in R.E. Ward, ed., Studying Politics Abroad (Boston, Little, Brown and Co., 1964). 5. See J.S.Coleman, ed., Education and Political Development (Princeton, N.J., Princeton University Press, 1965), pp. 18 if. On the general theoretical position, see Gabriel A.Almond's analysis in G.A.Almond and J.S.Coleman, eds., The Politics of the Developing Areas (Princeton, N. J., Princeton University Press, 1960), p. 11 and pp. 17ff. 6. In E. Allardt and Y. Littunen, eds., Cleavages, Ideologies and Party Systems (Helsinki, Westermarck Society, 1964). 7. S.M.Lipset, The First New Nation (New York, Basic Books, 1963); W.N.Chambers, Political Parties in a New Nation (New York, Oxford University Press, 1963); and Richard Merritt, 'Nation-Building in America: The Colonial Years', in K.Deutsch and W.Foltz, Nation-Building (New York, Atherton Press, 1963). 8. See three articles in The Historical Journal (Cambridge): Oliver MacDonagh, 'The Nineteenth Century Revolution in Government: A Reappraisal', I, 1 (1958); Henry Parris, 'A Reappraisal Reappraised', III, 1 (1960); and G. Kitson Clark, 'Statesmen in Disguise', II, 1 (1959). 9. Cf. J. LaPalombara's contributions in J. LaPalombara, ed., Bureaucracy and Political Development (Princeton, N.J., Princeton University Press, 1963) with R.Braibanti, 'The Civil Service of Pakistan', South Atlantic Quarterly, 58(2) (1959); R. Heussler, Yesterday's Rulers (Syracuse, N.Y., Syracuse University Press, 1963); and R.Rose, 'England: A Traditionally Modern Political Culture', in L.Pye and S.Verba, eds., Political Culture and Political Development (Princeton, N.J., Princeton University Press, 1965), pp. 112-122. 10. 'Preface', in C.Geertz, ed., Old Societies and New States (New York, The Free Press of Glencoe, 1963), p. vi. 11.L.Pye, 'The Concept of Political Development', in The Annals, Vol. 358 (1965), pp. 5if. 12. See B.M.Russett, et al., World Handbook of Political and Social Indicators (New

128

13.

14. 15. 16.

17.

Richard Rose

Haven, Conn., Yale University Press, 1964), especially the table of contents and pp. 293-303. The above points are illustrated in J.S.Coleman, 'Conclusion', in G.A. Almond and J.S.Coleman, op. cit., especially pp.532-533 and Tables 1-6. The operational significance is made clear for the analysis of individual countries historically in R.Rose, op. cit., pp. 86-105. B. M. Russett, op. cit., p. 3. A. S. Banks and R. B. Textor, op. cit., p. 7. American political scientists writing about American politics usually treat the system as unique. Given the high quality and the substantial amount of literature on American politics, perhaps encouragement should be given to an assessment of the relevance of these findings for comparative research. Robert E. Ward and Dankwart Rustow, The Political Modernization of Japan and Turkey (Princeton, N. J., Princeton University Press, 1964), pp. 463 ff.

CHAPTER VIII

The empirical and statistical basis for comparative analyses of historical change LEE BENSON University of Pennsylvania

'The unfortunate peculiarity of the history of man is, that although its separate parts have been examined with considerable ability, hardly anyone has attempted to combine them into a whole, and ascertain the way in which they are connected with each other. In all the other great fields of inquiry the necessity of generalization is universally admitted, and noble efforts are being made to rise from particular facts in order to discover the laws by which those facts are governed. So far, however, is this from being the usual course of historians that among them a strange idea prevails, that their business is merely to relate events, which they may occasionally enliven by such moral and political reflections as seem likely to be useful.... history has been written by men so inadequate to the great task they have undertaken, that few of the necessary materials have yet been brought together. Instead of telling us those things which alone have any value... the vast majority of historians fill their works with the most trifling and miserable details.... This is thereal impediment which now stops our advance. It is this want of judgment, and this ignorance of what is most worthy of selection, which deprives us of materials that ought long since to have been accumulated, arranged and stored up for future use [emphasis added]. In other great branches of knowledge, observation has preceded discovery; first the facts have been registered, and then their laws have been found. But in the study of the history of Man, the important facts have been neglected, and the unimportant ones preserved. The consequence is, that whoever now attempts to generalize historical phenomena, must collect the facts, as well as conduct the generalization [emphasis added]. He finds nothing ready to his hand. He must be the mason as well as the architect; he must not only scheme the edifice, but likewise

130

Lee Benson

excavate the quarry. The necessity of performing this double labour entails upon the philosopher such enormous drudgery, that the limits of an entire life are unequal to the task; and history, instead of being ripe as it ought to be, for complete and exhaustive generalizations, is still in so crude and informal a state, that not the most determined and protracted industry will enable anyone to comprehend the really important actions of mankind, during even so short a period as two successive centuries'. 1 That cri du coeur came from Henry Buckle in 1857 when he wrote the introduction to his great study, History of Civilization in England. It both expressed his conviction that the comparative study of human behavior over time and space was the best way to develop the science of social enquiry and explained why he had been forced to abandon his 'original scheme' to reconstruct the history of 'general civilization'. We need not take Buckle's observations literally to recognize the contemporary relevance and force of his basic argument. Although comprehensive theories of human behavior are perhaps best generated by individuals, they ultimately derive from, and depend upon, coordinated, rational division of scholarly labor. Unless we achieve such division, we are unlikely to get the comparable cross-temporal and cross-cultural studies needed to test and refine comprehensive theories of human behavior - particularly when they require vast quantities of varied data not previously 'accumulated, arranged, and stored-up for future use*. Specifically, the following proposition seems reasonable: If we hope to develop comprehensive theories concerning the 'processes of nation-building and nation integration', we must achieve some significant degree of coordinated, rational division of labor. That proposition emphatically does not imply creating an organization to coerce scholars working on nation-state problems to divide their labor according to some master-plan, even if we had some master among us convinced that he had 'the plan'. UNESCO, the ISSC, the ICSSD, however, might appropriately sponsor an organization to facilitate some degree of voluntary coordination and division of labor. That proposition probably finds widest agreement when it remains most abstract. Disagreements multiply, no doubt, when we get down to specifics. What is to be done? Who should do what? What organizational innovations can most effectively aid specialists to determine: 1. what to do; 2. who does what; 3. how can specified work best be done and in what order? To those questions, alas, I can only give elementary and general answers. This paper primarily is designed to serve as a springboard for more elevated discussion of questions relevant to Theme II of the Conference, 'the comparative analysis of historical change'. (Getting jumped on is, of course, a time-honored, main function of springboards.) Since the questions posed above essentially represent variants of the three sets of questions posed by the organizers of the Conference, the paper is organized according to the standard format they suggested.

The empirical basis for comparative analyses of historical change

131

1. Which lines of comparative research appear to have been the most promising so far, and why? To answer that question comprehensively requires far greater mastery of the relevant literature than I possess. Apart from other considerations, my answer necessarily is fragmentary and tentative because I can claim acquaintance with only a small fraction of the large number of studies presently available in English on 'nationalism', 'national growth', 'nation-building', 'emerging nations', 'political modernization', 'political cultures', 'political development', 'comparative polities', and the like. It seems reasonable to think, however, that the bewildering proliferation of terms used to designate more or less related phenomena reflects, and perhaps partially explains, the underdeveloped status of research on 'comparative polities', particularly the branch dealing with 'historical' nation-state building, and political development, i.e., change over relatively long periods. Concepts apparently are cloudy, researchers revel in semantic free play, and cumulative research suffers. Lucian Pye, at a 'Conference on Political Development', recently sponsored by the University of Pennsylvania, cited ten different definitions now extant in the scholarly literature (they by no means exhausted the list). After identifying three characteristics common to most definitions (equality of participation, capacity of government, differentiation and specialization of political institutions and structures), he suggested that systematic research might permit us: to distinguish different patterns of development according to the sequential order in which different societies have dealt with the different aspects of the development syndrome [emphasis added]. In this sense development is clearly not unilinear, nor is it governed by sharp and distinct stages [emphasis added], but rather by a range of problems that may arise separately or concurrently [emphasis added].2 If I correctly understand Mr. Pye's noteworthy attempt to bring order out of confusion, it differs significantly in at least two respects from the solution proposed by Karl Deutsch - if I correctly understand Mr. Deutsch. 3 1. Pye delineates a general system of analysis applicable to the development of any type of political community - empire, city-state, tribe, nation-state, and the like. Deutsch focuses on a single type, the nation-state. 2. Pye's system, if used in research on historical political development, eventually might permit us to establish categories to which we could assign different nationstates. These categories would emerge, however, only retrospectively - after the 'facts' were in, so to speak. That is, unless I have badly misunderstood him, we could establish categories to assign specified societies only after we had integrated the results of a great many studies of the particular historical development of individual societies. Moreover, even if societies had developed similar solutions to specified political problems, they would not be grouped together unless they had

132

Lee Benson

also dealt in the same ''sequential order' with a wide range of problems 'that may rise separately or concurrently'. In sharp contrast, Deutsch's scheme postulates a predetermined set of relatively few categories, identified by a relatively small set of (presumably) strategic variables common to all nation-states. Moreover, it explicitly is based upon the 'stages' concept explicitly discarded by Pye. Pye's discussion of the concept of political development is illuminating. I doubt, however, that his system of analysis can provide an economical or manageable framework for coordinating research on historical change in 'the processes of nation-building and national integration'. It does not identify strategic variables relevant to those processes and, in effect, precludes establishment of a standard classification system until the entire history of a number of nation-states has been reconstructed. It does not seem to provide a general framework, therefore, for scholars interested in designing their research to facilitate comparisons of different nation-states over time. In contrast, a system based on stages of national political development, identified by specified variables, might serve as a framework within which we could advance a number of promising lines of comparative research. The political stages in Deutsch's system of analysis seem to be conceived hierarchically; in other respects, they differ from those used in evolutionary, organic theories of social change. Societies need not pass through a fixed, unilinear sequence of stages, he postulates, and they can make both forward and backward 'leaps' (my term). His key assumptions seem to be: 1. a specified society at a specified time can be accurately identified as being at a specified stage of political development; 2. it is useful to do so. Granting his assumptions, we could assign different societies existing at the same time (or period of time) to different stages, and different societies existing at different times to the same stage. Clearly, if this procedure could be applied, it would give us a far larger number of case studies of political development than if attention were confined to contemporary nation-states, or to the history of each nation-state treated as a single case. Like Marx, Rostow, and others whose analyses are based on the concept of stages, Deutsch uses a few variables which can be combined to designate stages of political development. Summarizing his summary observations on the processes of nation-building and national development, Deutsch noted: Open or latent resistance (by some proportion of the population living in the national territory) to political amalgamation into a common national state; minimal integration to the point of passive compliance with the orders of such an amalgamated government; deeper political integration to the point of active support for such a common state but with continuing ethnic or cultural group cohesion and diversity; and, finally, the coincidence of political amalgamation and integration with the assimilation of all groups to a common language and culture - these could be the (four) main stages on the way from tribes to nation.4

The empirical basis for comparative analyses of historical change

133

In effect, Deutsch delineated four stages identified by an index combining three variables: 1. demands by government on population (e.g., taxes, regulation, military service); 2. compliance with governmental demands by public (population); 3. cultural integration of population. These four stages, however, all follow establishment of a nation-state, an impersonal, autonomous political entity with more or less distinct territorial boundaries. Should they be described as stages of 'nationbuilding', or stages of 'national political development', or do the terms make any difference? If we called them stages of 'national political development' and restricted the term 'nation-building' to phenomena and processes occurring during the period which terminates with establishment of a nation-state, we would, I believe, reduce semantic confusion and might increase conceptual clarity. For example, prior to July 4, 1776, we might say that 'American nation-building' occurred in some British colonies on the North American mainland; after that date, 'American national political development' occurred. Apart from semantic considerations, I suggest that the two terms be differentiated because I assume that establishment of a nation-state structure is the great central dividing line in the political evolution of any society which eventually 'adopts' that form. Even an abortive attempt at independence can have long range consequences, e.g., the Confederate States of America. What analysts may perceive as the same phenomena, I suggest, actually appear as qualitatively different to participants when they occur in the different contexts of the non-existence of a national political structure and the 'visible' existence of such a structure. Why then blur these differences by using 'nation-building' and 'national political development' as synonymous terms?0 If the semantic distinctions proposed above are accepted (or merely granted for the sake of argument), Deutsch's sequence should logically be extended to include 'stages of nation-building'. For example, although my terms are clumsy and the concepts primitive, we might say that the sequence begins with 'consolidation of core territory', followed by 'transition to nation-state form'. In one case, consolidation might occur as a complex set of 'centralizing' processes mainly directed by dynastic kings over the centuries; in another, as a relatively simple set of 'colonizing' processes directed mainly by agents of European imperial powers over a few decades. The 'details' (sic) could differ widely but the outcomes would be essentially the same, territorial consolidation in a single political system. Whether that particular category is defensible does not matter; the point is that it (or a substitute) would permit us to assign different societies at different time periods to the same stage of nation-building. But the most promising lines of comparative research can now best be pursued, I believe, within the context of 'national political development', and it therefore seems useful to examine more closely the index suggested by Deutsch's observations. Both methodological and operational considerations militate against including

134

Lee Benson

'cultural integration' in an index designed to identify political stages. 'Governmental demands' and 'public compliance' can reasonably be used to measure political development, cultural integration cannot. Depending upon our objectives, we might treat levels of cultural integration as either dependent or independent variables in relation to levels of political development. But nations which score the same on a 'demand-compliance' index may, if they happen to differ widely in the heterogeneity of their populations, differ widely in their levels of cultural integration. Let us assume, for example, what we know is empirically not true but theoretically could be true: France, India, and the Soviet Union now score exactly the same on a demand-compliance index and differ widely in their level of cultural integration. (That assumption could be disputed only if we made the a priori claim that cultural integration bears a one-to-one relationship to demand-compliance.) Couldn't we nevertheless reasonably predict that by 2065 the populations of those societies would continue to differ widely in their levels of cultural integration? Why then use cultural integration to measure political development? Even if methodological objections were waived, operational objections remain against inclusion of cultural integration in an index to identify political stages. Unless we confine research to the very recent past, valid and reliable indices of cultural integration are extremely difficult to devise. And even if they could be devised, the data needed responsibly to use them - or other 'attitudinal' indicators of 'nationalism' - in historical research could be obtained only at a very high cost and over considerable time. Thus, rather than helping us now to begin to make systematic comparisons of the processes of national political development, including cultural integration in an index of political stages would seriously delay us. 'Governmental demands' and 'public compliance' are, of course, complex concepts. Particularly when used to construct an index of political stages for nations which vary widely in 'vertical power distribution' (degree of federalism), they pose conceptual and methodological difficulties. These difficulties could be overcome, I believe, by 'teams of specialists', i.e., intellectually compatible historians and nonhistorians, working together on the political development of a specified nation. Such teams could, in the relatively near future and with relatively modest resources, develop a reasonably good demand-compliance index for a specified nation and identify its stages of political development. In developing an index of political stages, I think it justifiable at present to relax strict methodological standards and not insist that precisely the same indicators of 'demand' and 'compliance' be used for all nations, or for any one nation over its entire history. The immediate task is to develop a common framework within researchers can systematically begin to make 'cross-temporal cross-cultural' comparisons. We can best do this, I believe if we establish a standard classification system of political stages; we would bog down badly if we first try to specify a standard set of indicators for researchers to use in applying that system.

The empirical basis for comparative 2. Classification system for political

analyses of historical

change

135

stages

Since a concrete proposal may at least stimulate counter-proposals, I propose establishment of a standard classification system having six categories (stages) of national political development. It uses a three point scale to measure 'demands': 'low', 'moderate', 'high'; a two point scale to measure 'compliance': 'low to moderate', 'moderate to high'. Combining the two components, we have the following six stages of national political development: Stage 1. 2. 3. 4. 5. 6.

TABLE I. Political Stages Government Demand-Public Compliance Index Low demand, low to moderate compliance Low demand, moderate to high compliance Moderate demand, low to moderate compliance Moderate demand, moderate to high compliance High demand, low to moderate compliance High demand, moderate to high compliance

Researchers, I assume, could responsibly use that classification system and empirically identify the sequence of political evolution (broadly conceived) of all (or most) nations. Crudely put, the question remains: So what? Not quite so crudely: How would this help us to generate, test, and refine fruitful theories about the processes of national political development? It would do so, I suggest, by giving us a systematic political basis for grouping and differentiating nations other than comparing their specified characteristics at a given time, e.g., in 1960 some nations had a 'competitive electoral system', others had various types of 'non-competitive' systems. To compare nations on a strict chronological scale only permits statements about the distribution of phenomena at specified times, or periods; it neither really permits us to ask nor answer interesting questions about the processes of national political development. In contrast, the proposed system of stages might help us to deal with such questions as : Can general patterns of 'movement' from one stage to another be found? Do causal relationships exist among a range of political and non-political data? Can the history of 'politically developed' nations be responsibly used to make predictions about the outcomes if specified courses of action are adopted in 'politically underdeveloped' nations with specified characteristics? To be concrete: Does history 'show' the different patterns of political participation and stability indicated in Table II below? Stage 1 1 5 6

TABLE II. Political Stages, Participation and Stability Demand Compliance Popular Participation Low Low Low Low High Low High Low High High High High

Political Stability High Low Low High

136

Lee Benson

Whether such patterns have actually occurred is, for our present purposes, irrelevant. Again the relevant point is that the proposed classification system provides a framework for collecting, ordering, and analyzing the data needed t o determine whether such patterns have existed and whether the indicated relationships are causal or spurious. (No implication is intended, of course, that the hypothetical table contains all the requisite data to test relationships.)

