Class and politics in twenty-first century Norway: a homology of positions and position-taking

Debates about the death of class were partly fuelled by the confusion surrounding the contemporary relationship between class and politics. If at some point the relationship seemed straightforward – lower positions in the social hierarchy were associated with left-wing politics, and higher ones with the right-wing – that neat order of things was disrupted by a differentiation of class relationships as well as the political landscape. Rather than splitting society to an ever greater extent into ‘two great classes directly facing each other’ (Marx and Engels 2002 [1888]: 219–20), the twentieth century produced a complex kaleidoscope of positions and position-taking. The old-style politics of life chances now coexists with what has been dubbed ‘new’ or life politics, concerned with ‘how we should live in a world where everything that used to be natural (or traditional) now has in some sense to be chosen, or decided about’ (Giddens 1994: 90–1). According to one powerful account, this does not mean that class is dead but that it has been buried alive by ‘cultural voting’ for life politics (van der Waal et al. 2007).

In terms of class analysis, how should the connection between class and politics be understood today? Debates about the very nature of contemporary class divisions also seem warranted. Several scholars have maintained that class continues to account for attitudes and/or voting along the classical economic left vs. right divide (at least to some extent). Life politics (or ‘new politics’, ‘postmaterialism’ or ‘liberal vs. anti-liberal’), however, requires an explanatory variable of its own. Often, level of education is seen as the appropriate measure (see Houtman 2001), but Goldthorpe has argued that status should be used (Chan and Goldthorpe 2007). Implicit in these approaches is the view that not only is class a purely economic phenomenon; it should also be construed and analysed as a principle independent of cultural factors.

We follow Harrits et al. (2010) in applying Pierre Bourdieu's multidimensional understanding of social class to the political landscape of late modernity. Although the influence of Bourdieu is widespread, applications of his specific conceptualisation of social class have been in remarkably short supply. While maintaining that class is fundamentally shaped by the economic structures of capitalism, Bourdieu regarded social class divisions as irrevocably social and cultural. A hallmark of his understanding is that social class divisions are manifested in a multidimensional distribution of forms of capital. Crucial to his model was division by the composition of capital, denoting the difference in the relative weight of cultural to economic capital (Bourdieu 1984). Remarkably, that division is absent from the most extensive follow-up of Bourdieu's Distinction (Bennett et al. 2009), as well as the recent ‘new model of social class’ inspired by Bourdieu (Savage et al. 2013).

Harrits et al. drew on Rosenlund's interpretation of Bourdieu's homology thesis (2009) in constructing two analytically independent spaces: one of political attitudes, the other of social class divisions. They found their two-dimensional political space corresponded to their two-dimensional space of class divisions. In line with Distinction (Bourdieu 1984: 451–3), left-wing economic views were inversely related to the actors’ volume of capital, but positively correlated with the relative preponderance of cultural capital. Expanding upon Distinction, Harrits et al. found libertarian views of life politics to be positively associated with the volume of capital, and especially cultural capital (Harrits et al. 2010: 18–19). We follow a similar strategy but measure social space by using a new class scheme which distinguishes classes and class fractions by both the volume and composition of capital (Hansen et al. 2009).

Our empirical case is Norway. Due to a variant of the ‘frozen’ five-party model, an influential labour movement and a social democratic party that has dominated post-war governments, Norwegian politics shares many basic characteristics with that of its Nordic neighbours (Lipset and Rokkan 1967; Berglund and Lindström 1978; Demker 2006). Arguably, however, due to the establishment of a socialist green and a populist right-wing party in the 1960s and 1970s, ‘new politics’ has restructured the political space of Norway much in line with a Western European pattern (see esp. Kriesi et al. 2006). Although the Norwegian party system is arguably dominated by three major parties – Labour, the Conservatives and the Progress Party – it also accommodates several smaller parties – the Christian Democrats, the Liberal Party, the Centre Party (former the Farmers’ Party), the red-green Socialist Left Party and the socialist/communist party ‘Red!’ (formerly the Red Election Alliance).¹

