² Cf. Cooley, Charles H., Social Organization (New York, 1909)Google Scholar, especially chs. 3 and 4. For a review of current theories and research on the role of primary groups, see Shils, Edward, “The Study of the Primary Group,” in Lerner, Daniel and Lasswell, Harold, The Policy Sciences (Stanford, 1951), pp. 44–69Google Scholar; Katz, Elihu and Lazarsfeld, Paul F., Personal Influence (Glencoe, Illinois, 1955)Google Scholar, chs. 2–5; also Cartwright, Dorwin and Zander, Alvin, Group Dynamics: Research and Theory (Evanston, Illinois, 1953)Google Scholar, passim. Two valuable works dealing with small group influence on the formation of political attitudes have become available since this manuscript was prepared: Hyman, Herbert H., Political Socialization: A Study in the Psychology of Political Behavior (Glencoe, 1959)Google Scholar; and Verba, Sidney, The Experimental Study of Politics: The Contribution of Small Group Experiments in Leaderthip to the Understanding of Political Leadership, unpublished dissertation, (Princeton University, 1959)Google Scholar.

³ Cf. Berelson, Bernard R., Lazarsfeld, Paul F., and McPhee, Wm. N., Voting (Chicago, 1954), pp. 88–9, 92–3, 96–7, 120–2Google Scholar; Lazarsfeld, P. F., Berelson, B. R., and Gaudet, Hazel, The People's Choice (New York, 1948)Google Scholar ch. 15; Campbell, Angus, Gurin, Gerald, and Miller, Warren E., The Voter Decides (Evanston, Illinois, 1954), pp. 199–206Google Scholar; Milne, R. S. and Mackenzie, H. C., Straight Fight (London, 1954), pp. 44–5, 123–5Google Scholar; Maccoby, Eleanor, Matthews, Richard E., and Morton, Anton S., “Youth and Political Change,” Public Opinion Quarterly, Vol. 18 (Spring, 1954), pp. 23–9CrossRefGoogle Scholar; Alice S. Kitt and David B. Gleicher, “Determinants of Voting Behavior,” ibid., Vol. 14 (Fall, 1950), pp. 393–412; Benney, Mark and Geiss, Phyllis, “Social Class and Politics in Greenwich,” Brit. J. of Sociology, Vol. 1 (1950), pp. 324–7CrossRefGoogle Scholar; Helfant, Kenneth, “Parents' Attitudes vs. Adolescent Hostility, in the Determination of Adolescents' Sociopolitical Attitudes,” Psychological Monographs, Vol. 66 (1952)CrossRefGoogle Scholar.

⁴ Kitt and Gleicher, op. cit., pp. 401–2; Berelson et al., op. cit., pp. 120–2; on the effects of “cross-pressures” of various types on the decision to vote, see Lazarsfeld, Berelson, and Gaudet, op. cit., pp. 56–64. However, in their study of the 1952 election, op. cit., pp. 202–3, Campbell et al. found no connection between level of participation and membership in groups of “divided political loyalty.”

⁵ Lazarsfeld, et al., op. cit., p. 142; Campbell et al, op. cit., p. 202.

⁶ For a review of theories concerning the psychology of small group influence, see Riecken, Henry W. and Homans, George C., “Psychological Aspects of Social Structure” in Lindzey, Gardner (ed.), Handbook of Social Psychology (Cambridge, Massachusetts, 1954), Vol. II, pp. 786–832Google Scholar; Katz, Daniel, “Social Psychology and Group Processes,” Annual Review of Psychology (1951), Vol. 2, pp. 137–172CrossRefGoogle ScholarPubMed; Hovland, C. I., Janis, Irving L. and Kelley, H. H., Communication and Persuasion (New Haven, 1953)Google Scholar, ch. 5. An important attempt to elaborate and to apply a theory of group structure and influence has been made by Festinger, Leon, Schachter, Stanley, and Back, Kurt, Social Pressures in Informal Groups (New York, 1950)Google Scholar, ch. 9; also by Homans, George C., The Human Group (New York, 1950)Google Scholar. For summaries of small group theory as it applies more immediately to the research problem considered below, see Riecken, Henry W., “Primary Groups and Political Party Choice,” in Burdick, E. and Brodbeck, A. J., American Voting Behavior (Glencoe, 1959)Google Scholar, ch. 8; and Katz and Lazarsfeld, op. cit., pp. 34–115.

⁷ See Shils, E. A. and Janowitz, Morris, “Cohesion and Disintegration in the Wehrmacht in World War II,” Public Opinion Quarterly, Vol. 12 (1948), pp. 280–315CrossRefGoogle Scholar; also Newcomb's, T. M. study of the Bennington College community in Personality and Social Change (New York, 1943)Google Scholar, and Whyte's, W. F.Street Corner Society (Chicago, 1943)Google Scholar.

⁸ For experimental data bearing on these tendencies, see Schachter, Stanley, “Deviation, Rejection, and Communication,” Journal of Abnormal and Social Psychology (1951) 46, pp. 196–207CrossRefGoogle Scholar; Kurt Back, “Influence through Social Communication,” Ibid., pp. 9–23; Kelley, H. H. and Volkhart, E. H., “The Resistance to Change of Group Anchored Attitudes,” American Sociological Review (1952) Vol. 17, pp. 453–465CrossRefGoogle Scholar; Festinger, L. and Thibaut, John, “Interpersonal Communication in Small Groups,” reprinted in Swanson, G. E., Newcomb, T. M., and Hartley, E. L., Readings in Social Psychology (New York, 1952), pp. 125–134Google Scholar. For evidence that group influence may be exercised even in the absence of cohesiveness, see the landmark experiments by S. E. Asch and M. Sherif, brief descriptions of which are available, ibid, pp. 2–11, and 249–262.