3. Historical archives of national political

data

Among other things, Table II indicates how the proposed system of stages might provide a framework for advancing the line of research which, in my opinion, holds out the greatest promise for the comparative analysis of national political development. I refer here to a solution which has recently been found to the problem posed a century ago by Henry Buckle: How can scholars effectively generalize from historical phenomena if they 'must (simultaneously) collect the facts, as well as conduct the generalization... scheme the edifice... (and) excavate the quarry'? The solution has been to free individuals from the crushing burden of collecting 'the facts' of national political development by recognizing that it is a social task which must be socially performed and paid for. In other words, the solution has been to establish what might be called 'Historical Archives of National Political Data'. Stein R o k k a n has thus far, to my knowledge, made the most comprehensive and effective use of historical data in theoretically-oriented studies of national political development. In his essay on 'Electoral Mobilization, Party Competition and National Integration', he persuasively presents the case for establishing historical archives for political data which, in Buckle's words, 'ought long since to have been accumulated, arranged, and stored up for future use...' Describing the archive he and his associates established to permit them to study systematically Norwegian political development, Rokkan observes: We are currently at work on the development of an historical archive of ecological data on Norwegian politics and hope in this way to be able to pursue much more detailed analyses of variations between localities in the rates and directions of change. This punched-card archive was originally built up to allow multi-variate analyses of local variations in turnout and party strength for the elections from 1945 onwards, but efforts are now under way to extend the time series for each local unit. We are also making efforts ot extend the range of data for each unit: we have so far punched on decks for each commune not only data from local and national elections but also data from censuses, from educational, agricultural, industrial and fiscal statistics, data from a church attendance count, data on local party organizations and memberships as well as on nominees to party lists for parliament. We have found such data archiving an essential tool in our cooperative research work and we hope in the years to come to expand the scope of our archive both backwards to the earliest partisan contests and forwards to the oncoming local and national elections. We think our experiences justify us in recommending that similarly conceived archives be set up in other contries of the

The empirical basis for comparative analyses of historical change

137

West and we are convinced that the greater control of the data masses achieved through such archiving will facilitate systematic comparisons of rates of development in different countries [emphasis added].6 I strongly endorse Rokkan's recommendation. His experiences in Norway correspond to those we have encountered in the United States in establishing an historical archive for American political data. Aided by 'seed money' grants from the Social Science Research Council and subsequent development grants from the National Science Foundation, the archive is a joint project of the Inter-University Consortium for Political Research and a committee of the American Historical Association. It is located in the headquarters of the Consortium at the University of Michigan, Ann Arbor, Michigan. Apart from direct benefits to researchers, the project to establish the archive has brought about a significant degree of enthusiastic cooperation among a large number of American political scientists and political historians. By developing organizational innovations - the Inter-University Consortium for Political Research and the American Historical Association Committee - to centralize resources, American political researchers have begun to coordinate and rationally divide their labor. Our highly rewarding experiences thus far suggest that social organization makes it possible to avoid the frustrations of Henry Buckle, a 'Renaissance Man', who nobly but futilely, tried single-handedly to study the processes of nationbuilding and national political development. In addition to organizational innovations, modern data processing technology and techniques give contemporary researchers interested in cross-cultural crosstemporal comparisons enormous advantages over their nineteenth century precursor, Henry Buckle. Thus, the American historical archive now underway is designed for maximum computer use in all phases of operations and data are being stored in machine-readable form suitable for computer retrieval, ordering, and analysis. At present, the American archive has two main foci, electoral behavior and legislative behavior. In other words, to use now familiar concepts and terms, data is being collected relevant to both the 'input' and 'output' components of the American political system from 1789 to date ;7 eventually the archive will be extended to include data prior to 1776. More specifically, in addition to data relevant to changes over time and space in the composition of the electorate, turnout of eligibles, social and territorial bases of partisan support, all Congressional roll call votes from 1789 to-date are being recorded on IBM cards. Moreover, computer programs are being developed to aid analyses of electoral behavior and legislative behavior, as well as the interrelationships between electoral and legislative behavior. This paper has grown much too long and, in any event, it probably is unnecessary to describe in greater detail the American data now scheduled for collection. In some respects, they probably are directly comparable to historical data Rokkan

138

Lee Benson

and his associates have collected for Norway. But the schedule of collection reflected particular interests of American researchers and was not drawn up to facilitate cross-cultural comparisons. This observation dramatically highlights both the great opportunities and dangers presented by current activities in a number of Western countries to create historical archives of national political data. Given modern data-processing technology and techniques, we now have the potential capacity to create the historical data archives required for systematic cross-cultural and cross-temporal studies. To translate potential capacity into actual capacity, however, requires a high degree of international cooperation and coordination. Unless we immediately achieve such cooperation and coordination, we can safely predict that different national archives will fail to collect comparable types of data, will collect them out of phase, will use different standards of evaluation, and different - probably incompatible - storage and retrieval systems. It seems imperative, therefore, immediately to establish an organization effectively linking researchers in different countries now in the process of creating historical archives. To that problem I now turn. Hand III. What can be done to accelerate the accumulation, standardization and evaluation of information and data for comparative analysis of historical political change? What can UNESCO, the ISSC and the ICSSD most profitably do to advance that type of comparative research and what concrete steps should be taken over the next 10-year period?

At this early stage of development, the simplest and best answer to these complex questions, seems to me to propose that an appropriate international agency take responsibility for organizing a 'Committee on Historical Archives of National Political Data'. To permit immediate attention to be given to problems which will become far more difficult if delays are experienced in working out solutions, I suggest that, to begin with, the Committee consist of men representing national archives now in existence. The suggestion derives from our experience in creating an American historical archive. We began operations by organizing a communication network formally linking individuals and institutions engaged in similar activity. This procedure gave us detailed, comprehensive information on what actually was being done in a variety of places. We then found ourselves in a better position to decide what should be done when, as well as how best to do it. Moreover, we found that in the process we had developed the relationships and the organizational forms which helped us to implement our decisions, and to stimulate increasing numbers of researchers and institutions to participate in the work and extend the range of activities and resources. Put another way, my suggestion is to bring together, as quickly as possible, representatives of institutions which already have experience in coping with the complex problems inherent in collecting and using historical data to study the

The empirical basis for comparative analyses of historical change

139

processes of national political development. In the absence of such an organized group, detailed answers to questions II and III probably will derive from particular experiences and interests, and thus highly likely to be crude, fragmentary, and shortrun. At any rate, my answers would now have that character. To avoid the fallacy of misplaced (or mistimed) concreteness, therefore, I think it best to restrict myself to proposing that the first step be taken of what must be a long-range program, namely, rapid formation of an international 'Committee on Historical Archives of National Political Data'. 8

NOTES

1. Henry Thomas Buckle, History of Civilization in England (from the Second London Edition, New York, 1903, originally published 1857), 1 :pp. 3, 166-167. The quotations are spliced together from different pages but are in context and accurately reflect Buckle's position. His book might reasonably be described as the first systematic, cross-cultural, cross-temporal comparative study of the processes of nation-building and national integration. See in particular Vol. 1, Chapters 9 and 10. 2. Pye's paper on 'The Concept of Political Development' was published in the Annals of the American Academy of Political and Social Science (March, 1965), pp. 1-14. The entire issue is devoted to 'political development', and I am grateful to its special editor, Professor Karl von Vorys, for providing me with the advance set of galley proofs from which I quote. I am also indebted to him, and to Professor Henry Teune, both of the University of Pennsylvania, for helping me to clarify some of the ideas presented in this paper - to the extent, at any rate, that they have been clarified. 3. Karl W.Deutsch, in the introductory chapter to K.W.Deutsch and William J.Foltz, eds., Nation-Building (New York, Atherton Press, 1963), pp. 1-16. 4. Ibid., pp. 7-8. 5. A recent newspaper article dramatically illustrates the point. The headline read, 'Macedonians Hunt Own Identity in Complex of Balkan Peoples', and the account begins, 'The young hitchhiker said his father was Bulgarian and his mother Greek, but 'I am a Macedonian'. That sense of national identity, the account makes clear, was crystallized by the creation of 'the Republic of Macedonia, a constituent member of the Yugoslav Communist State'. David Binder, New York Times (February 15, 1965). 6. Stein Rokkan, 'Electoral Mobilization, Party Competition and National Integration', in Joseph LaPalombara and Myron Weiner, eds., Political Parties and Political Development (Princeton, N. J., Princeton University Press, 1966). 7. David Easton, A Framework for Political Analysis (Englewood Cli ffs, Prentice-Hall, 1965); Gabriel Almond and G. Bingham Powell, Jr., Comparative Politics: a Developmental Approach (Boston, Little, Brown & Co., 1966). The approaches taken by Easton and Almond differ, but I have profitted heavily from both their pioneering attempts to develop a general theoretical framework. 8. As this volume goes to press, two conferences are being held by the AHA Committee, designed to establish such a Committee (among other purposes).

C H A P T E R IX

Research possibilities using aggregate political and social data HAYWARD R. ALKER JR Yale University

In the Sisyphus legend some of us identify with Sisyphus and some of us align ourselves with the gods controlling the forces of gravity. Some believe the myth to have a gloomy message, while others rebel with Sisyphus and remain defiant, perhaps believing the next mountain they climb may differ in some hopeful respect from the last. Such, of course, is the belief of the scientific explorer. The purposes and organization of the present paper* Identifying with the rock-pushers, I should like personally to speculate about research possibilities using aggregate political and social data. Since many of my own research concerns derive from the work of Karl Deutsch and Harold Lasswell, co-founders of the Yale Political Data Program, the paper may also serve to report some of the research interests underlying that Program, including itsfirstsubstantial publication, the World Handbook of Political and Social Indicators.1 Because of the considerable attention given to the World Handbook by papers in the Comparing Nations volume, and by such commentators as Goran Ohlin and Ralph RetzlafF,2 I shall also try at least to mention what appear to be the main arguments concerning the utility of comparative socio-political research using aggregate data. One can conveniently summarize the main concerns of social scientific analysis as description, explanation, and evaluation. As any short paper must be highly selective, I propose to illustrate the contribution of research based on aggregate data with several examples of each kind of analysis. In this regard a major descrip* In writing this paper I gratefully acknowledge comments by Karl Deutsch and Richard Merritt, and the continuing additions by Harold Lasswell to my understanding of the scope and purposes of political analysis.

144

Hayward R. Alker Jr

tive problem in aggregate research is measuring important but latent national dispositions or propensities. Explanatory problems to be discussed below include ascertaining the limits of (and explanations for) deviations from universal relationships, comparing hypotheses across different levels of analysis, and developing causal theories of political and social change. The classical evaluative question, how to compare the relative merits of differently constituted political systems, also will be briefly considered. As should be very clear throughout my exposition, descriptive, explanatory and evaluative research concerns are closely intertwined. Measuring latent national dispositions Data is empirical information processed and refined so as to measure theoretical concepts.3 Thus measurement, as a scientific task, produces data that is both theory-relevant and theory-laden. Among others, James Coleman argues that most important variables in social science - such as anomie, racial segregation, social class, social status and behavioral interdependence - must be considered as latent phenomena for which accurate indicators are often difficult to obtain. 4 Coleman calls these propensity-like characteristics 'dispositions' and the general measurement problem that of 'disposition measurement'. The literature on cross-national and cross-cultural dispositional measurement is immense and rapidly growing. It is perhaps sufficient here to suggest something of the variety of techniques involved and the kinds of theoretically important variables they have been used to measure. Each example represents a different way of further refining 'partially processed' aggregate data. Karl Deutsch's papers show a remarkable sensitivity to some of the basic ideas of Paul Lazarsfeld's latent structure, class, and profile analysis. In one he has suggested relatively simple procedures for developing from aggregate data and judgmental ratings a 'country profile' index for comparatively assessing national political performance; in another he has drawn on Lazarsfeld's notion of interchangeable indicators in order to get a second order probabilistic measure of a nation's social mobilization based on first order intercorrelations of socioeconomic variables.5 In another place, Deutsch and Solow have developed a differential equation model with which one can measure assimilation rates in regions where the treatment of ethnic minorities has been or may become a major political problem, e.g., Quebec, India, Finland, or the United States.6 Deutsch and Savage have developed, and Leo Goodman has generalized, a null statistical model of transaction flows that is relevant to the study of economic interdependence and political integration.7 Following the lead of several demographers and psychologists, Alker has used factor analysis to measure and compare national foreign policy predispositions in the United Nations. 8 Rudolph Rummel and Bruce Russett are each currently applying a kind of multidimensional scaling technique to cross-national data

Aggregate political and social data

145

in the hope of linking 'social distance' among nations to their tendencies toward cooperation or conflict. Quincy Wright's A Study of War contains a number of such hypotheses. One important, almost obvious, characteristic of recent aggregate research on national and cultural propensities is that many researchers have been able to use data collected for a great variety of purposes for their own particular and distinct concerns. The multifunctional utility of such data is often brought about through the application of further index construction or dispositional measurement techniques tailored to these other concerns. The need for the enhanced availability of such data in data repositories accessible to a variety of scholars has been increasingly recognized in the international social science community. 9 Perhaps a more important conclusion can be drawn from the great variety of substantive concerns, more or less automated analytic methodologies, and more or less adequately processed aggregate data resources.10 Comparative scholars need a more sophisticated metatheory about which dispositional measurement techniques are most likely to be appropriate for what kind of substantive problems and which varieties of aggregate data. Some issues of this sort have already been raised in the World Handbook and the Comparing Nations volume. Generally speaking, they include: a) improving validation techniques for indices constructed in the presence of possibly nonrandom errors of measurement; b) making more explicit the statistical and causal assumptions involved in choosing various automated latent variable measurement procedures; c) in particular, developing a more theoretically advanced statement of the relative merits in different cross-national research contexts of nominal, ordinal and interval measurement; d) improving ways of making longitudinal inferences from cross-sectional results; and e) developing methodologies that do not assume that certain subcultures in a population obey the same kind of structural relations as do others. 11 Many of these problems are common to all the social sciences, but I expect aggregate researchers may find some answers that are specially relevant to their small samples with frequently high intercorrelations and nonrandom measurement errors. Greater insights, more widely shared, into these and similar topics would significantly enhance the measurement payoffs of cross-national and crosscultural research.