For some time, the nexus of class and politics was conceived of as fairly straightforward. The less well-to-do would opt for the redistribution of wealth and social welfare of the Left, whereas the better-off would deem their interests best served by the Right. According to Lipset's classical formulation: ‘More than anything else, the party struggle is a conflict among classes’: as the parties were essentially seen as vehicles for class interests, the link between class and politics could be explained by the ‘simple economic self-interest’ of voters (Lipset 1960: 223, 229)

That neat state of things was fractured by two parallel processes: the rise of the ‘new’ middle classes and the plethora of new political questions in the post-war period. Parkin's (1968) pioneering study of middle-class radicalism within the Campaign for Nuclear Disarmament underlined that such radicalism was in fact quite distinct from the bread-and-butter politics of the labour movement. Environmental concerns, the threat of nuclear war, gay rights, racism and xenophobia and other issues of a more ‘non-material’ nature redrew the political landscape. The advent of ‘post-materialist’ politics (Inglehart 1990) and the politics of lifestyle, as opposed to that of life chances (Giddens 1994), were seen as either supplanting or supplementing the old-fashioned clash between socialism and laissez-faire economic liberalism. Many scholars have thus argued that political attitudes should be understood two-dimensionally, as involving a classical economic left-right axis and a ‘post-materialist’, ‘new-political’ value dimension (Borre 1995; Clark and Lipset 1991; Flanagan 1982). This was accompanied by a marked decline in traditional class voting patterns in the post-war decades (Lipset 1981; Nieuwbeerta and De Graaf 1999; Nieuwbeerta 2001; Evans 1999).

For much of the twentieth century, sociologists were content to conceive of class divisions as dichotomous. Among the non-Marxists, this took the form of the divide between manual and non-manual labour (see the brilliant, if scathing, review in Parkin 1978). If this view was ever satisfactory, it gradually became less so. The ‘new middle class(es)’ emerged from key developments in twentieth century capitalism. The rise of joint-stock ownership paved the way for much larger organisations with more complex divisions of labour. This intersected with technological developments requiring greater expertise, but also with the expansion of the public sector as an employer, creating ‘structural demand’ for new types of white-collar workers. This development was common to capitalist societies, but there was considerable variation in the specific form and extent (Giddens 1973: 177–97). Particularly crucial is the burgeoning category of various managers, experts and professionals that Goldthorpe regarded as constituting a ‘service class’ (Goldthorpe 1982). This triggered an urge to devise more complex class schemes (Goldthorpe et al. 1980; Wright 1985; Wright 1997).

Most accounts of the decline in class voting were premised on the increasingly unsatisfactory view of class and politics as dichotomous: socialist vs. laissez-faire and manual vs. non-manual (Goldthorpe 1996). One diverse group of scholars interpreted these changes as the outcome of processes of individualisation (Beck and Beck-Gernsheim 2002), the ‘hyper differentiation of lifestyles’ (Pakulski and Waters 1996) and a ‘culture shift’ (Inglehart 1990). For these thinkers, political leanings were ‘less organised by class and more by other loyalties’ (Clark and Lipset 1991: 408)

According to Clark and Lipset, this meant that existing theorisations of class were ‘in need of substantial reformulation’ (1991: 397). Arguably, this was accomplished to some extent by the emerging Bourdieusian approach to class (Rosenlund 2009; Harrits et al. 2010; Faber et al. 2012; Atkinson and Rosenlund 2013; Savage 2000; Savage et al. 2005; Devine et al. 2005; Bennett et al. 2009; Savage et al. 2013; Skeggs 1997). To some extent, these contributions make common cause with ongoing work in sociology and political science that contributes to reconceptualising class and the methods traditionally used to measure it (Evans 1999; Manza and Brooks 1999; Oesch 2006, 2008; Kriesi 2010). The Bourdieusian approach, however, is distinct in regarding social class as not limited to the economic sphere, but as also shaped by the workings of various fields (which, under the label of situs, was seen by Clark and Lipset as undermining class divisions). In this approach, it makes little sense to contrast the explanatory power of education and cultural capital with that of social class (contra Stubager 2010; Houtman 2001; Sullivan 2001) as the former plays a crucial role in constituting the latter.

