¹ Pulzer, Peter G. J., Political Representation and Elections: Parties and Voting in Great Britain (New York: Praeger, 1967), p. 98Google Scholar.

² For example, see Nordlinger, Eric A., The Working Class Tories: Authority, Deference and Democratic Stability (Berkeley and Los Angeles: University of California Press, 1967)Google Scholar and McKenzie, Robert T. and Silver, Allan, Angels in Marble: Working Class Conservatives in Urban England (Chicago: University of Chicago Press. 1968)Google Scholar.

³ For example, see Abrams, Mark, “Party Politics After the End of Ideology,” in Allardt, Erik and Littunen, Yrjö, eds. Cleavages, Ideologies and Party Systems: Contributions to Comparative Political Sociology (Helsinki: Westermarck Society, 1964), pp. 53–63Google Scholar. My analysis demonstrates that Abrams is wrong in his assertion that “Middle class deviants [i.e., those who vote Labour] … are likely to be at least second generation middle class” (pp. 57–58).

⁴ The sampling procedures are described by Butler, David E. and Stokes, Donald E., Political Change in Britain: Forces Shaping Electoral Choice (New York: St. Martin's Press, 1969), pp. 449–462Google Scholar. Although the first wave included 2,009 respondents, the total N in Table 1 is 1832. I have excluded from the analysis those for whom the social class of the “person to be graded” was not available, or for whom no data about the “social class of father when respondent was a child” were available. I have also excluded respondents for whom direction of party identification was coded as “not applicable.”

⁵ Butler and Stokes, p. 104.

⁶ Butler and Stokes, pp. 97–98.

⁷ See Butler and Stokes, p. 97. If Butler and Stokes had classified lower nonmanual employees as middle class, they would have classified 30 per cent of their sample as objectively mobile.

⁸ To be more precise I assigned persons to a class on the basis of the “social class of the person to be graded.” In most cases that person was either the respondent or the respondent's husband. Butler and Stokes also assigned persons to a class on the basis of the occupation of the head of household.

⁹ The rationale for Butler and Stokes' decision is discussed in Butler and Stokes, pp. 68–73. I have employed the more generally used manual, nonmanual distinction. When studying the relationship of social mobility to partisan choice, mobility appears to be a threshhold phenomenon. (See Goldberg, Arthur S., “Social Determinism and Rationality as Bases of Party Identification,” American Political Science Review, 63 (03 1969), 22–25CrossRefGoogle Scholar.) My analysis of Butler and Stokes' data shows that mobility within either the manual or nonmanual occupational categories contributes little to partisan change. Persons mobile across the manual, nonmanual threshhold differ markedly in their partisan preferences from members of their class of origin.

¹⁰ This may be seen if one recalculates the data in Table 2. Among the 274 respondents with working-class Conservative fathers, 36 per cent were upwardly mobile; among the 520 respondents with working-class Labour fathers, only 21 per cent were upwardly mobile.

¹¹ Butler and Stokes introduced controls for the partisan preferences of the respondents' parents, where-as I controlled for the partisanship of the respondents' fathers.

Butler and Stokes found that upwardly mobile respondents with Labour parents were more likely to support the Conservatives than were downwardly mobile respondents with Conservative parents to support Labour. As the data in Table 2 show, I found the reverse to be true. This is only a minor discrepancy, however. In the first place, both my finding and the Butler and Stokes finding are based upon small N's. Is my analysis, there were only 50 downwardly mobile respondents with Conservative fathers. Butler and Stokes did not report the N's upon which their percentages were based, but the N of downwardly mobile respondents must be small since they classified only two per cent of their entire sample as downwardly mobile. In the second place, by restricting my analysis to the 1963 panel I have slightly overrepresented Labour's normal strength, since 1963 was a period of peak Labour popularity in the public opinion polls.

¹² Kenneth H. Thompson found a similar pattern in his analysis of British survey data collected in 1962. See “A Cross-National Analysis of Inter-generational Social Mobility and Political Orientations,” Comparative Political Studies, 4 (04, 1971), 8Google Scholar. However, Thompson was not able to add controls for the respondents' parents' party. Paul R. Abramson and John W. Books found a similar pattern in their analysis of a survey of British youth conducted in 1963. See “Social Mobility and Political Attitudes: A Study of Intergenerational Mobility among Young British Men,” Comparative Politics, 3 (04, 1971), 420–426Google Scholar. These patterns held even after they controlled for the respondents' fathers' party.

¹³ There is only one exception to this pattern. Upwardly mobile respondents with Conservative fathers are more likely to support the Conservatives than are nonmobile middle-class respondents with Conservative fathers.

¹⁴ Butler and Stokes do not present data on the partisan preferences of nonmobile respondents. Nonetheless, a careful comparison of their data on the partisan preferences of mobile respondents with their data on the partisan preferences of all working and middleclass respondents reveals a pattern substantially similar to the one I report.

¹⁵ This may be seen if one recalculates the data in Table 1. Thirty-nine per cent of the respondents did not support the party of their class; of these respondents, 36 per cent were socially mobile.

¹⁶ There is a possible alternative explanation for upwardly mobile respondents being more likely to support the Labour party than nonmobile middle-class respondents are. Upwardly mobile persons tend to occupy lower occupational positions within the middle class than nonmobile middle-class persons hold. Differences between mobile and nonmobile middle-class respondents might be the result of residual variation within the middle class. But these differences are only in small part the result of residual variation. Even within occupational grades, upwardly mobile respondents were less likely to support the Conservatives than were nonmobile middle-class respondents.