The decline and rise of class voting? From occupation to culture in Australia

Abstract

Conventional wisdom has long held that class is declining as an influence on voting. More recently, new conceptions of class, focusing on the ownership of economic assets and the possession of social and cultural capital, have challenged this view. This article evaluates these arguments in two ways. First, we examine trends in the impact of traditional measures of class on the vote in Australia from the 1960s to the present day. Second, using a 2015 national survey that measures different aspects of class voting, we assess for the first time the relative effects on the vote of occupation, assets, and social and cultural capital. The results show that while occupation has declined and is now unimportant, the ownership of both assets and cultural capital are major influences on the vote. We argue that the impact of class on the vote has not declined, but rather transformed itself in new and different ways, which has important long-term implications for party support.

Keywords

Australian politics class voting electoral behaviour social class voting behaviour

Few topics in political sociology have generated as much interest and debate as social class and its political consequences.¹ The relentless rise of class voting in the early 20th century defined the politics of most of the advanced industrial democracies and served to consolidate two-party conflict between labour and the trade unions on one side, and capital and business on the other. The unprecedented material affluence of the 1960s and 1970s appeared to weaken the long-established pattern of class voting, resulting in upward social mobility among the working class and a blurring of class lines. Such was the scale of this change that Dalton (1996: 186) called the decline of class voting almost a ‘conventional wisdom’ in comparative politics.

But is this apparent decline a truism, as many assert? Recent research has questioned whether class has really declined as a political force in the advanced democracies, or whether it has simply transformed itself in new and different ways. This revision of the class voting thesis centres less on occupational class and its political consequences and more on its social, economic and cultural context. These revisions have followed two main approaches. The first is the economic assets that citizens possess, usually categorized by the relative political risk that they incur for the owner. Assets held in savings accounts pose least risk from policy changes by an incumbent government (‘sovereign risk’), while assets held in the form of stocks or property investment pose most risk (‘market risk’). The electoral importance of asset ownership increases with the degree of political risk that a particular asset holds for its owner, with voters selecting the party that they believe will support policies which are most beneficial to their ownership of the asset in question. (see e.g. Hellwig and McAllister, 2017; Lewis-Beck and Nadeau, 2011; Lewis-Beck et al., 2013).

The second approach to the revision of class centres on the ownership of social and cultural capital. Drawing on the work of Bourdieu and others, cultural capital theories of class emerged from criticisms of the then prevailing emphasis on occupational classifications – what has been termed the ‘employment-aggregate’ approach to class analysis (see e.g. Crompton, 1998; Savage, 2000). This cultural approach to class is concerned less with the relative position the person occupies in the employment hierarchy than the quality and quantity of her social networks and on her ability to appreciate cultural goods, such as music and art. These social networks and cultural skills interact with one another to create power relationships which, in turn, influence a voter’s political attitudes and choices (Savage et al., 2013). The cultural capital definition of class has proved to be very influential in sociology (see e.g. Prieur and Savage, 2013; Savage et al., 2015).

This article tests the hypothesis that the often-observed and documented decline in class voting may instead reflect changes in the nature of social class in post-industrial society, rather than on a decline in class voting per se. In particular, there have been no studies which have tested the relative importance of social and cultural capital with regard to voting. Using Australia as a case study, we evaluate the impact of the various definitions of class – broadly categorized into occupation, assets and culture – on the vote, analysed over almost half a century. The overtime data come from national election surveys conducted in Australia since 1967. To evaluate the specific electoral impact of cultural definitions of class measured against older definitions, a national survey conducted in 2015 is used.

The article proceeds as follows. The first section presents the major theories of class voting, from occupation in the immediate post-war years to culture in the beginning of the 21st century. The second section examines the social and cultural context of class, derived from the work of Bourdieu. The third section provides an overview of the debate about class voting in Australia, together with survey evidence which shows that traditional measures of class voting have indeed declined since the 1960s. The fourth section outlines the data and measures used in the analyses, while the fifth shows the impact of the various class measures on voting. The conclusion evaluates the significance of the findings for the study of class voting in post-industrial society.

Traditional theories of class voting

For most of the 20th century, class voting was viewed as a ‘democratic class struggle’, a term first coined in 1943 and later popularized by Seymour Martin Lipset in his famous 1960 book Political Man (Anderson and Davidson, 1943; Lipset, 1960; see also Korpi, 1983). The term has come to epitomize the link between objective material position and party support, the former dichotomized into two competing social groups, the middle and the working class. On one side of the class division were those who relied on their labour for a living and voted for a party of the left that espoused economic redistribution through the state. On the opposing side were those who relied on capital and voted for a party of the right which sought to limit the economic role of the state. This apparently settled pattern of class voting was confirmed by Alford’s 1963 study of class and voting in the Anglo-American democracies, Party and Society. Alford concluded that Britain, and to a lesser extent Australia, were class-polarized polities, while Canada and the United States were less class-bound (Alford, 1963: 289–90).

Starting in the 1970s, questions began to be raised about the role of class, particularly as post-war affluence and increased social mobility affected skilled manual and non-manual clerical workers. Upward social mobility, in turn, was driven by the expansion of university education and the growth in professional and managerial employment (Manza et al., 1995: 143–4). Studies have generally found that upwardly mobile manual workers tend to adopt a middle position between their class of origin and their class of destination, but that subsequent generations are more likely to acculturate to their class of destination (see, for example, de Graaf et al., 1995). The once stable and accepted pattern of class voting – the working class voting for a party of the left and the middle class a party of the right – appeared to be breaking down.

