¹ Cf. Robert A. Dahl's notion of “the political stratum”—“a small stratum of individuals much more highly involved in political thought, discussion, and action than the rest of the population.” See Who Governs? (New Haven: Yale Univ. Press, 1961), p. 90 Google Scholar. For brief introductions to the enormous literature on “elite theories of politics,” see: Bottomore, T. B., Elites and Society (New York: Basic Books, 1965)Google Scholar, and Parry, Geraint, Political Elites (New York: Frederick A. Praeger, 1969)Google Scholar.

² Greenstein, Fred I., Personality and Politics (Chicago: Markham Publishing Company, 1969), pp. 7 and 29 Google Scholar.

³ There is, of course, an interactive relationship between environment and attitude, for each affects the other through enormously complex feedback loops.

⁴ For examples of this approach to the study of elites, see Deutsch, Karl W. et al., France, Germany, and the Western Alliance (New York: Charles Scribner's Sons, 1967)Google Scholar; Free, Lloyd A., Six Allies and a Neutral (New York: Free Press, 1959)Google Scholar; Seligman, Lester G., Leadership in a New Nation (New York: Atherton Press, 1964)Google Scholar; and Bell, Wendell, Jamaican Leaders: Political Attitudes in a New Nation (Berkeley: University of California Press, 1964)CrossRefGoogle Scholar.

⁵ For examples of this approach to the study of elites see: Barber, James David, The Lawmakers (New Haven: Yale Univ. Press, 1965)Google Scholar; Wolfenstein, E. Victor, The Revolutionary Personality (Princeton: Princeton Univ. Press, 1967)Google Scholar; Payne, James L., Patterns of Conflict in Colombia (New Haven: Yale Univ. Press, 1968)Google Scholar; and Edinger, Lewis J., Kurt Schumacher: A Study in Personality and Political Behavior (Stanford: Stanford Univ. Press, 1965)Google Scholar.

⁶ A number of other scholars have been interested in elite political culture. Nathan Leites drew on intensive, but impressionistic interviews with deputies in Fourth Republic France to speculate On the Game of Politics in France (Stanford: Stanford Univ. Press, 1958)Google Scholar. Lucian Pye described the political culture of Burmese political and administrative elites, although again his intriguing findings are painted with an impressionist's brush. See his Personality, Politics and Nation-Building (New Haven: Yale Univ. Press, 1962)Google Scholar. Some recent research on politics in developing nations has stressesd elite political culture. See, especially, Scott, James A., Political Ideology in Malaysia (New Haven: Yale Univ. Press, 1969)Google Scholar, and, more briefly, Quandt, William B., Revolution and Political Leadership: Algeria, 1954–1968 (Cambridge, Mass.: The M.I.T. Press, 1969)Google Scholar. A similar focus under a different alias has been suggested by George, Alexander, “The ‘Operational Code’: A Neglected Approach to the Study of Political Leaders and Decision-Making,” International Studies Quarterly, XIII (1969), 190–222 CrossRefGoogle Scholar.

⁷ Pye, in Pye, Lucian W. and Verba, Sidney (eds.), Political Culture and Political Development (Princeton: Princeton Univ. Press, 1965), p. 16 CrossRefGoogle Scholar.

⁸ Cf. Brown, Bernard E., “Elite Attitudes and Political Legitimacy in France,” Journal of Politics, 31 (1969), p. 421 CrossRefGoogle Scholar.

⁹ Pye, , “Personal Identity and Political Ideology,” in Marvick, Dwaine (ed.), Political Decision-makers (New York: The Free Press, 1961), p. 297 Google Scholar. The imbedded quotation is from Erikson, Erik H., Young Man Luther (New York: W. W. Norton & Co., 1958), p. 21 Google Scholar.

¹⁰ The evidence and much of the argument of this paper are taken from my dissertation, Politicians and Politics: Themes in British and Italian Elite Political Culture, submitted to Yale University in 1970 and shortly to be published by Yale University Press. For additional methodological information and some additional supporting evidence relevant to this paper, see that source, especially Chap. II. The subject of differences between Italian and British politicians has been avoided in this paper for reasons of space. These differences are, however, of considerable interest and are treated at some length in the larger study, especially at pp. 294–317.

