In the wake of significant political realignments over recent decades, the structure of European political spaces has again become a focal point of academic debate. While some authors argue that structural transformations have generated a new dimension of political conflict (Warwick Reference Warwick2002), it is commonly believed that critical junctures of globalization have rearticulated the existing structure of economic and cultural dimensions towards heuristics of integration and demarcation (Bornschier Reference Bornschier2010a; Hutter and Kriesi Reference Hutter and Kriesi2019). On the one hand, the economic dimension has evolved from a state–market dichotomy to a broader conflict about the scope, boundaries and deservingness of social solidarity (Mau Reference Mau2003; van Oorschot Reference van Oorschot2006). On the other hand, the cultural dimension has become increasingly centred on conflicts over immigration, European integration and the demarcation of welfare (Bornschier Reference Bornschier2010a). This two-dimensional structure is assumed to be undergirded by new structural antagonisms between winners and losers of globalization within national electorates (Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006, Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008; Teney et al. Reference Teney, Lacewell and De Wilde2014; Zollinger Reference Zollinger2024).
Empirical analyses on the supply side of politics have shown that party systems have congealed into this rearticulated two-dimensional structure in every European democracy (Hooghe et al. Reference Hooghe, Marks and Wilson2002; Wheatley and Mendez Reference Wheatley and Mendez2021). The underlying argument is that new critical junctures have had a homogeneous impact on political competition, and that national parties have embedded emergent political issues along the two established dimensions of contestation (Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006; Rovny and Edwards Reference Rovny and Edwards2012). While this commonality between party systems suggests the existence of a shared ‘basic structure’ on the supply side of European political spaces (see, for example, Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006; Marks et al. Reference Marks, Hooghe, Nelson and Edwards2006; Rovny and Edwards Reference Rovny and Edwards2012), research on the ideological structure of voters – which reflects the demand side of political conflict – has been more scant. Consequently, the question remains whether this convergence of European party systems is reflected in a common structure of demand-side political spaces as well.
This study aims to evaluate the existence of such a ‘demand-side basic structure’ by investigating the commonalities and divergences between eight national political spaces in Europe. This comparative approach will consider both the structure of voter attitudes, or the ‘dimensionality’, as well as the social and political rootedness, or the ‘structuring’, of the national political space. In doing so, the article contributes to the literature in two ways. First, it examines whether the ideological structure of European electorates corresponds to the common structure found in their respective party systems. I argue that, rather than converging into a basic structure, voters’ ideological conflicts are shaped by demand, supply, or contextual differences, and that, consequently, national electorates should be framed by unique political dimensionalities. Second, the article sheds light on the social and electoral anchoring of political dimensionality by comparing the position of party electorates and social groups across political spaces. While earlier scholarship has mainly studied dimensionalities independently from social structure (e.g. Wheatley Reference Wheatley2015; Wheatley and Mendez Reference Wheatley and Mendez2021), this article additionally investigates how current European politics pits distinct groups of voters against each other in individual countries.
The analysis will be based on a comparison of the demand-side political spaces of Belgium, Germany, Spain, Hungary, Italy, Poland, Sweden and the United Kingdom. All of these countries present important variations in terms of political institutions, party systems and social structures and therefore resemble a most different systems design. Using non-linear principal component analysis (Gifi Reference Gifi1990) on data from the 2016 edition of the European Social Survey (ESS), the findings show that political attitudes have coalesced into a common structure of economic and cultural dimensions in all eight cases (as well as in a group of 10 additional countries included in the Supplementary Material). While the analyses provide empirical support for a common dimensionality of European voters, the results simultaneously indicate that national partisan and social groups often vary in their positioning within this two-dimensional structure. Taken as a whole, these findings thus demonstrate that demand-side political conflicts across Europe are similar in terms of ideological composition, but different in the coalitions of voters which they divide.
Theoretical background
Political dimensionality in Europe
The basic assumption that politics can be framed along multiple ideological dimensions has been at the heart of political sociology for many decades. While some authors have characterized political competition in Europe as being one-dimensional or multidimensional, the most commonly used model of political dimensionality consists of two (orthogonal) dimensions: an economic dimension, rooted in debates about resource distribution and market regulation, and a cultural dimension, revolving around ‘rule and belonging’ and the recognition of social diversity (Borbáth et al. Reference Borbáth, Hutter and Leininger2023).
