¹ See Lipset, Seymour Martin, Political Man: The Social Bases of Politics (Garden City, N.Y.: Doubleday, 1959), 285–309 Google Scholar for an extensive summary of findings through the mid-1950s.

² Robert E. Lane provides a useful summary of postwar economic changes, as well as a provocative discussion of the possible political consequences of affluence, in “The Politics of Consensus in an Age of Affluence,” The American Political Science Review, 59 (December, 1965), 874–895 CrossRefGoogle Scholar. In particular, see pp. 884–889 for Lane's discussion of the effects of affluence upon partisan choice.

³ These data were made available by the Inter-University Consortium for Political Research. The Consortium bears no responsibility for the analyses and interpretations presented here.

⁴ See Campbell, Angus, Converse, Philip E., Miller, Warren E., and Stokes, Donald E., The American Voter (New York: Wiley, 1960), pp. 356–361 Google Scholar.

⁵ For a discussion of status polarization, see Campbell et al., pp. 344–346.

⁶ Campbell et al., pp. 356–357.

⁷ Campbell et al., p. 357.

⁸ Campbell et al., p. 357.

⁹ Campbell et al., p. 359.

¹⁰ Alford, Robert R., Party and Society: The Anglo-American Democracies (Chicago: Rand McNally, 1963), pp. 219–249 Google Scholar.

¹¹ Alford, p. 226. Alford provides more recent analysis, which includes results through 1964, in “Class Voting in Anglo-American Democracies,” in Party Systems and Voter Alignments, ed. Lipset, S. M. and Rokkan, Stein (New York: Free Press, 1967), pp. 84–88 Google Scholar.

¹² Alford, , Party and Society, pp. 224 Google Scholar.

¹³ Social class was defined on the basis of the head of household's occupation. If the head of household was not employed, class was defined on the basis of his occupation when he worked. Professionals, semi-professionals, self-employed, managerial, supervisory, and other white-collar workers were classified as middle class; skilled workers, semi-skilled workers, unskilled workers, service workers, and protective service workers were classified as working class. (This dichotomous division is identical to that employed by Alford in his analysis of the S.R.C. surveys.)

¹⁴ Direction of vote is based upon the postelection report of how the respondent voted. Persons who did not vote, or who voted for minor parties, have been excluded from these figures, although the effects of including Wallace voters in the 1968 analysis are reported below.

¹⁵ The term age-cohort is used to designate any group of persons born during the same time period. Except for the 1952 and 1968 surveys, in which year of birth was recorded, year of birth has been calculated from data about respondent's age.

To maximize longitudinal comparability the arbitrarily defined ten-year cohorts employed in the 1948 survey were used, even though more detailed data about age were available for subsequent surveys. For persons born after 1923, eight-year cohorts were used. Ideally four-year cohorts should have been employed since presidential elections are held every four years, but the small Ns precluded this refinement.

¹⁶ As blacks are largely working class and heavily Democratic, to include blacks in the analysis would increase the proportion of working-class respondents who voted Democratic. Including blacks thus increases the relationship of social class to party choice, having its greatest impact upon that relationship when black turnout is high and when the proportion of white workers voting Democratic is low. Including blacks has little effect on the overall relationship between class and voting except in 1968, and these effects are noted below (see note 23).

¹⁷ For a discussion of this measure see Alford, , Party and Society, pp. 79–86 Google Scholar.

¹⁸ The fivefold ordinal classification of occupational ranking was as follows: professionals and semi-professionals; self-employed, managerial, and supervisory; other white-collar; skilled and semi-skilled workers; unskilled workers, service workers, and protective service workers.

¹⁹ Campbell and his colleagues used a tau-b to measure this relationship. It was impossible to replicate precisely the measure of occupational status used by the University of Michigan authors since there are some ambiguities in their description of their measure. However, my overall measures of status polarization for the total population closely approximate their results.

²⁰ See Alford, , Party and Society, pp. 352–353 Google Scholar.

²¹ In the calculation of levels of class voting in 1944, service workers were classified as working class. In his analysis of the 1944 N.O.R.C. surveys, Alford classified service workers as middle class. Regardless of whether service workers are considered working class, middle class, or are excluded from the analysis, respondents between the ages of 30 and 39 (born between 1905 and 1914) manifested higher levels of class voting than did the other cohorts studied. A strong cautionary note is in order. Given that the 1944 sample was of the civilian population, the youngest cohort sampled in 1944 may be highly atypical since a large proportion of the men were in the armed forces.

²² A measure of status polarization comparable with that used in this study cannot be calculated, since the N.O.R.C. did not use the same fivefold occupational ranking employed by the Survey Research Center.

²³ As noted above (see note 16), including blacks has an impact on the association between class and voting in the 1968 election. Including nonwhites raises class voting in 1968 to + 15, status polarization to .18.

²⁴ Among working-class white voters, 15 per cent voted for Wallace; among middle-class white voters, 10 per cent did.

²⁵ If nonwhites are included, class voting rises to + 11, status polarization to .14.

²⁶ Rintala, Marvin, “Political Generations,” in International Encyclopedia of the Social Sciences (New York: Macmillan, 1968), vol. 6, p. 93 Google Scholar. Such experiences, as Rintala notes, can create divisions within generations as well as between generations.

²⁷ See Converse, , “Of Time and Partisan Stability,” Comparative Political Studies, 2 (July, 1969), 139–171 CrossRefGoogle Scholar. See also, Campbell, et al., The American Voter, pp. 125–128 Google Scholar.