Ascertaining the limits of universality Having concluded a brief review of several recent attempts to describe national propensities with a plea for a more sophisticated metatheory of index construction, I now want to suggest another class of research problems in some ways similar to this one. These problems concern the development of metatheories that explain what are often called 'regional' variations or departures from universal statistical

146

Hayward R. Alker Jr

relationships. Put more generally, when do certain contextual variables change the relationships among other aggregate variables, and why do they do so?12 Let us now be more specific. In a section of the World Handbook entitled 'Regionalism versus Universalism in Comparing Nations', I reviewed a statistical method - covariance analysis - for testing the universality of particular relationships suggested by some of the simple correlations presented earlier in the book. It turned out that some variables that were universally uncorrelated nonetheless has regionally significant associations, while other universal relationships had no validity within particular geographic or cultural regions. Thus an almost nonexistent correlation between McClelland's achievement motivation scores and per capita income hides a fairly strong positive correlation between these variables among Latin American countries and a moderately negative association among European countries (a finding suggesting that in economically advanced countries textbooks do not accurately reflect levels either of achievement motivation, or of successes, even if they do so in certain underdeveloped ones). Similarly it was shown that a universally small negative correlation between proportion of Catholic population and proportion of labor force in agriculture hides a much stronger negative association among these variables among Afro-Asians, and a stronger positive relationship among Latin Americans. Apparently the Weberian hypothesis about the retarding effect of Catholicism is universally inappropriate, very wrong when Catholicism indicates colonial and industrializing imperialism, but still of some value when comparing entire countries whose cultures approximate European conditions of perhaps a hundred years ago. 13 Like much of the research in the Comparing Nations volume, this analysis was supposed to suggest a number of measurement and explanation problems concerning the indicators used in the World Handbook. Obviously the first of these is the need to ascertain, validate, and explain the latent meaning of various indicators in different sociocultural and political contexts. In such a case the validity of our finding certainly depends on the number and variety of indicators giving approximately the same rankings of nations. Lazarsfeld, for example, refers to this phenomenon as the'interchangeability of indicators, meaning that indicators of the same latent variable will both be positively intercorrelated with each other and exhibit similar correlations with distinct related phenomena. Donald T. Campbell has similarly suggested that measurement validity comes from coincident indicators determined according to different methodologies and giving results distinguishable from other known conceptual variables.14 Table 1, for example, shows the beginnings of such an analysis. The variables include the logged Usui-Hagen estimates of 1957 per capita GNP (variable 1),15 as used in the World Handbook, and private per capita consumption figures derived from official exchange rates (variable 2). We can compare these with nonmonetary or 'real' indices of national economic development (variables 3 and 4) validated by

Aggregate political and social data

147

Wilfred Beckermann against the painstakingly gathered real product consumption estimates of Gilbert and his associates. 16 Especially when we use logged Usui-Hagen G N P per capita estimates, as was done in the World Handbook, it becomes clear that the Usui-Hagen index correlates very highly (r = 0.95, r 2 = 0.89) with Beckermann's real G N P per capita index and even with his conceptually quite different real private consumption figures. With either logged or unlogged variables, the Usui-Hagen index does considerably better than the official exchange rate as an approximation to the 'real' nonmonetary G N P per capita figures derived by Beckermann. It is worth noting that in fact the 0.95 correlation would probably be even higher if data for the same years were being compared and the Beckermann estimates were not, as he avows they are, 'rough estimates' based on private consumption figures and a correction factor. Such a high figure is clearly within a 10-15% range of expected or permissible error in cross-national comparisons focusing on general trends and patterns. Variables correlating over 0.90 with each other must of necessity have similar relationships with additional variables indexing distinct phenomena. That they do is evidenced in Table 1 by similar correlations of three economic indicators with

TABLE 1:

Selected Correlations * of four indicators of national economic development in I960. (Correlations below the diagonal are for logged data, those above are not.)** 1) UsuiH a g e n estim a t e s of GNP PC

1 Usui GNP per cap. 2 Real Priv. Cons. 3 Official Priv. Cons. 4 Real GNP per cap. 5 Vote Level 6 Domestic group violence

2) Becker- 3) I n d e x of m a n n ' s inprivate dexof'real' consumpprivate tion at consumpofficial extion c h a n g e rates

4) Beckerm a n n ' s estimates of 'real' G N P per capita

5) V o t e level

0.45 0.49 0.45 0.49

1.00 0.91 0.92 0.95 0.53

0.96 1.00 0.89 0.99 0.51

0.96 0.93 1.00 0.84 0.56

0.93 0.99 0.87 1.00 0.51

-0.45

-0.51

-0.23

-0.50

6) Domestic g r o u p violence

-0.43 -0.50 -0.31 -0.52

31 < N < 56 * Table entries are product-moment correlations, each of which may be interpreted as a regression slope for two standardized variables. Squaring these numbers gives the fraction of variance explained by the linear regression of either of the variables onto the other. ** Sources for variables 1: Usui and Hagen, World Income, 1957; variables 2,3, 4: Beckermann, op. cit., columns 1 and 2 of Table 5, column 3 of Table 6; variables 5 and 6: Russett et a!., op. cit., pp. 84-6, 99-100. Variable 5 is never logged, variable 6 always is.

148

Hayward R. Alker Jr

data on levels of voting turnout and domestic group violence, two of the 10 or 20 explicitly political variables in the World Handbook. Only the official exchange rate data seriously differ in their correlations with such variables. Correlations in a range like that of 0.45 ± 0.05 probably fall within the range of transformation, transcription and measuring error in a sample with an N of 30 or 50 that includes between 10 and 30 underdeveloped countries. (Certainly one should not be surprised by error margins of 10 % in this regard.) In summary then, we find that besides being highly intercorrelated, three rather different indicators of national economic development give strikingly similar results when correlated with representative aggregate political variables. These results greatly increase our assurance, within modest error margins, of their validity as cross-national indicators of economic activity. 17 As suggested above, however, the regional significance of such indicators may differ considerably. As well as generally confirming the relatively high correlation between the Usui-Hagen and real G N P indices, Table 2 points strikingly to distinctive relations between economic and political indicators in underdeveloped, primary commodity economies. 18 For all four of the indicators involved, among the developed countries we find high or higher relations between economic development and voting participation. Thus apparently a 'threshold' or intensification effect

TABLE 2 :

Correlations among selected economic and political indicators for developed and developing nations. (Correlations for economically developed countries are above the diagonal.)*

1 Usui GNP PC 2 Real Priv. Cons. 3 Official Priv. Cons. 4 Real GNP PC 5 Vote Level 6 Domestic Group violence

1) Usui G N P PC

2) Real Priv. Cons.

3) Official Priv. Cons.

4) Real G N P PC

5) Vote Level

6) Domestic Group Violence

1.00 0.69 0.88 0.89 0.23

0.95 1.00 0.83 0.99 0.13

0.88 0.84 1.00 0.77 0.37

0.91 0.99 0.74 1.00 0.04

0.33 0.31 1.37 0.37

-0.36 -0.47 -0.16 -0.68

0.09

0.02

0.39

0.33

9 < N < 31 * Data for variables 1, 2, 3, 4, and 6 have been subjected to logarithmic transformation. Sources for variables 1 through 6 are described beneath Table 1. The 'developed-developing' distinction is based almost completely on the political self-labeling of the 'Committee of 77' developing countries at the United Nations Conference for Trade and Development. See Volume 1 of the Proceedings of the Conference, 'Final Act and Report', 1964, p. 66.

Aggregate political and social data

149

of economic development must exist before per capita income and popular political activity become positively related. More striking, however, is the result for all four indicators that income levels among 'developed' or non-primary commodity countries are negatively correlated with violence levels, while they are positively correlated with them for the underdeveloped or developing ones. It is here that the real nonmonetary index most decisively shows its superiority over the Usui figures, not in giving qualitatively different results, but in indicating the magnitude of the reversal effect that seems to occur once development has reached and passed a certain threshold. Note in this regard remarkable 'real GNP/violence' correlation of - 0.68 among developed countries and the radically different + 0.33 figure for developing countries Further intriguing higher order questions are thus raised both about the meaningfulness of weak and strong universal statistical relationships and the determinants of regional departures from them. They include first an attempt to delineate 'regions' characterized by the same kinds of statistical and causal interrelationships. 19 Secondly, there is the universalist's preference to discover those 'second- or thirdorder' variables that are in fact causing the regional or cultural shifts from negative to positive statistical relationships. Thirdly, there is the need to develop and formalize mathematically comprehensive theories including contextual and higher order intensification and reversal effects. Clearly much of the literature on pattern variables of political development, stages and crises of economic and political development is enormously suggestive in this regard.

Comparing different levels of analysis Closely related to the urge toward universality is the drive toward general empirical generalizations that hold for many different levels of analysis. One can also try to understand and explain why some such generalizations do not hold across different levels of analysis. Personal research strategies may differ, but desires to borrow or discover the more attractive features of systems at different levels of analysis are almost universal. 20 For example, consider scholarly shifts in levels of analysis related to problems of war and conflict. The aftermath of a major war between peculiar coalitions of monolithic and pluralistic regimes brought renewed scholarly attention by Adorno, Lasswell and many others to the authoritarian and democratic personality. Students of American politics, for example, have spent a good deal of effort trying to explain how a relatively competitive national party system persists on top of 50 states, the majority of whose party systems are not terribly competitive, all representing a public strongly committed to general democratic principles if not always to their specific applications. 21 Myrdal is one of the most eloquent social scientists, but

150

Hayward R. Alker Jr

certainly not the first, to note that vicious circles of poverty and prejudice trouble local communities as much as international ones.22 Aggregate social and political data have been used to corroborate, modify, or reformulate these and other hypotheses relating different levels of analysis. A number of already completed studies indicate possibilities of this sort. 23 Michael Haas, for example, has found domestic violence levels to increase and then decrease several years before the outbreak of international conflict.24 The author of the World Handbook found international inequalities in the possessing of economic product to be larger than every one of twenty different intranational income distributions for which data were available. Coincident with cross-sectional and historical analyses indicating a trend toward decreased inequality within more highly developed countries, a provisional projection to 1975 showed a similar international trend. (Errors in these figures are, however, likely to be considerably larger than those discussed earlier because they rely on absolute figures and not logged intercorrelations.) Alker and Russett, in their previously cited work, studied degrees of ideological and socioeconomic foreign policy polarizations evidenced in United Nations voting behavior, contrasting them with polarization levels of consensus, and distributions of power in various national systems. Reflection about the substantive concerns, data resources, and methodological procedures of these and similar writings again raises a number of research possibilities. Putting the topic of similarities and differences among various levels of analysis in a somewhat more Lasswellian vein, the basic question is: Which collective characteristics (or identities) influence which individuals or group actors to behave in what particular ways toward which other objects and actors? Research possibilities of this sort clearly must rely on simultaneously available and comparable data from various levels of analysis, including both intranational, national and supranational units. Therefore it is gratifying to note that along with projects like the World Handbook and World Politics in the General Assembly, oriented mostly toward national units, there are a number of data collection efforts directed toward supranational and subnational units. 25 Along with aggregate data on different levels of analysis, we need methodologies for determining which actors and characteristics at which levels of analysis ought to receive our greatest attention as explanations of political and social phenomena. Specifically, we need to know to what extent our beginning and final units of analysis have subjective and/or objective validity,26 whether we can infer anything about one level of analysis from data available only for another one, 27 and how parsimoniously to identify foci of activity at different levels of analysis.28 Certainly for all of these purposes, many of the dispositional measurement techniques previously discussed would be of high relevance. Theoretically, higher order propositions explaining why some cross level hypotheses are true, while others are not, are very much to be desired. Ecological distor-

Aggregate political and social data

151

tions of individual behavior and tendencies for some variables to be approximately constant at sufficiently aggregated or disaggregated levels would both be plausible statistical reasons for such results. More substantively satisfying explanations would include various contextual factors intensifying or reversing cross-level relationships, or the greater availability of rewards associated with certain collective orientations.29 That the courses of war and conflict have been associated with individual, group and systemic explanations assures continuing concern among social scientists with different levels of analysis in order better to understand and promote desirable patterns of behavior. Developing causal theories of political and social change The above discussion of the limits of universality and on cross-level differences are both couched in theory-building language. Each problem in a way represents no more than a flexible compromise between the competing claims of universal, multi-level generality on the one hand and fidelity to the empirical details of political situations on the other. A more optimistic hope is that metatheories may soon become available that explain interregional and cross-level differences or similarities. In any case, data from a large range of situations, and at different levels of aggregation are an essential ingredient in making such research decisions. Clearly, our theory-building purpose would also be well served if we were more explicitly to include a time dimension in our analyses. Like differing regions and system levels, the various historical contexts underlying statistical relationships undoubtedly reflect differentially active causal variables. Historical comparisons are necessary because there always exist variables undetectable at a particular time, place, or systemic level. Another major argument for explicitly collecting time series data is the greatly increased precision and predictive power that temporally specific theories give to our analyses. Benefiting from a more clearly defined periodicity in their data, as well as more fully elaborated quantitative theories, economists have developed many of the tools and aspirations necessary for the study of political and social change. These include a causal modeling tradition, 30 a derivative econometric concern with interdependent, structural relationships,31 and the recent development of extremely complex and sophisticated computer simulation models.32 One partial application of economists' procedures may be sufficient to suggest their relevance to other social scientists. It concerns the use of four interdependent 'causal' equations that attempt partly to capture national political processes. The need for and utility of aggregated empirical data for the construction and validation of this kind of model should become particularly apparent whether or not this particular model is intuitively, impressionistically, or objectively satisfying.33 Figure 1 presents a partially verified theory of causal relations among some of

152

Hayward R. Alker Jr

FIGURE

1: A Developmental and Equilibrium Theory of Relationships Between Communist Voting and Democratic Government in Economically Advanced NonCommunist Countries*

A. Arrow Diagram

simultaneous Xlt a2lXlt +

political

interpendencies

+ X2t a 3 2X 2t + X3t a 4 2X 2t + a43X3t +

ai4X4t X«

earlier socio-economic conditions + + + +

+ bl2Z2t-2 + bl3Z3t-3 = b2lZlt-l = b3lZlt-l + b32Z2t-2 = =

Ult U2t U3t U4t

* Partial empirical tests of several theories related to the above are given in Hayward R. Alker, Jr., 'Causal Inference and Political Analysis', in Joseph Bernd, ed., Mathematical Applications in Political Science (Dallas, Southern Methodist University Press, 1966).

the major variables often associated with national political development: competitive opinion media, parties, and elections (in a word, polyarchy); high levels of popular political participation ¡levels of communist voting; and low levels of internal violence. Also included are variables such as popular income levels (indexed by per capita G N P data), urbanization percentages, and literacy rates, variables often thought to be related to political development. Controversy has repeatedly focused on the causes and effects of communist voting in economically advanced and non-communist countries. Many would argue it negatively affects political development; others would argue its functional role as a legitimized protest vehicle. The model in Figure 1 was designed to make more explicit and to test some of these arguments while at the same time taking into account social and economic factors frequently considered to influence political phenomena. Its central 'feedback cycle' hypothesizes that random increases

Aggregate political and social data

153

in communist voting will increase levels of polyarchy, which then lead to increased political participation and decreased domestic group violence; decreased violence levels in turn decrease and thus tend to restore communist voting levels to their previous proportions. Negative initial deviations, on the other hand, also lead to restorative feedback tendencies. As such, the model is a stable one to the extent that negative feedback compensates for initially 'destabilizing' tendencies. That communist electoral competition might in fact be stabilizing argues for the toleration of communist participation in democratic polities. Among 36 economically advanced, non-communist countries, the simple correlation between communist voting (in all cases below 25 % of the total) and an index of polyarchy is a small one. However, when we control explicitly for the posited causal relations between these as well as other socio-economic and political variables, we find quite a strong positive net effect of communist voting levels and polyarchical tendencies! The 'functional' interpretation of such expressions of discontent is thus partly supported. So is the 'stabilizing' view implicit in some theories of competitive coexistence within non-communist countries. Dynamically speaking, such models as this one have enormous potential for helping us to theorize about and measure political and social development and decay.3* Causal relations may involve negative (restorative) or positive (deviation amplifying) feedback; they can be used to model processes of equilibrium, growth, and decay. In particular, 'vicious circle processes', such as arms races, racial prejudice, or the growing gaps between rich and poor, are only a special class of dynamic processes susceptible to causal analysis. In combination with aggregative data, such models allow empirical verification, falsification and magnitude estimation of some of the major developmental hypotheses characterizing contemporary political theories and ideologies. Evaluating the performance of political and social systems Traditionally the social sciences have always included descriptive, explanatory and evaluative investigation. It would be an inaccurate reflection of the state of research in these areas, therefore, not to mention here some of the most exciting areas of theoretical and empirical work even more closely related to normative concerns than the examples already discussed. A classical kind of normative question concerns the relative 'merits' or 'efficiency' of different kinds of political systems. Such justifications usually appeal in part to facts, facts that are often controversial and likely to be distorted. Exploring and testing the empirical components of such arguments and making the evaluations more precise and explicit can help increase reality orientations among participants in such disputes. Take for example, claims that a certain kind of government, more or less plural-

154

Hayward R. Alker Jr

istic, more or less socialistic, has the advantage of producing or representing egalitarian, happy, and prosperous citizenry, //"data on such intranational phenomena as the number of individuals imprisoned or killed were available, along with evidence that clearly established the causal factors associated with these deprivations, e.g., political suppression or foreign intervention, a number of ideological arguments might require considerable revision. In the World Handbook, data on domestic group violence, on the average level of income, and the inequality within each country of the sharing of that income are most suggestive in this regard. A particular hypothesis of this sort for which relevant aggregative data have already been analyzed, is that 'the greater the degree of inter-party competition within a political system, the more liberal the social welfare measure that system will adopt'. 35 Drawing on earlier theoretical formulations by V.O. Key, Jr., and aggregate data on 46 American states, Richard Dawson and James Robinson find numerous positive rank order correlations (c. 0.60) among a variety of indicators that support the hypothesis. (The indicators include the percentage of times political control within a state has been divided between parties or dominated by one party, and levels of per pupil education expenditures, unemployment and old age assistance.) The most intriguing part of the Dawson-Robinson study, however, comes from the decision further to examine alternate explanations for their obviously impressive and normatively gratifying positive correlations. Thus they hypothesize and in fact find that economic development indicators (per capita income, urbanization and industrialization) are positively correlated with both the inter-party competition indices and the social welfare ones. At least their first finding, presaged by S. M. Lipset's work, indicated economic development to be a partial prerequisite for political democracy. The hypothesis then follows, and it is supported quite strongly by a reanalysis of the original positive correlations controlling for economic development, that the degree of inter-party competition does not possess a strong intervening influence between socio-economic factors and liberal welfare programs. That is, the originally hypothesized relationship is at least in part a spurious one because of the implicit double effects of economic development. Recent developments in game theory and activity analysis provide an even more ambitious framework with which to compare the relative 'efficiency' for realizing various values of differing politico-economic decision-making mechanisms. In particular, mathematical economists have proved some more or less startling equivalences among apparently different allocative procedures. When applied in a partially aggregated fashion to political systems, these findings help increase the comparability for evaluative purposes of differing forms and procedures of political organization taking into account the particular goals, resources and structural limitations ofparticular countries.3e

Aggregate political and social data

155

Summary and Conclusions This paper has ranged selectively over a number of substantive issues and related quantitative procedures that have seemed particularly appropriate for aggregate cross-national and cross-cultural analysis. Put simply, these research possibilities can be summarized as descriptive, explanatory, and evaluative tasks. Topics of discussion found both in the work of members of the Yale Political D a t a Program and in critical discussions of the World Handbook of Political and Social Indicators included latent national dispositions, geographical or cultural domains of generality, subnational through supranational levels of analysis, causal processes, and normatively evaluated organizational efficiency. In each case, a general class of theoretical problems was suggested as worthy of further research by those interested in aggregative analysis. Although by no means exhaustive, such topics were presented because their metatheoretical aspects became clear from reviews of earlier cross-national and cross-cultural research. The remarkable confluence of concerns among behavioral and social scientists, and equally auspicious developments in relevant quantitative methodologies may yet match the gravitational pulls of human limitations, of which Sisyphus was very much aware.