These insights led to a renewed interest in the measurement of class, through the development a more sophisticated occupational schema compared to the simple manual versus non-manual dichotomy that prevailed in the 1950s and 1960s.² Goldthorpe (Erikson and Goldthorpe, 1987, 1992) initially proposed a seven-category class model,³ identifying occupational groups according to a series of employment-related criteria, including how far they permitted economic advancement, their level of economic security, and the degree of workplace autonomy. Later revisions took into account the occupational structures of other countries, as well as changes in the nature of work, resulting in a revised eleven-category schema. Applying this schema to class voting in Britain has suggested that, rather than following a long-term decline, class voting has been characterized by ‘trendless fluctuation’ (Evans et al., 1991; Heath et al., 1991).⁴

Two main criticisms have been made of the Goldthorpe schema (see e.g. Savage et al., 2013: 222ff). First, it has been argued that occupation alone is inadequate to define a person’s class position; occupation is more suited to a descriptive analysis of social mobility and social stratification than to class per se. Second, occupation may not provide an explanation as to how economic inequality comes about, since it lacks any rigorous theoretical underpinning (Savage et al., 2005: 37–9). Nevertheless, the Goldthorpe measure of class has been highly influential in the study of class voting, especially in Britain. The schema has the advantage of being easy to operationalize in a survey, and identifies occupational groups with apparently different political priorities and interests, which in turn can be mobilized by the major political parties.

A further refinement in class measurement has come from Wright, who draws on Marx and Weber to highlight the different interests and outlooks of various social classes (Wright, 1985, 2005). His central argument is that the complexity of industrial society has caused a fragmentation in social class. The consequence of this is that the material interests and lifestyles of social groups cross-cut traditional class interests, weakening class polarization. This, Wright argues, has resulted in a middle class which has grown in size while at the same time lacking any sense of their class position (Wright, 1989: 3). Wright’s response has been to identify how social groups are able to exploit their market position to extract more favourable conditions from employers (see also Baxter and Western, 2001). By considering a person’s economic assets in addition to her occupation, Wright’s work anticipates Savage and Devine’s work in the Great British Class Survey which we examine in the next section.⁵

While there have been many criticisms of the general decline of class voting thesis, three points consistently stand out. The first concerns the measurement of class voting (see e.g. Nieuwbeerta, 1996; Nieuwbeerta and de Graaf, 1999). ‘First generation’ measurement techniques such as those employed by Robert Alford (1963) measured ‘absolute’ class voting and are sensitive to changes in the levels of party support.⁶ Such changes in party support can significantly affect the strength of the relationship between class and party, thus providing a biased estimate of class voting (van der Waal et al., 2007). As an alternative to these ‘first generation’ measures, odds ratios or log odds ratios have been used to measure relative levels of class voting (Evans, 2004).

A second criticism has been that values represent an emerging cleavage that is gradually replacing class. As outlined in the work of Inglehart (1990, 1997), post-material values, reflected in public concerns about environmental, cultural and quality of life issues, are replacing social class as the major political cleavage in the advanced democracies. Inglehart’s argument is that the once-dominant materialist value systems of the advanced democracies are being replaced by post-materalist values, through a process of generational replacement or, as Inglehart calls it, a ‘silent revolution’ (Inglehart, 1977). This is leading to the electoral rise of green and environmental parties to cater for these new post-material demands, and to the decline or transformation of traditional class-based parties.

A third criticism has focused on the supply side. Political parties have a key role in mobilizing support around particular social divisions, so they in effect can ‘choose’ which cleavages to politicize. In the case of Britain, Evans and Tilley (2012; see also Evans, 2000) argue that class dealignment has been caused by the restricted range of parties voters can choose from. This has occurred due to ‘the centrist movement of the parties, especially the Labour Party … and a consequent decline in the impact of voters’ positions on inequality and redistribution on their part choice’ (Evans and Tilley, 2012: 974). While this explanation has relevance in explaining the decline of class in Britain, it remains to be seen how far it can explain the same phenomena in a cross-national context.

In the next section we examine how these criticisms, and the earlier research which generated them, have provided a foundation for a new cultural approach to class analysis, with important implications for class voting.

Social and cultural theories of class

Social and cultural theories of class have been seen as a corrective to occupational definitions of class. Moving beyond the association between class and occupation, which is central to the Alford and Goldthorpe categorizations, this approach is associated with Bourdieu’s (1984) distinction between economic, social and cultural capital. Economic capital is defined by wealth and income, while cultural capital is the ability to appreciate cultural goods, such as music and art. Social capital is defined in terms of the quality and quantity of a person’s social networks, which in turn is based on social resources theory (for a review, see Lin, 1999). This theory argues that social networks provide a key set of resources which enhance status attainment. All three types of capital overlap in complex ways, and in the British context have been used to identify distinct classes which in some cases transcend occupational categories (Savage et al., 2005, 2013).

The theory identifies the capital, assets and resources that a person possesses as the main driver of her class position and of the relative degree of inequality that she will experience during the course of her life. The idea of moving beyond occupation has its origins in other approaches, such as the concept of human capital (Becker, 1964) and more recently social capital (Putnam, 2000). Bourdieu’s concept of capital has added a further refinement, by arguing that individuals can deploy their capital in relationships across a range of fields, of which economic relations is one (Bourdieu, 1984, 1997). This approach appeared to provide an advance on stratification theories of class, which relied mainly on the categorization of occupations. By taking into account the broader context of an occupation, it promised a more comprehensive account of social class.

Bourdieu’s theory has been used by Savage and a range of collaborators to propose three advances in our understanding of social class (Savage et al., 2005). First, the theory has moved the field away from an exclusive emphasis on economic relations, by placing class in a social and cultural context. Second, it takes into account institutional arrangements which might affect economic (or other) relations, notably with respect to the operation of a free market. Third, the theory recognizes that class cannot be removed from its cultural context, which in turn has implications for power relations. Overall, Savage and his collaborators argue that the garnering of resources over time across these areas permits the accumulation of capital which can then be converted into power (Savage et al., 2005: 45).