¹¹ Putnam, op. cit., pp. 29–38. Of an initial randomly-drawn sample of 110 MPs in Britain, I was able to interview 85 percent; of the initial random sample of 106 Italian deputati I interviewed 78 percent. More detailed analysis reveals that except for a slight under-representation of incumbent Ministers, these samples do not deviate significantly from the respective universes in terms of any social, personal, or political characteristic for which data are available. All respondents (with the exception of a few Welshmen) were interviewed in their native language.

¹² Throughout this paper I systematically report the inter-coder reliability coefficients for all judgments about which there might reasonably be some doubt. The statistic used is tau-beta, a measure of ordinal correlation between the two independent judgments. In the coding process the independent judgments of the two coders were confronted, and the coders arrived at an agreed final judgment; the statistical analyses reported in this paper are based on these consensual judgments of each interview. For additional discussion of this method of coding, see Putnam, op. cit., pp. 49–59.

¹³ Much of the methodological literature on content analysis is highly relevant to the present research. A concise but comprehensive introduction to this literature is the article by Riley, Matilda White and Stoll, Clarice S. in The International Encyclopedia of the Social Sciences (New York: Crowell Collier and Macmillan, Inc., 1968), Vol. III, pp. 371–377 Google Scholar. An excellent discussion of the problems of coding can be found in Cartwright, D. P., “Analysis of Qualitative Material,” in Festinger, Leon and Katz, Daniel (eds.), Research Methods in the Behavioral Sciences (New York: Dryden Press, 1953)Google Scholar, chap. 10. For a good overview of the methodological problems of “impressionistic” coding, see Selltiz, Claire et al. Research Methods in Social Relations, rev. ed. (New York: Holt, Rinehart and Winston, 1959)Google Scholar, chap. 10.

¹⁴ See Minar, David W., “Ideology and Political Behavior,” Midwest Journal of Political Science, V (1961), pp. 317–331 CrossRefGoogle Scholar; Naess, Arne, Democracy, Ideology, and Objectivity (Oslo: University Press, 1956)Google Scholar; and Barnes, Samuel H., “Ideology and the Organization of Conflict,” Journal of Politics, XXVIII (1966), pp. 513–530 CrossRefGoogle Scholar.

¹⁵ Sartori, Giovanni, “Politics, Ideology, and Belief Systems,” American Political Science Review, 63 (1969), p. 398 CrossRefGoogle Scholar.

¹⁶ A number of scholars use such a definition, even though they appear to be discussing what Sartori terms “ideology in politics.” See, inter alia, Spengler, Joseph J., in Braibanti, Ralph and Spengler, Joseph J., Tradition, Values, and Socio-Economic Development (Durham, North Carolina: Duke Univ. Press, 1961), pp. 3–56 Google Scholar, and Harry Johnson, “Ideology and the Social System,” in The International Encyclopedia of the Social Sciences, op. cit., Vol. 7, pp. 76–84.

¹⁷ See Gross, Feliks (ed.), European Ideologies (New York: Philosophical Library, 1948)Google Scholar, and Burks, R. V., “A Conception of Ideology for Historians,” Journal of the History of Ideas, X (1949), pp. 183–198 CrossRefGoogle Scholar.

¹⁸ For such a conception see, inter alia: Lane, Robert E., Political Ideology (New York: The Free Press, 1962), p. 14 Google Scholar, and Geertz, Clifford, “Ideology as a Cultural System,” in Apter, David E. (ed.), Ideology and Discontent (New York: The Free Press, 1964), pp. 48ff.Google Scholar

¹⁹ The literature is gargantuan. A brief, but representative bibliography is given in David E. Apter (ed.), op. cit., pp. 329–334.

²⁰ Shils, Edward, “Ideology and Civility: On the Politics of the Intellectual,” Sewanee Review, LXVI (1958) pp. 450–480 Google Scholar. All later references to Shils in this paper are to this article. For a later, and essentially unchanged version of his views, see his “The Concept and Function of Ideology,” The International Encyclopedia of the Social Sciences, op. cit., vol. 7, pp. 66–75.