Political dimensions draw the blueprints of party competition but equally serve as normative reference points for the public at large. For voters, political dimensions are latent polarities of interconnected issues that structure consistent ideological thinking, each of which is delineated by a collectivist and individualist endpoint (Huber et al. Reference Huber, Jankowski and Juen2023; Rovny and Whitefield Reference Rovny and Whitefield2019). On the economic dimension, which Seymour Martin Lipset (Reference Lipset1960) argues to be the expression of ‘democratic class struggle’, capitalism is set against a socialist economy, dividing the middle classes (mainly the petite bourgeoisie and the self-employed) from manual workers. By the same token, the cultural dimension represents the antagonism between progressiveness, emphasizing cultural diversity and self-expression, and conservatism, which advocates cultural monism and stable authority (Middendorp Reference Middendorp1978). These dimensions thus set the parameters of political conflict in the national community and serve as a cognitive shorthand to reinforce voters’ ‘established interpretation of what politics is about in the specific country’ (Bornschier Reference Bornschier2010b: 62).
The normative conflicts spanned by political dimensions remain susceptible to social change. In this connection, the current article starts from the premise that political spaces have transformed as a result of recent structural transformations. Indeed, despite earlier theories describing Western European political conflict as ‘frozen’ (Lipset and Rokkan Reference Lipset and Rokkan1967), there is ample evidence to suggest that this dimensional structure has changed in recent decades. As major societal shifts related to globalization and individualization disrupted the pre-existing order of the nation state, novel political issues came to the fore; ultimately resulting in a new phase of political realignment. These transformations expressed themselves primarily in a de-closure and de-structuring of the social contract – where welfare provisions and political rights were reserved exclusively for national citizens – pressuring the established cultural compromise and setting the stage for new parties to mobilize (Abts Reference Abts2012; Mau Reference Mau2003). The ramifications of these social developments provoked a wholesale transformation of national political spaces that has since become structurally anchored in a new cleavage between so-called ‘winners’ and ‘losers’ of globalization (Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006, Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008; Teney et al. Reference Teney, Lacewell and De Wilde2014). However, multiple studies also attest to a rearticulation of pre-existing political dimensions, which occurred at the level of party competition (Bornschier Reference Bornschier2010a; Hooghe and Marks Reference Hooghe and Marks2018; Hutter and Kriesi Reference Hutter and Kriesi2019; Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006, Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008), as well as between voters (Delespaul et al. Reference Delespaul, Meeusen, Abts and Swyngedouw2025; Häusermann and Kriesi Reference Häusermann, Kriesi, Beramendi, Häusermann, Kitschelt and Kriesi2015; Swyngedouw Reference Swyngedouw1995; Wheatley and Mendez Reference Wheatley and Mendez2021).
The focus of this article lies on the dimensionality of voters. In this respect, academic attention has primarily been centred on the (transformation of the) cultural dimension and the related emergence of green and radical right parties. This transformation unfolded first through a process of value liberalization in the 1960s, which brought new ‘postmaterialist’ issues onto the political agenda and subsequently embedded debates about alternative lifestyles, political participation and the environment within the cultural dimension (Inglehart Reference Inglehart1977). Second, as a result of globalization and increased migration flows, the cultural conflict was enlarged by further ideological contestation about immigration and European integration (Bornschier Reference Bornschier2010b). Both phases of transformation relate to the nation state and its capacity to draw internal and external membership boundaries – thus recalibrating the cultural dimension towards a conflict of ‘integration’ versus ‘demarcation’ (Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008), or, alternatively, of ‘universalism’ versus ‘particularism’ (Swyngedouw Reference Swyngedouw1995). These heuristics have since become powerful cues among European voters and their salience has eclipsed that of the traditional economic dimension (Grande et al. Reference Grande, Schwarzbözl and Fatke2019; Hillen Reference Hillen2023).