This approach was applied in the Great British Class Survey, which aimed to identify these new class distinctions in British society (Savage et al., 2013). Using a large web-based opt-in survey conducted by the BBC in 2011, combined with a nationally representative sample survey, the investigators identify seven classes which they argue more ‘effectively capture the role of social and cultural processes in generating class divisions’ (Savage et al., 2013: 220). The seven classes range from an elite at the top of the hierarchy, possessing very high economic capital, high social capital and very high cultural capital and encompassing 6% of the population, to a ‘precariat’ at the bottom, with poor economic capital and low social and cultural capital and encompassing 16%. Applying the same methodology to the 2015 survey used in this article, Sheppard and Biddle (2017) come to a very similar class distribution in Australia.

Criticisms of this new theory of class have been methodological as well as theoretical. In terms of method, Mills (2014) argues that the seven classes simply reflect the measurement technique used (and to some extent the data collection), and that the occupational categories developed by Goldthorpe still represent the best available measure of class.⁷ Theoretical criticisms have focused on the absence of a clear distinction between the middle and working classes, which has underpinned class analysis since industrialization. In particular, Bradley (2014) argues that several of the groups that are identified are not classes as traditionally understood but ‘class fractions’ or subgroups. Other criticisms concern how the schema relates to gender (Bradley, 2014) and race (Rollock, 2014). Despite these concerns, the schema provides a useful alternative to other class definitions. Perhaps more importantly from the perspective of this analysis, we know little about how these distinctions relate to voting. Before turning to that question, we examine long-term patterns of class voting in Australia.

Class voting in Australia

The relationship between class and party choice in Australia has been the subject of considerable research over more than half a century. The first empirical studies conducted in the 1960s and early 1970s confirmed the political importance of social class, based on the socioeconomic interests of the nascent political parties at the end of the nineteenth century. In the first analysis using national opinion surveys, Alford (1963: 173) concluded that Australian politics had been ‘dominated by class cleavages before and since its foundation as a nation’. Writing a decade later, Don Aitkin and Michael Kahan (1974: 477) characterized occupational class ‘as the rock on which the party system rests and a major influence on electoral behaviour’. Aitkin elaborated on this argument when he concluded that while class voting was stable, ‘Australian politics is the politics of parties, not of classes’ (1982: 142).

The debate about class voting was further developed by David Kemp (1978). Using a range of commercial opinion polls, Kemp argued that class voting had declined consistently during the 1960s and 1970s. The argument was based on theories of embourgeoisement, then popular in Europe, which predicted the steady decline of traditional blue-collar occupations in the face of increasing affluence and consumerism. Kemp argued that the existing political parties, and particularly the Labor Party, would become increasingly irrelevant as voters sought to satisfy their political needs elsewhere.

Kemp’s thesis generated considerable scholarly controversy. One theme was what was actually meant by class and how it could be measured reliably (Goot, 1994). A second theme has re-evaluated the empirical evidence using public opinion polls over a longer time-span than was available to either Aitkin or Kemp. Jones and McAllister (1989) conclude that there was no decline in class voting in Australia until after the middle of the 1960s and since then the decline has been more gradual than originally thought. Weakliem and Western (1999: 627), using the same data, argue that while there had been a decline in the political importance of class, it was occasioned by ‘a realignment, or qualitative change in the relationship between class and vote’. And a third theme has been the development of more complex models of occupation. Kelley and McAllister (1985) examined conflict aspects of class – principally the supervision of others within the workplace, self-employment, and employment in the government sector – and found that these aspects of occupation were at least as important in predicting the vote as the simple blue-collar/white-collar divide (see also Charnock, 1997; McAllister, 2003; McAllister and Bean, 2000, 2006).

Studies conducted over extended periods have also shed light on the changing socioeconomic bases of politics. Goot and Watson (2007) employ an overtime analysis of the Australian Election Studies surveys to show how the political loyalties of occupational groups have changed over time. By pooling the surveys conducted during the Hawke/Keating Labor years (1987–93) with the surveys conducted during the Howard Coalition years (1996–2004), Goot and Watson conclude that there was a significant shift in upwardly mobile or ‘aspirational’ blue-collar voters to Howard in 1996, securing him the election. They go on to argue that ‘the Coalition’s occupational base and that of Labor are increasingly similar’ (Goot and Watson, 2007: 269) and that Howard’s success lay in his populist appeal, which transcended occupational groups.

There is general agreement within the research that there has been a post-war decline in the electoral importance of class. However, there is disagreement about the extent of the change and when it occurred. Jones and McAllister (1989) see the change as occurring during the 1960s, although the process was not an even one, a finding which is confirmed by Weakliem and Western (1999) and Goot (1994). This is in contrast to Kemp (1978), who argued that the post-war decline in class voting has been much more consistent. Extending the analyses into the early 2000s, Charnock (2005) shows that the largest decline in class voting took place in the post-war period up to 1960, and since 1990 class voting has been largely stable.

To test the decline in class voting thesis and the timing of any changes, Figure 1 uses surveys conducted since 1967 to calculate odds ratios for four occupational groups: manual versus non-manual; supervisors versus non-supervisors; government employees versus private sector employees; and self-employed versus employees. The results show that there was a steep decline in class voting in the late 1960s, with some stabilization from the 1990s onwards. The traditional class measure (manual versus non-manual) shows a high but declining level of class voting in the late 1960s, with the decline stabilizing in 1979 and remaining largely unchanged from 1990 onwards. There is a similar, but weaker, trend for supervisors. The data for government employees and self-employed workers does not start until 1979, but confirms the post-1990 trend. The main exception is the spike in the electoral importance of the self-employed in 1996 (preceded by a smaller increase in 1993), which reflects the surge in support from this group for the Liberal-National Coalition in the 1996 election. Figure 1 also shows that aspects of occupation other than the traditional blue-/white-collar divide have been more important in predicting the vote since the early 1990s.

