¹ The concept is variously defined in the literature according to the particular research focus of the scholar using it in his work. Thus Deutsch, Karl W. associates elites with key communication and command functions: The Nerves of Government (New York: The Free Press, 1963), 154–160Google Scholar, whereas Lasswell, Harold defines as elites those who enjoy the greatest proportion of values in a society: Politics: Who Gets What, When, How (New York: Meridian Books, 1960)Google Scholar. Elsewhere elites have been defined as those who hold high position and perform the function of making major social decisions in a society; see, for example, Aron, Raymond, “Social Structure and the Ruling Class,” The British Journal of Sociology, 1 (03, 1950), at p. 9Google Scholar; Dahrendorf, Ralf, Class and Class Conflict in Industrial Society (Stanford: Stanford University Press, 1959), 303–307Google Scholar; Mills, C. Wright, The Power Elite (New York: Oxford University Press, 1959), 3–5Google Scholar; Nadel, S. F., “The Concept of Social Elites,” International Social Science Bulletin, 7 (1956), at p. 414Google Scholar; and Lasswell, Harold, Lerner, Daniel, and Rothwell, C. Easton, The Comparative Study of Elites (Stanford: Stanford University Press, 1952), p. 6Google Scholar. Mosca's, Gaetano broad category of “ruling class” embodies all three criteria: The Ruling Class (New York: McGraw-Hill, 1939), p. 50Google Scholar; while Keller, Suzanne emphasizes the criterion of wielding moral responsibility: Beyond the Ruling Class; Strategic Elites in Modern Society (New York: Random House, 1963), p. 4Google Scholar. For a good review of the literature see, Beck, Carl, Malloy, James M., and Campbell, William R., A Survey of Elite Studies (Special Operations Research Office, The American University, 1965)Google Scholar.

² We recognize that expressions of individual attitudes are usually distorted by coding procedures and that survey data—especially in the case of elite samples that are usually relatively small—must be interpreted with caution. We also recognize that a distinction exists between underlying attitude and expressed opinion, though, following common practice, we treat the two as synonymous in this article.

³ For an exception, see Bell, Wendell, Hill, Richard J., and Wright, Charles R., Public Leadership (San Francisco: Chandler Publishing Company, 1961)Google Scholar.

⁴ Just as the national political system may be conceptualized as functioning in an environment—the international system—so the local political system can be seen to function in the environment of the national system. While each remains susceptible to description and analysis in terms of a set of self-contained, integrated systemic roles, the central empirical fact of political decision making in the American small community is its permeation by state and federal structures. For the most part local political values are in effect authoritatively allocated by superior governmental units in the United States as elsewhere. See, Vidich, Arthur J. and Bensman, Joseph, Small Town in Mass Society (New York: Anchor Books, 1960)Google Scholar, and the abstracts by Rossi, Clelland, Arensberg, Sjoberg, and Warren, , in Press, Charles (ed.), Main Street Politics, Policy Making at the Local Level: A Survey of the Periodical Literature Since 1960 (East Lansing, Michigan: Institute for Community Development, 1962)Google Scholar. A good deal of research on this subject is now in progress and may yield significant findings on the interdependence and similarity of national and sub-national elite structures.

⁵ See, for example, Vidich and Bensman, op. cit., Dahl, Robert A., Who Governs? (New Haven: Yale University Press, 1963)Google Scholar; and Agger, Robert E.et al., The Rulers and the Ruled (New York: Wiley, 1964)Google Scholar. For convenient summaries of community elite studies see, Press, op. cit.; and Bell, Hill, and Wright, op. cit. An interesting synthesis of the stratification and pluralistic approaches has recently been attempted by Presthus, Robert in Men at the Top; A Study in Community Power (New York: Oxford University Press, 1964)Google Scholar.

⁶ See Lasswell, Harold D., “Agenda for the Study of Political Elites,” in Marvick, Dwaine (ed.), Political Decision-Makers (New York: The Free Press, 1961), 264–389Google Scholar; Bottomore, T. B., Elites and Society (New York: Basic Books, 1964)Google Scholar; and Keller op. cit.

⁷ Some recent exceptions are Bell, Wendell, Jamaican Leaders: Political Attitudes in a New Nation (Berkeley: University of California Press, 1964)Google Scholar; Pye, Lucian W., Politics, Personality and Nation Building (New Haven: Yale University Press, 1962)Google Scholar, Ch. 5; LeVine, Victor T., The Conflict of Elite Generations in the Context of Independence in French Speaking Africa (Hoover Institution, forthcoming)Google Scholar; Frey, Frederick W., The Turkish Political Elite (Cambridge, Mass.; MIT Press, 1965)Google Scholar; and Schlesinger, Joseph A., Ambition and Politics (Chicago: Rand McNally, 1966)Google Scholar.