²¹ Shils is, naturally, not the only student of politics to make this argument, but his discussion is the most comprehensive and provocatively phrased and will be used here to represent a long list of contributions by equally distinguished scholars. Other notable examples include Lipset, S. M., Political Man (London: Mercury Books, 1960)Google Scholar, and Bell, Daniel, The End of Ideology (New York: Collier Books, 1961)Google Scholar. The list is potentially endless. See the references in Waxman, Chaim, The End of Ideology Debate (New York: Funk and Wagnalls, 1968)Google Scholar.

²² Waxman, op. cit.

²³ LaPalombara, Joseph, “Decline of Ideology: A Dissent and an Interpretation,” American Political Science Review, LX (1966), pp. 5–16 CrossRefGoogle Scholar, and Seymour Martin Lipset, “Some Further Comments …” ibid., pp. 17–18.

²⁴ The issue is often raised implicitly, however, as in the LaPalombara-Lipset interchange.

²⁵ Naess, op. cit., p. 196.

²⁶ Sartori's contributions focus on elements 3 and 4 of our list, viz., rigidity of beliefs and emotional intensity, sometimes with an admixture of element 5, distortion of reality, (cf. “Ideology … implies a doctrinaire and somewhat unrealistic way of framing political issues.” Sartori, , “European Political Parties: The Case of Polarized Pluralism,” in LaPalombara, Joseph and Weiner, Myron (eds.), Political Parties and Political Development [Princeton: Princeton Univ. Press, 1966], p. 158 Google Scholar. Unfortunately, there is no direct evidence on this point in our data, since strictly speaking, what is required is repeated interviews with the same respondents during an extended period of time in which changes in the “real world” require changes in beliefs. The selection of element 3 as the core of a definition of “ideological” is slightly unorthodox, although in many discussions it plays the role of one of our B _n 's, i.e., a concomitant of “ideology.” In any event, Sartori's conception of ideology is intriguing and deserves some attempts at operationalization.

²⁷ Others who have proposed this interpretation include Spiro, Herbert, Government by Constitution (New York: Random House, 1959), pp. 178–210 Google Scholar; Sidney Verba, “Comparative Political Culture,” in Pye and Verba, op. cit., pp. 544–550; and Waterman, Harvey, Political Change in Contemporary France (Columbus, Ohio: Charles E. Merrill Publishing Company, 1969), pp. 113–138 Google Scholar et passim. Both Verba and Waterman, however, use “political style” (in this limited sense) as only part of their definitions of “ideological” behavior; they also include some of the B _n 's in their definitions.

²⁸ References after each quotation simply indicate the country (“I” and “B”) and respondent number for the interview cited.

²⁹ The tau-beta reliability coefficient—the tau-beta correlation between the judgments by the two coders—was between .43 and .51 for “generalizer-particularizer” and between .37 and .43 for the “inductive-deductive” scale. Of course, there is a methodological difficulty involved in reliably separating judgments on these two factors. But unless one were prepared to argue that the two factors are not closely related to an “ideological style” of policy analysis, the conclusions of this paper would be unaffected by this methodological difficulty.

³⁰ Note that this characteristic concerns not mere passing references to historical events, but rather the use of an historical context in explicating the problem.

³¹ The reliability coefficient for “historical context” ranged between .46 and .60 (2 percent—5 percent were placed at opposite ends of the three-point scale). The reliability coefficient for “moralizer” ranged between .48 and .58 (5 percent were placed at opposite ends of the three-point scale).

³² The reliability coefficients for the stylistic characteristics listed in this paragraph are as follows:

Reference to Group Benefits: tau-beta = .68 (15 percent rated at opposite ends of the three-point scale)

Reference to Acceptability: .37 (8 percent)

Reference to Practicality: .32 (11 percent)

Reliability coefficients have not been calculated for “reference to tradition” and “reference to cost,” but they probably fall in the same range as those given above.