Figure 1.

Occupational class and voting, 1967–2016.

One caveat to these results is how occupation is measured across the surveys. This is done using the ‘head of household’ measure, which assumes that the family is the relevant unit and assigns class position to that of the male (if present), thus excluding women’s work. This measure has generated heated debate on both theoretical and empirical grounds (for reviews, see Tach, 2014; Sorensen, 1994). Theoretically, feminist scholars have argued that it effectively excludes one half of the population from consideration. Empirically, it is argued that increasing female labour force participation has created measurement error. Since class matters because it captures material circumstances, changes in family formation and labour force participation may affect the level of class voting if it relies on an imprecise concept. While there is undoubtedly measurement error in the patterns observed in Figure 1, other research has shown that the effects are generally small, and limited to the past two decades (Evans and Tilley, 2017: 54ff; see also Sorensen, 1994: 35–7). Since our conclusion is that the main changes in Australian class voting occurred in the late 1960s and 1970s, we assume that the error introduced by using the head of household measure does not invalidate our conclusion.

The available research suggests that most of the decline in class voting occurred in the 1960s, and since then voting has remained largely stable. This conclusion is confirmed by the results in Figure 1. However, the results also show that aspects of occupational class other than the manual/non-manual distinction have been at least as important in shaping the vote, and in some cases – such as the self-employed and supervisors – consistently much more important. This leaves open the possibility that the post-1960s decline in class voting is less a decline per se, than the emergence of different aspects of class which have become politically salient. Before we turn to this possibility, the next section outlines the data and method for the analyses that follow.

Data, measurement, method

Data

The data come from the Class in Australia survey which was a national survey conducted by telephone between 13 and 27 July 2015 as part of the regular ANUpoll series. The response rate was 21% and the survey has been weighted to adjust both for telephone status (landline and mobile) and the demographic characteristics of the population. The survey data and the full technical report are publicly available from http://ada.edu.au/.

Measurement

The variables used in the analysis are listed in Table 1. The dependent variable is voting intention and is based on the question: ‘If a federal election for the House of Representatives was held today, which of the following parties would you vote for?’ The main independent variables are as follows. Occupation is measured by dummy variables representing whether the respondent was a professional, a manager, a trade/technical worker, non-manual worker, manual worker or unemployed. Economic assets are measured by a cumulative scale from zero to four combining ownership of a house (either outright or by paying mortgage), an investment property, shares, and savings of $A40,000 or more. These assets represent the major aspects of ownership within the electorate (Lewis-Beck, Nadeau and Foucault, 2013).

Table 1.

Variables, coding, means.

Variable	Coding	Mean
Vote
Liberal-National vote	1 = yes, 0 = no	0.46
Green vote	1 = yes, 0 = no	0.18
Occupation
Professional	1 = yes, 0 = no	0.20
Manager	1 = yes, 0 = no	0.10
Trade, technical	1 = yes, 0 = no	0.08
Non-manual worker	1 = yes, 0 = no	0.19
Manual worker	1 = yes, 0 = no	0.08
Unemployed	1 = yes, 0 = no	0.04
Assets	From 0 to 4	1.27
Social capital
Professional	From 0 to 100	42.51
Non-manual	From 0 to 100	40.96
Manual	From 0 to 100	21.18
Cultural capital
High	From 0 to 1	0.40
Music	From 0 to 1	0.52
Youth	From 0 to 1	0.49
Family	From 0 to 1	0.85
Controls
Gender	1 = female, 0 = male	0.49
Age	Deciles	4.72
Graduate, postgraduate degree	1 = yes, 0 = no	0.38
Diploma	1 = yes, 0 = no	0.16
Trade, TAFE	1 = yes, 0 = no	0.10
Married	1 = yes, 0 = no	0.67
Children ⩽18 in household	1 = yes, 0 = no	0.23
Urban resident	1 = yes, 0 = no	0.68

Note: The four assets are ownership of a house, an investment property, shares, and savings; see text for details. The composition of the social and cultural capital scales are given in the text. N = 859 respondents.

Source: Class in Australia survey, 2015, weighted data.

Social capital is measured by asking the respondents if they knew someone who was employed in each of 18 occupations⁸ and is represented by three scales, coded according to the mean occupational status, on a zero to 100 scale, of the occupations in question.⁹ The three scales were identified by a factor analysis of the 18 occupations; this revealed three underlying factors, broadly corresponding to non-manual, professional and manual occupations, respectively.¹⁰ The four cultural capital scales are the mean number of activities the respondent engaged in, coded from zero to 1.

There is an extensive debate about how to measure cultural capital. Savage et al. (2013: 226–7) use an inductive approach to identify different types of cultural capital based on 27 separate activities. The survey used here was unable to ask about such a wide range of activities, so a subset of 15 activities was included in the questionnaire. Factor analysis was then used to identify four underlying patterns of activities: high culture, music, youth and family.

The control variables represent the main aspects of social structure and, with the exception of age (which is coded in deciles),¹¹ are all measured by dummy variables. Initial analyses included a much wider range of control variables, covering such aspects as birthplace and family income. Since preliminary analyses showed that a smaller group of control variables produced almost exactly the same findings, in the interests of parsimony the list was restricted to the six control variables used here, namely gender, age, education, marital status, children and urbanization.

Method

The multivariate analyses reported in Table 2 use multinomial logistic regression to predict the vote. The dependent variable is divided between the Liberal-National Coalition and the Greens, with Labor forming the excluded category. Voters for other minor parties or independents are excluded, producing an N of 859 respondents. In estimating the multinomial logistic model we use STATA’s weighting command and the sampling weights provided in the datafile. To ensure that our standard errors are not biased the models are run utilizing all the cases (see Williams, 2018 for a discussion.)

Table 2.