⁸ For example, Rosenau, James N., National Leadership and Foreign Policy (Princeton: Princeton University Press, 1963)CrossRefGoogle Scholar; Matthews, Donald R., United States Senators and their World (New York: Vintage Books, 1960)Google Scholar; Hunter, Floyd, Top Leadership, U. S. A. (Chapel Hill: University of North Carolina Press, 1959)Google Scholar; Frey, op. cit.; Guttsman, W. L., The British Political Elite (New York: Basic Books, 1963)Google Scholar; Deutsch, Karl W. and Edinger, Lewis J., Germany Rejoins the Powers (Stanford: Stanford University Press, 1959)Google Scholar; Lasswell, Harold D. and Lerner, Daniel (eds.), World Revolutionary Elites (Cambridge, Mass.: MIT Press, 1965)Google Scholar; and Lloyd Warner, W. L., The American Federal Executive (New Haven: Yale University Press, 1963)Google Scholar.

⁹ Deutsch, , The Nerves of Government, 155–56Google Scholar. See also, Matthews, Donald R., The Social Background of Political Decision-Makers (Garden City, N. Y.; Doubleday and Company, 1954)CrossRefGoogle Scholar; Brzezinski, Zbigniew and Huntington, Samuel P., Political Power USA/USSR (New York: Viking Press, 1964)Google Scholar; and Janowitz, Morris, “The Systematic Analysis of Political Biography,” World Politics, 4 (04, 1954), 405–412CrossRefGoogle Scholar. Although it has been rightly objected that most studies dealing with recruitment and elite composition are primarily descriptive and fail to explain why particular elites were recruited and why they act the way they do, information about social background characteristics may well be a prerequisite for this sortof explanatory analysis. See Seligman, Lester G., “The Study of Political Leadership,” in Eulau, Heinz, Eldersveld, Samuel J., and Janowitz, Morris (eds.), Political Behavior; A Reader in Theory and Research (Glencoe, Illinois: Free Press, 1956), p. 178Google Scholar.

¹⁰ For example, Harold D. Lasswell asserts, “Elite members who prepare to play active political roles usually become differentiated at an early age from their more passive contemporaries. Such differences of perspective affect motivation to acquire the base values appropriate to politics, notably the skills essential to leadership ….” See Lasswell, Harold D. and Lerner, Daniel (eds.), World Revolutionary Elites, 12–14, 21Google Scholar. See also, Glick, Henry Robert, “Political Recruitment in Sarawak: A Case Study of Leadership in a New State,” Journal of Politics, 28 (February, 1966), 81–84CrossRefGoogle Scholar. For more general recent treatments of political socialization, see, inter alia, Lane, Robert E., Political Life (Glencoe, Ill.: Free Press, 1959)Google Scholar; Dawson, Richard E. and Prewitt, Kenneth, Political Socialization (Boston: Little Brown, forthcoming)Google Scholar; Almond, Gabriel A. and G. Powell, Bingham, Comparative Politics: A Developmental Approach (Boston: Little Brown, 1966), Ch. 3Google Scholar; Hyman, Herbert, Political Socialization (Glencoe, Ill.: Free Press, 1959)Google Scholar; and Greenstein, Fred I., Children and Politics (New Haven: Yale University Press, 1965)Google Scholar.

¹¹ Thus, according to Lasswell and Lerner, “… elite perspectives can be accounted for by exploring the perspectives that were incorporated on the road toward active power or at least to ward an adult status…. Not only predispositions but environments count”: op. cit., p. 18. The ultimate step is the inference of elite behavior patterns from collective attitude patterns. But since that process goes beyond the socialization experiences which concern us here, it will not be considered in this paper.

¹² Lasswell and Lerner (eds.), World Revolution ary Elites.

¹³ Heckscher, Gunnar, in contrast, seeks to approach social background in elite studies from a more problem-oriented, functional perspective. See his The Study of Comparative Government and Politics (London: Allen and Un win, 1957), 108–120Google Scholar.

¹⁴ Lasswell, Harold D. and Lerner, Daniel (eds.), World Revolutionary Elites, 6–8Google Scholar.

¹⁵ The search for attitudinal predictors does not necessarily imply, of course, a commitment to a determinism of the type underlying some socio-economic and psychological theories. (See Moore, Wilbert E., “The Utility of Utopias,” American Sociological Review, 31 (12, 1966), 765–773CrossRefGoogle Scholar. While most men are creatures of habit and believe and behave in patterned ways, such patterns and their antecedents are admittedly complex and ephemeral. Ours, however, is not an argument against the possibility of their study, but rather a plea for a probabilistic approach dictated by the complexity of multivariate elite data.