²⁸ Obviously, many individuals will have formative socialization experiences at later stages in their life. Some individuals will experience career mobility from one class to another. The generational model assumes that for most persons late adolescence and early adulthood will be especially significant in shaping partisan attachments, and that for many persons these attitudes will be relatively stable.

²⁹ As is apparent from the data in this article, the net relationship between class and party for any given cohort will vary from election to election. Even for the Depression cohort that relationship was weak in 1956 when the conditions of that election contributed to a low relationship between class and party.

³⁰ Based upon recall questions in the 1952 survey, 42 per cent of the voters born between 1924 and 1931 voted in 1948. (Among 1952 voters born between 1924 and 1927, 68 per cent had voted in 1948.) Even among those who had voted in 1948, class voting in 1952 was only + 6. Since the Ns for this subset of 1948 voters are low, one should be cautious in making inferences from this finding. But there is little evidence that participation in the 1948 election was a formative experience contributing to high levels of class voting among first-time participants in that election.

³¹ It should be clear from this discussion that there could be circumstances under which these two explanations would lead to similar predictions. In such cases a longitudinal comparison could not distinguish between the explanations.

³² For an excellent discussion of the methodology of cohort analysis, see Hyman, Herbert H., Secondary Analysis of Sample Surveys: Principles, Procedures, and Potentialities (New York: Wiley, 1972), pp. 274–290 Google Scholar.

For a cohort analysis that tests the explanations developed here within the context of European political behavior, see Abramson, Paul R., “Social Class and Political Change in Western Europe: A Cross-National Longitudinal Analysis,” Comparative Political Studies, 4 (July, 1971), 131–155 CrossRefGoogle Scholar.

³³ A basic assumption of cohort analysis is that one has randomly selected members of each cohort. All six S.R.C. surveys were based upon national random samples. Even when a random sample is used, however, the probability that subgroups will represent the universe of persons who belong to those subgroups is less than the probability that the total sample will reflect the total universe of persons from whom the sample was drawn. On the other hand, when the same relationships occur among a subgroup in numerous surveys based upon separately conducted samples one can have more confidence that the relationships that result have not occurred by chance. (It should be noted that more than half the respondents sampled in 1960 were drawn from a panel survey begun in 1956.)

³⁴ The low relationship of social class to partisan choice among the two oldest cohorts in 1960 results from high levels of Republican voting among working-class Protestants. But I do not have a satisfactory explanation for the high relationship of class to party choice among the youngest cohort.

³⁵ Norval D. Glenn combined the results of five Gallup surveys conducted shortly after the 1968 election and his results are based upon Ns substantially larger than those in my study. He found that among whites between the ages of 21 and 29 white-collar workers were more likely to vote Democratic than blue-collar workers, a difference which became more pronounced if Wallace voters were included in the analysis. See Glenn, Norval D., “Class and Party Support in the United States: Recent and Emerging Trends.” Public Opinion Quarterly, 37 (Spring, 1973), 1–20 CrossRefGoogle Scholar.

³⁶ There is some historical basis for this division. Persons born before 1924 were first eligible to vote before or during World War II; persons born after 1923 first entered the electorate after the War. But the generational explanation obviously does not suggest sharp cut-off points between persons born in one year and those born the next, since the age at which formative socialization occurs will vary from individual to individual.

³⁷ In 1952, the decline in class voting was more the result of the low level of support for Stevenson among the young working class. In all subsequent elections there were only negligible differences in levels of Democratic voting among workers born before 1924 and those born after 1923.

³⁸ Such differences are obvious for 1952, 1956, 1964, and 1968 merely by inspecting the appropriate figures. Even in 1960, however, the mean level of Democratic voting among middle-class voters born after 1923 was marginally higher than that of middle-class voters born before 1924. Data for 1948 also support this trend, but the Ns are too small to be meaningful.

³⁹ For a comparison of the social bases of partisan support in these two elections, see Converse, Philip E., Miller, Warren E., Rusk, Jerrold G., and Wolfe, Arthur C., “Continuity and Change in American Politics: Parties and Issues in the 1968 Election,” The American Political Science Review, 63 (December, 1969), 1083–1105 CrossRefGoogle Scholar.

⁴⁰ These percentages include persons born between 1940 and 1943 who were not included in Figure 5.

⁴¹ These percentages include persons born between 1884 and 1893 who were not included in Figure 6.

⁴² See Blau, Peter M. and Duncan, Otis Dudley, The American Occupational Structure (New York: Wiley, 1967), pp. 231–238 Google Scholar.

⁴³ In this analysis being Catholic provides an indirect measure of having ethnic origins that contributed to voting Democratic. Although the Survey Research Center surveys for 1964 and 1968 provided detailed data about ethnic origins, there was a tendency for respondents to report themselves as “American only” if both parents were born in the United States. When ethnic origin was used as a variable, the younger middle class was only marginally more likely to have Irish, East European, and Southern European origins than was the older middle class.

⁴⁴ The four items used measured support for federal government involvement in aiding education, promoting jobs and Prosperity, supporting school integration, and guaranteeing public accomodations for Negroes. (The pecific items were Variables 66, 78, 100, and 408 in 1964 and Variables 60, 66, 75, and 78 in 1968.)

Respondents scoring “high” on the index generally favored federal government activism in three or more of these areas; those scoring “low” generally opposed government involvement in three or more of these areas.