Our analytical strategy consists of constructing a space of political position-taking and then studying its interconnection with the structure of social space, measured using an adapted version of the Oslo Register Data Class scheme, a class scheme inspired by Bourdieu's model of social space. We use the fourth round of the European Social Survey (ESS), conducted in 2008, as it asked questions about ‘old’ and ‘new’ politics. The ESS is a random probability sample of all persons aged 15 and over. We use only the data for Norway and restrict our sample to the working population aged 18–67 (n = 1112). Post-stratification weights, developed by the ESS team, are applied in all analyses to correct for selection bias as well as sample skew in terms of sex, age, education and region. Data and documentation are available from http://www.europeansocialsurvey.org/.

To construct the space of position-taking, we use Multiple Correspondence Analysis (MCA). MCA works bottom-up, building a model from the distributions observed, as opposed to testing a pre-defined model against the observations (Le Roux and Rouanet 2010; Greenacre 2007). We run the MCA on an ‘individuals x categories’ matrix, with the rows representing individuals and the columns all categories of the questions. This procedure simultaneously calculates distances between individuals based on how different or similar they are in their response profiles, and distances between the categories based on how different or similar they are in the ‘composition’ of individuals. This creates a cloud of individuals and a cloud of categories, both pertaining to the same analysis (Le Roux and Rouanet 2010). In the former, individuals are found more closely together the more similar their responses are. In the latter, categories appear more closely the more often they are selected by the same individuals.

MCA allows for the use of supplementary variables for structured data analysis. Supplementary variables do not affect the structure of the solution as they are given no weight in the calculation of the distances. In our case, the structure of the space is determined by the relationship between political attitudes and our variable measuring social class is thus supplementary. This allows us to inspect the mean point and dispersion of each class category in the political space. The position of each social class category is thus in a sense a function of its members’ political views; this can to some extent be likened to a visual regression analysis (Lebart et al. 1984).

Bourdieu's arguments about a homology between class conditions and lifestyles have proven controversial. Commentators interpret it as implying a ‘one-to-one correspondence’ of one's social position and lifestyle (Coulangeon and Lemel 2009: 48) so ‘that social stratification […] and cultural stratification map onto each other very closely’ (Chan and Goldthorpe 2005: 194).²

It may seem more reasonable to expect a looser correspondence. Homologies emerge through the mediation of the habitus – the generative formulae ‘which underlie each of the classes of practices and properties’ (Bourdieu 1984: 126). Habitus is the internal structure of actors. Political position-taking, as well as lifestyle practices, are products of the intersection of habitus, field and capital, expressed in the somewhat confusing formula: ‘[(habitus) (capital)] + field = practice’ (Bourdieu 1984: 101). The habitus is crucially – but not exclusively – shaped by early experiences (and hence, social origin). Widespread social mobility (in absolute terms) means that any region of social space will be populated by a mix of people who originated from that class as well as people from other backgrounds, implying some heterogeneity of habitus and, accordingly, of practices. In other words, this essentially historical mediation arguably implies that there can be no perfect congruence between positions and forms of position-taking via dispositions.

Bourdieu's intention was that homology should refer to the relationship between different universes or structures. Therefore, our research design should be geared towards comparing structures. In this respect, we have been inspired by the trail-blazing work of Lennart Rosenlund (2009). Rosenlund constructed one space of lifestyles and one space of social positions through two, independent correspondence analyses. Once both spaces had been constructed, Rosenlund could compare their structures by projecting coordinates from one space onto the other. We follow the same principle here: our space of position-taking is constructed independently of our class scheme, and we investigate the homology by projecting our ‘social space classes’ onto the political space once its structure has been established.