NOTES

1. Russett, Alker, Deutsch, and Lasswell, World Handbook of Political and Social Indicators (New Haven, Conn., Yale University Press, 1964). Other early publications related to early Data Program work are H.R. Alker and B. Russett, World Politics in the General Assembly (New Haven, Conn., Yale University Press, 1965); and R. Merritt and S.Rokkan, eds., Comparing Nations: The Uses of Quantitative Data in Cross-National Research (New Haven, Conn., Yale University Press, 1966), based on a September, 1963 conference sponsored by the International Social Science Council and the Yale Political Data Program. The Merritt-Rokkan volume is of particular value in suggesting ways of meshing survey results with nonsystematic observational information, content analysis findings, and aggregate data. 2. G.Ohlin, 'Aggregate Comparisons: Problems and Prospects of Quantitative Analysis Based on National Accounts' (see below, pp. 163-170); and R.Retzlaff, 'The Use of Aggregate Date in Comparative Political Analysis', Journal of Politics, 27 (1965), pp. 797-817. 3. A richly suggestive view of the evolution of scientific ideas about measurement is Harry Woolf, ed., Quantification: A History of the Meaning of Measurement in the Natural and Social Sciences (Indianapolis, Ind., Bobbs-Merrill, 1961). 4. J.S.Coleman, Introduction to Mathematical Sociology (New York, Macmillan, 1964), pp. 1-92. See also P. Lazarsfeld, 'A Conceptual Introduction to Latent Structure Analysis', in P. Lazarsfeld, ed., Mathematical Thinking in the Social Sciences (Glencoe, 111., The Free Press, 1955). 5. K.W. Deutsch, 'Toward an Inventory of Trends and Patterns in Comparative and International Polities', American Political Science Review, 54 (1960), pp.34-57;

156

Hayward R. Alker Jr.

'Social Mobilization and Political Development', American Political Science Review, 55 (1961), pp. 494-514. W. A. Gibson, 'Three Multivariate Models: Factor Analysis, Latent Structure Analysis, and Latent Profile Analysis', Psychometrika, 24 (1959), pp. 229-252. One of Retzlaff's most constructive suggestions in his previously cited review is that Deutsch's work on index constructions could be furthered through the use of factor analytic methods, themselves a special case fo latent structure analysis. That this was an intention of the World Handbook's authors is abundantly clear in the citations and discussions in its introduction. 6. See especially the Appendices of K.W.Deutsch, Nationalism and Social Communication, Revised edition (Cambridge, Mass., M.I.T. Press, 1966). A related but broader study of alternative paths toward regional or international isolation or interdependence is Harold Lasswell, 'The World Revolution of Our Time', in D. Lerner and H.D. Lasswell, eds., World Revolutionary Elites (Cambridge, M.I.T. Press, 1965). 7. The best theoretical introduction to this literature is K.W.Deutsch, 'Toward an Inventory...'. The statistics are concisely presented in L. Goodman, 'A Short Computer Program for the Gross Analysis of Transaction Flows', Behavioral Science, 9 (1964), pp. 176-86. Most recent applications are cited and discussed in more detail in H. R. Alker and D. Puchala, 'Trends in Economic Partnership in the North Atlantic Area', in a book obviously relevant to the present discussion, J. D. Singer, ed., Quantitative International Politics (New York, Macmillan, 1967). Alker and Puchala find a surprising degree of coincidence between changes in economic interdependence and variations in political integration. 8. H.R. Alker, Jr., 'Dimensions of Conflict in the General Assembly', American Political Science Review, 58 (1964), pp. 642-57. Variables of interest include (changes in) commitments to use and obey supranational peacekeeping capabilities, cold war and anti-colonial predispositions, and the extent to which various groups of nations fail to distinguish among these policy differences. The validity of the factor model for analyzing this kind of data, an important but infrequently asked question, is discussed in H.R. Alker and B.Russett, World Politics in the General Assembly, Chapters 2, 7, and 13, and in H. R. Alker, 'The Long Road to International Relations Theory: Problems of Statistical Nonadditivity', World Politics, 18 (July 1966), pp. 623-655. P. M. Gregg and A. S. Banks have applied Q-type factor analysis to Banks and Textor's Cross-Polity Survey and have found interesting distinctions among classes of national political systems in 'Dimensions of Political Systems: Factor Analysis of A CrossPolity Survey'. American Political Science Review, 59 (1965), pp. 602-614. Using factor analytic techniques A.W.Gouldner and R.A.Peterson, Notes on Technology and the Moral Order (Indianapolis, Bobbs-Merrill, 1962), have similarly distinguished Apollonian and Dionysian cultural patterns in cross-cultural data. 9. See Parts I and IV of Comparing Nations. W. Beckermann's monograph International Comparisons of Real Income (Paris, OECD Development Centre, September 1965, No 4), cited approvingly by Ohlin, is one of the best examples in the economic sphere of just this point. In order to construct and validate a nonmonetary index of real incomes, he used multiple regression estimation techniques and publicly available data on real consumption figures, population size, steel and meat consumption, cement production, domestic mail volumes, radios, telephones and road vehicles per head. Certainly each of these data series was collected for a different set of purposes, none of which was to help validate an indirectly calculated index of real consumption levels.

Aggregate political and social data

157

10. The issue of inaccuracies in World Handbook data has been somewhat confused by the failure to distinguish the relative magnitude and significance of transcription errors, choices as to desired variable transformations, failures of the original measurements to meet their stated purposes, and problems associated with using such more or less adequately processed information for purposes quite different from those of the original investigator. See, for example, Ohlin, op. cit.; Banks, 'Review of Russett, et al., World Handbook of Political and Social Indicators', American Political Science Review, 59 (1965), pp. 144-146; and Russett's reply, 'A Note on the Evaluation of Error and Transformation in Data Analysis', American Political Science Review, 59 (1965), pp. 444-446. 11. Recent partial results in these areas have not all been widely circulated. But just listing a few of these should suggest how much more similar research is needed. They include: a) Rummel's finding that error in some aggregate cross-national data seems to be related either to restrictions on free reporting or to low levels of economic development or to both of these factors; b) the demonstration by Blalock and others that factor analysis and regression analysis are equivalent to a model assuming 'independent, additive causes' behind the causally unrelated manifest variables (perhaps a natural way of thinking about the underlying causes of separate UN roll-call votes), while stepwise regression and simplex analyses are more appropriate when (as is done in some sections of the World Handbook) we want to assume that 'developmental causal sequences' underlie our data; c) Guttman's brilliant discovery of a partly automated multidimensional scaling procedure whereby ordinal proximity information can be parsimoniously reduced to interval scale maps revealing simplexlike and circumplex-like statistical relationships; d) Coleman's explication of the equilibrium assumptions necessary for giving cross-sectional results longitudinal significance; e) my own efforts in defining ecological regionalism to weaken the local independence assumptions of latent profile analysis to allow for locally significant statistical interrelationships; or the design by Sonquist and Morgan of a partly automated large-sample 'tree-building' program that can be used to detect subculturally distinct developmental sequences. 12. S.Rokkan, 'The Comparative Study of Political Participation: Notes Toward a Perspective on Current Research', in A. Ranney, ed., Essays on the Behavioral Study of Politics (Urbana, University of Illinois Press, 1962), refers to such questions as 'third order macro-micro' theory; Lasswell's many relevant discussions center on the 'contextual principle', e.g., The Future of Political Science (New York, Atherton Press, 1963). See also Erwin K.Scheuch, 'Cross-National Comparisons Using Aggregate Data: Some Substantive and Methodological Problems', in the Comparing Nations volume. 13. In the World Handbook I suggest that multiplicative universal models can sometimes be formulated that capture such intensification or reversal effects of economic or cultural advancement. See my Mathematics and Politics (New York, Macmillan, 1965), Chapters 5 and 6 for more examples and references to this kind of problem. 14 See D. T. Campbell and D. W. Fiske, 'Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix', Psychological Bulletin, 56 (1959), pp. 81-105; P.Lazarsfeld and M.Rosenberg, eds., The Language of Social Research (Glencoe, 111., The Free Press, 1955); and K.W.Deutsch, 'Communication Codes for Organizing Information', Behavioral Science, 11 (1966), pp. 1-17. 15. H.Usui and E.E.Hagen, World Income, 1957 (Cambridge, M.I.T., Center for International Studies, 1959).

158

Hayward R. Alker Jr

16. See Wilfred Beckermann, op. cit. Beckermann's study is one of the best examples of cross-national comparisons that I know of, precisely because it pays careful attention to index construction problems and because it experiments with regional differences among nonmonetary variables as estimates of real national income. 17. Ohlin comes to much different conclusions on the basis of what is apparently a more impressionistic analysis of the World Handbook. The two problems that 'most seriously impair the possibility and validity of comparisons among nations on the bases of (national) accounts' are 1) the problem of 'how to delimit economic activity in developed or underdeveloped market or planned economies; and 2) that of 'making real comparisons between incomes and product estimates for different economies with different price systems'. As to the first problem, 'no one can minimize the seriousness of this question' of measuring economic activity in underdeveloped countries in monetary terms. As to the second, Gilbert, Kravis, et al., found that the magnitude of error involved in using official exchange rates to estimate GNP's was something like 50% in comparing the United States with several European nations; while Kuznets found US and Chinese income ratios perhaps 100 % off. As citations to the work of Gilbert and Kravis and suggested error margins (which range from 3 % to 50 %) in the World Handbook indicate, such errors might indeed exist in absolute per capita figures. But a logged correlation (or even a ranked one) between Usui-Hagen data and non-monetary estimates of real economic product derived from the work of Gilbert and Kravis, reveals that the Usui-Hagen figures average within 11% of each other (in a linear logarithmic sense); intercorrelation errors are also roughly in the 10 % range. ApparentlyjUsui-Hagenfigures are considerably better than 'official exchange rate data' that both Ohlin and the World Handbook authors (but not Beckermann) confuse them with. The Lasswellian multivalued data organization scheme in the World Handbook is intended to suggest that national 'development' should not merely be evaluated in term of monetary activity, but also in terms of the volume and degree of sharing of political, social, familial, medical, religious, artistic, technological and intellectual accomplishments. Most certainly the authors of the World Handbook would want to be the last to underestimate Chinese or European achievements in any of these regards. 18. Ohlin worries about the validity of monetary indicators in underdeveloped countries and seems somewhat more justified (r = 0.89, r2 = 0.79 between the 'real' GNP and the Usui figures for an N of 12). But given the probable non-normality of the underdeveloped sample data and the larger measurement error undoubtedly involved in both variables, the result is still high enough for us to expect that total measurement error in an logarithmic-linear transformation sense is considerably less than 20 %. A much more serious and fascinating problem, as Table 2 indicates, is the radical difference in the pattern of correlations among economic and political indicators among underdeveloped countries. This suggests that even if accurately measured, economic development has causes and consequences among such countries greatly different from those affecting more highly developed ones. Relevant moderating variables possibly affecting these relationships include inequalities and social mobility in the national system. 19. Allardt's work, to cite a single additional example of this possibility already mentioned in the World Handbook, shows how regional factor analyses can be used to distinguish the structural bases of 'emergent' and 'traditional' radicalism: Erik Allardt, 'Patterns of Class Conflict and Working Class Consciousness in Finnish Polities', in E. Allardt

Aggregate political and social data

159

and Y. Littunen, Cleavages, Ideologies, and Party Systems (Helsinki, Westermarck Society, 1964). 20. General systems theory is perhaps the most ambitious approach of this sort, in trying to find propositions that hold for everything from physiological and interpersonal to international systems. Articles in General Systems, Yearbook of the International Society for General Systems Research, are of mixed quality to be sure. Impressive advances, however, have also been made, including W.R.Ashby, Design for a Brain (New York, J.Wiley, 1960); K.W.Deutsch, The Nerves of Government: Models of Political Communication and Control (New York, The Free Press of Glencoe, 1963); J. G. Miller, 'Living Systems: Basic Concepts', 'Living Systems: Structures and Process', and 'Living Systems: Cross-level Hypotheses', Behavioral Science, 10(1965),pp. 193-237 and 337-411; D.Easton, A Systems Analysis of Political Life (New York, J.Wiley, 1965). 21. See V.O.Key, Jr., Politics, Parties and Pressure Groups, 3rd ed. (New York, T.Y. Crowell, 1953); R.Dahl, Who Governs?: Democracy and Power in an American City (New Haven, Conn., Yale University Press, 1961). 22. G.Myrdal, Economic Theory and Underdeveloped Regions (London, Gerald Duckworth, 1957), Chapter 2. 23. Chadwick Alger, for example, has analyzed international systems using Almond's functional scheme for comparing developing national ones, and in the same article found a number of fruitful analogies with the politics of primitive societies. C.Alger, 'Comparison of Intranational and International Polities', American Political Science Review, 57 (1963), pp.409-16. See also R. Masters, 'World Politics as a Primitive Political System', World Politics, 16 (1964), pp.595-619. 24. M. Haas, Some Societal Correlates of International Political Behavior, Studies in International Conflict and Integration, Ph.D. thesis, distributed by Stanford University, Stanford, Calif., May, 1964. 25. A number of subnational archives and analyses of survey and aggregate data are described in Comparing Nations, especially Part IV, and in the December 1965 issue (Vol. 4, No. 3) of Social Science Information. Starting with plans developed in 1963 and 1964, the Yale Political Data Program has been interested in developing certain aggregate data archives for subnational units. Stephen V. Stephens, for example, has collected and is now analyzing several hundred data series on the American states with a primary interest in studying the structure of American federalism. Many of the studies collecting or generating aggregate date at the supranational level have already been mentioned. In addition to Quincy Wright's and L. F. Richardson's classic work on the causes, participants and violence levels of international wars, Raymond Tanter's ambitious attempt at Northwestern University to collect and analyze structural differences in over one hundred regional and universal international organizations is especially suggestive of further research possibilities. It should be noted that subnationally generated Gini coefficients for national income inequalities are some of the most interesting and suggestive data in the World Handbook. Clearly, it is not 'too much to hope', as Ohlin surmises, that subnational data will become available on the same scale as data in cross-national data bank programs! 26. See, for example, D.T.Campbell,'Common Fate, Similarity, and Other Indices of the Status of Aggregates of Persons as Social Entities', Behavioral Science, 3 (1958), 14-25; H.C.Selvin and W.O.Hagstrom, 'The Empirical Classification of Formal Groups', American Sociological Review, 28 (1963), pp. 399-411; and Coleman, Introduction, Chapter 14.

160

Hay ward R. Alker Jr

27. Leo Goodman's several innovations regarding cross-level inferences from ecological regressions are summarized in O. D. Duncan, R. P. Cuzzort, and B. Duncan, Statistical Geography (New York, The Free Press of Glencoe, 1961). See also R.Boudon, 'Propriétés individuelles et propriétés collectives: un problème d'analyse écologique', Revue Française de Sociologie, 4 (1963), pp.275-299; and H.M.Blalock, Jr., Causal Inferences in Nonexperimental Research (Chapel Hill, University of North Carolina Press, 1964), Chapter 5. These writers and Retzlaff all argue that ecological units often give meaningful comparisons. 28. Classical and modern analysis of variance and covariance techniques have frequently been helpful in this regard. See D. MacRae, Jr., and J. A. Meldrum, 'Critical Elections in Illinois: 1888-1958', American Political Science Review, 54 (1960), pp.669-683; D.E.Stokes, 'A Variance Components Model of Political Effects', in J.M.Claunch, ed., Mathematical Applications in Political Science (Dallas, Texas Southern Methodist University Press, 1965); L.Kish, 'A Measurement of Homogeneity in Area! Units', Bulletin de l'Institut International de Statistique, 31 (1961), pp. 1-10. 29. K. Waltz, Man, the State and War (New York, Columbia University Press, 1959). Similarly, James N.Rosenau, 'Pre-Theories and Theories of Foreign Policy', in R.B.Farrell, ed., Approaches to Comparative Politics and International Relations (Evanston, 111., Northwestern University Press, 1966), presents a number of intriguing hypotheses about the effect of national size, economic development, and political competitiveness as moderators of individual, group, and cultural factors influencing foreign policies. 30. Blalock's Causal Inferences in Nonexperimental Research is an extremely attractive introduction to the causal inference literature because of its lucidity, its concern with important substantive problems and its methodological sophistication in the areas of the theoretical validation, index construction, ecological and cross-sectional analysis. For these reasons Blalock's articles were cited in a number of places in the World Handbook as suggestive of exciting research possibilities. More advanced work, include Ando, Fisher and Simon, Essays on the Structure of Social Science Models (Cambridge, Mass., M.I.T. Press, 1963); H. Wold, Econometric Model Building: Essays on the Causal Chain Approach (Amsterdam, North-Holland Publishing Company, 1964) and R. Boudon, L'analyse mathématique des faits sociaux (Paris, Pion, 1966). 31. Work on 'interdependent' causal systems initiated by Haavelmo, and greatly enriched by Koopmans and others, departs from the hierarchical causal ordering assumptions of the Frisch-Tinbergen-Simon-Blalock approach, to which it reduces in a number of special cases. See, for example, S.Valavanis, Econometrics: An Introduction to Maximum Likelihood Methods (New York, McGraw-Hill, 1959); and L.R.Klein, R.J. Ball, A.Hazelwood, and P.Vandome, An Econometric Model of the United Kingdom (Oxford, Basil Blackwell, 1961). (My paper on causal inference, cited below Figure 1, illustrates and discusses both the causal ordering and the interdependent approach.) Fortunately, attractive model validation procedures exist for each of these approaches. Hierarchical models allow the investigator to read off falsifiable predictions from a large variety of models ; prior estimates of coefficient signs and magnitudes can be used for models (such as Figure 1) of the interdependent type. 32. G. H. Orcutt, M. Greenberger, J. Korbel and A. M. Rivlin, Microanalysis of Socioeconomic Systems: A Simulation Study (New York, Harper & Row, 1961); E.P.Holland with R.W.Gillespie, Experiments on a Simulated Underdeveloped Economy: Development Plans and Balance of Payments Policies (Cambridge, Mass., M.I.T. Press, 1963).

