Sie sind auf Seite 1von 10

American Association for Public Opinion Research

A Suggested Index of the Association of Social Class and Voting Author(s): Robert R. Alford Source: The Public Opinion Quarterly, Vol. 26, No. 3 (Autumn, 1962), pp. 417-425 Published by: Oxford University Press on behalf of the American Association for Public Opinion Research Stable URL: http://www.jstor.org/stable/2747230 Accessed: 09/11/2010 11:45
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aapor. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Association for Public Opinion Research and Oxford University Press are collaborating with JSTOR to digitize, preserve and extend access to The Public Opinion Quarterly.

http://www.jstor.org

LIVING

RESEARCH

Living Research is a section of the Quarterly reserved for brief reports of research, discussions of unsolved problems, presentations of neat methodological tricks, and other items that arise out of the daily work of researchers. It provides an opportunity for discussion in print of questions and results that may not seem to warrant a full-blown article. Researchers will find in this section a place to exhibit data which are not adequate to substantiate important generalizations but seem promising in their implications or suggest lines of further theoretical exploration. Succinct case histories are welcomed, as well as hypotheses and insights that may be useful to other students of public opinion. Notes published here are not intended to rank below the regular articles in quality or significance, but are distinguished by their shiorter length, greater informality, and more tentative nature.

A SUGGESTED INDEX OF THE ASSOCIATION OF SOCIAL CLASS AND VOTING


BY ROBERT R. ALFORD* In order to construct an index of the association of social class and voting behavior in two-party parliamentary systems, a suitable measure of social class must of course be found, and voting patterns must be classified. An appropriate measure of social class in modernized societies depends entirely on the theoretical purposes and assumptions of a particular research problem. No single social characteristic of individuals can adequately measure their economic life chances, their community prestige, or their power over other individuals. In addition to this general difficulty, social changes now most marked in Westerr societies have created severe problems for the analysis of the influence of social class upon political behavior. The decreasing size of a visible working class and its apparently decreasing organizational solidarity deprive single indexes of social class of much significance. Neither subjective class identification nor a composite measure of objective class position (a measure, for example, combining education, income, and occupation) is of great help, because they hide the very complications which should be analyzed-the discrepancies between these measures, and the different kinds of attitude and behavior which the discrepancies produce. By trying to establish the "best" measure of social class, researchers may reify the concept, even if they are conscious of its complexities. When they find that such a composite measure
* The author is Assistant Professor of Sociology, and Associate Director of the Survey Research Laboratory, at the University of Wisconsin.

418

PUBLIC OPINION QUARTERLY

does not predict behavior very well, they may conclude too readily that social class is no longer of much importance.1 Where the stratification order is not the principal object of study (the dependent variable), occupation can probably be used as a convenient way of measuring objective social class position. For a comparative study of voting behavior, occupation is probably the best single indicator.2 But the effect of other status characteristics should be examined as well, not to show the deficiencies of occupation as an index of class position but to clarify the complex ways in which people in roughly similar social positions may differ politically. If the changing importance of status or class position for political behavior is of concern, then comparing groups defined in a roughly similar way at a number of points in time should reveal any tendency to come together politically. If the political meaning of being in a "middle class" or a "working class" occupation has shifted or differs from country to country, it can be discovered by such research.3 The best empirical measure of class probably will be different if the problem is to locate individuals within a given stratification system than if the main concern is to locate strata characterized by a number of predominant attributes. Much discussion of the methodological problems of defining classes assumes that the main problem is that of
1 A recent stuidy of class and party in a Norwegian village found that occupational differences remained the chief basis of stratification and political cleavage even though substantial differences of neither income nor prestige were present (George K. Park and Lee Soltow, "Politics and Social Structure in a Norwegian Village," American Journal of Sociology, Vol. 67, 1961, pp. 152-165). This suggests that, under some conditions, using occupation as an index of stratification may lead to somewhat different relationships than if income or prestige were used, and possibly may give a more static picture (which need not mean a more distorted one). By comparison, a study of political behavior in some future United States election might find that if income were used as the index of stratification, no relation would be found between it and political behavior. Such a result would show that this particular criterion of stratification was no longer relevant for attitudes and behavior, not that stratification itself did not exist or have consequences. In a period of social change, the validity of indexes of major structural phenomena such as stratification may need to be reconsidered over a period of time. 2 The authors of The American Voter note that, among the objective indicators of class, "occupation tends to predict political attitudes and behavior most efficiently" (A. Campbell, P. E. Converse, W. E. Miller, and D. Stokes, The American Voter, New York, Wiley, 1960, p. 344). See also, A. H. Birch, Small Town Politics, London, Oxford University Press, 1959, pp. 104-106, for a brief discussion of the problems of classifying occupations into "classes." 3 For discussions of changes in the class structure of the United States, see Kurt Mayer, "Recent Changes in the Class Structure of the United States," in Transactions of the Third World Congress of Sociology, 1956, pp. 66-8o, and Ely Chinoy, "Social Mobility Trends in the United States," American Sociological Review, Vol. 20, 1955, pp. 18o-186. Similar patterns of change have been foreseen for most of the two-party parliamnentary nations, which are among the wealthier nations in the world.

