DEFINING COMPARATIVE EDUCATION: CONCEPTIONS

The last decade has witnessed not only a vast burgeoning of the literature in and about comparative education, but also a radical change in the rationales, methods, and goals of the field. Whether this change in landscape has been for the better or the worse I shall leave for colleagues to judge. But imagine the situation of R. V. Winkle, professor of comparative education, who had fallen asleep at the end of 1959, to awaken again only in 1970. His slumbers would have commenced with his subject dominated by the works of Kandel, Hans, Lauwerys, and Rossello. He would have been aware of only the barest intimations of a more deliberately social-scientific approach. On awakening he would have found a new style of work bidding strongly to take over the field, though without having won over by any means all of its practitioners. In any event, he would have had a formidable reading assignment awaiting his attention! Much of the justification for undertaking comparative education studies prior to 1960 was in terms of their potential either for countering parochialism or ethnocentrism, or for assisting in the improvement of education at home. Basically, researchers and writers were asking such questions as: What is characteristically French about the French secondary school curriculum? or, What is happening in German schools that we might profit from? The theme of recent work may perhaps be seen as a progressive transfer of attention from country characteristics to problems, and from problems to the specification of relationships and the formulation and testing of theories. This is not to suggest, of course, that the new style has found universal and unquestioned acceptance, or that the previous genre of work is without merit. On the contrary, we continue to see, and shall continue to want studies with such titles as, "Higher Education Reform in Germany", "The Technical School in the Dominican Republic", "Local Initiatives in Pre-School Education in the Soviet Union", and so on. Moreover, all is not plain sailing in the new mode. The conceptual and practical problems of conducting theory-oriented comparative research are not only not immediately and obviously tractable, but are also being widely aired.1 In this change of emphasis comparative education is clearly following a course already charted in economics, sociology, and political science. Economics has ventured furthest, perhaps. It has now left far behind its earlier preoccupation with the identification and description of economic institutions and has become a complex endeavour to explain and predict behaviour connected with making choices among alternatives. Sociology, similarly, has moved beyond the description and classification of social units to analysis and prediction of their interaction. And, just at the present time, some of the most fruitful work of relevance for comparative education is currently appearing from political scientists pursuing a cross-national approach.2 Clearly, these parallel developments have not occurred simply by chance: they express a common reaction to a common set of methodological potentialities and problems. The challenge to move from the particular to the general, from identificationdescription-classification to hypothesis-testing, theory building and prediction is pervasive. One test of the progress of a science is its acquisition of a terminology. In developing "technical terms", words are often borrowed from everyday use, and then more precisely defined for technical purposes. One thinks immediately of the use in physics of the term "velocity" (with its essential connotation of direction as distinct from the unvectored concept, "speed"); or, in economics our attempt to define "demand" as "ability and willingness to pay", and not simply to retain its common meaning of "need" or "desire". Indeed, on occasion the most far-reaching result of scientific study of a phenomenon appears to be the recognition of a new, more powerful, albeit more limited, definition of a term. Consider what is happening to the term "comparative" in the title that denotes our field. I believe that we are about to move rather rapidly away from the everyday meaning of the word to a much more technical meaning. This rather radical redefinition of the term "comparative education" will involve at once a limitation and an extension of its scope. The impulse toward limitation will arise because we have come to realise that many studies that happen to use international and foreign data are not to be considered "comparative" simply by virtue of that fact; and the impulse toward extension will occur because many studies conducted on the basis of data drawn from within a single country nevertheless have a valid claim to be considered comparative, once we define the term in a way that reflects the function of comparison in systematic explanation. Clearly, while this process is continuing we can expect a rather lively controversy on just what the term should and does mean. In part, I suppose, this is what this conference is about. The summary of our deliberations may legitimately expect to record what is happening to the nature of our field, and if we are optimistic, we can even hope to influence it. Comparative education has mistakenly come to be identified either with the study of education in another country, or with studies using data drawn from more than one country. This view of what constitutes comparative education enjoys the sanction of both common usage and common sense. One finds out what is going on abroad and compares it with what is happening at home, often with a practical programme of amelioration in view.3 Certainly, many essays in comparative education are of this type. Alternatively, one uses a collection of multi-national data to identify, describe, and compare relationships (usually correlations) within education, or between education and other social phenomena.4 Again, I must emphasise that to call such studies "comparative" agrees with common sense and usage. But the weakness of that position is that it establishes