¹⁶ See, for example, Wahlke, John C., Eulau, Heinz, Buchanan, James, and Ferguson, Leroy C., The Legislative System (New York: Wiley, 1962)Google Scholar; Wendell Bell, Jamaican Leaders; Deutsch, Karl W., Edinger, Lewis J., Macridis, Roy C., and Merritt, Richard L., France, Germany and the Western Alliance (New York: Scribners, 1967)Google Scholar; James N. Rosenau, National Leadership and Foreign Policy; Seligman, Lester G., Leadership in a New Nation (New York: Atherton Press, 1964)Google Scholar; and Bonilla, Prank and Silva Michelena, José A. (eds.), Studying the Venezuelan Polity (Cambridge, Mass.: MIT Press, forthcoming)Google Scholar.

¹⁷ The program was designed for use on discrete data and employs a “logical,” as opposed to a “statistical” criterion for multivariate analysis, to group data into deterministic models. This alone appeared to make it preferable to factor analytic techniques, which are further limited by the element of arbitrariness introduced with axis rotation. Robot data screening performs a selective search of the data in stepwise fashion, utilizing all possible combinations of N-1 variable predictors at each new level of combinations, until no further combinations of independent variables are found which would raise the predictability of the dependent variable. It allows for manipulation of the significance level and includes an entropy measure of uncertainty derived from information theory. See Sterling, T.et al., “Robot Data Screening: A Solution to Multivariate Type Problems in the Biological and Social Sciences,” Scientific Applications: Communications of the A.C.M., 9, 7, (07, 1966), 529–532Google Scholar; Sterling, T. D., et al., “Robot Data Processing Techniques for Mul tivariate Epidemiological Predictions,” Annals of the New York Academy of Sciences, 126 (August 6, 1965), 779–794CrossRefGoogle Scholar; and Sterling, T. D. and Pollack, S. V., Computers and the Life Sciences (New York: Columbia Press, 1965)Google Scholar.

¹⁸ Deutsch, Karl W., Edinger, Lewis J., Macridis, Roy C., and Merritt, Richard L., Arms Control and European Unity: Elite Attitudes and Their Background in France and the German Federal Republic (New Haven: Yale Political Data Library, 1966)Google Scholar; and Deutsch, Karl W., Edinger, Lewis J., Macridis, Roy C., Merritt, Richard L., and Voss-Eckermann, Helga, French and German Elite Responses, 1964: Codebook and Data (New Haven: Yale Political Data Library, 1966)Google Scholar. See also Deutsch, Karl W., “Integration and Arms Control in the European Political Environment: A Summary Report,” this Review, 60 (06, 1966), 354–365Google Scholar.

¹⁹ The combination of reputational and positional approaches yielded an initial group of 441 French and 650 German leaders which for operational purposes was understood as representative of elites in both political systems. Cf., Deutsch, Edinger, Macridis and Merritt, op. cit., Ch. 1.

²⁰ Dahrendorf, Ralf, “Recent Changes in the Class Structures of European Societies,” Daedalus. 93 (Winter, 1964), at p. 242Google Scholar.

²¹ Categories for this instrument were based upon Lewis J. Edinger's previous studies of West German elites. They included many of the most frequently employed background characterietics in elite studies.

²² As a first step in the secondary analysis of such data recorded on punch cards, it is often necessary to recode the information, rearranging it in such fashion as might be more amenable to the new questions that are to be asked. Since we sought to make considerably greater comparative use of the Yale data than was intended when it was originally collected and coded, we had to recode most of it so as to develop commensurable categories for both nations.

²³ In addition to treating the French and West German elites as separate populations (in which case one must invoke a caveat emptor in so far as the samples are analogous rather than exact) we have sometimes treated the French and German respondents as an analytic universe. Not as a sample of all French and German elites, that is, but rather as an elite population within which we may examine the relationships between social background variables and attitudinal patterns.

²⁴ The number of latent attitudes (11) used was the same for both the French and German leaders. However, with regard to manifest attitudes less data was available for the French (39 attitudes) than for the Germans (48 attitudes). With regard to the biographical or social background data, there was considerably less data available in the French case (11 variables), than for the German (40 variables). We felt that for theoretical reasons it was worthwhile to utilize the greater body of German data, which will explain differences in our tables. Whenever the French and German respondents were compared, however, this was done using only the attitudes and background variables for which adequate data were on hand for both.