At the same time, pressures for austerity and welfare state maturation transformed redistributive politics into a zero-sum game, in which the (efficient) provision of social rights became increasingly debated in terms of productivity (Häusermann and Kriesi Reference Häusermann, Kriesi, Beramendi, Häusermann, Kitschelt and Kriesi2015; Mau Reference Mau2003). While this transformation did not directly erode support for the agency and scope of government redistribution, it did place more emphasis on ‘good spending’, efficiency and the unintended consequences of social benefits (Mau Reference Mau2003). In this context, the universalist redistributive design of the welfare state has become more strongly politicized on the economic state–market dimension (Abts et al. Reference Abts, Dalle Mulle, Kessel and Michel2021), with welfare sceptics campaigning to preserve benefits for either the deserving or the needy (Häusermann and Kriesi Reference Häusermann, Kriesi, Beramendi, Häusermann, Kitschelt and Kriesi2015). Moreover, immigration and the blurring boundaries of national solidarity challenged the welfare state in its ‘external bounding’ (Abts Reference Abts2012; Ferrera Reference Ferrera2005). The universal provision of welfare benefits thus became contested by an identitarian ‘welfare populist’ countermovement, championed by the radical right (Abts et al. Reference Abts, Dalle Mulle, Kessel and Michel2021; Häusermann and Kriesi Reference Häusermann, Kriesi, Beramendi, Häusermann, Kitschelt and Kriesi2015).
It is plausible that the content and composition of these two rearticulated dimensions varies across countries. After all, political transformations will usually unfold at the national level, where established actors and institutions can rearticulate new political issues into distinct ideological packages (Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006). The national political environment – shaped by demand-side, supply-side and contextual factors – therefore remains key to understanding the structure of voter conflicts across Europe (e.g. de Wilde et al. Reference de Wilde, Langsaether and Ozdemir2023; Jackson and Jolly Reference Jackson and Jolly2021; Manow Reference Manow2015; Wheatley and Mendez Reference Wheatley and Mendez2021). The country-specific interplay between these three factors can determine how issues are articulated, how salient they are to voters and the way in which they cluster with other issues to form coherent political dimensions.
The first set of demand-side factors covers all social conditions that shape political dimensions. These primarily include the different attitudes and social structures of European societies. Because the rate of social change can differ between countries, we can expect political transformations to have occurred in stages. In this sense, cross-national differences in dimensionality should result from the variable pace of structural transformations across Europe. Céline Teney et al. (Reference Teney, Lacewell and De Wilde2014), for instance, show that the national degree of globalization influences how issues of EU integration are embedded within the cultural dimension. In the same vein, Anna Pless et al. (Reference Pless, Tromp and Houtman2023) find that polarization over moral issues decreases with secularization, while conflict over immigration broadens. The main implications from these studies are that both dimensions can develop asymmetrically across countries, as long as structural change has not (yet) reconfigured national politics. The composition of dimensions thus reflects the ‘old’ structure wherever national social structures have remained stable, and the ‘new’ structure wherever these have transformed.
In addition, the behaviour of national political elites matters for structuring the issues that emerge from social change. Indeed, in moulding the political landscape, parties and politicians at the supply side compete in ‘a struggle over issue linkages’ and more specifically determine ‘the dimensional configuration of [the] political space’ (Rovny and Edwards Reference Rovny and Edwards2012: 59). When faced with novel political issues, the programmatic strategies of parties can determine the structure of voter conflicts. If mainstream parties engage in a dismissive strategy, these new issues may be taken up by issue entrepreneurs and form a dimension of their own. If, on the other hand, parties pursue an accommodative strategy, they will try to incorporate these newly introduced issues within the existing dimensional fold, organizing them to maximize their capacity to mobilize group interests (Meguid Reference Meguid2005). The crux is that the dynamics of national party competition amalgamate issues into ideological conflicts that are coherent in the specific country. Parties may as such perpetuate dimensions in some countries but trigger their transformation in others (Bornschier Reference Bornschier2010b).
Finally, the national political institutional context determines the opportunity structure in which political elites can exercise their agency. This wider context determines ‘which ideas are considered “sensible”, which constructions of reality are seen as “realistic”, and which claims are held as “legitimate” within a certain polity at a specific time’ (Koopmans and Statham Reference Koopmans, Statham, Giugni, McAdam and Tilly1999: 228). The ideological structure of national electorates can accordingly be determined by the broader environment in which dimensions are formed: institutional factors such as welfare regimes can, for example, shape public opinion on the scope, burden and distribution of public welfare (Henjak Reference Henjak2010; Jaeger Reference Jaeger2006; Svallfors Reference Svallfors2013), while historical legacies and critical junctures may establish distinct issue linkages in the national political space (Bornschier Reference Bornschier2009; Chaisty and Whitefield Reference Chaisty and Whitefield2015). Factors such as electoral systems, institutional access for new parties or national histories can therefore frame the conditions for political restructuring (Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008).