To measure attitudes to both ‘old’ and ‘new’ politics, 13 questions have been analysed (see Table 1). Most of them are direct questions about positions in terms of political issues – such as whether one thinks it is the responsibility of the government to ensure full employment. However, to measure attitudes to environmentalism, we have used an item asking respondents whether someone who thinks caring for the environment is important is similar to them. We have assumed that people who support ‘green’ politics would answer ‘like me’ or ‘very like me’.

Previous experience has demonstrated that an MCA conducted on Likert-scale questions produces a separate axis which distinguishes those who position themselves at the far end of the scales (very strongly agree, very strongly disagree) from those who place themselves more closely to the centre (agree or disagree). This may be interpreted as demonstrating an intensity of opinion but it may also be an artefact. In this analysis, we have avoided this problem by recoding all Likert-scale questions, grouping the categories for ‘strongly agree’ with plain ‘agree’ and the same for ‘(strongly) disagree’.

Following Bourdieu, we conceive of class as structured two-dimensionally. To operationalise this, we use the Oslo Register Data Class Scheme (ORDC) (Hansen et al. 2009). The main classes are differentiated hierarchically based on their total amount of capital, whereas the middle and upper classes are seen as cross-cut into fractions by the type of predominant capital. This produces a basic 11 class scheme constructed based on occupations, ‘generally a good and economical indicator of [one's] position in social space’ (Bourdieu 1987: 4). For the economic middle-class fractions, we use household income to differentiate between the higher and lower groups.³

Although economic issues are most salient in shaping this axis, several other issues are also distributed across the first dimension. For instance, this is the case with the view that immigration makes the country a better place to live in and opposition to harsher prison sentencing. On the right, there is support for harsher prison sentencing, the view that immigration makes the country worse, the view that gays and lesbians should not be free to live as they wish and the ‘neither’ position on whether the government should reduce inequalities,

The second axis appears to represent a liberal vs. anti-liberal dimension or it might reflect attitudes to ‘new’ politics. At the top of the map, there is opposition to the following issues: schools should teach children to obey authority; harsher sentences; women should work less for the sake of the family and support for immigration. At the bottom, there are above average contributions from the categories indicating mixed views as to whether homosexuals should be free to live as they choose, whether women should work less for the sake of the family, whether immigration makes the country worse and whether there is support for harsher sentencing.

However, some economic issues also emerge as important. On the top, there is the view that the government should not aim to reduce inequalities; that small inequalities are not favourable; and that full employment is not the government's responsibility. At the bottom, there is support for small inequalities and that the government should aim to reduce them.

The interpretation of planes 1–2 seems relatively clear. The upper-left quadrant consists of seemingly left liberal views: a mix of liberal values and support for increased taxation. The lower-left quadrant resembles the ‘traditional’ left-wing: leftist or socialist positions on economic issues are pronounced. The upper-right quadrant appears to be right-wing liberal, with an emphasis on opposition to the economic policies of the Left. The lower-right quadrant seems to be right-wing and anti-liberal, characterised by opposition to liberal values.

When inspecting the mean points of classes in the political space, it is striking that the class categories align in exactly the same pattern as in Figure 1. At the top of the map there are the upper classes, with the working classes at the bottom. To the left there are the cultural capital fractions, whereas to the right there are the economic capital fractions.

Accordingly, the first, horizontal axis, contrasting socialist with laissez-faire views, corresponds to the capital composition axis of the social space. The second, vertical axis, separating liberal from anti-liberal views, corresponds to the capital volume axis of the social space. The pattern identified here is consistent with the homology thesis: the structure of the political space corresponds to that of the social space.