Aggregate political and social data

33.

34.

35.

36.

161

Constructing and validating simulation theories such as those developed by H. Guetzkow, in Alger, Brody, Noel, and Snyder, eds., Simulation in International Relations: Developments for Research and Teaching (Englewood Cliffs, N. J., Prentice-Hall, 1963) clearly requires enormous amounts of aggregated and partially disaggregated data on national political systems. Obtaining such data was one of the original purposes of Guetzkow's Dimensionality of Nations Project, now directed by Rudolph Rummel at Yale. For a progress report see Rummel's chapter, 'The Dimensionality of Nations Project' in the Merritt-Rokkan volume. Sources and index construction techniques are described more fully in my 'Causal Inference' paper; a detailed argument about these details is not my purpose here. Obviously, however, further specification of the variables, relationships and time lags appropriate to the model is desirable. One of the exciting things about quasi-econometric techniques - an obvious interest of all World Handbook authors - is the progress being made. Without any knowledge of new ways of entangling collinearities, relating cross-sectional and longitudinal results, ascertaining behavioral relevance, and knowing when and how to disaggregate established correlations, one might otherwise find Ohlin's quite appropriate list of problems in the area a bit depressing. Cross-culturally, a number of authors have studied developmental patterns using explicit causal inference procedures. H.Blalock, 'Correlation Analyses and Causal Inferences', American Anthropologist, 62 (1960), pp. 624-631, using correlation coefficients published by Driver and Massey, finds a direct developmental sequence relating matridominant division of labor to matrilocal residence to matricentered land tenure to matrilineal descent; he also finds evidence for a direct causal link between division of labor andland tenure patterns. See also R.Peterson, 'Simplex: A Mathematical Model for the Analysis of Social Change', The Sociological Quarterly, 5 (1964), pp. 264-271; and R.Naroll, 'A Fifth Solution to Galton's Problem', American Anthropologist, 66 (1964), pp. 863-867. R.E.Dawson and J.A.Robinson, 'Inter-Party Competition, Economic Variables, and Welfare Policies in the American States', Journal of Politics, 25 (1963), pp. 265-289. The quote is taken from p. 270. Relevant earlier work by V. O. Key, Jr., and D.Lockard include Key's Southern Politics in State and Nation (New York, Knopf, 1951); and Lockard's New England State Politics (Princeton, N.J., Princeton University Press, 1959). Using the many classifications of national systems, such as by Banks and Textor, Lipset, Almond and Coleman, similar international hypotheses are similarly testable. Of course different generalizations might be expected to hold at this level, but the validity of Lipset's proposition relating economic development and political democracy at both levels (see below) also suggests some generalizations will have cross-level validity. It is possible to interpret some of the most recent work on linear and non-linear programming in terms akin to classical arguments about political economy. Compare, for example, the chapters on the theory of demand, the firm and its objectives, market structure, general equilibrium and welfare economics, theory of distribution, inputoutput analysis, activity analysis, game theory and decision theory in W. J. Baumol, Economic Theory and Operations Analysis (Englewood Cliffs, N.J., Prentice-Hall, 1961); and Dahl and Lindblom, Politics, Economics and Welfare (New Haven, Conn., Yale University Press, 1950). Some of the most intriguing classical and recent results have stressed the economic

162

Hayward R. Alker Jr equivalence of'socialist'allocation decisions in accord with an overall objective function subject to resource constraints and 'capitalist' profit maximizing by decentralized firms aware only of their own technology and the market prices of the resources they use and produce. Baumol and Fabian, 'Decomposition, Pricing for Decentralization, and External Economies', Management Science, 11 (1964), pp. 1-33, for example, offer for the first time 'a systematic approach to the planning problems resulting from the presence of external economies and diseconomies (...divergence between private and social and costs and benefits) and the consequent limitations which characterize suboptimal decisions made independently by subdivisions of the economic unit in question' (p. 20).

CHAPTER X

Aggregate comparisons: problems and prospects of quantitative analysis based on national accounts GORAN OHLIN OECD Development Centre, Paris

Aggregate national statistics serve general purposes ot description and comparison and invite the testing of hypotheses about social systems and their change. Today such hypotheses tend to centre around problems of economic and social development, and there is clearly a strong demand for overall measures of economic activity, whether for purposes of assessing welfare, leveis of production, national power, or simply the 'stage of development'. Actual national accounts to supply such measures exist in only some seventy countries, and are passably reliable from a purely statistical point of view in less than half of them. Estimates have been published on a more comprehensive scale, notably by Usui and Hagen, and nothing is a better testimony to the thirst for this kind of information than the frequent use of their figures; they are also the ones included in the World Handbook by by Bruce Russett et al.1 Before I pass to the problems raised by this situation, I want to make a general observation about the nature of the indices available for national comparisons. Some national statistics are fairly straightforward physical data, and others are what Stein Rokkan has called 'process produced' outcomes of national administration, but even indices that habit makes us consider self-evident are often constructs that testify to the primacy of theory over observation. Consider total population - a simple magnitude in terms of its definition, but curious nevertheless in its addition of all members of a society: infants, children, and adults, men, women. The original interest in population was manifested by governments anxious to know numbers of able-bodied men. taxable households or some such thing, bur not the total population. Even today, the paucity ot information about population in, say, many African countries, reflects above all a lack of interest in this dimension of society.

164

Goran Ohlin

'Total population' expresses the most egalitarian notion of man, but this is not why it has imposed itself. I submit that it imposed itself principally as a result of the attempts to understand demographic change by studying births and deaths. From the beginnings of systematic demography it was obvious that it was convenient to introduce this exhaustive concept of scale which, unlike more arbitrary sub-groups, is also unambiguously defined by life and death themselves. Only in terms of demographic theory, conceiving of individuals moving through life, is the total number of coexisting individuals an obviously interesting magnitude. Other demographic variables: such as age, specific rates of fertility and mortality, only underline the point that theory and purpose have preceded the collection of data. The interest in life-insurance provided the impulse for the study of mortality while lively interest in fertility came only with its decline. International cooperation began early in the field of demography, and progress towards comparability was immeasurably advanced by the pressure of rationality and the needs for useful and internally consistent concepts to describe the demographic process. Now, in the case of national economic accounts, the situation is quite similar in one respect, and quite different in another. It is similar in the dependence on theory: the concepts of gross national product and national income have not been arrived at from a 'common sense' notion of income. Far more clearly than in the case of total population, they are intellectual constructs and they result from considerable advance in economic thought and in the analysis of economic structures. When the situation is different, it is because demographic theory is of universal applicability because the basic biological aspects of reproduction are quite uniform. There are nagging doubts, however, about the applicability of the constructs of national accounting, which have been elaborated in economically advanced countries, to 'underdeveloped' societies, and also there are very real problems involved in comparing national incomes even when the computations are standardized and the countries are similar. These and other problems of international comparisons of national economic accounts are well known, but they are so often referred to as mere sources of error that it is necessary to stress that their primary effect is to make the truth complex and not just to remove us from it. Actually, enormous strides have been made in the postwar decades towards conceptual precision and international comparability, but these also advanced our insights about the limits to such precision and comparability. This is not the place to enter into the technical problems involved in, for instance, the elaboration of the United Nations and the OECD systems of national accounts, nor would I be competent to do so. But I cannot pass over entirely the two problems which I think most seriously impair the possibility and validity of comparisons among nations on the basis of such accounts. The first is the problem of how to delimit economic activity. Product and income concepts require that a boundary line is drawn between economic and other activi-

Problems and prospects of quantitative analysis based on national accounts

165

ties, and in developed economies it is natural to draw this line in rather close accordance with the scope of the market economy. The treatment of the government sector raises problems that remain controversial, but outside of that sector it is only exceptionally thought appropriate to impute economic values to activities that do not enter the market, as for instance the rental value of owner-occupied housing or the unsold part of the product of one's 'own trade,' as the UN system puts it. In underdeveloped countries where the market plays a smaller part, imputation becomes necessary on a very large scale if income and product are to have any meaning at all. But if one imputes economic value to activities carried on within households in underdeveloped economies, international comparability would require a good deal more imputation, for instance of housewives' services, in developed ones. There is virtually no limit to the imaginable imputations if all services that individuals perform for themselves or others (or at least those that may be performed at a price) should be included in the concept of economic activity. Conversely, virtually all consumption, private and public, might be construed as a cost of production and excluded from final product or income. The concept of income is thus a convention, and the question is whether it is a useful convention for comparisons of developed and underdeveloped economies, or for that matter between market economies and planned ones. No one can minimize the seriousness of this question. Professor Frankel has gone so far as to deny even the possibility of measuring economic activity in underdeveloped societies in money terms which he thinks is 'to do violence to the governing principles of social organisation and evaluation in them', for the same reasons that make us abstain from measuring services performed for one another by members of a household.2 Irving Kravis, on the other hand, has argued that in spite of institutional differences all societies face the same basic economic problems, and has tried to indicate criteria for a distinction of economic activity, a dividing line between work and play, which would require the inclusion of many activities now left out of account in both underdeveloped and developed economies as well as the exclusion of some in the latter.3 Several years ago, Kuznets experimented with similar ideas after noticing that per capita incomes in the poorest countries were patently below the subsistence minimum in the West, which suggested that the borderline of economic activity had been wrongly drawn. As an illustrative experiment, he adjusted an estimate of Chinese national income in the early 1930's to compensate both for omissions in the Chinese accounts and for 'grossness' in the US concept, and concluded that the ratio between the US and the Chinese per capita level could be halved by such corrections.4 A related question concerns the treatment of foreign economic activity in underdeveloped countries where the 'territorial' and the 'national' product concept some-

166

Goran Ohlin

times differ by a factor of two or more. In the case of the petroleum economies this is not the only problem, but their peculiarities are well illustrated by the fact that in the World Handbook the series for GNP per capita is led by Kuwait with an income higher than that of the United States, and arrived at by an 'informed guess' about its oil production summarized in one sentence by its authors. 5 The second problem is that of making real comparisons between income and product estimates for different economies with different price systems. The international real product comparisons undertaken by Gilbert and Kravis and their associates at the OEEC in the 1950's indicated the magnitude of the error involved in using exchange rates for such comparisons, even among industrial countries. When the GNP of the United States and European countries were weighted by identical price systems, the choice of price system made a substantial difference the use of US prices resulted in a smaller differential than the use of European ones - but in both cases real per capita GNP in the European countries (relative to US) was much higher than indicated by a conversion at official exchange rates. The geometric averages of the former were about 50 per cent higher than the exchange rate conversion.6 A great number of attempts have been made in recent years to extend the scope of such real comparisons by different methods, e.g. by impressionistic corrections as in the figures of Rosenstein-Rodan, 7 by extrapolation of the discrepancies between exchange rates and purchasing power parities found by Gilbert and Kravis, 8 or by the use of non-monetary indicators of economic and social development. 9 The discrepancies between the tentative results of such work are enormous, and they do not seem to me to leave room for any complacency. The World Handbook faces the problem squarely and then dismisses it by citing the estimates of Usui and Hagen which were based on current exchange rates, expressing the pious hope that at least rankings will not be significantly affected. Even this is far from certain: the comparison of tentative estimates on different bases shows extensive reversals of rank. 10 No doubt more work will be done in this field in the near future, but at the moment the state of real income comparison on international scale cannot be termed very satisfactory. The usefulness of the available information obviously depends on the purpose: analyses of the structure of the national accounts are less impaired by the conversion problem. But for general comparisons of nations the presently available material seems ill suited, and I think Viner's warning against diffusing 'dangerous counterfeits of knowledge' must be taken quite seriously.11 As I have indicated, the problems of using national accounts for international comparisons are made particularly acute by our interest in the underdeveloped countries, and this quite apart from the poor quality of their primary statistical information. I think this focus of attention also has a bearing on the general problem posed by the use of the nation as a unit of observation. Doubts about this

Problems and prospects of quantitative analysis based on national accounts

167

practice are sometimes countered by the reply that after all nation states produce national statistics. Kuznets has also suggested that sovereign states do possess a distinct historical heritage and a unified government which influences their economic growth significantly, and that there is thus a positive case for considering the state as a unit of study.12 In many of the underdeveloped countries these cases are weak. In a 'dual economy' with a sharp cleavage between a small modern sector and a large subsistence sector, national averages may be quite irrelevant to whatever process of social change there is. On the criterion of historical legacy and unity, the reasons for considering the new states as individual units are often slight. Moreover, they often do not yet produce statistics in significant amounts, and the time to consider the desirable form of their statistics is now. Already a decade ago, the proposal was made that the subsistence sector should be singled out in national accounts. However, it is probably too much to hope that data for smaller units than nation-states would be available on the scale contemplated in data bank programmes. Nevertheless, it is one thing to use the nation state as a unit of study, and not quite the same thing to use it as a unit of comparison. I am profoundly ill at ease with correlation and multiple regression which treat each nation as an observation of equal weight regardless of whether it is a dwarf or a giant. Is there not a strong case for assigning weights in testing associations between different indices? The use of simple correlations between the indices in the World Handbook are, if I have understood matters correctly, intended as a convenient and computerized first step in the inspection of quantitative data. When it is so modestly presented, it is difficult to object to it in spite of the impropriety of using product-moment correlations on this kind of data. I must confess, however, that I found the results a bit disappointing and not sufficiently fertile in gestating hypotheses to warrant the continuation of such groping. Generally speaking, one finds that the variables in the 'economic-development cluster' are all closely associated, others much more loosely or not at all. This association between various reasonably obvious indices of economic and social development was discussed by Kuznets a long time ago. On that occasion, he stressed the dangers of translating close statistical association into significant causal relationships - 'a logical trap that should be avoided lest it lead to intellectual sterility and to a dangerously mechanistic approach to policy implications.' On the subject of income he added: 'The realization of the extent to which per capita income level is only a symbol of a whole socio-economic complex of conditions is important for any intelligent policy approach, as well as for analysis of causes.'13 No doubt the authors of the World Handbook would be the first to agree to such words of wisdom, but I cannot help suspecting that some practitioners of the new 'political arithmetic' might in their enthusiasm be led to forget them unless frequently reminded. When the correlations and associations within the economic-development com-

168

Goran Ohlin

plex of variables is very high, this does not, then, mean that we are any closer to an understanding of the process of development. On the contrary, it raises the acute and classical problems of econometrics of how to cope with collinearity. In which direction are we then moving on? Here I think there is a difference between the approach of economists and that of other quantifying social scientists concerned with the broad aspects of social change. Economists are surely inclined to proceed first to a more disaggregated level of the national accounts, to break down rates of growth, not only to a per capita level but to some 'per unit of input' level, to seek out behavioural aspects like savings-ratios, to combine a cross-sectional approach with that of time-series, to be suspicious of data from single years, etc. Kuznets's painstaking inventory of the quantitative aspects of economic growth of nations represents one form of this approach. 14 Econometric analyses are usually led into time-series analyses but a pertinent exception is Minhas's study of international trade in which the assumption is that trading countries share the same technological possibilities so that each country reveals one point in the production functions. 15 Those who seek to relate development variables to non-economic indices seem to proceed in the opposite direction in their use of national income figures: instead of disaggregating they combine them with others, either by reducing a great number of series into few by factor analysis, or by multiple regressions. The first of these techniques which constructs mysterious patterns or factors of a vague general character out of many well-defined series and does so on purely statistical criteria strikes me as so utterly alien to an economist's way of looking at things that 1 cannot say what, if anything, it contributes to our understanding, but this is no doubt because I fail to grasp it fully. 16 As for 'multifactor explanations' of social variables, such as those presented in the World Handbook for birth rates, mail per capita, etc. (pp. 311 ff.), 1 find them of very great interest, although I always wonder when a multiple-regression fit of this kind is to be considered good enough. Is an r 2 of .50, say, high enough to force us to take a model seriously? Nor do I like the nagging suspicion that a number of other variables might yield better correlations, even though they might be nonsense correlations. Presumably, these calculations are tests of hypotheses plausible from the start, but how conclusive are they? Statistical decision theory suggests, I believe, that the test for rejecting or accepting a hypothesis should take into account the gains from being right and the loss from being wrong. Perhaps the gain from a provocative hypothesis about broad relations between social variables is so great and the evil consequences of being wrong so small that it would be shame to let a beautiful theory be murdered by ugly facts if it can reasonably be avoided. There would then be reason to accept even models that explain only a small amount of total variance. But no proposition can be really important unless it is also important that it is not wrong, and although it may be superfluous I want to

Problems

and prospects

of quantitative

analysis based on national accounts

169

suggest that tests of significance become more necessary the more remote a model is from a relatively rigorous theory such as economics provides for market behaviour. However, many economic phenomena have no economic theory, and it is clear that any systematic attempt to link economic and noneconomic information is of enormous interest. All of the economic theory of public finance, for instance, does not begin to explain what determines the acceptable level of taxation in a country, and here the kind of socio-political analysis of nation-building that the Political Data Progam wishes to promote is obviously relevant. Similarly, all speculation about basic mechanisms in economic and social development tends to steer us towards noneconomic factors. Nevertheless, I believe the appropriate dimension for use of national economic accounts in such contexts is not GNP or national income but far more disaggregated, and I suspect that fruitful hypotheses in this domain will have to grow out of the consideration of micro-economic and partial aspects of the growth problem instead of relating themselves to an overall and more or less symbolic index of growth, such as that provided by GNP or national income per capita.