²⁵ It is worth emphasizing that even the back-ground variable predicting the greatest number of attitudes allowed us to do so for only twelve, that is, 25 percent, of all 48 attitudes.

²⁶ We use scope rather than strength as a summary measure because for purposes of the present discussion—demonstrating that some background variables have more relevance than others for elite attitudes in a national political system—the number of attitudes a background variable predicts seems more to the point.

²⁷ These were responses to the following inter-viewer questions: (1) What development in domestic policy are likely to bring about a change in German foreign policy? (2) Would the recognition of East Germany ease international tension? (3) How would recognition of the Oder-Neisse line affect efforts toward German reunification? and, (4) What is the greatest threat to German security at the present time?

²⁸ Since only eleven French background categories were examined, we prefer not to extend our interpretation to argue that because fewer French elite attitudes were related to background factors, background as such was less related to French attitudes than to German. However, this limitation imposed upon the present analysis by the nature of the data employed merely emphasizes the need to expand and coordinate data collection for more extensive cross-national comparisons.

²⁹ Those attitudes predicted by more back-ground variables (three to five) than other attitudes were responses to the following questions put by the interviewers: (1) Are you content with the present governmental system in France? (2) Are you satisfied with the government's foreign policy measures? (3) Which group or groups do you feel have gained in political power over the last few years? (4) Which features of French foreign policy, if any, are likely to remain after deGaulle? (5) Do you expect an appreciable change in the relations between France and the countries of Eastern Europe during the next few years? and, (6) Is an independent nuclear deterrent necessary for a nation's prestige in the world?

³⁰ By combining the French and German data we are once again limited to examining only those background variables (11) and those attitudes (39) for which we have comparable information. Moreover, we can observe a drop in predictability in the combined groups for only those background predictions that were already found to be significant when the two national elites were considered separately.

³¹ The expectation was that if the ratings were accurate, a respondent rated low on degree of information in the Latent Schedule should have given a high number of “D.K.” answers in the Manifest Schedule, and vice-versa. While our check indicated a relatively low number of interviewer errors and appeared to justify an assumption of high accuracy for the “Latent” estimates, we should note that an estimate of this nature is purely quantitative, while it is likely that qualitative judgments were to some measure a component of the original ratings.

³² We had expected the association to be greater because the Latent Attitude Schedule listed much more diffuse orientations than the Manifest. However, the Latent Attitude data seems comparatively weak and limited; it includes only about one-quarter as many items as were covered by the Manifest Schedule. We are presently exploring this problem in greater depth.

³³ Moreover, these alienated elites are generally-oriented beyond the boundaries of France: not only are they on the whole supra-nationalists, but their cognitive focus extends to a “larger European” and “Atlantic” outlook, i.e., to a diffuse internationalism.

³⁴ Our procedure was as follows: starting with the question on which responses by both French and German elites showed the greatest consensus (i.e., where the greatest number of respondents agreed on one or another position) we separated out the groups of greatest agreement. We then culled from these those who were in highest agreement on a second question as well and continued this dichotomizing selection process through all eight questions.

³⁵ Our measure of association was Pearson's contingency coefficient. If background characteristics were distributed indiscriminately, or in the same way, in both Groups I and II which were in absolute disagreement on three attitudes, then they could not have been related to these attitudes and the coefficient would be zero. As the coefficient increases, however, the background characteristic increasingly discriminates between the two groups, and is therefore a more reliable indicator in predicting attitudinal group membership.

³⁶ Two control groups of the same size as Groups I and II were randomly selected from each of the entire French and German elite samples. They moved in the opposite direction from Groups I and II for the Germans, and remained about the same as the French entire elite sample or moved but slightly in the direction of the French Groups I and II.

³⁷ In the case of the data used here predictive strength often proved acceptable, but the scope of prediction was frequently lower than might have been expected and did not seem to conform to any obvious patterns beyond those we suggested rather tentatively. Moreover, the relevance of the diffuseness or specificity of the attitudes in question remained to be resolved. Despite our contrary findings in comparing latent with manifest attitudes among French and German elites, we suspect that quantitative forecasting may be more successful with diffuse than with specific attitudinal responses.

³⁸ This was underscored in our study of attitudinal cohesion, while the pervasive effects of nationality emphasized the theme from a different perspective.

³⁹ Some indications of a possible approach were suggested to us by our finding that adult socialization experiences—particularly those isolated by occupation and party affiliation categories—were consistently highest in relationship to attitude, and that those attitudes most frequently predicted by the background data usually were related to salient political issues. We realize, of course, that the data base for these particular findings was extremely limited and our conclusions correspondingly tentative.