In short, all this suggests that the dimensionality of demand-side political spaces remains the combined product of national social structures, how these are politicized by parties, and the subsequent interaction with the prevailing political opportunity structure. Although a more thorough investigation of the isolated role of these three factors would certainly produce a fruitful avenue of inquiry, the present study will take a step back by providing an exploratory comparison of demand-side political spaces across Europe. Given the well-documented differences in demand, supply and national context, my first hypothesis maintains that the dimensionality of European electorates differs across countries – and hence does not coalesce into a common ‘basic structure’ at the demand side. This inquiry extends both to the number of dimensions required to describe political conflict and to the composition of the dimensions themselves.
Hypothesis 1: The political dimensionality of European electorates differs across countries.
The structuring of political dimensions
The social and political rootedness of these dimensions is likely to vary cross-nationally as well. This stems from the premise that electorates and social groups are positioned in the political space along national cleavage structures, which are historically distinct across Europe (Lipset and Rokkan Reference Lipset and Rokkan1967; Rokkan Reference Rokkan1970). The ‘structuring’ of national political spaces should accordingly reflect historical conflicts and long-standing party alignments. We can therefore expect dimensions to divide different coalitions of voters in different countries, based on the durable sociopolitical linkages that are established by national cleavages.
Cleavages refer to struggles between ‘winners’ and ‘losers’ of major structural transformations that are mobilized in ideological, identitarian and political-organizational terms (Bartolini and Mair Reference Bartolini and Mair1990). In their pioneering work on cleavage formation, Seymour Martin Lipset and Stein Rokkan (Reference Lipset and Rokkan1967) identified four historical cleavages that manifest in various constellations across Europe: (1) a religious cleavage between confessional and secular voters, (2) a centre–periphery cleavage between ‘national’ and ‘peripheral’ cultural groups, (3) a class cleavage between production workers and owners, and (4) an urban–rural cleavage between primary and industrial producers. Whether these four cleavages became mobilized depended on nationally established power dynamics and the strength of existing alignments – two contextual factors able to mitigate the ‘loss’ endured by the socio-structural coalitions at risk. Broadly speaking, the religious cleavage would materialize wherever Church authorities risked losing their traditional privileges during processes of nation-building; the centre–periphery cleavage wherever centralizing efforts met resistance from subordinate peripheral groups; and the urban–rural cleavage wherever farmers had not yet been mobilized by religious parties (Lipset and Rokkan Reference Lipset and Rokkan1967). Only the class cleavage has been ingrained in every European political space (Bartolini and Mair Reference Bartolini and Mair1990). In Eastern Europe, communist rule prevented the structuring of similar cleavages to those in Western Europe (Bornschier Reference Bornschier2009).
The structuring of new cleavages is likely to vary across countries as well. Indeed, the political reactions to globalization ‘are bound to manifest themselves above all at the national level’, which means that ‘the composition of the groups of winners and losers [of globalization] varies between national contexts’ (Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006: 921–922). The new cultural cleavage can as such be expressed in terms of a middle-class divide between sociocultural professionals and managers (Kriesi Reference Kriesi1998), a new class divide between sociocultural professionals and production workers (Oesch and Rennwald Reference Oesch and Rennwald2018), or a universalism–particularism divide between education groups (Bornschier Reference Bornschier2010b) – depending on the impact and politicization of globalization in the national context and the exit options available to the affected groups. The positioning of social groups and party electorates in the national political space should thus be framed by the form and salience of the new cultural cleavage.
In conclusion, we can expect that the imprint of these national cleavage variations is reflected in the different structuring of political spaces across Europe. More specifically, the presence of a cleavage in national politics should – as cleavage theory predicts – result in a larger ideological distance between corresponding partisan and social groups along relevant issue dimensions. According to the theory set out above, the national context should not only matter for the composition of political dimensions but also for its sociological anchoring and configuration within the political space. These expected differences in structuring underpin the second hypothesis.
Hypothesis 2: The structuring of European political spaces differs across countries.
Data and analytical strategy
To test whether these hypotheses hold water, this study will compare the dimensionality and the structuring of national political spaces across eight European countries. The analyses are based on survey data from the 2016 round of the European Social Survey (ESS8). Although there are more recent editions of the ESS available, this round includes a rotating module on climate and welfare state attitudes, which makes it possible to tap into a broader range of salient sociopolitical issues. For the ensuing analyses, only EU member states are selected (including, at that time, the UK), in order to adequately understand the structuring role of attitudes towards EU integration (n = 38,239). The results are corrected by a preconstructed analysis weight to account for different selection probabilities.