NOTES

1. Mikoto Usui and Everett E. Hägen, World Income, 1957 (Cambridge, Mass., 1959). 2. S.Herbert Frankel, 'Concepts of Income and Welfare in Advanced and UnderDeveloped Societies with Special Reference to the Intercomparability of National Income Aggregates', in Income and Wealth, Series III (Cambridge, 1953), p. 163. 3. Irving B. Kravis, 'The Scope of Economic Activity in International Income Comparisons', in Problems in the International Comparison of Economic Accounts, Studies in Income and Wealth, Vol.20 (Princeton, N.J., Princeton University Press, 1957). 4. 'National Income and Industrial Structure', repr. in S.Kuznets, Economic Change (New York, W. W. Norton, 1953). 5. Usui and Hägen, op. cit., p. 21. 6. Milton Gilbert and I. Kravis, An International Comparison of National Products and the Purchasing Power of Currencies (Paris, OEEC, 1954), pp. 22-23. 7. P.N.Rosenstein-Rodan, 'International Aid for Underdeveloped Countries', Review of Economics and Statistics (May, 1961). 8. J. P. Delahaut and E. S. Kirschen, 'Les revenus nationaux du monde non-communiste', Cahiers Économiques de Bruxelles (avril 1961). 9. M.K.Bennett, 'International Disparities in Consumption Levels', American Economic Review (September, 1951). 10. For a survey of the situation in this field and a bibliography, see Wilfred Beckermann, International Comparisons of Real Incomes, Development Centre Studies (Paris, OECD, 1966). 11. Problems in the International Comparison of Economic Accounts (Princeton, N.J., Princeton University Press, 1957), p. 395.

170

Goran Ohlin

12. Simon Kuznets, 'The State as a Unit in the Study of Economic Growth', Journal of Economic History, 9 (1951), pp. 24-41. 13. Simon Kuznets, Economic Change, p. 225. 14. See the series of papers on 'Quantitative Aspects of the Economic Growth of Nations', in Economic Development and Cultural Change, beginning in 1956. 15. B. S. Minhas, An International Comparison of Factor Costs and Factor Use (Amsterdam, North Holland Publishing Company, 1963). 16. Brian J. L. Perry, 'An Inductive Approach to the Regionalization of Economic Development', in N. Ginsburg, ed., Essays on Geography and Economic Development (Chicago, University of Chicago Press, 1960).

C H A P T E R XI

Aggregate comparisons: the validity and reliability of economic data PHYLLIS DEANE Cambridge University

When the United Nations Statistical Office produced its first international yearbook of national income statistics in 1948, it showed national income and expenditure data for 39 countries. The issue published in 1964 contained data for 107 countries, and for most of them there were growth data measuring the progress of national income, and national income per head, over the last decade or so. Today there is scarcely a nation state in the world which does not produce an annual national income calculation set up in internationally standardised social accounting from, on lines recommended, and kept under review, by expert international committees. This represents a truly remarkable triumph of international statistical co-operation, for national income statistics provide a distillation and interpretation of a complex collection of detailed statistical information on the country to which they relate. It is not surprising that social scientists in general have seized on them as a valuable type of datum against which to set their comparative assessments of social and political phenomena in different countries. There is some danger, however, that the achievement will be overestimated by those who want to take the figures as given and to use them uncritically. For they are not in themselves comparable. Most economists and statisticians who have either had a share in compiling national income statistics, or have had occasion to depend on them for analytical purposes, will be constantly aware of the fact that they are ambiguous measures of economic phenomena, and that their quality is a direct function of the statistical resources on which they are based. They are invariably estimates of quantities and values which cannot be - and indeed ought not to be rigidly defined for all countries and all uses. In some cases they are based on such inadequate statistical

172

Phyllis Deane

reporting systems that the publication of annual national income aggregates can be only the purest of window-dressing, if not a deliberate instrument of propaganda. It is the international abstracts of national income statistics that should be regarded with the greatest suspicion by the uninitiated, for they are full of undocumented pitfalls. There is a long list of reasons why the figures they show are not to be taken at their face value and are not strictly comparable with each other, in spite of the conscientious efforts of their compilers to make them so. It should be sufficient to mention a few of these reasons. 1. The problem of the degree of estimation The first is that the national income accounts are drawn up in a variety of ways and that abstracts of statistics give no hint of the methodological background to the individual figures. In some countries there is a co-ordinated statistical reporting system in which all or most of the basic statistics which go into a national income aggregate are collected with this end specifically in view. In others the members of a national income unit, or ad hoc research team, take responsibility for adjusting statistics, collected for quite other purposes, to the form required. Many developing countries lack economic statisticians competent to compile or use national income statistics. In these countries an estimating procedure developed originally by a national income expert, supplied on short term assignment in some technical assistance scheme, may be applied mechanically by statistical assistants who have no understanding of the meaning of their calculations, and often with no purpose in view other than that of making a stylised return to some international agency. Of course there are worse examples than this - for example the case where the need to supply politicians with 'evidence' of economic growth, or to persuade aid-giving agencies that their donations are having a satisfactory effect on economic performance, determines the nature of the statistics produced. But the innocent errors and misconceptions are almost certainly more common, and quite as misleading in their results, as the deliberate manufacture of false statistics. 2. The problem of defining economic activity A second reason why the published aggregates are not comparable is that they differ in definition as between countries. This is inevitable and often desirable. Economists define the concepts they want to measure in the light of two considerations - the analytical use to which they intended to put them and the quantitative data it is feasible to collect. Both would normally differ as between countries. For example, whether the rents attributable to owner-occupied housing should be included as part of national income depends partly on whether the users of the accounts regard

The validity and reliability of economic data

173

this as a useful item to measure, and partly on whether there exists enough information to make an estimate possible. In a country where the bulk of rural housing consists of mud huts, built by their occupiers and neighbours on a mutual aid basis, it is highly unlikely that it would be possible or sensible to attempt an estimate. In developing countries which are in the process of transition from a largely-subsistence to a largely-exchange economy, there exists a wide range of economic activities which might be thought eligible for inclusion in the national income accounts in some countries and not in others. It is not possible to prescribe international standards to cover all these possibilities. Nor is it possible for the compilers of international abstracts either to specify them all in footnotes or, if they can identify them, to adjust the figures: for they do not have the basic information with which to make such adjustments. There are also difficult problems of definition which arise for the accounts of developed and developing countries alike. The definition of capital expenditure, for example, or of depreciation may be subject to accounting conventions concerning the line drawn between repairs and new investment, or concerning the expected length of life of capital assets, which may vary substantially as between countries as well as between national income investigators or users.

3. The problem of defining the economy So far I have been considering differences in the possible definitions of, or classification of, particular economic activities. Ambiguities may also arise in defining the limits of the economy which may be dealt with differently in different countries or even at different periods within the same country. Should one evaluate the oil exported by a producing country inclusive or exclusive of the profits of the company distributing it on the world market? Are the incomes of immigrant workers or businessmen or soldiers to be regarded as part of the national income of the country in which they are legally domiciled or the country in which they are working? In the year when some of the big international copper companies operating in Northern Rhodesia changed their country of domicile from the United Kingdom, the national income of Northern Rhodesia went up by 25 per cent though little had changed except the companies' formal relationship with the tax collectors. Algeria in the 1950's had one of the highest rates of national income growth in the world because the income of French soldiers who were then pouring into the country were included as part of the national income.

4. The problem of the average A further reason why international comparisons of national income aggregates, and particularly national income aggregates expressed in per head terms, may be

174

Phyllis Deane

grossly misleading is that the extent of the aggregation problem differs for different countries. The significance of an income average depends of course on the dispersion of actual incomes about that average. In countries where incomes are unequally distributed, the average may have little meaning as an index of either the welfare or the productivity of the typical inhabitant. The extreme example of this problem of aggregation is of course the little oil state of Kuwait, which is one of the richest countries in the world in terms of national income per head although its inhabitants are amongst the poorest in the world. 5. The price problem Finally, probably the most important reason for mistrusting the validity of international comparisons based on the national income estimates which appear in international abstracts, is what might be called the price, or weighting, problem. A national income calculation represents a summation, in money terms, of all the economic goods and services produced by the economy in question. This summation is made by applying a set of money prices to the heterogeneous mass of real goods and services produced. It is well known that different individuals put different values on these goods and services, but for the purposes of a national income calculation it is convenient and meaningful to accept the valuation of the national market as expressed in the prices of a particular year. If we want to compare the national incomes of two countries, however, we have to apply the set of prices prevailing in one of them (or possibly an artificial set of valuations representing an approximation to an 'average' for both) to both of them. The prices in effect constitute a method of 'weighting' the quantities of goods and services. If the value systems of the two countries are similar (which generally means that they are at a similar stage of economic development) the comparison may give similar results whichever set of prices is adopted: and in this case one might be able to approximate the result by the short-cut method of applying the exchange rate between the two countries provided that the prices of goods entering into international trade reflect faithfully their prices in the home markets. But if the two countries differ at all significantly in the relative valuations they put on the goods and services they produce (which need not necessarily mean that they are at different stages of economic development, only that tastes differ in the two countries) then it is possible to get contradictory results from applying each of the possible systems of prices weights. In other words, if we make the comparison in terms of the prices of country A we may find A has the higher average national income and if we make it in terms of the prices of country B we may find B has the higher income. Moreover, in a world where the prices of goods entering into international trade are frequently distorted by subsidies and tariffs, or where currency values are influenced by systems of exchange control, there is no reason to expect the rates at which national currencies

The validity and reliability of economic data

175

exchange to have any relation whatever to their purchasing power in domestic markets. Conclusions What all this amounts to is not that social scientists should abandon altogether the use of national income estimates as key economic indicators, nor that international comparisons of national income data are completely impossible. One must conclude, however, that since national income estimates are not so much factual records as interpretations of incomplete economic statistics, they cannot be accepted without reference to the viewpoint of the interpreter and the quality of the underlying statistics. It is safe to say that a national income statistic that is worth considering as an economic datum will always appear originally in a source containing an explanatory text specifying the definitions and methodology of the estimates. The user who does not go back to the original source is dealing in unknowns. A second conclusion is that international comparisons of national income data whichdomorethanillustratethe obvious, which seek, for example, to rank countries in a definite order, cannot be read off from unprocessed tables of national income aggregates. An international comparison that is worth anything at all is the endproduct of a specific research project. Examples of researches which have yielded acceptable results in this field are those which were produced under the auspices of OECD a few years ago comparing national output, productivity and purchasing power for certain European countries.1 They involved a detailed re-processing of some of the raw statistics on which official national income estimates were based in order to bring them into internationally comparable form. A great deal more research needs to be done in this field however, before one can list the majority of the countries of the world in rank order of either their productivity or their standards of living.

NOTE

1. See for example Milton Gilbert and Irving B.Kravis,

An International Comparison of and the Purchasing Power of Currencies (Paris, OECD, 1954); Milton Gilbert and Associates, Comparative National Products and Price Levels (Paris, OECD, 1958); and Deborah Paige and Milton Gilbert, A Comparison of National Output and Productivity of the United States and the United Kingdom (Paris, OECD, National

1959).

Product

CHAPTER XII

The cross-cultural use of sample surveys: problems of comparability ERWIN K. SCHEUCH University of Cologne

1. Progress as a new problem

identification

A review of publications based on cross-cultural surveys makes it at first appear doubtful that there has been definite progress in research methodology and practice.1 Published research certainly does not show a neat pattern of continuous ascent to ever higher levels of methodological sophistication. However, I hope to show that nevertheless such progress exists in a specific - although somewhat frustrating - way. In surveying surveys, one can point to some technological innovations. I think, however, that the major progress has been to increase the awareness of the real sources of difficulties. In this way, the use of surveys in cross-cultural comparisons influences our understanding of research techniques and methodology in general. This is then my main theme: progress as changes in the awareness of problem areas, as the spreading realization that an earlier identification of the sources of difficulties was much too simple. Up to this day, the difficulties encountered in cross-cultural comparisons tend to be perceived as problems of research technology. However, it is now increasingly being realized that the main problems are methodological in the more limited sense of this term. This change in the perception of problem areas is being extended also to the somewhat self-contained field of survey research. Cross-cultural surveys are beginning to be viewed as just one technical form of the method of comparison, and under this perspective difficulties in research technology often lead to the realization that actually both method and theory may be wanting. Regress to theory and method may still frequently permit the solution of problems that, understood

Cross-cultural use of sample surveys

177

merely as problems of research technology, were difficult to handle, and this in turn will advance substantive theory. This orientation is just beginning to be acceptable - and perhaps this statement is more an expression of hope than a description of reality. In the cross-cultural use of surveys, the standards of technical perfection are already below the standards for good work within a particular country. And in the phase of interpretation it sometimes seems as though cross-cultural comparisons via survey research were less a source for new insight and more an empirical buttressing of pre-conceived notions about differences among social and political systems.2 However, this is merely a criticism of the behavior of researchers, and not an indictment of a method. Quite the contrary: if a better understanding of methodological problems in the cross-cultural use of surveys is achieved, this should lead to an immediate and dramatic advancement in our social science knowledge.