³³ Reliability coefficients have not been calculated for “reference to (named) ideology” and “reference to (past or future) utopias,” but they are fairly high, probably in the range of tau-beta = .50.

³⁴ The mean communality is about .40; that is, about 40 percent of the total variance is common to the set of characteristics, and 60 percent represents purely idiosyncratic or random variation in each trait. Of course, the total communality is attenuated as a function of the unreliability of the individual coding judgments.

³⁵ For the sake of brevity, I present here only the global factor analysis, based on both national samples combined. A much more detailed analysis based on each sample separately confirms this factor structure in all essential respects. See Putnam, op. cit., pp. 242ff. A final reservation must be noted. The factor structure we have discovered is, of course, based ultimately on the coders' judgments. It is possible that this structure reflects not the basic dimensions of political style in some objective sense, but rather the basic dimensions along which coders analyze political style. That is, we may be seeing the esssential structure, not of politicians' behavior, but of coders' behavior. There is no final way of refuting this possibility, except by repeating the study with separate coders independently rating each stylistic dimension. Naturally, that would be enormously expensive. Every attempt was made in the present study to instruct coders to judge each item independently, and the double-coding procedure provides some protection against such biasing. Moreover, there are a few features of the factor structure—such as the presence of “historical context” in the second rather than the first factor—which run counter to our expectations, and therefore counter to the presumptive direction of biasing. In any event, even if the possibility discussed here were wholly correct, it would affect only the dimensional analysis in the present section. The Ideological Style Index would still reflect significant and reliable differences in the way politicians discuss issues. The next section of the paper offers further validation of the ISI.

³⁶ Operationally, the Ideological Style Index is the sum of a respondent's ratings on each of the four component variables, standardized and weighted according to the factor loading of each on Factor 1. Of course, the ISI is dependent in part on the nature of the issue being discussed. Discussions of economic planning tended to be more “ideological” than average, discussions of urban transport somewhat less than average. Since time constraints meant that only two issues were discussed in any given interview, each respondent's precise ISI rating for this study was, obviously, a function to some extent of the random choice of issues. But there is a high correlation between a respondent's ISI score on “his” issue and his score on the “suggested” issue; about two-thirds of those rated “high” on one issue were also rated “high” on the other, regardless of what those issues were. We are, in other words, dealing with a relatively stable feature of an individual politician's style of analyzing different topics.

³⁷ To simplify the presentation of results, I will often contrast “high” and “low” scorers on the Ideological Style Index and several other indices to be presented later in the paper. In all cases “high” and “low” refer to respondents above or below the mean score for both national samples combined.

³⁸ The intercoder reliability for this measure was tau-beta = .57. Only 8 percent of all respondents were rated more than one point apart by the separate coders. This measure is based on a section of the interview quite separated from the discussions of policy. Thus, the risk is slight that either the respondent's discussion of party differences or the coder's judgment of that discussion was directly influenced by the “style” of the earlier policy analysis.

³⁹ The measure was introduced in Campbell, Angus, Converse, Philip E., Miller, Warren E., and Stokes, Donald E., The American Voter (New York: John Wiley & Sons, Inc., 1960), pp. 216–265 Google Scholar. Similar measures have been used in Eldersveld, Samuel J., Political Parties: A Behavioral Analysis (Chicago: Rand McNally Company, 1964), pp. 196–198 Google Scholar, and Barnes, Samuel H., Party Democracy: Politics in an Italian Socialist Federation (New Haven: Yale Univ. Press, 1967), pp. 162–167 Google Scholar. Slight differences in operationalization make direct comparisons of my data with that generated in these other studies difficult, but my two samples of national elites both seem considerably more “ideologically sensitive” then any of the local elites or mass publics studied in these previous projects.

⁴⁰ The tau-beta correlation between the two independent ratings of each interview was .51; only one respondent in six was placed more than one point apart on the five-point scale by the two coders.

⁴¹ I am quite sure, on the basis of independent evidence from outside the interview itself, that this respondent genuinely holds the views quoted here.