Dimensions of class and the vote.

	Model 1				Model 2				Model 3
	Lib-Nat		Green		Lib-Nat		Green		Lib-Nat		Green
	β	(SE)	β	(SE)	β	(SE)	β	(SE)	β	(SE)	β	(SE)
Occupation (non-labour force)
Professional	0.06	(.29)	−0.02	(.37)	−0.07	(.29)	−0.04	(.38)	−0.24	(.30)	−0.16	(.39)
Manager	0.03	(.33)	−0.25	(.49)	−0.19	(.33)	−0.28	(.51)	−0.44	(.36)	−0.37	(.53)
Trade, technical	0.07	(.41)	0.87	(.52)	0.03	(.43)	0.85	(.52)	0.10	(.44)	0.99	(.52)
Non-manual	−0.01	(.28)	−0.12	(.41)	−0.08	(.29)	−0.14	(.41)	−0.29	(.29)	−0.18	(.44)
Manual	−0.17	(.39)	−0.24	(.57)	−0.16	(.39)	−0.23	(.57)	−0.14	(.44)	−0.07	(.57)
Unemployed	−0.37	(.54)	0.67	(.57)	−0.45	(.55)	0.64	(.57)	−0.04	(.52)	0.60	(.60)
Assets	– na –		– na –		0.34**	(.10)	0.03	(.13)	0.29**	(.10)	0.00	0.14
Social and cultural capital
(Social)
Professional	– na –		– na –		– na –		– na –		0.00	(.01)	0.01	(.01)
Non-manual	– na –		– na –		– na –		– na –		0.02*	(.01)	0.01	(.01)
Manual	– na –		– na –		– na –		– na –		−0.01	(.01)	0.00	(.02)
(Cultural)
High	– na –		– na –		– na –		– na –		−0.07	(.40)	0.25	(.54)
Music	– na –		– na –		– na –		– na –		−1.33**	(.36)	0.66	(.51)
Youth	– na –		– na –		– na –		– na –		0.25	(.35)	1.00*	(.46)
Family	– na –		– na –		– na –		– na –		1.17**	(.43)	−1.20*	(.58)
Controls
Gender (female)	0.06	(.19)	0.02	(.25)	0.01*	(.20)	0.03	(.25)	0.02	(.20)	0.07	(.27)
Age	−0.00	(.01)	−0.02*	(.01)	−0.01	(.01)	−0.02*	(.01)	−0.01	(.01)	−0.01	(.01)
Qualifications (none)
Degree	−0.23	(.22)	0.95**	(.31)	−0.37	(.23)	0.94*	(.32)	−0.52*	(.25)	0.76*	(.34)
Diploma	0.01	(.27)	0.10	(.41)	0.00	(.27)	0.11	(.41)	−0.13	(.27)	0.14	(.40)
Trades	0.43	(.32)	0.45	(.47)	0.32	(.32)	0.44	(.47)	0.24	(.33)	0.40	(.45)
Married (single, widow, divorced)	0.14	(.20)	−0.32	(.27)	−0.03	(.21)	−0.34	(.28)	−0.14	(.22)	−0.36	(.30)
Children ⩽18 in household (none)	−0.34	(.24)	−0.74*	(.32)	−0.39	(.24)	−0.74*	(.32)	−0.46	(.24)	−0.79**	(.32)
Urban resident (rural)	−0.44*	(.19)	0.18	(.27)	−0.46*	(.19)	0.17	(.27)	−0.41*	(.20)	0.10	(.29)
Constant	0.62	(.47)	−0.12	(.64)	0.90	(.48)	−0.09	(.65)	−0.02	(.73)	−1.22	(.91)
(N)	(859)				(859)				(859)

p<.01, *p<.05.

Note: Multinomial logistic regression showing partial coefficients and standard errors (in parentheses) predicting the probability of voting Liberal-National or Green against the excluded category, Labor. For dummy variables, the excluded category is shown in parentheses.

Source: Class in Australia survey, 2015, weighted data.

Class and the vote

While we know much about how occupation and the ownership of economic assets structures the vote, little is known about the electoral implications of social and cultural capital. In developing the theory, Bourdieu argued that society was organized around the exercise of power, with politics at its core (Swartz, 2013). While Bourdieu saw political power as the key to understanding society, he was interested in how power was distributed across society and exercised by the social groups that possessed it, rather than in how it shaped political parties and election outcomes. This is one reason why there have to date been no studies which examine the role of social and cultural capital in predicting the vote; the analysis presented here fills this gap.

In order to test how these various forms of class influence the vote, Table 2 presents the results of a multinomial logistic regression analysis, which shows the impact of class on the vote, divided between the Liberal-National Coalition and the Greens, measured against the excluded category, Labor.¹² In order to evaluate the impact of each group of class characteristics on the vote – occupation, assets and social and cultural capital – three models are estimated, progressively adding each class group into the equation. This enables an estimate to be made of the relative importance of occupation in the absence of assets and capital, and the relative importance of assets in the absence of capital. The results of this analysis show that occupation matters little regardless of what controls are added. Further, comparing the fit of the base model excluding the occupational class dummies with one including occupational class dummies does not significantly improve the model fit.¹³ These results confirm the view that the political consequences of occupation are now negligible

Judged across the three groups of class characteristics, cultural capital is by far the most important and consistent factor shaping the vote. Within this broad category, engaging in family activities is a significant predictor of support for both parties compared to Labor, but the effect is around three times as strong for Liberal-National voters as compared to Green voters. This effect is net of age, marital status and children in the household. There is also an effect for musical activities, with Liberal-National voters being less likely to engage in them compared to Labor voters; there is no significant effect for Green voters compared to Labor voters. In contrast to cultural capital, social capital is relatively unimportant, save for an effect for non-manual social networks among Liberal-National voters, who are more likely to know individuals in these occupations compared to Labor voters.