2. Changes in the identification of problem areas The type of cross-cultural research we are discussing here, and its problems, are mainly a phenomenon of the period since World War II. However, this should not lead to the frequent misperception that cross-cultural research as such is something new. In the 19th century sociology was largely comparative in orientation, relying both on historical and on cross-cultural comparisons. During the second half of that century a major empirical basis for sociological treatises was material collected by ethnographers, and while this material was cross-cultural by virtue of the subject matter, it remains significant that sociological theory was then understood to require empirical information from this inherently cross-cultural discipline. And what little social research proper was carried out then was likewise comparative in orientation. When Le Play collected family budgets to analyze the structure of the family, he checked his findings in France against data from Germany as a matter of course ; 3 as a matter of principle, Durkheim was not satisfied to test his theorems on anomie just by data from France: in his Rules of Sociological Method he declared the comparative method central to sociology. In this methodological orientation, Durkheim considered himself a direct descendant of John Stuart Mill; like Mill he maintained that the social sciences were essentially observational disciplines, and that in disciplines of that character, systematic comparisons were needed as an analogue of experimentation. Cross-cultural comparisons were of course not the only expression of this methodological orientation, but were widely considered to offer evidence of an especially convincing kind. The present cross-cultural surveys are not a direct continuation of this early emphasis, and this becomes obvious in the current methodological discussions on comparative research. Comparative sociology, and especially comparative empiri-

178

Erwin K. Scheuch

cal research, had to be discovered all over again. The very development of empirical research as we understand it today seems partly responsible for this hiatus in tradition. The Chicago school of social research (though not so much its founders, as, e.g., Robert Ezra Park) was fascinated with its own immediate environment. It placed an emphasis on the immediately observable, and developed techniques particularly suited to demonstrating within-culture variations. This social research was problem-oriented in the sense that 'muckraking' was. Quite understandably, a strong connection with social work came about, and in turn social workers were most influential in developing some of our techniques.4 This is especially true for the prevailing rules of procedure in interviewing, the set of 'do's and don'ts', something which I would like to call the 'folklore of interviewing'. The same exclusive concern with one's own society was of course characteristic of consumer research and opinion polling - two major sources of survey technology. In the familiar and unreflected environment of one's own society, a methodological reflection on the problems of relating sensory impressions to generalized statements appeared unnecessarily 'theoretical' (after all, the evidential meaning of indicators appeared obvious). Progress in research was then understood as a more reliable collection of observations. Given the further fact that empirical research was no longer carried out mainly by the scholar himself but that - for a variety of reasons - the collection and later the analysis of sensory impressions had to be delegated to relatively unskilled helpers, an emphasis on the codification of procedures was natural. This emphasis was in my opinion a real progress. During the earlier prevalence of more 'fundamental' questions, the reliability and empirical validity of actual research procedure was without doubt treated too lightly. Nevertheless, in reflecting about current cross-cultural comparisons, it is important to realize that during the late twenties research methodology became largely the discussion of research technology. As such, it was later exported to other countries. This emphasis on technology is probably nowhere more pronounced than in such a highly codified procedure as survey research today. Thus, the methodological problems of cross-cultural surveys were, and are still, seen primarily as problems of comparability only in a narrow sense. How to design a sample that in its concrete form can be used in all countries to be studied? How to standardize field work procedures internationally? How to ensure that questions are well translated? How to standardize categories, e.g., for background data? These are the prevailing ways of identifying problems in surveys outside one's own culture. Some progress has been made in answering such questions, and this experience seemed to call for fewer adjustments in our technology than was originally expected - e.g., one can conduct an interview in Europe pretty much in the same way one

Cross-cultural use of sample surveys

179

had learned to do in America. Actually, one of the most surprising experiences in doing cross-cultural survey work is that so much of the technology of the datacollection phase appears transferable. Yet this happy experience obscures the fact that in a more strictly methodological sense the difficulties of cross-cultural surveys are indeed quite formidable. Genuine differences in social structures exert themselves even if they permit a partial transfer of one's technology. Becoming aware of this forces us to spell out assumptions we do not need to spell out when working within our own culture. Intuition and research folklore are not necessarily a great help in analyzing data not from one's own society. In handling technical problems in cross-cultural surveys - and especially in the secondary analysis of survey material - we have once more to change from research technicians into methodologists. I now want to draw on some concrete examples in order to show how a better awareness of type of problems can help to turn the cross-cultural survey into the powerful tool for a generalizing social science that we have every reason to expect it to be. 3. Problem area I: Question meaning and problems of verbal communication Question wording was the first problem that attracted attention as the supposedly major difficulty in international surveys. However, if one contrasts the discussion of the Time International Survey in the 1948 volume of the Public Opinion Quarterly with the summaries of the 13th AAPOR conference in the same journal ten years later,8 a shift in emphasis will be obvious. The first concern had quieted down, and standard techniques were developed that gave researchers the feeling that they were in control of the difficulty. The situation is, by the way, analogous to the decline in the excitement originally generated by Hadley Cantril's Gauging Public Opinion, when he reported dramatic examples of the effects of non-neutral question-wording. The chief technique (especially for academic researchers) for controlling the correctness of translations has been borrowed from cultural anthropologists. Ideally, a foreign bilingual translates a questionnaire into his language and a bilingual from the researcher's home country prepares a retranslation; this version is then checked against the original wording. While this is undoubtedly a fine technique to check the ability of translators, it does little to control the chief problem in question wording: equivalence of meaning. Commercial researchers tend to be more sophisticated here and to take less refuge in the comfort that they were able to achieve literally or idiomatically correct translations. By now, a commercial institute making an international survey quite often has a master questionnaire drawn up by the particular unit handling the client. The individual participating agencies then translate and modify (!) the

180

Erwin K. Scheuch

instrument. If time and money permit, a general conference is called, where changes in questions are discussed that appeared necessary to achieve equivalence of meaning. Essentially, this is a qualitative procedure in which the individual researcher uses his best judgment as a criterion. However, in too many cases there has been insufficient awareness of the source of difficulties when developing equivalent forms of questions. These difficulties arise from the relation of language to reality, from structural differences between societies, which are reflected in translation problems. I. Let me point to some experiences which led me to postulate a number of problem factors; these are presumed to account for most of the difficulties experienced in question wording for international surveys. If we should succeed in identifying the main problem factors in international surveys, we should at the same time learn something more about the meaning of questions in general. In 1954 in Italy and France, Gabriel Almond found the term 'Communist Party' referred to quite different political entities and had a different content for the respondents. 7 If the term 'communist' is used in questions asking for membership and voting, this difference in the real meaning of the term obviously does not cause problems; but if a question using the term 'communist' is used as an indicator for a system of attitudes and beliefs, problems do certainly arise in a cross-cultural comparison. Using a semantic differential, Hofstatter explored the meaning of the word 'lonesomeness'. 8 He could show that this English word did not correspond to the presumed German equivalent, 'Einsamkeit'; even between the American and the British usage there were strong differences in the meaning of this term. An attitude test that included the term 'lonesome' would undoubtedly have a different indicator meaning in a cross-cultural comparison. Ruth Benedict found that she had to use a number of Japanese terms in order to account for the English term 'duty'. This signified that duty in Japanese society is probably no abstract norm, but denotes specific obligations in a particular relationship. 9 Conclusion 1. The concepts behind words are often delineated differently in different languages; the more abstract the concept, the greater the likelihood of differences. To manipulate the surplus emotional meaning of words that are presumed to have a straight cognitive meaning is one of the prime skills in question phrasing, but becomes a major problem in cross-cultural work. When we adapted the Bogardus social distance scale to Germany, we found that the term 'neighbourhood' carried distance implications that were not equivalent to the cognitive equivalent 'Nachbarschaft'. Similar problems were encountered when using the term 'your community'. While in most societies the subject matter referred to by the term 'socialized

Cross-cultural use of sample surveys

181

medicine* can be denoted by a value-free term, in the US there seems to be no simple term left any more that does not tap party preferences. Conclusion 2. Terms may have an emotional meaning only in one society and be quite technical in another. Questions referring to 'fair play' are extremely hard to render in other languages; it is the combination of the concept 'fair' with 'play' that, e.g., makes it impossible to give an adequate rendering in German. The same is true for the words esprit or Beamier. Conversely, there is no general term for husband in Japanese. In a technical sense, problems arising from terms unique to a language or the absence of a term in another can usually be solved by word combinations; this is obviously not true for the emotional implications. Conclusion 3. That certain words are unique to a language or altogether absent from it may signify that the phenomenon to which they refer is unique or absent. As Blood has pointed out, the discovery of unique terms and 'linguistic blanks' is an important substantive finding that should not be treated as calling for mere technical ingenuity.10 In the International Citizenship Survey by Almond and Verba,11 one of the questions was : 'Here are some important problems facing the people of this country. Which one do you feel is most important to you?' One of the choices offered was 'spiritual and moral betterment'. The combination of this with 'a country's problems' in the context of a political interview seems plain silly in Europe (and in a cognitive sense it may also be silly in the US, campaign promises of Mr. Goldwater notwithstanding). In the same survey respondents were asked to choose from the following statements in order to describe their feelings when going to the polls : '1. I get a feeling of satisfaction 2. I do it only because it is my duty 3. I feel annoyed - it's a waste of time'. The implication of this ordinal scale, that a hedonistic component is a very meaningful dimension of political participation - 'a feeling of satisfaction' - is in most European countries quite unwarranted; and in these countries joining the two statements 'I feel annoyed' and 'it's a waste of time' as equivalent expressions of displeasure is very problematical. To ask a Japanese respondent 'Where do you go on your vacation?' makes little sense since only the élite has the privilege of going on vacations. Conclusion 4. Questions that presuppose the combined existence of factors may - however well worded - appear quite silly when transported into other societies. Questions have to be 'realistic' in wording.

182

Erwin K. Scheuch

In some of the developing societies (and sometimes among faculty members) direct questions are considered improper or even threatening. In a methodological study we could show that bifurcated vs. open-ended questions are differentially suited to the style of reacting to stimuli with uneducated vs. highly educated respondents. Closed questions, which in effect impose upon the respondent the conceptual orientation of the researcher, can be rather risky in cross-cultural comparisons. The very notion that a question is a stimulus to obtain cognitively used information from a 'subject' was found to be alien even in parts of Western European populations. Conclusion 5. The format of questioning may carry cultural implications. Researchers in technologically and economically underdeveloped societies report that it is much more difficult to find words that are high in cognitive meaning than is true is Western industrialized societies. Specifically, it is presumably very hard to find terms for social roles that do not carry undesirable status connotations. Thus, it is reported for Java that there are sixteen terms for the wife of a partner in conversation, depending on whether the questioner is younger or older than his partner, lower or higher in status, a stranger or an acquaintance. A somewhat extreme example of the status implications of presumably cognitive terms is again to be found in the Almond-Verba study.12 In order to test partisanship, the respondents are asked how they would feel if their child married across party lines. Due to the high degree of association of parties with certain occupational and status groups in several European countries, the question has in effect the 'surplus meaning': 'How would you feel if your son or daughter married someone of lower, equal or higher status?' Consequently, adherents of higher status parties object much more strongly to their children's 'marrying down' than is true for the reverse. The degree of 'surplus meaning' which party labels carry is obviously a function of the closeness with which particular parties are associated with specific occupational groups. For an American it is probably difficult to perceive that party preference may be an indicator both for preference for a political content and for self-identification with occupational status groups. Conclusion 6. A language may be so constructed as to contain few terms referring to roles and group affiliation that would not carry strong status implications. In general, the cognitive loading of such terms appears to be lower, the lower the degree of cross-cutting group membership. Stern and d'Epinay report from Switzerland that, while the written language is uniformly more or less 'High German', the spoken 'Schwitzerdeutsch' has strong

Cross-cultural use of sample surveys

183

local differences.13 Accordingly, questionnaires are sometimes drafted in a sort of basic 'Schwitzerdeutsch' where the interviewer is then free to decide on the local idiom to be used (e.g., Ziirideutsch or Berner Deutsch). This is not just an example of a widely used technique in multilingual nations (such as India); High German is understood by all and used by most respondents. However, High German is largely restricted to mass communications and public functions, and smacks of 'officialese'. In a survey dealing with sexual behavior and attitudes we experienced major problems from the lack of a familiar language for this topic, a language that was neither vulgar nor medical. However, for the great majority a language dealing with sex in a matter-of-fact way is not available. Conclusion 7. The very choice of a language (or level of discourse) may have implications that are non-existent in another culture. Not in all societies is there a universal colloquial language, and even in countries where this is true, the language of colloquial discourse does not extend to all topics. It is hoped that these generalizations may be useful as caveats. Hopefully, they are also suggestive of a better understanding of the question-and-answer process within a particular society, where due to group differentiation analogous problems arise - although undoubtedly much less dramatic in impact and consequently less visible. In actual cross-cultural survey work, experiences such as those mentioned above have naturally led to an emphasis on remedies for the problems arising from lack of comparability in wording. II. In designing questionnaires for cross-cultural surveys, the emphasis has changed from insistence upon the correctness of literal translations to comparability of meaning. Usually, a purely qualitative procedure is used to achieve such comparability: the researcher in charge of a project informs his colleagues what he had in mind when phrasing a specific question, e.g. when he notices a non-literal translation of his original formulation. If time and money permit, the preferred procedure is a questionnaire conference after the first round of pretests. Of course, these qualitative checks are least effective if a question offers no problems in a literal translation. What is urgently needed as a routine procedure is a circulation of manuscripts in the analysis phase, affording cooperating researchers a chance to object to a questionable interpretation of previously unproblematical items. Unfortunately, the need for comparability of meaning is usually so far only perceived as a problem in the stage of design. We think indeed that comparability of questions as a problem poses itself especially in the analysis stage, and needs to be tackled here, too. (If the later notion of 'equivalence of indicators' is accepted, concern with the problem at this stage does not amount to crying over spilt milk.) However, so far there has been somewhat more interest in some devices which are to ensure comparability during

184

Erwin K. Scheuch

the stage of design. Notable here is a suggestion by Osgood to check comparability of meaning by using the semantic differential.14 Unfortunately, this is much too cumbersome a process to be used except for some key terms; even here it is, of course, only useful in checking the emotive connotations. Somewhat greater has been the practical impact of suggestions to use non-verbal devices in international surveys. Such instruments as the scalometer are by now real fads in cross-national studies. Hadley Cantril's study, 'Hopes and Fears for Self and Country', is the latest example; here one chief instrument was graphic.15 Some of these techniques are undoubtedly quite useful-but they are only a limited answer to our more serious difficulty: achieving equivalence of meaning. After all, visual stimuli are also a sort of language, and thus are not completely free from the problems mentioned before. An example is Cantril's latest multi-nation study, where he uses as one stimulus a ladder.16 This picture - or rather the device it refers to - was unknown to the Zulus. Subsequently, he employed with good success a picture of climbing successive terraces of a mountain, since ascending mountains was a widely shared experience. It is true that many elementary errors are still committed when transposing questionnaires into other cultures. Experienced researchers, however, can usually handle intuitively the problems of developing an instrument that will be a good •free translation', and will be reasonably easy to administer. The latter may be a very misleading criterion. One of the major problems in survey work - magnified in cross-cultural comparisons - is the fact that questions are answered even if they mean quite different things to different respondents. This in itself does not invalidate surveys by any means. Obviously, it is not the faultiness of an instrument as such that causes worry, but the degree of faultiness as measured against the uses made of the responses. This simple consideration is widely overlooked: the frequent European critiques of comparative research via surveys advance unrealistic standards of the needed perfection of empirical instruments; the practitioners of survey research tend to rest assured with the experience that most of the predicted difficulties just did not materialize and that one almost always gets 'results'. I do think that the meaning of the many difficulties we experience in question wording for cross-culturally comparative surveys is inadequately understood; I also think that there has been considerable progress in handling the practical problems. Beyond this advance in practices, I think there has been a major progress especially in research that is administered by commercial institutes: a willingness to depart from literal translations and to aim at equivalence in meaning. This I think is a major step - a step, however, that should be followed by the understanding that equivalence of meaning is only a special case of the general property of 'functional equivalence of indicator meaning'. This, however, presupposes a different outlook on the question-and-answer process in survey interviewing.

Cross-cultural use of sample surveys

185

4. Problem area II: Equivalence of indicators There is still, however, a good deal of uneasiness about putting equivalence of meaning before identical wording, especially in academic research. After all, haven't we learned what seemingly minor variations can do to the responses? Thus, the prevailing orientation is one of rather naive realism: if a question is not asked in the same way, we cannot compare results; if it has been asked the same way, we can. This may be defensible folklore of research, but it certainly is not good methodology. By now social scientists should have become accustomed to looking at questions as indicators - indicators that have a probabilistic relationship to a property one intends to measure. Indicators are interchangeable in terms of their functions, which are to express the property we want to ascertain. Hence, the criterion for maintaining that questions are comparable is not whether they are identical or equivalent in their commonsense meaning, but whether they are functionally equivalent for the purposes of analysis. I see major progress in cross-cultural survey work in that this notion is now beginning to be more generally accepted. Hopefully this will translate itself into research within one country. I. Functional equivalence may be an exciting concept, but it is of course somewhat difficult to apply in actual research.17 As a matter of fact, one would either need to check this property empirically, or have to use an empirically supported substantive theory - preferably both, if they are available. To give some examples: 1. As I mentioned above, several years ago in Germany we tried to adapt the Bogardus social distance scale. We found that, in Germany, the concept of neighborhood carries a different (less stringent) implication in terms of social intimacy. Thus, we had to try to find another item indicating the same position on the SevenStep Distance Scale. Of course, the criterion we finally settled for - 'Have as a greeting acquaintance' - did not have exactly the same position on the latent continuum (in terms of distance from the end points), but it did have the same ordinal position, and the total scales were equivalent. A somewhat different problem was encountered some twenty-five years ago by Stuart Dodd in the Near East, when he found that the Bogardus social distance items did not cover the total range of possible enmities.18 As you may remember, Dodd finally came up with an item 'I wish somebody will kill all these people' as an adequate measure for the total range of actually existing distances. 2. As just mentioned, Cantril had to use different versions of the scalometer in cultures as different as the Zulus and the United States.19 He encountered another and more difficult problem with his verbal stimulus: to imagine one's best conditions and worst conditions, five years hence. The ability to project hope and fear into the future for definite time intervals was - not surprisingly - quite culture-specific.

