Department of Sociology, Pennsylvania State University, University Park, Pennsylvania 16802
KEY WORDS: political sociology, class analysis, political behavior, political change
Over the last two decades, many social scientists have argued that the stable class politics of industrial capitalism is giving way to newer types of social and attitudinal cleavages. Some scholars have gone further to associate what they see as significant declines in the anchorings provided by class with the rise of new political movements, parties, and even politicians standing for office completely outside traditional party systems. Advances in class theory and statistical methods coupled with the availability of high quality data have led others to reexamine the issue. They have suggested that these arguments reflect a misreading of the empirical evidence and/or exaggerate the significance of these developments. We conclude that despite the absence of a clear consensus in the field, theories asserting a universal process of class dealignment are not supported.
INTRODUCTION
The sociological study of politics centers on two broad concerns: the relationship among social structures, social action, and political institutions (or more generally the interaction between âstatesâ and âsocietiesâ), and the social bases of individualsâ political behavior. The seminal postwar work in political sociology, Seymour Martin Lipsetâs Political Man (1960), for example, explicitly developed analyses in both of these areas. In recent years, however, the state tradition has become the predominant focus of political sociology (Alford & Friedland 1985, Barrow 1993, Haynes & Jacobs 1994: 71). This review considers the more neglected side of the classical tradition: the relationship between social divisions and political behavior, specifically the relationship between class and voting in the capitalist democracies of Western Europe, North America, and Australia.
The status of the âdemocratic class struggleâ (Anderson & Davidson 1943, Lipset 1981 [1960], Korpi 1983, Esping-Anderson 1994) has been the subject of vigorous recent scholarly debates. A number of analysts have suggested that the stable class politics of industrial capitalism is giving way to newer cleavages based on gender, identity, and values. Some scholars have gone further to argue that class dealignment, along with declining political salience of other social cleavages, is leading to increased political instability in the form of new political movements, new parties, and even politicians standing for office completely outside traditional party systems. Others have countered that conclusions of class âdealignmentâ are a misreading of the empirical evidence and/or an exaggeration of the significance of these developments. Instead of class dealignment, defenders of class analysis argue that the class/vote association is merely subject to patterns of âtrendless fluctuationâ or, in exceptional cases, class ârealignment.â Our task is to review and evaluate these debates and the evidence.
We begin our review with an overview of the early scholarship on class voting. In part two, we consider the evidence for the declining significance of class for electoral politics. Part three examines research that challenges these findings, insisting on the continuing significance of class. In both of these sections, we discuss and evaluate both theoretical and empirical studies based on single-nation or cross-national comparisons, although to keep the discussion manageable we limit ourselves to the English language literature. In part four we address the methodological issuesâincluding how to define classâthat partly frame the debate. Last, we review analyses of the class basis of nonvoting, an important topic for class politics but one often isolated from the analysis of vote choice.
We note at the outset that our discussion of the literatures on class and voting addresses two related topics that have each been the subject of heated debates. First, there are controversies over the proper way of measuring the level of class voting and in particular of assessing the presence or absence of trends in the association between class and voting. A variety of statistical approaches have been developed to assess trends in class voting, and these often produce different estimates of the magnitude and changes over time in class-vote association (Goldthorpe 1994, Weakliem 1995). Second, there are controversies over the proper interpretation of the observed trends (or nontrends). Researchers have offered different kinds of institutional, historical, and cultural analyses to account for empirical findings.
SCHOLARLY ROOTS OF CONTEMPORARY DEBATES
The nineteenth century socialist movements and parties spawned an enduring interest in the association between class and voting (Przeworski 1980, Przeworski & Sprague 1986). Marx argued that the industrial working class would form the backbone of the revolutionary transformation of capitalism. Many socialists and social democratic leaders later assumed, and their conservative opponents feared, that if workers won the franchise, they could lay the foundation for an electoral road to socialism. By 1895, Engelsâ infamous âIntroductionâ to Marxâs The Class Struggles in France, 1848â1850 could hail the steady progress made by social democratic parties in Western Europe and suggest that the socialist revolution could be brought about through the ballot box by workers voting for socialist parties (Engels 1978 [1895]).
The social scientific study of class voting is usually motivated by an interest in testing the assumption of a direct link between class and electoral behavior as posited by theorists and opponents of social democracy. Early attempts to study the class-vote link used ecological techniques to infer the voting preferences of different income groups (Siegfried 1913, Ogburn & Hill 1935, Ogburn & Coombs 1940). Anderson & Davidson (1943) used Northern California precinct registration data to analyze how occupation affected political preference and changes in party identification during the New Deal period. The advent of election surveys, however, made possible more direct testing of the class-vote link. Lazarsfeld and his colleagues at the Bureau of Applied Social Research at Columbia made landmark contributions in the 1940s (Lazarsfeld et al 1948, Berelson et al 1954). They compiled evidence of a social basis for political behavior in election surveys in Erie County, Ohio (1940) and Elmira, New York (1948). More influential still were the empirical findings and theories developed by the âMichigan Schoolâ (especially Campbell et al 1960). The American National Election Study (ANES) has been carried out biannually in the United States since 1948, and similar national surveys based on the Michigan model have been periodically carried out in virtually every western democracy since the 1960s.