In terms of the other class measures, assets are a significant predictor of the Liberal-National vote, with more ownership leading to a higher probability of Coalition support, net of other things; there is no parallel effect for Green voters. However, most importantly for the argument that class is not declining but simply transforming itself, none of the occupational measures is a significant influence on the vote. In particular, employment in a manual occupation is no more nor less likely to lead to either a Coalition or a Green vote compared to Labor. This contrasts sharply with long-established studies which find that occupational class is a major determinant of the vote. However, the results do confirm the findings of studies which show that asset ownership is a significant influence on the vote, even after a wide range of other factors have been taken into account (Hellwig and McAllister, 2017).

These patterns are further illustrated in Figure 2, which shows the predicted probabilities of voting for each of the three parties by the ownership of assets, and for three of the four cultural capital measures using the mean values for the other independent variables.¹⁴ The first three graphs for assets show their strong and consistent influence across all three parties. For the Liberal-National parties, the probability of securing a vote from someone with no economic assets is 38%, compared 70% for someone owning the maximum of five economic assets. Both Labor and the Greens lose votes among those with more economic assets, but the effect is not as large as that for the Liberal-National parties. The other graphs show that music cultural capital matters, in different directions, for the parties, while the effect of youth cultural capital is more modest. Family cultural capital positively influences the probability of voting for the Liberal-National parties, and negatively for the Greens. These effects are all the more striking since they control for a wide range of other factors.

Figure 2.

Predicted probabilities for class and party vote.

These results confirm that class voting has indeed changed in Australia. Where occupation was once the dominant factor shaping the vote – as the earlier results in Figure 1 illustrated – it now has no significant influence. The average marginal effects for different occupational classes are shown in Figure 2 and show little variation. Indeed, the only direct effect of economic position is based on the assets that a person holds, which is more likely to lead to a centre-right vote, as the theory predicts. In contrast to a person’s traditional class position, the cultural capital she possesses is now of major significance. But again, this finding has an important caveat: it is the types of cultural activities that she engages in rather than her social networks that matter for voting. In short, Bourdieu’s theory does matter for voting, but only in terms of culture. We explore the reasons for this and its implications for electoral behaviour in the conclusion.

Conclusion

The view that class voting has declined across the advanced democracies has become a truism in political sociology. The traditional explanation for this decline is the increasing complexity of modern society and its transition from an industrial to a post-industrial economy. While industrial society had clearly demarcated lines based on physical input and natural resources, post-industrial society relies on intellectual capabilities and is driven by new industries organized around education, information and knowledge (Powell and Snellman, 2004). The complexity of the post-industrial workplace and its associated lifestyle has undermined class identity and solidarity, to the extent that there is now a mismatch between a person’s economic position and her class self-identification (d’Hooge et al., 2017; Sosnaud et al., 2013). The net effect of these changes has been, it is argued, a weakening in the political significance of class.

Another interpretation of the class voting debate is that class has not declined, but simply transformed itself in order to reflect these new economic realities. As Clark (2001: 10) puts it: ‘it is better not to ask is class “alive” or “dead”, but rather what factors encourage more or less class salience?’ Van der Waal et al. (2007: 403) also express this view by declaring that ‘class is not dead – it has been buried alive under the increasing weight of cultural voting, systematically misinterpreted as a decline in class voting’. In this interpretation, what has occurred is that class concepts and measures have failed to keep pace with the rapid reorganization of the post-industrial workplace. While Bourdieu and others have theorized about what a post-industrial class landscape might look like, our research is the first attempt to examine what impact such a revision of class has on the vote, and to benchmark it against other, more traditional measures of class.

Our analysis has examined how this new conceptualization of class – measured by social and cultural capital – shapes the vote compared to another new concept, asset ownership, and an older one, occupation. The findings are unequivocal: cultural capital, reflected in the frequency with which a voter engages in musical and family-related activities, has a major impact on the vote, net of a wide variety of other factors. Musical activities predispose a voter to support the centre-left, while family activities predispose her to support the centre-right. Asset ownership also matters in voting, this time in predicting a centre-right vote. Emphasizing how far class voting has changed, occupation has no statistically significant effect on how people vote, net of these other class measures. While our results for the new class concepts cannot trace their effects over time, our results for occupation from the 1960s onwards show a steady decline in class voting to the early 1990s, and a stabilization thereafter. It would be reasonable to hypothesize that cultural capital began to replace occupation at about this time.

These findings represent what Jansen et al. (2013: 377) call ‘a bottom-up’ approach to class voting. In other words, we have examined class from the perspective of how it is organized within society. Equally plausible is the ‘top-down’ or party choice approach, which sees political parties as adapting their appeals in response to the perceived decline in class voting and the changing distribution of their core voters (Evans and Tilley, 2012). Evidence to support this view is the finding that the positions of parties and their degree of polarization influences the political salience of class (Jansen et al., 2013). Similarly, Hellwig and McAllister (2018) show that the relative economic policy positions of the parties, but particularly that of the centre-left party, matters in how they influence the strength of asset voting. The political choices that voters are offered undoubtedly matter in explaining how voters decide.

For political parties, the findings have a number of implications. First, the organization of election campaigns will become increasingly difficult as the electorate becomes more segmented – a process already well advanced with the rapidly expanding importance of social media. Second, an already volatile electorate with weaker partisan attachments may become even more volatile, as voters’ attachments to social groups weaken still further. And, third, where the parties decide to position themselves in relation to the values that underpin the main types of cultural activities will affect the votes they receive. While traditional class voting may have declined, the parties will have to adapt to and eventually exploit the new cleavages that will inevitably emerge.