LIVING RESEARCH

419

placing individuals.4 If the purpose is to determine the location and attributes of strata, then the intercorrelation of various attributes of status becomes a principal methodological line of attack upon the problem of identifying classes.5 The implication of this argument for the construction of an index of the association of social class and voting is that a number of single empirical indexes can probably be used which more or less efficiently indicate the existence of social strata with certain attributes (income, education, occupation, prestige, style of life, historical continuity, etc.). Whether survey data on income, education, subjective class identification, or occupation are adequate, and which item is the best such indicator is an empirical question, but several studies seem to show that the best predictor of other such attributes (both of individuals and of strata) is occupation. A comparison of nineteen different indexes of status has been made, using the method of factor analysis, and the variable most closely related to the first factor extracted was occupation. Although this conclusion was based only on United States data, it may hold for a number of Western societies.6 Such a single index has obvious shortcomings, because of the probable low level of consistency of various attributes of status (in the United States at least), but is particularly suitable for comparative research. The differences among countries, cities, or regions in the level of what has been called "status [or class] crystallization" constitute an important problem, but such differences do not preclude the use of a single measure of class position.7 First, as already mentioned, occupation is probably the best single predictor of other class characteristics (and this may well be true in other countries as well as in the United
4 Leonard Reissman's excellent analysis of Class in American Society (Glencoe, Ill., Free Press, 1959) is a case in point. He suggests that "four kinds of criteria have been used to identify social classes: 1) How the person lives, 2) What others think of him, 3) What he thinks of himself, and 4) What he does." All these criteria assume that the location of individuals is the key problem, not the location of strata. 5 See Werner Landecker, "Class Boundaries," American Sociological Review, Vol. 25, 1960, pp. 868-877, for one of the first empirical attempts to delineate strata by utilizing survey data. 6 Joseph A. Kahl and James A. Davis, "A Comparison of Indices of Socio-economic Status," American Sociological Review, Vol. 20, 1955, pp. 317-325. 7 No comparative research has been done using the concept of status or class crystallization, although it would certainly be possible. Existing studies have focused upon single cities or national samples without regard for regional or other area variations, and the failure to find more differences in behavior among persons with consistent and discrepant statuses may be due to the lack of comparative focus, not to the lack of importance of this as an independent variable. See Gerhard E. Lenski, "Statuscrystallization: A Non-vertical Dimension of Social Status," American Sociological Review, Vol. 19, 1954, pp. 405-413, and Irwin W. Goffman, "Status Consistency and Preference for Change in Power Distribution, American Sociological Review, Vol. 22, 1957, pp. 275-281.