The distances between class categories are mostly large enough to be notable and some of them should even be considered large (Le Roux and Rouanet 2010: 71–2).⁷ As for the lower-middle class, the differences are not quite as striking – neither among its fractions nor vis-à-vis other classes. This, however, is as should be expected:

But how strong is the correspondence between the two spaces? To assess this more systematically, a simple cross-tabulation has been drawn up. We have created clusters from our political space, producing groups of individuals as similar as possible in their attitudes expressed.⁸ This means that individuals are grouped together into clusters based on the similarity of their positions in the space. We use only the first three dimensions for our clustering.⁹

Table 2 provides a cross-tabulation of the social space classes with the clusters from the political space. Given the finding above, it is unsurprising that the overall association in the table is low, indicating a 21.9% correspondence between the two spaces. The Pearson chi-square shows this to be statistically significant. As the associations may be expected to be most pronounced at the extreme ends of the spaces, we use adjusted standardised residuals to assess to which cells – i.e. between which social space classes and political space clusters – a statistically significant relationship pertains. Values above 2 indicate a significant positive association, whereas values below −2 indicate a significant negative association (Agresti 2002: 80–2).

This supports the view that associations are most pronounced at the edges. The cultural upper-middle class is very strongly associated with the ‘New left’, with close to 60% of them in that cluster. Correspondingly, this class fraction is disassociated with the ‘New right’. The symmetric upper-middle class is positively associated with both variants of liberal views – more than a quarter of them in the ‘New left’ and the ‘Right liberal’ clusters alike, and disassociated from the ‘Old left’. The economic upper-middle class is strongly associated with the ‘Right liberal’ cluster, with 52.5% of them in it, disassociated from the ‘Old left’ and the ‘Neithers’.

Skilled workers are positively associated with the ‘Old left’, with more than a quarter of them in this category, disassociated from the ‘New left’. Unskilled workers are even more pronounced, positively associated with both the ‘Old left’ and ‘New right’, but also disassociated from the ‘New left’ and the ‘Right liberals’. As the lower-middle class fractions exhibit less clear profiles, the expectation that the relationship should be more pronounced at the edges is unequivocally supported.

Our results indicate that a homology does indeed apply to the relationship between the space of political position-taking and the social space of positions. Strikingly, our class categories aligned in exactly the same pattern in the political space as in the theoretical scheme. We have found quite similar patterns before, with different data, within the Norwegian middle class (Author A) as well as the population at large (Author B). Our results largely resemble the findings of Harrits et al. (2010) and De Keere's paper in this special issue. We think this strengthens the argument for a Bourdieusian approach to understanding social class today, seeing social class divisions as manifesting themselves in the interplay between different forms of capital. More specifically, our results suggest that the often overlooked second dimension of social space, the composition of capital, is essential to account for the way in which contemporary politics remains classed.

However, in the papers of Harrits and De Keere, the social space classes followed the diagonals of their political space (Harrits et al. 2010: 18, De Keere fig. 8). The same ‘diagonal homology’ was revealed in a Danish analysis of the space of lifestyles (Faber et al. 2012). Does this reveal a different nexus between class and politics in different countries? Two circumstances may imply that the difference could be assigned to the differences in research design. First, our constructions of political space may differ as they rely on different questions. However, considering the Danish case, this seems not to be the case, as the distributions of the political parties in the political space are similar in both analyses (Harrits et al. 2010: 20) and Norway and Denmark have quite similar party political systems (see Arter 2008 for an account of the Scandinavian party system model).¹¹ Second, we have constructed the social space differently. The Danish study relied on an MCA of indicators of capital, whereas we model it through occupational classifications. However, the difference in results is probably not attributable to that. Jakob Skjøtt-Larsen, a co-author of the Danish study, cooperated with us in projecting our ORDC classes onto their political space. This revealed the same patterns as in their original analysis. This may indicate that Norway, the UK and Denmark may in fact be different in this respect. This should be further investigated through a systematic comparison.