as the criterion for classification as a comparative study the mere presence or absence of foreign or multi-national characteristics of data, and by implication ignores, or even denies, the existence of a characteristically comparative method. We are hindered from asking a set of key questions: Are all inter-, cross-, or multi-national studies ipso facto comparative? Are all comparative studies necessarily either inter-, cross-, or multi-national? What, indeed, are the necessary and sufficient conditions for a study to be comparative? Does there exist a characteristic comparative approach to a problem? If so, what is it? Nations constitute one important set of systems that attract our attention, and we have employed so-called comparative studies largely to identify and describe the attributes of such national systems. We have ended up with "nominal" statements of the type: "In country A, the secondary school curriculum is such-and-such; while in country B. it is so-and-so; and in countries C, D, and E, it is something else." Or, we might say in quantitative terms: "In country A, the fraction of the GNP spent on education is high (7-8 percent);in country B, it is moderate (5 percent);in country C, it is low (21/2-3 percent)." However, as the social sciences have extended the range of questions they ask and as comparative studies (among them, comparative education) have matured, so we have begun to comprehend a fundamentally different role for comparison, whether conducted on the basis of national systems, or of other units. The key to this transformation in our thought lies in the attempt inherent in the social sciences to explain and predict, rather than merely to identify and describe. A simplified example may, perhaps, help illustrate the new emphasis in comparative work: Let us assume that we wish to explain (and, perhaps predict) the relationship between the size of a family's income and the probability of the children in the family enrolling in full-time post-secondary education. If we find mirabile dictu that this relationship is the same from country to country, then we have no need to proceed further. We can immediately make a general (that is, a non-system-specific) statement defining a relationship between family income and the probability of post-secondary enrolment that is valid without including the names of any countries. But matters are more complicated if we are faced with the more Iikely case in which relationships differ from country to country. For example, we might find that while all countries exhibit a positive relationship between these two variables, the correlation is very strong in some countries, only moderate in others, and rather weak in a third group. Or. putting it in the language of least-squares linear regression analysis, we find that our best fitting equation explains different proportions of the observed variance in different countries. Let us assume, too, that no amount of within-system adjustment of either the independent or dependent variables alters the fundamental fact that in different countries similar levels of family income are associated with (or, "produce") different probabilities of a family's children attending post-secondary institutions. This is the paradigm situation calling for employment of the comparative method. We now have to ask, what are the system-level factors that are at work, influencing the interaction of within-system variables? As we shift the level of analysis from consideration of within-system to system-level factors, we are engaged in trying out the effect upon these different within-system relationships of introducing additional, theoretically justifiable independent variables, in the form of system characteristics. We continue to do this until we can no longer (a) increase further the proportion of observed variance explained within each country; and, (b) reduce further the differences among countries in the proportions of observed variance explained. To continue with our example, we might try out in turn the effect of including among our explanatory variables such system-level factors as "degree of income inequality", "ratio of the number of secondary school graduates to the size of the corresponding age-cohort", "proportion of direct costs of postsecondary education defrayed from non-tuition sources of finance", and "recency of the post-secondary institutions enrolling 5 percent (or 10 percent) of the corresponding age-cohort". We stop when the inclusion of further theoretically justifiable system variables yields insignificant returns in terms of (a) and (b) above.5 Only at this point do we introduce the names of countries in explanation, ascribing the remaining differences in proportions of variance explained to the unanalysed or unanalysable peculiarities of the countries. In this explanatory model, country names are used to tag bundles of unexplained variance. The object of the exercise, then, is not, as in traditional comparative studies, to extend and enrich as far as possible, the connotational content of country names; instead, we seek to extend and enrich to the limit general "law-like", cross-system statements, bringing in country (that is, system) names only when our power accurately to generalise across countries fails. A comparative study is essentially an attempt as far as possible to replace the names of systems (countries) by the names of concepts (variables). In this style of comparative study, for the example we have taken, we might hope to make a statement of the type: In all countries, size of family income is positively associated with the probability of children in the family being enrolled in full-time post-secondary schooling, and differences in family income can explain at least one-half of within-country differences in the probability of enrolment. In those countries where income inequality is high and the proportion of costs defrayed from non-tuition sources is low, the explanatory power of differences in family income rises to at least three-quarters. Consideration of the fraction of the age-group graduated from secondary education, and the recency of growth of the postsecondary system does not improve explanation appreciably in any case except in the Soviet-type countries, where these factors do seem to be important. 