In a first step of the analysis, national political spaces are estimated based on a set of voter attitudes that represents salient political issues in domestic politics. To reconstruct these political spaces, the analysis employs non-linear principal component analysis (NLPCA), a dimensionality reduction technique that can reveal complex – possibly non-linear – relationships within the data (Gifi Reference Gifi1990). Compared to other dimensionality reduction techniques that assume variables to be strictly numeric (e.g. principal component analysis (PCA)) or nominal (e.g. multiple correspondence analysis (MCA)), NLPCA can handle variables of all different measurement levels simultaneously, taking into account the potentially more sophisticated non-monotonic and monotonic non-linear relationships between them (Linting and van der Kooij Reference Linting and van der Kooij2012). As such, the method will yield dimensions that better fit the structure of complex multi-measurement-level data. NLPCA requires prior specification of the variables’ analysis level through a process known as ‘optimal scaling’, which entails that the variable’s analysis level does not have to correspond to its original measurement level but can instead be tailored to fit the variable’s relationship with other variables in the data. If, after an initial transformation, the variable’s category quantifications (i.e. the distances on a vector between the category point and the origin) follow a non-monotonic or a monotonic non-linear pattern, it should be specified as nominal or ordinal, respectively. If the quantifications follow a linear pattern, the variable should be specified as numeric (Linting et al. Reference Linting, Meulman, Groenen and van der Kooij2007).Footnote 1
To uncover commonalities and divergences across political spaces in Europe, the dimensionality reduction is based on 12 key variables that represent salient political conflicts across national contexts. To operationalize traditional state–market conflicts, scales were built for egalitarianism and welfare provision, moral and economic criticism of the welfare state and beliefs in welfare state effectiveness.Footnote 2 While the first two scales address the agency and scope of welfare arrangements, the final three are related to more recent debates about their efficacy. Next, to assess the heterogeneous reactions to globalization, scales were developed measuring attitudes towards immigration, supplemented by three items on welfare chauvinism, support for EU benefits and support for EU integration. Finally, the model includes a scale on cultural liberalism and two items on individual responsibility for climate change reduction (climate: responsibility) and support for government actions to mitigate climate change by increasing taxation on fossil fuels (climate: policy action).Footnote 3
To investigate the European basic structure – as well as the way in which individual countries deviate from it – the following section first presents the dimensionality of the complete sample and subsequently shows respecified models for individual national political spaces. In a final step of the analysis, the structuring of dimensions is visualized by projecting party electorates and social groups as supplementary points in the national political spaces. To facilitate comparisons, the second and third part of the analysis focus on eight countries: Belgium (BE), Germany (DE), Spain (ES), Hungary (HU), Italy (IT), Poland (PL), Sweden (SE) and the United Kingdom (UK). These cases were selected to reconstruct a most different systems design. At the demand side, the eight countries vary in the degree of structural change experienced: for instance, Poland, Hungary and Italy are less secularized (Pless et al. Reference Pless, Tromp and Houtman2023), while countries in Western Europe have generally been more affected by globalization than those in the East (Hooghe and Marks Reference Hooghe and Marks2018). In terms of political supply, the sample on the one hand includes polities without (at the time) far-right (Germany, Spain) or new-left parties (Hungary), and on the other hand cases with agrarian and regional parties due to the presence of an urban–rural (Poland, Sweden) or centre–periphery (Belgium, Spain, UK) cleavage. Moreover, these cases provide dissimilar political opportunity structures, with all eight countries differing in terms of electoral systems, welfare regimes, experiences of democratic rule and traditions of coalition-building. Results for the 10 countries that were not presented in the comparative part of the analysis can be found in Appendices F and H of the Supplementary Material.
Results
The dimensionality of voter attitudes in Europe
The first NLPCA model based on the complete sample serves as a benchmark for subsequent by-country comparisons. An initial analysis of the variables at the least restricted analysis level (i.e. (spline) nominal, meaning that the variable categories are quantified without assuming an order) indicates that only two dimensions should be retained.Footnote 4 The least restricted model is also used to determine the ‘optimal’ analysis level of the 12 selected variables. Transformation plots (provided in Appendix B of the Supplementary Material) of the category quantifications in this baseline model reveal that only the immigration item should be analysed at a (linear) numeric level, while most other variables were specified at an ordinal or nominal analysis level (with spline transformations being applied in cases where a variable has many categories). Compared to a linear PCA model using the same variables, optimal scaling improved the model fit by 2.0%. However, the non-linear model provides (much) higher communalities for the individual variables – in particular for those attitudes related to redistribution and the welfare state.