Footnotes

Author’s note

The 2015 Class in Australia survey is available from the Australian Social Science Data Archives at . Our thanks for Murray Goot for comments on an earlier version of the article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

ORCID iD

Ian McAllister

Author Biographies

Ian McAllister is distinguished professor of Political Science at the Australian National University. He has been director of the Australian Election Study since 1987, a large national post-election survey of political attitudes and behaviour. His research covers Australian politics, comparative political behaviour and post-communist politics.

Toni Makkai is Emeritus Professor in the Centre for Social Research and Methods at the Australian National University. Prior to this she was the Dean of the College of Arts and Social Sciences at the ANU. Her research covers drugs and crime, crime statistics, regulation and compliance, and social and political attitudes.

References

Aitkin

D.A.

(1982) Stability and Change in Australian Politics. Canberra: ANU Press.

Aitkin

D.A.

Kahan

(1974) ‘Australia: Class Politics in the New World’, in Rose

(ed.) Electoral Behavior: A Comparative Handbook. New York: Free Press.

Alford

R.R.

(1963) Party and Society: The Anglo-American Democracies. Westport, CT: Greenwood.

Anderson

H.D.

Davidson

P.E.

(1943) Ballots and the Democratic Class Struggle. Stanford, CA: Stanford University Press.

Baxter

Western

(eds) (2001) Reconfigurations of Class and Gender. Stanford, CA: Stanford University Press.

Becker

(1964) Human Capital. New York: Columbia University Press.

Bourdieu

(1984) Distinction: A Social Critique of the Judgement of Taste. London: Routledge.

Bourdieu

(1997) ‘The Forms of Capital’, in Halsey

A.H.

Lauder

Brown

Wells

A.S.

(eds) Education: Culture, Economy, Society. Oxford: Oxford University Press.

Bradley

(2014) ‘Class Descriptors or Class Relations? Thoughts Towards a Critique of Savage et al.’, Sociology 48: 429–36.

10.

Charnock

(1997) ‘Class and Voting in the 1996 Australian Federal Election’, Electoral Studies 16: 281–300.

11.

Charnock

(2005) ‘Post-war Changes in the Influence of Social Structure on Australian Voting: The “Decline of Social Cleavages” Revisited’, Australian Journal of Political Science 40: 343–55.

12.

Clark

T.N.

(2001) ‘What Have We Learned in a Decade on Class and Party Politics?’, in Clark

T.N.

Lipset

S.M.

(eds) The Breakdown of Class Politics: A Debate on Post-industrial Stratification. Boulder, CO: Westview Press.

13.

Clark

T.N.

Lipset

S.M.

Rempel

(1993) ‘The Declining Political Significance of Social Class’, International Sociology 8: 293–316.

14.

Crompton

(1998) Class and Stratification. Cambridge: Polity Press.

15.

Dalton

R.J.

(1996) Citizen Politics: Public Opinion and Political Parties in Advanced Industrial Democracies. Chatham, NJ: Chatham House.

16.

de Graaf

N.D.

Nieuwbeerta

Heath

(1995) ‘Class Mobility and Political Preferences: Individual and Contextual Effects’, American Journal of Sociology 100: 997–1027.

17.

d’Hooge

Achterberg

Reeskenset

(2018) ‘Imagining Class: A Study into Material Social Class Position, Subjective Identification, and Voting Behaviour across Europe’, Social Science Research 70: 71–89.

18.

Erikson

Goldthorpe

J.H.

(1987) ‘Commonality and Variation in Social Fluidity in Industrial Nations, Part I’, European Sociology Review 3: 54–77.

19.

Erikson

Goldthorpe

J.H.

(1992) The Constant Flux: A Study of Class Mobility in Industrial Societies. Oxford: Clarendon.

20.

Evans

(1999) ‘Class Voting: From Premature Obituary to Reasoned Appraisal’, in Evans

(ed.) The End of Class Politics? Class Voting in Comparative Perspective. Oxford: Oxford University Press.

21.

Evans

(2000) ‘The Continued Significance of Class Voting’, Annual Review of Political Science 3: 401–17.

22.

Evans

Tilley

(2012) ‘The Depoliticization of Inequality and Redistribution: Explaining the Decline of Class Voting’, Journal of Politics 74: 963–76.

23.

Evans

Tilley

(2017) The New Politics of Class: The Political Exclusion of the British Working Class. Oxford: Oxford University Press.

24.

Evans

Heath

Payne

(1991) ‘Modelling Trends in the Class/Party Relationship’, Electoral Studies 10: 99–117.

25.

Evans

(2004) Voters and Voting: An Introduction. London: Sage.

26.

Goot

(1994) ‘Class Voting, Issue Voting and Electoral Volatility’, in Brett

Gillespie

Goot

(eds) Developments in Australian Politics. Melbourne: Macmillan.

27.

Goot

Watson

(2007) ‘Explaining Howard’s Success: Social Structure, Issue Agendas and Party Support’, Australian Journal of Political Science 42: 1–24.

28.

Heath

(1991) Understanding Political Change: The British Voter, 1964–1987. Oxford: Pergamon.

29.

Hellwig

McAllister

(2017) ‘The Impact of Economic Assets on Party Choice in Australia’, Journal of Elections, Public Opinion and Parties, https://doi.org/10.1080/17457289.2017.1408632

30.

Hellwig

McAllister

(2018) ‘Party Positions, Asset Ownership, and Economic Voting’, Unpublished ms.

31.

Inglehart

(1977) The Silent Revolution: Changing Values and Political Styles among Western Publics. Princeton, NJ: Princeton University Press.

32.

Inglehart

(1990) Culture Shift in Advanced Industrial Society. Princeton, NJ: Princeton University Press.

33.

Inglehart

. (1997) Modernization and Postmodernization: Cultural, Economic, and Political Change in 43 Societies. Princeton, NJ: Princeton University Press.

34.

Jansen

Evans

de Graaf

N.D.