6 For our present purpose, the crux of all this is the necessity at some point in the analysis to stop further withincountry analysis and to change the level of analysis to incorporate among-country variables. For this is the essential condition for a study to be classified as "comparative": data are collected at more than one level and analysis also proceeds at more than one level. With this criterion we can attempt answers to the questions posed above. Q. Does there exist a characteristic comparative approach to solving a problem, testing a hypothesis, formulating a theory? A. Yes. It involves formulating the analysis so that within-system relations are explained as fully as possible using within-system variabIes, comparing the characteristics and differences of such explanations across systems, and trying to explain these characteristics and differences by changing the level of analysis to take account of the operation of variables identified at the level of systems. Q. Are all comparative studies necessarily either inter-, cross-, or multi-national? A. No, although many are. National units commonly form the matrix for data collection and governments are willing to finance studies (either directly, or indirectly through the international agencies) as part of the international sport of competitive growthmanship. But we ought to insist that a study within, say, the United States of the relationship between family income and the probability of the family's children enrolling in post-secondary education, formulated in terms of South vs. non-South, or urban vs. rural areas, or Whites vs. Blacks, has an equal chance with an international study of employing the comparative approach, as defined above. 7 Q. Are all inter-, cross-, or multinational studies ipso facto comparative? A. No. Many studies use data from more than one country, but restrict the variables considered or the analysis employed to a single level, either within-system or whole-systems but not both. Thus, we have seen multi-national analyses of trends in educational expenditures that are restricted to juxtaposing country-level relationships (for example, percentages of GNP devoted to education), and there are multi-country studies of curriculum restricted to within-country univariates (for example, the amounts of time assigned to different school subjects). In the technical sense of the term that we have suggested above, such studies are not comparative. NOTES 1. See Bruce M. Russett et al. (1964), Part B "The Analysis of Trends and Patterns", especially "Multifactor Explanations of Social Change", pp. 311-321. Also R. Merritt and S. Rokkan (1966) Bernhard Dieckmann (1970), Dieter Berstecher (1970), S. Rokkan (1968), and A. Przeworski and H. Teune (1970). Some points presented in this paper rely heavily on Part One of the latter book. Each of the volumes cited here contains important bibliographies. [BACK] 2. See Przeworski and Teune (1970), D.E. Apter (1968). R.C. Macridis (1968), H.A. Scarrow (1969) and P. Shoup (1968). G.A. Almond and S. Verba (1965) remains a work of primary importance in the field of comparative political/ educational analysis, although see Sheuch's contribution in Rokkan (1968) for a critique of many aspects of the Almond and Verba work. [BACK] 3. The locus classicus is M.-A. Jullien's "Esquisse . . . " reprinted in S. Fraser (1964). [BACK] 4. See, for example, Michel Debeauvais (1970). [BACK] 5. Often, of course, we must stop short of this point, owing to lack of time and money. [BACK] 6. Such a statement might set the stage for trying to develop a cross-nationally valid theory of the link between family income and family demand for schooling in general, and not just for post-secondary education. [BACK] 7. Most participants at this conference are specifically concerned with the comparative study of educational phenomena based on national units. Perhaps, therefore, our field might be better termed "cross-national comparative education". This nomenclature would have the merit of implying the existence of other bases or units for undertaking comparative analysis. Not only would we want to retain links with comparative studies using other bases, but we would recognise the existence of a common logic underlying all comparative analysis, and be drawn to follow it in our work. [BACK] REFERENCES Almond, G.A., and S. Verba. The Civic Culture. Boston: Little, Brown, 1965. Apter, D.E. Some Conceptual Approaches to the Study of Modernization. Englewood Cliffs, N.J.: Prentice Hall, 1968. Berstecher, D. Zur Theorie und Technik des internationalen Vergleichs: Das Beispiel der Bildungsforschung. Stuttgart: Klett, 1970. Debeauvais, M. Comparative Study of Educational Expenditure and Its Trends in OECD Countries since 1950. Paris: OECD, 1970. Dieckmann, B. Zur Strategie des systematischen internationalen Vergleichs: Probleme des Datenbasis und der Entwicklungsbegriffe. Stuttgart: Klett, 1970. Fraser, S. Jullien's Plan for Comparative Education, 1826- 1827. New York: Teachers College, Columbia University, 1964. Macridis, R.C. The Comparative Study of Politics. New York: Random House, 1968. Merritt, R., and S. Rokkan (eds.). Comparing Nations: The Uses of Quantitative Data in Cross-National Research. New Haven: Yale University Press, 1966. Przeworski, A., and H. Teune. The Logic of Comparative Social Inquiry. New York: Wiley, 1970. Rokkan, S. (ed.). Comparative Research across Cultures and Nations. Paris/The Hague: Mouton, 1968. (Especially papers by H.R. Alker, Jr., L. Benson, A.J.F. Köbben, D. Lerner, G. Ohlin, E.K. Scheuch.) Russett, B.M. et al. World Handbook of Political and Social Indicators. New Haven: Yale University Press, 1964. Scarrow, H.A. Comparative Political Analysis. New York: Harper & Row, 1969. Shoup, P. "Comparing Communist Nations: Prospects for an Empirical Approach". American Political Science Review 62 (1968). Sumber: http://www.fe.hku.hk/cerc/1c.html