⁴² Obviously, there is some circularity to the correlation shown in Figure 1, for in part the judgment about the respondent's “conceptual scheme” is based on his discussion of the policy problems, as is the ISI. There is no completely accurate way of estimating the impact of this fact, but using the “cross-coder correlation technique” merely reduces, and does not eliminate, the relationship. The average correlation between one coder's judgment on the “conceptual scheme” variable and an Ideological Style Index based on the other coder's ratings of the two issue-discussions is r = .45. Note, in addition, that the other variables used to validate the ISI are not in any way subject to this possible methodological criticism.

⁴³ Shils, op. cit., p. 450.

⁴⁴ The intercoder reliability of this placement was quite high; the tau-beta was .76 and only 3 percent of all respondents were placed more than one point apart, while 63 percent were coded identically.

⁴⁵ The first variable is based on a summary judgment of the respondent's overall attitude to the socio-economic system of his country, using the five-point scale indicated in Table 7. The intercoder reliability of this judgment is tau-beta = .54; only 3 percent of all respondents were placed more than one point apart The second variable is based on a similar judgment of the respondent's overall attitude to the existing political institutions of his country, using the five-point scale shown in Table 8. The intercoder reliability of this judgment is tau-beta = .68; not a single respondent was placed more than one point apart by the independent coders.

⁴⁶ Shils, op. cit., p. 452.

⁴⁷ The reliability of this judgment is fairly high, given the diffuse character of the criteria employed. The tau-beta correlation between separate coders' judgments of the same interview was .34; 7 percent of all interviews were rated by the coders at opposite ends of the three-point scale.

⁴⁸ The intercoder reliability of this four-point scale was tau-beta = .53, with 9 percent of all respondents rated more than one point apart by the coders.

⁴⁹ The reliability of the code for “dichotomous thinking” was tau-beta = .62, with 1 percent rated at opposite ends of the three-point scale. The reliability of the code for “tolerance” was tau-beta = .49, again with 1 percent rated at opposite ends of the three-point scale.

⁵⁰ Methodologically, this Index is simply the factor score for each respondent on the dominant factor which emerges from a factor analysis of the matrix of intercorrelations among the four components of the Index. The mean intercorrelation among the four is r = .52 in Britain and r = .56 in Italy.

⁵¹ Christoph, James B., “Consensus and Cleavage in British Political Ideology,” American Political Science Review 59 (1965), p. 629 CrossRefGoogle Scholar.

⁵² Shils, op. cit., p. 452.

⁵³ The point here, of course, is not that ideologues are never dogmatic and hostile—some are and some aren't. Indeed, it may even be the case that some specific ideologies are disproportionately held by particularly intolerant people. I am, however, concerned here to dispute the contention that ideological thinking per se is intrinsically alien to openmindedness.

⁵⁴ One interesting approach to this problem would be to use biographies and (especially) autobiographies of political leaders. There are methodological difficulties with this approach, too, but it would be a useful supplement to more direct evidence.

⁵⁵ The correlations between age and the IPH in each country and the correlation between age and the ISI in Britain are all statistically significant at the .05 level or better.

⁵⁶ Only one Italian out of 83 mentioned anything of the sort. There are many attractions in Italian politics, but warm personal relations with colleagues do not seem to be among them, at least to judge by this evidence.

⁵⁷ See Putnam, op. cit., pp. 162ff.

⁵⁸ Here “West” refers essentially to the countries of Western Europe and North America. In a more detailed analysis we would, of course, need to consider more carefully differences among these nations, but it seems clear that there are important common features of the changes all are experiencing. At least this is true for Britain and Italy.

⁵⁹ For an outstanding, comprehensive summary of the various contributions to this literature, see Waterman, op. cit., passim.

⁶⁰ Lipset, S. M., “The Changing Class Structure and Contemporary European Politics,” Daedalus, XCIII (1964), p. 296 Google Scholar.

⁶¹ Lipset, “Some Further Comments …,” op. cit., pp. 17–18.