(2013) ‘Class Voting and Left–Right Party Positions: A Comparative Study of 15 Western Democracies, 1960–2005’, Social Science Research 42: 376–400.

35.

Jones

F.L.

McAllister

(1989) ‘The Changing Structural Base of Australian Politics Since 1946’, Politics 24: 7–17.

36.

Kelley

McAllister

(1985) ‘Class and Party in Australia: Comparisons with Britain and the United States’, British Journal of Sociology 36: 383–420.

37.

Kemp

(1978) Society and Electoral Behaviour in Australia. Brisbane: University of Queensland Press.

38.

Korpi

(1983) The Democratic Class Struggle. London: Routledge and Kegan Paul.

39.

Lewis-Beck

M.S.

Nadeau

(2011) ‘Economic Voting Theory: Testing New Dimensions’, Electoral Studies 30: 288–94.

40.

Lewis-Beck

M.S.

Nadeau

Foucault

(2013) ‘The Compleat Economic Voter: New Theory and British Evidence’, British Journal of Political Science 43: 241–61.

41.

Lin

(1999) ‘Social Networks and Status Attainment’, Annual Review of Sociology 45: 467–87.

42.

Lipset

S.M.

(1960) Political Man: The Social Bases of Politics. London: Heinemann.

43.

McAllister

(2003) ‘Border Protection, the 2001 Australian Election and the Coalition Victory’, Australian Journal of Political Science 38: 445–64.

44.

McAllister

Bean

(2000) ‘The Electoral Politics of Economic Reform in Australia: The 1998 Election’, Australian Journal of Political Science 35: 383–400.

45.

McAllister

Bean

(2006) ‘Leaders, the Economy or Iraq? Explaining Voting in the 2004 Australian Election’, Australian Journal of Politics and History 52: 604–20.

46.

McMillan

Beavis

Jones

F.L.

(2009) ‘The AUSE06: A New Socioeconomic Index for Australia’, Journal of Sociology 45: 123–49.

47.

Manza

Hout

Brooks

, (1995) ‘Class Voting in Capitalist Democracies Since World War II: Dealignment, Realignment, or Trendless Fluctuation?’, Annual Review of Sociology 21: 137–62.

48.

Mills

(2014) ‘The Great British Class Fiasco: A Comment on Savage et al.’, Sociology 48: 437–44.

49.

Nieuwbeerta

(1996) ‘The Democratic Class Struggle in Postwar Societies: Class Voting in Twenty Countries, 1945–1990’, Acta Sociologica 39: 345–84.

50.

Nieuwbeerta

de Graaf

N.D.

(1999) ‘Traditional Class Voting in 20 Postwar Societies’, in Evans

(ed.) The End of Class Politics? Class Voting in Comparative Perspective. Oxford: Oxford University Press.

51.

Powell

W.W.

Snellman

, (2004) ‘The Knowledge Economy’, Annual Review of Sociology 30: 199–220.

52.

Prieur

Savage

(2013) ‘Emerging Forms of Cultural Capital’, European Societies 15: 246–67.

53.

Putnam

(2000) Bowling Alone: America’s Declining Social Capital. New York: Simon and Schuster.

54.

Rollock

(2014) ‘Race, Class and “the Harmony of Dispositions”’, Sociology 48: 445–51.

55.

Savage

(2000) Class Analysis and Social Transformation. Buckingham: Open University Press.

56.

Savage

Warde

Devine

(2005) ‘Capital, Assets and Resources: Some Critical Issues’, British Journal of Sociology 56: 31–48.

57.

Savage

Devine

Cunningham

Taylor

Hjellbrekke

, et al. (2013) ‘A New Model of Social Class? Findings from the BBC’s Great British Class Survey Experiment’, Sociology 47: 219–50.

58.

Savage

, et al. (2015) ‘On Social Class, Anno 2014’, Sociology 49: 1011–30.

59.

Sheppard

Biddle

(2017) ‘Class, Capital, and Identity in Australian Society’, Australian Journal of Political Science 52: 500–16.

60.

Sorensen

(1994) ‘Women, Family and Class’, Annual Review of Sociology 20: 27–45.

61.

Sosnaud

Brady

Frenk

S.M.

(2013) ‘Class in Name Only: Subjective Class Identity, Objective Class Position, and Vote Choice in American Presidential Elections’, Social Problems 60: 81–99.

62.

Swartz

D.L.

(2013) Symbolic Power, Politics and Intellectuals: The Political Sociology of Pierre Bourdieu. Chicago: University of Chicago Press.

63.

Tach

(2014) ‘Social Mobility in an Era of Family Instability and Complexity’, Annals of the American Academy of Political and Social Science 657: 83–97

64.

van der Waal

Achterberg

Houtman

(2007) ‘Class Is Not Dead – It Has Been Buried Alive: Class Voting and Cultural Voting in Postwar Western Societies (1956–1990)’, Politics and Society 35: 403–26.

65.

Weakliem

Western

(1999) ‘Class Voting, Social Change, and the Left in Australia, 1943–96’, British Journal of Sociology 50: 609–30.

66.

Williams

(2012) ‘Using the Margins Command to Estimate and Interpret Adjusted Predictions and Marginal Effects’, The Stata Journal 12(2): 308–31.

67.

Williams

(2018) ‘Analyzing Complex Survey Data: Some Key Issues to Be Aware Of’, https://www3.nd.edu/~rwilliam/ Last revised January 20, 2018.

68.

Wright

E.O.

(1985) Classes. London: Verso.

69.

Wright

E.O.

(1989) ‘A General Framework for Analysis of Class Structure’, in Wright

E.O.

(ed.) The Debate on Classes. London: Verso.

70.

Wright

E.O.

(2005) ‘Foundations of a Neo-Marxist Class Analysis’, in Wright

E.O.

(ed.) Approaches to Class Analysis. Cambridge: Cambridge University Press.