186

Erwin K. Scheuch

In the US, five years indicated a planning span familiar to respondents in their everyday life; in some primitive societies, a season, or at most a year, was the timespace in which respondents normally operated. Cantril finally resorted to a more variable time-reference to achieve equivalence of function: a referent for fears and hopes. In a cross-cultural survey comparing role differentiation between husband and wife in the United States and Japan, a battery of household tasks had to be translated from English to Japanese.20 Of the eight items used in America, only one could be used literally, and one further after considerable rewording. For the six other items, the activities did not occur in Japanese households or meant different things there, and functional equivalents had to be found. 3. In a cross-cultural survey of the goals and methods of child-training it was found that rates of physical punishment varied among cultures and within societies among classes.21 It was not possible to measure the amount of disciplinary activity using this question as the main indicator. Even a middle-class sociologist living in a homogeneous middle-class suburb and teaching middle-class students would be aware of within-culture variations enough to know that beating one's children and/or wife has considerably less emotional significance in the lower classes. Therefore, a battery of indicators had to be designed which would really measure the extend of disciplinary activities as a latent concept, rather than merely reporting the prevalence of some overt acts. II. Logically the same problem of 'equivalence of meaning' has been discussed for many years under a different label - but the discussion has taken the same course I have just outlined for the debate on question meaning. I am referring here to the attempts to standardize internationally so-called 'background characteristics', i.e. questions that are routinely employed as independent variables in breakdowns (such as age, income, education, community size, etc.). There has certainly been no lack of effort in several meetings of WAPOR, ISSC, and various international chains of commercial institutes. Recommendations were passed by international committees to standardize background variables. I have to report now that in this sense there has been little progress in international survey work. It has not even been possible to agree on common age-groupings. Why these difficulties in agreeing upon common background characteristics and why the lack of success even with such seemingly trivial conventions as common age-groupings? This time lack of cooperation is not due to the frequent idiosyncracies of any group of scientists. There is a better reason, and awareness of it can be viewed as progress. In discussing internationally the 'best' age breaks it became more and more obvious why age is really used, and when used why it is effective. Certainly the attribute, age, is usually not interesting as a physical characteristic; rather age-groupings are employed to denote powerful and general role differences - differences which can pattern behavior across many different situa-

Cross-cultural use of sample surveys

187

tions. Often age has been used when 'stage-in-the-life-cycle' as a sequence of role configurations was meant. However, if physical age is used as an indicator for different social roles, then the same physical age will denote different roles and different stages in the life cycle in different cultures. Hence no agreement on a standardization. Age is just one clear example. 'Old' as a social definition is obviously related to a different physical age in various societies. Thus, 40 years of age will define a different position for a woman in Sicily than for a woman in southern California or Florida. Actually, the same relationship between physical and social age does not even hold within the same country. A 45-year-old worker is nearing old age; a physician of the same age has just about reached full maturity. Another example is reported by Stern and d'Epinay. 22 They show that in Switzerland a town of 5,000 people is considerably more of a center of commerce and cultural activity - is in other words a different social unit - than is true for a town of the same size in France. If community size is used as an index of urbanization and relative access to central facilities, a town of 50,000 in Sicily may be equivalent to a town of 2,000 people in the midwestern United States. In striving for standardization of background characteristics in the sense of identity of labels, researchers have been worrying about the wrong problem. Again there was the frequent danger in cross-cultural research of equating formal identity of procedures with equivalence of an indicator's meanings. Actually some disagreement about the exact cutting points in a quantitative variable is not so important in the first place as long as the range of variability is comparable (except of course in dealing with very different societies). Much more difficult is the standardization of qualitative variables, or quantitative variables qualitively expressed - and if such standardization exists, the interpretation of the categories is difficult too. A peasant in Europe is still something different from an American farmer; if one compares responses for both groups, much of what is done actually shows that similar labels refer to different groups, rather than demonstrating cross-cultural differences between the responses of otherwise comparable groups. While it is a substantive finding of considerable interest to show that occupational groups do not exactly match their counterparts in other countries, this fact does not help much when employing occupation as a background variable. However, if a specific hypothesis exists as to why occupation should affect certain responses, we again can decide on functional equivalents. If one wants to test whether persons in occupations with easy access, little hope for upward mobility, and low job security show a lower preference for political movements that stand for social reform via evolution, one might take an unskilled worker in the US and compare him with a plantation worker in an underdeveloped country. Consider even the seemingly comparable definitions of educational categories in

188

Erwin K. Scheuch

the US and Europe. The last session of the ISA Subcommittee on Stratification and Social Mobility amply demonstrated the difficulties arising from the different social consequences attached to the same educational label. 23 As Mark Abrams pointed out during the 1962 World Congress of Sociology, college education means something vastly different in the US from what it means in England. 24 I am not referring here to real or imagined differences in the quality of education, but to its differing consequences for one's social position. A college education in England still denotes that one is a member of a select cultural minority which will interact frequently with other minorities across professional boundaries. In the US, the rarity of a college education is by now about as great as the rarity of a high school diploma was a generation ago. Thus in comparing the United States and England one might compare the college educated in England with the graduates of major private colleges in the US. A baccalauréat in France and a Matura in Germanspeaking countries have no exact counterpart in American education. Again, this is not meant as a reflection on quality, but refers to differences in the meaning of education. European educational systems just happen to be elitist, while American education is not. Once these difficulties are properly conceptualized (in part I have only made the implications of practices explicit), many of the problems of insuring comparability become much easier to manage. 5. Problem area III: The respondent as a unit in design and analysis Largely as a result of difficulties experienced in designing samples for underdeveloped countries, another problem area begins to emerge: assumptions about the respondent as a unit in survey research. These are implicit in the procedures that we customarily use. In making such assumptions explicit, technical problems in field work and in other stages of research once again turn into problems of methodology. Specifically we find ourselves discussing a long by-passed methodological problem, in part already posed by Durkheim: the character of data in sociological research as a basis for deriving general statements. 25 I. In accordance with the historical context in which surveys developed, an important implication of survey research, or at least of prevailing survey design, is that all members of a population matter, are largely interchangeable as units (i.e. as carriers of attributes), and exhibit the properties under investigation. That all units matter, and that individuals are the relevant units for the purpose of a s t u d y - t h i s is by no means self-evident; in many societies and for many purposes this assumption is quite wrong. When in Southern Italy, e.g., a rural woman is approached and asked to give her opinion, she not only often feels unable to do so, but also exhibits signs of feeling threatened when pressed to perform that peculiar operation that we call 'giving an

Cross-cultural use of sample surveys • 189 opinion'. A frequent reaction to the question 'how do you feel about...? is 'we here feel...'. In other words, the individual tends to report in terms of the prevailing opinions within his group; he is reluctant to dissociate himself mentally from group membership or to make a distinction between individual opinion and group consensus. Similar experiences have been reported from India. A salesman type of interviewer may still be able to extract some kind of answer as demanded by the questionnaire. Yet how are we to make use of such a response? T.H.Marschall in 1950 suggested a model for what he called 'extension of citizenship' to ever more groups of a society, until finally all physical members were to be regarded as citizens in a substantive sense.26 The degree to which this process has actually taken place in a society can be considered a (latent) limiting condition for the prevailing type of survey research. Actually, it is a surprising phenomenon that, especially in the US, most people are willing to voice an opinion on nearly any imaginable topic, from resurrection to the effects of subsidizing medical care for the aged by social security funds, from weapons systems to the effects of a balanced budget. And equally surprising is that these responses are often treated as somehow referring to fact or deciding controversial contentions. 'Is there in your opinion life on Mars?'-such a question does not strike most respondents as strange, and as we know from some wellknown earlier studies, a majority is quite willing to have definite views on nonexistent consumer items or on an imaginary 'metallic metals act'. 27 In societies with a legitimate stratification system, individuals define differently the range on which they feel competent to voice opinions. An extreme case appears to be France, for which Mark Abrams reports that in the middle classes only so-called serious questions, defined as involving fantasy on a rather cosmic scale, are considered acceptable.28 These same respondents are less willing to voice opinions on, e.g., government decisions that they define as subjects for the experts. The latent communality explaining the differential willingness to voice an opinion and to consider questions as reasonable seems to be the respondent's self-definition of competence. Obviously we need more empirically substantiated propositions about relevance and reactions of our units in survey research. Such propositions about respondents' self-definition and behavior - and more generally a clarification of the very nature of data in survey research - are easier to generate in cross-cultural surveys, where differences in the contexts from which we draw our units of study (i.e. respondents) are greater and thus are more easily conceptualized. Though these differences are responsible for some of our greatest problems, they can at the same time contribute to the great advance which cross-cultural comparison can mean for survey research. II. Whether the relevance we attribute to each unit - explicitly or implicitly in our procedures - holds cross-culturally is one of the chief problems underlying sampling for comparative research. This is a new problem-formulation, and one

190

Erwin K. Scheuch

that has emerged from conducting surveys in the developing countries. These experiences are also relevant for research in industrialized and urbanized countries, although this fact as yet is rarely seen. Sampling was one of the earliest problem areas to be recognized as such in comparative survey research. As is true for the other methodological aspects of cross-cultural surveys the problem definition has changed over the years, starting with concern for identical procedures. Again we meet the common-sense notion that comparability is assured only if the same procedures are applied in all countries. Yet on purely theoretical grounds this should strike us as a very peculiar notion. Whatever the sampling procedure: either a sample is representative and will permit inferences as to the composition of the sampled population (usually a nation-state) or it is not. There is no one sample design that insures representativeness under all conditions; it can usually be achieved through a variety of designs. Of course there is often an optimal sample for any specific purpose and population - but only in terms of relative cost. To insist on identical sampling procedures as a condition of comparability shows little confidence in samples a a tool of inference, and constitutes a misplaced trust in some of its concrete features. This trust is misplaced because concrete sample designs for human populations - at least by implication - represent adjustments of the model of the simple probability sample to specific topics and structural properties of the universe that is being studied. Rather than insisting on identical procedures, one should postulate that, if significant aspects of the universes to be studied differ, the designs should differ accordingly. One important qualification is necessary. In terms of sampling theory, insistence on identical procedures is unnecessary and sometimes harmful. However, we have to consider losses in probability sampling - and we known that the chances of losing persons within various population groups vary with different procedures. For various Western countries the usual rate of loss is between 10% and 20% of the original sample; in some subgroups, such as metropolitan populations, losses are often around 25 %. These figures hold cross-culturally. Since such sizable groups of the sample are not covered, one needs to be reasonably sure that there are comparable types of losses; comparable procedures in the last stages of a multi-stage design may be one way of making comparability of losses more likely. Even then it makes absolutely no sense to insist routinely on uniform procedures. The prevailing method of nation-wide probability sampling in the U.S. happens to be a type of area sampling, which in the final stages calls for rather cumbersome estimates of population composition (for stratification purposes) and the tedious procedure of 'prelisting.' If we compare conditions in the US with those in some European countries, we realize that this sample design is well suited to handle some problems specific to the US: unavailability of population lists (there is no compulsory registration), wide spacing of the census (at 10-year intervals), relative homo-

Cross-cultural use of sample surveys

191

geneity of neighborhoods and specifically of the houses within these neighborhoods. There is little reason to insist on this type of sample for Europe. There are laws forcing everyone to register his residence; the local administrations are infinitely more standardised; heterogeneity within apartment houses is high; and, in particular, there are often very efficient local statistical offices. In Germany and Holland, for example, there are excellent registers for all units in the population (in several German cities one IBM card per person contains all the data necessary for the usual stratification procedures). In France, registers have mainly been found deficient merely in cities between 10,000 and 100,000. In several European countries we have even had the choice between several sampling lists, enabling us to draw according to our needs a sample either of households or of individuals. Some years ago I used an area sample in Germany partly to check which population groups would be under-represented in random samples from files. The area sample yielded only 2 % of persons we would have missed in a conventional file sample, and it did so at a much higher cost and with higher sampling losses. The area sample is just one practical design for coping with certain peculiarities of one's universe. There is certainly no reason to behave as if it were the 'natural' design for population samples - as is done sometimes in American texts. Cross-cultural surveys have stimulated the fantasies of a few sample specialists. Some of their innovations in cross-cultural work might be used more widely in domestic research as well. In particular the work of the Indian Statistical Bureau in this immensely differentiated sub-continent has resulted in new designs (e.g., 'interpenetrating samples').29 Deming's population sample for Greece and subsequent proposals based on similar experiences have been published in part and are reasonably well-known among sampling specialists.30 One of the networks of commercial institutes, well-known for its insistence on random sampling, is presently using a revised version of Deming's master sample for survey work in South America. This design again presupposes only a minimum of available information. Repeatedly, another design has been proposed which makes very few assumptions about existing knowledge: the so-called random-path, or random-walk design. So far this design appears not to have found a continuous usage - probably because of the high demands it makes the interviewer's honesty. Essentially, this randomwalk design consists in sampling points from a grid. These points merely define the beginning of a cluster of households. A detailed prescription governs the way in which one proceeds along a chain of households and arrives at the one to be sampled. Since the distance between the starting point and the selected households is defined in number of households in between, different population density is automatically neutralized. This sample largely dispenses with the necessity of first stratifying by size and then prelisting. In research practice (especially within international networks of commercial agencies) there have been repeated attempts to develop a sample that can be

192

Erwin K. Scheuch

applied with only minor variations in all countries. Thus, one of the networks of commercial institutes works at present with a modified multistage area sample. There have also been recurring attempts to standardize the controls in quota samples. In these attempts there is a more or less intuitive awareness that uniformity of procedures is not the best strategy; but there is an insistence on using the same master plan. This insistence on some uniformity in principles of design with variability in procedures is advisable; it is essential in cross-cultural surveys tracing the importance of particular variables. Ceteris paribus, the stronger the descriptive aspects of a cross-cultural survey, the less important is uniformity in the principles of sample design; the more analytical purposes are stressed, the more important is the identity of principles of design; identical procedures, however, are usually a misplaced concreteness. The Almond and Verba project offers an excellent opportunity to exemplify what we consider a permissible - and often necessary - range in variability for cross-cultural surveys; it also makes obvious the conditions under which the need for uniformity of sampling principles occurs. 31 This study in five very different societies is characterized by the use of the same basic type of sample, a wide variety of designs, uniformity and diversity of procedures, and a very considerable divergence in the degree to which sample designs could actually be implemented. The Almond-Verba study was actually carried out by commercial survey research institutes of the five countries studied. All these institutes adhered closely to their normal procedures - procedures that in the countries concerned were usually considered to represent good technical standards. These institutes were part of a major commercial chain, and in years of cooperative research they had agreed on considerable similarity of sampling designs. Consequently, there were both a basic similarity due to past cooperation and a divergence according to the traditions of particular institutes. All institutes used a multistage, stratified probability sample that, up to the ultimate sampling stages, conformed essentially to the principles of area sampling. The number of stages in the area sampling phase differed, and so did the nature and variability of units. To characterize, first, the variability of selection up to the smallest contiguous area: in Germany the psu's were the 30,000 communities, from which 100 communities were selected in one step. In Italy, the first stage units were the 92 provinces from which 13 were included in the sample as the universe for the second stage; here 49 communities were selected. In the United Kingdom three stages were used to arrive at the smallest contiguous area: from 630 parliamentary constituencies an (unknown) number was selected; from each of the remaining constituencies two wards were chosen, and from each of those one polling district. In the United States, the sample followed closely the standard procedures in the US versions of area sampling: first-stage units were the 3,000metropolitan areas and non-metropolitan

Cross-cultural use of sample surveys

193

counties; second-stage units in urban areas were census tracts, while in rural areas 'localities' were used; in the third stage the sampling units were either blocks or segments (i.e. equivalent non-urban small geographical units); altogether, 491 small geographical units were thus selected. The most complicated design was employed in Mexico; although it is not quite clear what the actual procedures were, it seems that in some way 27 cities over 10,000 inhabitants constituted the sampling frame, from which in one step city blocks were chosen in a systematic fashion. There is no doubt that in all cases the design either conformed to the principles of probability sampling or came close to them, although the standard error of the procedures (or the cluster effects involved in multistage sampling) should differ dramatically. This cluster effect up to the smallest contiguous geographical area seems smallest in Germany, largest in Mexico and Italy - but the data reported to not permit even a rough calculation of size. Quite different were also the procedures followed in the actual determination of respondents within these smallest geographical areas. In Germany, Italy, and the United Kingdom the register of electors was used, but in a different way in all three countries. In Germany the name of an elector was merely used to locate a household, within which the actual respondent was identified by using a table of random numbers. In the United Kingdom respondents over 21 years of age were immediately identified and respondents below voting age were selected in a two-stage process akin to the one followed in Germany (instead of random numbers, the birthday procedure was used). In Italy the respondents above legal voting age were identified in a onestage process as in the U K ; those below voting age were taken from households in which an older respondent had already been selected. Both in Mexico and in the United States, the selection of the usu's was preceeded by a 'prelisting' within blocks or segments. In the USA, dwelling units were prelisted within all blocks and selected first; a prelisting of all adults within a dwelling unit was then prepared and a table of random numbers applied to identify the actual respondent within the du. In Mexico selection within blocks proceeded immediately to the household as the next sampling unit, and within households the individuals were identified via random numbers. Thus, procedures in this second phase of sampling differed strongly by the sizes of the initial universes (communities or blocks), the number of steps taken to identify the respondent within the blocks, and the method for identifying respondents - two procedures sometimes being used simultaneously. The cluster effects were probably smallest in the UK and in Italy, probably largest in Mexico and the United States. Both in Mexico and in Italy, the chances of some of the respondents were more strongly affected by the size and character of households than was true in the other countries. In this very summary description we have purposely left out a description of the - sometimes quite elaborate - stratification procedures on several stages of the sampling process. However, they undoubtedly tended to counteract some of the

194

Erwin K. Scheuch

cluster effects and risky weighting operations. It still remains doubtful whether, in view of the rather small sample size the reduction was generally sufficient to allow the use of the - anyhow a bit questionable - formula for the computation of standard errors that Almond and Verba list.32 For highly skewed distributions we would guess that this mathematical expression may often lead to an underestimation of sampling errors, and this becomes quite relevant for some of the differences between countries when in several of them the p