Table 1 provides an overview of the two-dimensional solution for the non-linear model after rotation. This model accounts for 38.2% of the variance – which is a reasonably high proportion when analysing sociopolitical attitudes (cf. Häusermann and Kriesi Reference Häusermann, Kriesi, Beramendi, Häusermann, Kitschelt and Kriesi2015; Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008). The importance of a variable to a dimension can be represented either in terms of component loadings (i.e. the correlations between the quantified variable and the component), or in terms of contributions (i.e. the variable’s contribution to the component’s eigenvalue). In addition, the VAF (or ‘variance accounted for’) in the variables describes the proportion of its variance that can be explained across the principal components.
Rotated Component Loadings
Notes: Variable principal normalization with varimax rotation; bold indicates component loading > 0.4; VAF = variance accounted for; Wfs. = welfare state.
This initial model bears close resemblance to the anticipated basic structure. The first component, which accounts for 19.3% of the variance, corresponds to the presupposed cultural dimension: it pits cosmopolitan and universalist respondents against culturally conservative particularists, with large contributions from attitudes related to globalization (immigration, EU integration and welfare chauvinism) as well as from attitudes related to lifestyles and the quality of physical and social life (cultural liberalism and environmentalism). The second component is slightly less salient (19.0%) and combines traditional conflicts over redistribution (egalitarianism and welfare provision) with newer conflicts over welfare arrangements (moral and economic criticism of the welfare state and welfare state effectiveness). It is interesting to note that the item on EU-wide social benefits (‘EU: solidarity’) loads more strongly on the economic dimension, while the more general question related to support for EU integration has a cultural connotation. While this demonstrates the multidimensionality of EU integration issues observed elsewhere (e.g. Otjes and Katsanidou Reference Otjes and Katsanidou2017; Wheatley and Mendez Reference Wheatley and Mendez2021), both attitudes on climate change remain embedded in the cultural dimension, even though one of the items (‘Climate: policy action’) explicitly refers to potential economic burdens in the fight against climate change. Conversely, moral and economic criticisms of the welfare state both cluster in the economic dimension, despite relating to different dimensions of political competition (Abts et al. Reference Abts, Dalle Mulle, Kessel and Michel2021; Roosma et al. Reference Roosma, van Oorschot and Gelissen2014).
To visualize the dimensional structure of the data, the quantified variables can be projected in a vector or a centroid model.Footnote 5 In Figure 1, the top pane depicts the vector coordinates of the highest and lowest quantifications of the 12 structuring variables (i.e. the variable categories with the largest distances to the origin on the best-fitting vector). On the horizontal cultural dimension, vector coordinates range from conservatism (left) to progressivism (right), while the vertical economic dimension contrasts redistributive (top) and market-liberal (bottom) outlooks. Similar to what was observed in Table 1, the vector coordinates show some divergence on the cultural dimension: pro-environmental attitudes are more strongly correlated to economically right-wing positions, while attitudes in favour of cultural liberalism and globalization are more strongly associated with economically left-wing positions. This result suggests that environmentalism – which is theoretically unrelated to traditional concepts of tolerance and hierarchy – may not (yet) have converged fully with the cultural dimension (cf. Kenny and Langsæther Reference Kenny and Langsæther2023).
European Political Space: Variables in the Vector Model with Highest and Lowest Quantifications (Top) and Centroid Coordinates for Countries (Bottom)
The bottom pane of Figure 1, in turn, shows the centroid coordinates of the countries retained in the initial analysis. These categories are projected as supplementary points and therefore do not contribute to the estimation of the dimensions. Here we see that national electorates in Europe occupy quite different positions along the two-dimensional structure. Overall, countries are positioned on an East–West divide ranging from left-authoritarian to right-libertarian positions, with southern European electorates (i.e. Italy, Portugal and Spain) appearing more ideologically divided.
The dimensionality of national political spaces
In order to test the first hypothesis, the second step of the analysis compares the commonality of this dimensional structure across eight national political spaces. This comparison requires optimal scaling of the 12 analysis variables for country-specific subsets of the data, following the same process as for the initial NLPCA presented above. Scree plots indicate that the two-dimensional model is an acceptable fit in each of the eight countries.Footnote 6 However, considerable heterogeneity in the countries’ relational structure is observed when assessing the variables’ optimal scaling level (i.e. whether they should be considered as linear, ordinal or nominal, see Appendix D).Footnote 7 The results of the country-specific non-linear PCA models with optimal scaling are presented in Tables 2a and 2b.