420

PUBLIC OPINION QUARTERLY

States). Second, the very imperfection of occupation as an index is an integral part of stratification in this type of society and political system. The prosperity, high level of social mobility, and relative wealth of the countries with two-party parliamentary systems imply that a relatively high proportion of their populations is moving upward or downward, is in contact with other social classes or persons from other social origins, and has "contradictory" class attributes and experiences. For the purposes of constructing an index of the association of social class and voting, an additional manual-nonmanual distinction has some theoretical justification. The move across the manual-nonmanual "line" seems to have a similar meaning in most Western societies. The shift from a blue-collar to a white-collar job has more significance-personally and socially-than a shift within each category, although the relative prestige and life chances of various occupations differ considerably within either broad category.8 But, more important, the remaining contrasts between middle class and working class in styles of life, education, and values probably justify calling the manual-nonmanual division a "class" distinction which would be blurred if the constant empirical focus were upon political differences between professionals and white-collar workers, or skilled and unskilled laborers. Such differences are important, however, and the focus on class neglects many important differences within the middle-class and the working-class electorates.9 This usage of the term "class"is similar in one respect to Max Weber's usage, which distinguishes a class from a "community" or self-conscious status group. "Classes"merely represented possible and frequent bases for communal action to Weber, who specified that a class exists when a "number of people have in common a specific causal component of their life-chances and this component is represented exclusively by economic interests in the possession of goods and opportunities for
8 This generalization is probably most questionable for shifts from skilled worker to owner of a very small business. Also, the problem of how to classify farming occupations or farm-nonfarm occupational shifts in terms of social mobility has not been satisfactorily solved. Since agrarian politics is usually distinctive in most countries, for purposes of an empirical index probably those social strata most characteristic of industrial societies-persons in manual and nonmanual occupations-can be examined without regard for the farming population. 9 The further decision to dichotomize the various occupational groups into manual and nonmanual can also be justified on practical grounds, particularly if comparative secondary analysis of existing surveys is contemplated. Specific occupational categories are classified differently in various surveys, and the manual-nonmanual distinction is the most unambiguous one for comparative purposes. Also, the data are much more easily handled when only two social strata are compared, instead of the eight or ten occupations usually available in a survey.

LIVING RESEARCH

421

income, and is represented under the conditions of the commodity or labor market."10 In the sense that the "specific causal component" of life chances is probably more disadvantageous for skilled workers than for clerks and small businessmen, the manual-nonmanual distinction suggested as the measure of class is a reasonable one. But, by the same token, the limitations of using the manual-nonmanual occupational division are clear. The distinction is too general, because sales clerks are not in the same "class situation," in Weber's sense, as professionals or executives of large businesses, nor are the latter in the same situation. The use of the term "class" here does not imply consciousness of membership in a social class. Whether or not manual or nonmanual workers are conscious of a collective identity, and whether they associate this collective identity with a political party representing a collective interest, are of course important empirical questions but ones which should be kept separate from that of the degree of political divergence of objectively defined occupational strata. We may now mention briefly the problem of classifying voting patterns. Since voting as such can be analyzed at any political level, the index of class voting to be suggested can be applied to communities, regions, or nations (or to any social group within these areas). Since the parties in the kinds of class and political systems referred to here usually represent "Left" and "Right" issues on a number of dimensions (using the terms to divide parties in their stands on welfare, taxation, regulation of business and trade unions, and other similar issues), for purposes of empirical comparison parties can usually be divided in this way. Where two major parties are the main contenders for power, it is assumed that dividing the vote in that way fairly represents the fundamental political division in the electorate. Since minor parties exist in most "two party" systems, some decision must be made concerning the character of their appeals, parliamentary alignments with a major party, and legislative policies, in order to classify them as part of a Right or Left political alliance. For purposes of an empirical index, such a decision might vary from country to country, and from election to election.
AN INDEX OF CLASS VOTING

The extent to which manual and nonmanual strata divide in their support for political parties can be summarized in a numerical index of class voting. As with an index of class itself, any such measure has
10 Max Weber, Essays in Sociology, edited by H. Gerth and C. Wright Mills, New York, Oxford University Press, 1946, p. 181.