The seminal postwar voting research studies offered three distinct hypotheses about the association between an individualâs class location and partisan preference, hypotheses that have continued to inform contemporary research. The simplest model emphasized material interests as the foundation for class voting. For example, Upset and his colleagues argued in 1954 that class voting is a matter of ââŠsimple self-interest. The leftist parties represent themselves as instruments of social change in the direction of equality; the lower-income groups will support them in order to become economically better off, whereas the higher-income groups will oppose them in order to maintain their economic advantagesââ (Lipset et al 1954:1136; also Downs 1957: Ch. 3). The assumption that material class interests provide the foundation for class voting has remained the standard explanation shared by virtually all analysts to some degree or another (Evans 1993:263). But the empirical fact that many members of a particular class do not vote according to narrow class interests (for example, Heath et al 1991:68â69 note that members of a class rarely provide more than 60% of their votes to their natural class party), however, suggests that further theorization of the mechanisms linking class membership and voting are needed (Lipset et al 1954:1136, Scarbrough 1987, Weakliem & Heath 1994a).
The âColumbia Schoolâ (Lazarsfeld et al 1948, Berelson et al 1954) provided an explanation for votersâ decision-making that went beyond the simple invocation of interests. The core of their findings was the discovery of very high levels of stable partisanship on the part of voters, and that voters susceptible to changing their votes in the context of the campaign were those least interested in politics (Lazarsfeld et al 1948:69). They explained this seeming anomaly by emphasizing the importance of the cumulative effects of the historical experiences of social groups and the reinforcing effects of relatively homogenous social networks (Lazarsfeld et al 1948:137â149, Berelson et al 1954:88â109). Class voting in this model reflects the strength of common experiences of key historical moments and the reinforcing effects of intra-class friendship networks and social organizations (see also Frankel 1991 and Weakliem & Heath 1994a for recent applications).
The most influential of the postwar approaches to understanding political behavior was that of the Michigan School, in particular the analysis developed in The American Voter (Campbell et al 1960). The Michigan School viewed the mechanisms of voting as occupying positions along a funnel that represented both a causal and a temporal ordering (Campbell et al 1960:24â25). Social structural variables, including class origins and occupation, were seen as operating at the mouth of the funnel, leading to the social-psychological attributes (primarily political attitudes and partisan identification) at the narrow end of the funnel that ultimately predicted vote choice. Emphasizing a lack of ideological awareness and political sophistication on the part of most Americans (Converse 1964), the Michigan School viewed class voting as requiring a level of political sophistication unavailable to a significant portion of the electorate. Voters who were aware of their class location and who could make active use of class voting heuristics were much more likely to cast a vote consistent with their class (Campbell et al 1960: Ch. 13).
The accumulation of national election surveys since the 1960s has permitted detailed examination of theses about the association of class voting over extended time periods. Through the early 1970s, most findings suggested that class had a strongâthough variableâinfluence on voting behavior (Lipset 1981 [1960], Alford 1963, Lipset & Rokkan 1967a, Rose 1974a). Lipset & Rokkanâs influential theoretical synthesis (1967b) argued that two revolutions, the national revolution and the industrial revolution, initiated everywhere processes of social differentiation and conflict. The two revolutions produced four basic sets of cleavages: 1) church(es) vs the state; 2) dominant vs subject cultures; 3) primary vs secondary economy; and 4) employers vs workers (1967b: 14). The precise political articulation of these cleavages varied from country to country, depending on geopolitical structures and the timing of political and economic development, but all countries were subject to the same basic pattern. Further industrialization led to the decline of most types of social cleavage other than class, magnifying the importance of the democratic class struggle and âfreezingâ the cleavage structure (Lipset & Rokkan 1967b; see also Rose & Urwin 1970, Bartolini & Mair 1990).
THE DECLINING SIGNIFICANCE OF CLASS?
A series of setbacks experienced by social democratic or working-class parties, the emergence of religious, nativist, and regionalist political movements (cleavages long thought to have been depoliticized), and the sudden appearance of ânewâ issues such as ecological and antinuclear concerns have given rise to a wide range of challenges to class-based models of politics. A variety of labels have been offered to characterize this new environment: Scholars have variously claimed to find evidence of the âdeclining political significance of classâ (Clark et al 1993), a âloosening of social structureâ (Burler & Kavanaugh 1984:8), an âopening upâ of the electorate (Rose & McAllister 1986), the emergence (especially among younger cohorts) of a ânew politicsâ or âpostmaterialistâ orientations (Ingelhart 1977, 1990, Baker et al 1981, Dalton 1988), a general process of âdealignmentâ (Sarlvik & Crewe 1983, Nie et al 1981), or âthe decline of cleavage politicsâ (Franklin et al 1992a). In this section we review the empirical foundations of the hypothesized declining significance of class and the attempts to explain it.
Empirical Evidence of Declining Class Voting
Figure 1 replicates a graph that has been presentedâwith only minor variationsâin numerous publications as evidence of the decline of class voting (e.g. Lipset 1981:505, Inglehart 1984:30, Dalton 1988:157, Huckfeldt & Kohfeld 1989:3, Minkenberg & Inglehart 1989:85, Clark & Lipset 1991:403, Piven 1992a:7, Clark et al 1993:312). The graph is based on an extension of Alfordâs (1963,1967) deceptively simple index of class voting, first introduced in the 1960s. The âAlford index,â as it has come to be known, measures class voting as the percentage of âpersons in manual occupations voting for Left partiesâ minus âthe percentage of persons in nonmanual occupations voting for Left partiesâ (Alford 1963:79â80). The figure suggests that in many different countries support for left parties by manual workers has declined while employers and nonmanual employeesâ support for those same parties has increased. (We analyze the statistical properties of t...