Component Loadings and Total VAF for Country-specific NLPCAs: Cultural Dimension
Notes: Variable principal normalization with varimax rotation; bold indicates component loading > 0.4; VAF = variance accounted for; Wfs. = welfare state.
These results point to the existence of a common dimensional structure in Europe. Contrary to the first hypothesis, political attitudes appear to have coalesced into a similar cultural and economic dimension in each of the eight selected countries. To start, the cultural dimension is clearly shaped by attitudes towards immigration and EU-integration in all countries. These two attitudes are in most cases supplemented by variables related to cultural liberalism and welfare chauvinism, except in Belgium (where cultural liberalism is not embedded in the cultural dimension) and in the UK (where welfare chauvinism is more strongly correlated to the economic dimension). More specific cross-country differences emerge with regard to the two variables on environmentalism. In the cases of Poland and Hungary – as well as in the post-communist societies not included in the analysis – environmentalist attitudes are only weakly related to the cultural dimension, while they are clearly integrated in this dimension in West European political spaces. At first glance, these findings reflect the different structures of post-communist party systems, where environmental issues are detached from typical cultural issues (Chaisty and Whitefield Reference Chaisty and Whitefield2015; Rohrschneider and Miles Reference Rohrschneider and Miles2015; Rohrschneider and Whitefield Reference Rohrschneider and Whitefield2012).Footnote 8 Conversely, cultural liberalism contributes more strongly to the cultural dimension in the two Eastern European countries in comparison to more secular countries such as Belgium or the UK. While unable to test this link directly, this result would seem to confirm the general claim that the degree of secularization – or of structural change more broadly – indeed shapes different dimensionalities across Europe (Pless et al. Reference Pless, Tromp and Houtman2023).
We can observe more general cross-national regularities on the economic dimension (Table 2b). In each of the selected cases, this dimension is structured by the items on egalitarianism, welfare provision, welfare state effectiveness and both aspects of welfare state criticism. In Germany and Poland – the largest contributor and receiver of EU funds, respectively – the item on EU solidarity contributes substantially to the economic dimension as well. The item on welfare chauvinism contributes to the two dimensions in Belgium and the UK, steering the economic dimension more clearly towards heuristics of producerism and deservingness. Against expectations, different welfare regimes do not appear to structure welfare state attitudes differently (cf. Jaeger Reference Jaeger2006; Svallfors Reference Svallfors2013; Van Hootegem Reference Van Hootegem2022).
Component Loadings and Total VAF for Country-specific NLPCAs: Economic Dimension
Notes: Variable principal normalization with varimax rotation; bold indicates component loading > 0.4; VAF = variance accounted for; Wfs. = welfare state.
It is worth noting, finally, that the variance explained by the two dimensions (VAF) varies marginally between countries. In Hungary, Italy and Sweden, the cultural dimension is the main line of political contestation among voters. In the cases of Belgium, Germany, Spain, Poland and the UK, on the other hand, the economic dimension represents the more dominant political conflict. Although previous research has indicated that the cultural dimension has become more salient in explaining voting behaviour (Jackson and Jolly Reference Jackson and Jolly2021), these results suggest that both dimensions are roughly equal in their capacity to structure voters ideologically. More fundamentally, the findings also contradict recent work concluding that (certain) European political spaces are uni- or multidimensional (Kenny and Langsæther Reference Kenny and Langsæther2023; Toshkov and Krouwel Reference Toshkov and Krouwel2022; Wheatley and Mendez Reference Wheatley and Mendez2021).
The structuring of national political spaces
In broad strokes, the previous analyses can be summarized by two observations: first, that demand-side political spaces are two-dimensional in each of the selected countries, and second, that those two dimensions correspond to a large degree to the cultural and economic dimensions of the anticipated European basic structure. Although the results laid bare minor contextual differences between countries, this common dimensionality appears to reject the first hypothesis. In order to further uncover the social and partisan groups divided by these dimensions (Hypothesis 2), Figure 2 projects party electorates and national social structures in the country-specific NLPCAs of the eight selected countries.Footnote 9
National Political Spaces Including Social Groups and Party Electorates
(Continued.)
(Continued.)
(Continued.)