422

PUBLIC OPINION QUARTERLY

inherent limitations as well as advantages, depending on the problem of concern. The suggested index of class voting is computed very simply as follows: subtract the percentage of persons in nonmanual occupations voting for "Left" parties from the percentage of persons in manual occupations voting for "Left" parties.l The hypothetical figures given in the accompanying example indicate that the statistic computed as shown has not changed in value during the three elections from 1948 to 1956. Therefore, according to this index, class voting has not changed in the period 1948 to 1956, despite a move away from the Left party in both social strata.12
Per Cent Voting Left Manual occupations Nonmanual occupations Difference 1948 70 50 +20 1952 60 40 +20 1956 50 30 +20

Use of this index assumes that it is the gap between the voting patterns of manual and nonmanual occupations that is vital for assessing class voting, not the over-all level of Right or Left voting. The problem of how to interpret a shift to the Right or the Left is not basically that of determining the choice of an index of class voting, since the problem of interpretation would remain regardless of which index is chosen. This particular index embodies the assumption that the deviation of either manual or nonmanual stratum from a 50-50 split
11 The logic behind the use of such an index is that of Donald J. Bogue in his computation of a "coefficient of dissimilarity," measuring the "total amount of dissimilarity between any two percentage distributions in which the two sets of percentages are distributed by the same classes and refer to the same units" (see Donald J. Bogue, The Structure of the Metropolitan Community, University of Michigan, Horace H. Rackham School of Graduate Studies, 1950, p. 72). Here, an index of class voting is suggested only for dichotomies, but nothing in the nature of the index prevents it from being used for more numerous categories. Another use of the same index is presented in 0. D. Duncan and Beverly Duncan, "Residential Distribution and Occupational Stratification," American Journal of Sociology, Vol. 6o, 1955, PP. 493-50312 Campbell et al., op.cit., utilizing survey data for an analysis of American voting behavior, compute an index of the association of social class and vote and call it "status-polarization." The concept of "status-polarization" used in The A merican Voter is a social-psychological one, in line with the theoretical focus of the Michigan voting studies upon the factors affecting individual voting decisions. Status-polarization, in their definition, refers to the extent of "identification" of individuals with a class or status group. Occupation, religion, and education are treated as "external" characteristics of individuals which may or may not indicate an identification with a social group. This theoretical orientation produces a concern with what the study calls "short-term fluctuations" of the importance of status or class factors in voting behavior.

LIVING RESEARCH

423

for one of the major parties is the important fact relevant to an assessment of class voting. Adding up the deviation (with regard to the sign) produces this index. Although the method of computation just given does not make this clear, a reinterpretation of the above example will. For the 1952 figure, for example, the manual stratum split 60-40 for the Left and Right parties, respectively, i.e., lo per cent more in their expected (Left) direction than a 50-50 split. The nonmanual stratum split 40-60 in the opposite direction, i.e., lo per cent more in their expected (Right) direction than a 50-50 split. Adding these two lo per cent deviations from a 50-50 split in each stratum, we get an index figure of 20, which is exactly the same as the simpler computation given. But what if the voting figures upon which the index was based were as follows?
Per cent voting left: Manual Nonmanual Difference 20 0 +20

Clearly this is a radically different situation from the one above. The political significance of class has changed, as has the whole political system, and that is just the point. Here we have the virtual destruction of support for one major party, and complete unanimity in one stratum. The index is meaningless for such a situation, even though it shows an unchanged figure of +20. The conditions affecting the social bases of the parties have changed to such an extent that any simple statistic such as this lacks relevance to an understanding of what has happened. But, assuming for the moment that no such catastrophic change has occurred, it is still difficult to interpret the change. The lack of any Left voting among nonmanuals may mean that they are extremely class-conscious, and are voting Right for pure class reasons. On the other hand, the 20 per cent of manual workers voting Left may mean that the rest have taken on middle-class aspirations and are expressing them politically by voting Right. A total decline in the significance of social class as a determinant of voting need not be implied even by such outlandish figures as these. One stratum may be extremely class-conscious and voting for a party representing its felt interests; another stratum may be voting for the same party for completely nonclass reasons. Any empirical index, regardless of its particular construction, applies only to a given set of social and political conditions. But, given such conditions, why compute the index in this particular way? To answer this question, we must bring in further assumptions concerning the character of this type of political system.