Our results are also in line with the expectations suggested by the research into the matter of liberal vs. anti-liberal, where the conventional understanding is that liberal views accompany high education and/or social position. Those following Kohn (1969) explain this in terms of people's employment: workers are less liberal because of their lack of self-direction. Others follow Lipset (1959) in explaining liberal attitudes in line with one's level of education, claiming that high education produces liberal attitudes and low or no education the opposite (Svallfors 2006: 121–43; Houtman 2003: 24–65).

It is surprising, however, that the classical economic left vs. right dimension relates solely to the capital composition dimension and not to the volume of capital. Both working-class categories appear just slightly right of centre, far from the radicalism of cultural capital. It is, of course, unsurprising that the economic capital fractions hold laissez-faire views but the absence of an association between the volume of capital and economic left-right views is at odds with the received wisdom about the link between class and politics. Two provisos must be offered here, however: the working-class groups are clearly much bigger than the middle-class fractions in the present study. Obviously, the larger the category, the more it will gravitate towards the average. Moreover, we should note a significant conceptual qualification: our notion of hierarchy is synthetically composed of the total volume of capital, whereas the conventional understanding of class politics makes more sense if one restricts it to an economic hierarchy (as class politics was understood to be a conflict of economic interests).

However, our working-class groups were divided (Table 2). Both groups have considerable left-leaning contingents and the unskilled workers in particular exhibited a positive association with both ‘Old left-wing’ and ‘New right-wing’ politics. This indicates that the Norwegian working class is politically split which raises the question of how this internal heterogeneity can be accounted for sociologically. This topic calls for further work.

Bourdieu pointed out a peculiar political alignment between the cultural fraction of the dominant class and the workers (both leaning left). Our analysis indicates that this must be substantially qualified. It is true that the cultural upper fraction is markedly leftist and can in this regard thus be seen as aligning with what might be a left-leaning fraction of the unskilled working class. The working class is markedly separated from their potential bedfellows by their lack of enthusiasm for, and in some cases outright dismissal, of liberal views. Judging from the heated controversies generated by questions of immigration and gender equality, this is not likely to be a gap easily bridged.

This raises the question of whether, and how, radical politics are integrated in upper-middle class culture and how an aversion to it might be built into working-class culture. We would claim that political attitudes can be seen as a political taste, on par with broader cultural tastes. Vegard Jarness has shown how the economic middle-class fractions rich in capital and the working class are united in their contempt for the intellectual radicalism-cum-liberalism and besserwisser attitude of the cultural capital fractions, as if they always know better about everything from literature to climate change. Conversely, the culturally liberal politics, involving what might be seen as ‘tolerant’ attitudes, of the cultural capital fractions in no way precluded them from issuing harsh dismissals of other groups, employing aesthetic and political criteria. In their eyes, working-class Progress Party supporters were seen as tacky philistines and racist homophobes: undistinguished cultural tastes and undistinguished political tastes were widely seen as two sides of the same coin (Jarness 2014; Jarness 2015; Jarness 2013).

To some extent, ‘New left’ views could be understood as part of the culture of the class fractions richest in cultural capital. Just as their tastes for cultural and material goods distinguish them from both the working class and the fractions rich in economic capital, so too do their political views. Moreover, economic leftism, aimed at diminishing economic inequalities and limiting the power of money, challenges economic capital. Conversely, right-wing parties in Norway promote a politics that limits social spending and restricts funding for cultural institutions and grants to artists and so forth – this can be seen as amounting to an attack on the power and autonomy of cultural capital.

No potential conflict of interest was reported by the authors.

Magne Paalgard Flemmen is post-doctoral research fellow in sociology at the University of Oslo. His research interests lie in the areas of class analysis, class theory, social stratification and inequality more broadly, with particular attention to the empirical and theoretical relevance of the Bourdieusian model of social space. His recent work focuses on social mobility in the upper classes and the cultural and political aspects of class divisions.

Hedda Haakestad is a researcher at FAFO and a Phd candidate at the Centre for the Study of Professions at Oslo and Akershus University College of Applied Sciences.