⁶² A similar point is made in the case of France by Waterman, op. cit., p. 125, under the heading “the importance of being doctrinal.” A failure to distinguish between the decline or modification of certain specific ideologies and the decline of an ideological style of politics is a grave weakness in the analysis of “ideology” in Lerner, Daniel and Gorden, Morton, Euratlantica: Changing Perspectives of the European Elites (Cambridge, Mass.: The M.I.T. Press, 1969)Google Scholar.

⁶³ Inglehart, Ronald, “Post-Bourgeois Radicalism in France,” unpublished manuscript, p. 51 Google Scholar. This manuscript offers an extraordinarily appealing theory of changing patterns of political cleavage in Western societies, based on the assumption that as basic physiologic and economic needs are satisfied with increasing ease and regularity, drives for belongingness and self-fulfillment will become dominant. See also his “The Silent Revolution in Europe: Intergenerational Change in Post-Industrial Societies,” American Political Science Review (forthcoming).

⁶⁴ Dahl, Robert, “Epilogue,” in Dahl, (ed.), Political Oppositions in Western Democracies (New Haven: Yale Univ. Press, 1967), p. 401 Google Scholar.

⁶⁵ The argument of this paper received interesting confirmation from the behavior of the new Conservative Government elected in Britain in 1970. The consensus of British commentators during its first year in office was that it had been the most “ideological” British Government since the immediate postwar period. It is tempting to suggest that this movement away from traditional Tory political style might have been predicted from the evidence and analysis of the present study, completed more than a year before the Heath Government took office.

⁶⁶ Waterman, op. cit., p. 215.

⁶⁷ Lane, , “The Decline of Politics and Ideology in in a Knowledgeable Society,” American Sociological Review XXXI (1966), pp. 649ff.CrossRefGoogle Scholar

⁶⁸ Waterman, op. cit., p. 123.

⁶⁹ Lipset, for example, has said of “Weltanschauung politics” that “… by definition, such political views do not include the concept of tolerance.” (Emphasis added.) “Some Social Requisites of Democracy: Economic Development and Political Legitimacy,” reprinted in Cnudde, Charles F. and Neubauer, Deane E. (eds.), Empirical Democratic Theory (Chicago: Markham Publishing Company, 1969), p. 173 Google Scholar.

⁷⁰ See, for example, MacRae, Duncan Jr.: “Ideology as a reasoned view of the world is not inconsistent with tolerance for alternative ideologies in the same political system.” Parliament, Parties, and Society in France, 1946–1958 (New York: St. Martin's Press, 1967), pp. 329–330 Google Scholar. See also Samuel H. Barnes, “Ideology and the Organization of Conflict,” op. cit., p. 524.

⁷¹ Rejai, M., Mason, W. L., and Beller, D. C., “Political Ideology: Empirical Relevance of the Hypothesis of Decline,” Ethics LXXVIII (07, 1968), p. 309 Google Scholar.

⁷² Lindblom, , The Intelligence of Democracy (New York: The Free Press, 1965), pp. 137–138 Google Scholar.

⁷³ de Jouvenal, Robert, Le Republique des camarades (Paris: Bernard Grasset, 1914), pp. 58, 62 Google Scholar, as quoted in Waterman, op. cit., p. 120. This thesis is supported by a recent study of Swedish decision-making in the area of metropolitan reform, which traces the effectiveness of the Stockholm reformers precisely to the “pragmatism” and “particularization” which characterize the Swedish style of policy analysis. Anton, Thomas J., “Incrementalism in Utopia: The Political Integration of Metropolitan Stockholm,” Urban Affairs Quarterly V (1969), pp. 75–76 Google Scholar.

⁷⁴ This summary is drawn from Lindblom, The Intelligence of Democracy, op. cit., pp. 143–151, and from Lindblom, Charles E. and Braybrooke, David, A Strategy of Decision (New York: The Free Press, 1963), pp. 81–110 Google Scholar.

⁷⁵ Duncan MacRae, Jr. loc. cit. (June, 1967), pp. 3–12.

⁷⁶ Shonfield, , “The Pragmatic Illusion,” Encounter (06, 1967), pp. 3–12 Google Scholar.

⁷⁷ Shonfield, op. cit., pp. 10–11.

⁷⁸ Shils, op. cit., p. 465.