424

PUBLIC OPINION QUARTERLY

The index reflects the assumption that both major parties respond to issues salient to the voters and attempt to gain as much support as possible; neither party tries to isolate its social base from political influence. Parties change with the changing importance of issues and are susceptible to influence from social currents and conflicts within the electorate. Clearly, this is true only of certain societies and in certain historical periods. Computing the index as a simple percentage difference between the Left support of the two social strata assumes that it is easy to change parties in these political systems, and that it is no harder for the average Left vote of either stratum to go from 40 per cent to 50 per cent than from 50 per cent to 6o per cent. This point deserves some elaboration, since it reflects an important assumption relative to the problem of change in the level of class voting. The index suggested assumes that major social strata tend to be affected similarly by political and social currents to the Right and to the Left, that, in other words, a high level of consensus exists. No social group is impervious to national social trends. Evidence to this effect has been found in at least two United States studies. An early study of voter registration in Santa Clara County, California, found that a trend toward Democratic registration occurred among all occupational groups, not just those presumably benefiting most from the New Deal.13 More recent studies of voting change have shown the reverse shift to have occurred in the 1948 to 1956 period, when all occupational groups moved in a Republican direction.14 If this is the way that political shifts have occurred in this type of political system, then only a movement in both strata toward 50 per cent for each party would be a true decline of the importance of social class as a systematic factor differentiating the support of the parties. This would be a move away from the normal predispositions in both strata. Actually, as already pointed out, changes in other directions could occur. The Left vote could decline steadily in both strata, past the 50 per cent mark. This might mean that both strata were becoming middle-class in their values, but that the Left party was stubbornly maintaining its identity as the party of workers. Or, conversely, both strata might be radicalized and move away from both traditional parties toward revolutionary parties, if the old parties maintained conservative appeals. The suggested index of class voting must not be overinterpreted to determine the importance of class issues in the programs of the parties or the meaning of class issues to the voters. Whether a high level of
13 See D. Anderson and P. E. Davidson, Ballots and the Democratic Class Struggle, Stanford, Calif., Stanford University Press, 1943, p. 370. 14 Campbell et al., op.cit., pp. 346-347.

LIVING RESEARCH

425

class voting (as measured by this index) is related to class appeals and a high level of class-consciousness is an important question, but no inferences from the index can be made. It measures simply the extent to which social strata, as defined by manual and nonmanual occupations, diverge in their support of a major political party. Comparative research by the author using this index has shown that the rank order of class voting in four Anglo-American countries is Britain, Australia, the United States, and Canada (as shown in the accompanying table).'5 This cross-national difference is independent of
CLASS VOTING IN

FouR ANGLO-AMERICAN 1946-1959*

COUNTRES,

Country Great Britain Australia United States Canada

LowestIndexof Highest Index of No. of Polls Class Voting Class Voting Based on 38 29 16 2 44 39 28 16 6 10 8 9

* For Great Britain, the Labor Party was used, for Australia, the Australian Labor Party, for the United States, the Democratic Party, for Canada, the CCF and Liberal Parties. Where two parties were

classifiedas "Left,"their votes in each stratum were combined.The exactquestions askedin each survey,the occupational divisions used, the datesof polls, andthe numbers of casesin manualand nonmanual are given in Alford,op.cit. The surveyswere taken at occupations various timesbetween1946 and 1959.Questions referred in mostcases to votingintention in a nationalelection,but past vote was used in a few instances.Galluppolls were the main sourceof data. The research was supportedby Ford Foundationgrants for studies of comparative politicalbehavior. regional, religious, age, and sex controls, and holds when other indexes of class are used. This difference seems to be associated with a number of aspects of the social and political systems of these nations. The consistency of the results may indicate that an index such as the one suggested does measure a relatively stable characteristic of two-party parliamentary and open-class systems.
15See R. Alford, "Class-Voting in the Anglo-AmericanPolitical Systems,"in S. M. Lipset and Stein Rokkan, editors, ComparativePolitical Systems, New York, The Free Press of Glencoe,forthcoming.

Das könnte Ihnen auch gefallen