⁹ Festinger, Schachter, and Back, op. cit., pp. 168–176; Katz and Lazarsfeld, op. cit., pp. 53–6.

¹⁰ The theoretical statement and summary of research bearing on these processes is set forth most thoroughly by Festinger, Leon, A Theory of Cognitive Dissonance (Evanston, 1957)Google Scholar.

¹¹ See Sherif, loc. cit.

¹² Festinger and Thibaut, loc. cit.; Schachter, loc. cit.

¹³ Katz and Lazarsfeld, op. cit., p. 62.

¹⁴ The interview schedule is too lengthy, and the manner of constructing the indexes too complex, to be described in detail in this paper. Information concerning them may be had upon request, however, through the Laboratory for Research in Social Relations, University of Minnesota.

¹⁵ The classification of voters as stable, moderate, or unstable was determined from their responses concerning: present party preference, first vote, first presidential vote, presidential candidates supported in 1944, 1948, 1952, present preferences in congressional and state elections, self-description of voting habits and consistency of party support, etc.

¹⁶ The effect of such factors on stability of preference was explored in an unpublished study by Herbert McClosky and Norris C. Ellerston, carried out through the Laboratory for Research in Social Relations. The degree to which voters responded to these factors was found to be determined largely by the stability of their party attachments.

¹⁷ Unless otherwise indicated, all differences discussed are statistically significant at, or beyond, the .05 level.

¹⁸ Apparently these relationships are also somewhat affected by the strength of the family's indoctrination, the more strongly indoctrinated favoring the party of their parents more frequently and consistently than the weakly indocrinated. The differences are not large, however, and are only significant at the .07 level. Possibly our measure of “strength of family influence,” resting as it does on retrospective reporting, is too insensitive to give full scope to this variable.

¹⁹ The “index of family reinforcement,” which was built to measure this variable, takes into account the initial and current party outlook of each family member, siblings as well as parents. It includes the direction and intensity of the initial indoctrination and of all adult members of the family at present, all of which are scored in relation to the respondent's party preference. The “intensity” scores, in turn, take into account levels of political interest, frequency of political discussion with parents and siblings, strength of party attachment among the several family members, etc.

²⁰ The measure of “physical distance” took into account place of residence of each family member, the actual geographic distance between them, and the average number of times respondent saw each of them. The cohesiveness measure was comprised of a scale of items testifying to various aspects of family harmony and solidarity, which was principally developed, and generously made available to us, by Jansen, Luther T.. See his “Measuring Family Solidarity,” American Sociological Review (December, 1952), Vol. 17, pp. 727–33CrossRefGoogle Scholar. We should have liked, of course, to allow “cohesiveness” to vary while controlling for “physical distance” (and conversely), so tha t we could measure the effect of each of them separately. The size of the samples did not make this feasible, however, so we paired them in this way.

²¹ Lazarsfeld, Berelson, and Gaudet, op. cit., ch. 3. The Index of Political Predisposition combines three criteria—occupation, religion, and rural-urban residence—in ascertaining a voter's “natural” party predisposition. Ours was an urban sample, which meant that we could not employ the rural-urban criterion at all. Our index also differs from Lazarsfeld's in that its SES rating took into account income as well as occupation. Furthermore, we did not rely on the Catholic-Protestant distinction alone, but took into account the rate of church attendance and the Republican or Democratic predispositions of the specific Protestant denominations. Finally, our index also used education as a criterion. Each of these was weighed and combined into a single life-style score for each respondent. In determining the weights for the index we are indebted to Angus Campbell and the Survey Research Center for making available to us in advance data on the demography of party affiliation from their 1952 election survey.

²² For voting patterns arising from religious affiliation, see Allinsmith, W. and Allinsmith, B., “Religious Affiliation and Politico-Economic Attitude: a Study of Eight Major U. S. Religious Groups,” Public Opinion Quarterly, Vol. 12 (1948), pp. 377–389CrossRefGoogle Scholar.

²³ Cf. Campbell, et al. , The Voter Decides, p. 201Google Scholar; Maccoby et al., op. cit.; Lazarsfeld, et al. , The People's Choice, pp. 140–5Google Scholar.

²⁴ The Elmira study found, for example, that voters, and especially uncertain or undecided voters, avoid talking with people who hold different views. See Baxter, R. H., “Interpersonal Contact and Exposure to Mass Media during a Presidential Campaign,” unpublished dissertation (Columbia University, 1951)Google Scholar.

²⁵ The multiple correlation for stability, family reinforcement, spouse, and peer group preference is .49, even if social distance is omitted. This is consistent with our assumption that social distance variables can, for many purposes, be viewed as indirect measures of primary group membership.

²⁶ Key, V. O. and Munger, Frank, “Social Determinism and Electoral Decision: the Case of Indiana,” in Burdick, and Brodbeck, , American Voting BehaviorGoogle Scholar, ch. 15.

²⁷ Maccoby et al., op. cit.; Berelson et al., op. cit.