The projected social groups first correspond to structural divisions laid out by the three traditional cleavages: the religious cleavage (active churchgoing versus not religious), the urban–rural cleavage (urban versus rural residence) and the ‘old’ class cleavage (production workers versus small business owners).Footnote 10 The three additional divisions reconstruct more recent oppositions related to globalization and tertiarization: the middle-class cleavage (sociocultural specialists versus managers), the ‘new’ class cleavage (production workers versus sociocultural specialists) and the universalism–particularism cleavage (high versus low educated) (cf. Lachat and Dolezal Reference Lachat, Dolezal, Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008; Oesch and Rennwald Reference Oesch and Rennwald2018).Footnote 11
Despite the shared structure of political attitudes, the results of this final analysis suggest that national dimensions often divide different coalitions of voters across countries. A first difference relates to the importance of the three traditional structural oppositions, which have retained salience in some countries but lost importance in others. Notably, religious oppositions are more polarizing in Poland and Hungary, which are comparatively less secularized than the six other (Western) European countries. According to classic cleavage theory, this religious divide should be strongest in the countries where Church elites had opposed nation-building – which had indeed been the case in Eastern European countries where churches had provided one of the few platforms to voice opposition against communist repression (Kurek and Fałkowski Reference Kurek and Fałkowski2022). Furthermore, religious oppositions appear to cross-cut most other divides in the West, but reinforce them in the East. Part of this distinctive alignment may be due to the lack of a Polish or Hungarian Christian-democratic party able to appeal to confessional working-class voters (Kalyvas and van Kersbergen Reference Kalyvas and van Kersbergen2010).
The old class divide appears to have lost relevance in most countries. While this cleavage once represented a ‘standardizing element across the variety of Western European party systems’ (Rokkan Reference Rokkan1970: 130), small business owners and production workers have since adopted similar culturally conservative outlooks in the national political space (Oesch and Rennwald Reference Oesch and Rennwald2018). It is noticeable that social-democratic electorates have lost the connection to production workers in all countries but Spain (PSOE) and Sweden (S), where the old class divide is strongest. Analogously, the profile of small business owners overlaps with centre-right electorates in both countries. By comparison, the urban–rural divide appears salient in most countries and usually overlaps with new class and educational oppositions. Polarization along this divide is highest in Sweden – where the urban–rural cleavage has historically been politicized by the agrarian Centerpartiet (C) (Manow Reference Manow2015) – and, surprisingly, in Belgium, where this divide is traditionally not considered to represent a structuring cleavage (Delespaul Reference Delespaul2024).
The three more recently mobilized structural oppositions vary significantly across countries as well. The middle-class divide is ideologically most polarizing in Germany and Sweden, where historically established liberal parties have long been able to mobilize managerial classes. In all selected cases, it is evident that sociocultural professionals and managers are heterogeneous in their economic orientations, while they usually have quite comparable preferences for cultural universalism (cf. Marchesi Reference Marchesi2022; Oesch and Rennwald Reference Oesch and Rennwald2018). Furthermore, the new class cleavage is strongly polarizing in all eight cases but does not differ substantively from fiercer antagonisms between high- and low-educated voters. In Belgium, Germany, Spain and Italy, these two divides are primarily associated with the cultural dimension, while in the four other cases, they are characterized by ideological polarization on both dimensions.
The particular alignment of partisan and social groups in the UK, Poland and Hungary warrant more elaborate discussion. First, structural and political divisions are ordered on a single polarity in the UK, which may be a result of its two-party system and corresponding cleavage structure. The main political division between Conservative and Labour electorates is as such aligned with a structural division between the low-educated, production workers and small business owners on the one hand, and the high-educated and sociocultural professionals on the other. In the same manner, we see that the political spaces of Hungary and Poland are divided into, at the time, ‘government’ and ‘opposition’ blocs. In Poland, the position of the Law and Justice (PiS) electorate corresponds to a homogeneous structural coalition of the low-educated, production workers and rural and Catholic voters. Polish opposition parties, on the other hand, rely on a structurally more divided support base, with the left-wing Zjednoczona Lewica (ZL) electorate closer to urban voters and sociocultural professionals, and the centre-right Civic Platform (PO) electorate closer to the high-educated and managers. In Hungary, finally, the Fidesz electorate is closest to the mean points of production workers and – surprisingly – of urban and non-religious voters; two groups that are markedly more conservative in Hungary than elsewhere in Europe.
Overall, the results of Figure 2 confirm that the structuring of European political spaces is subject to significant national variations. In line with the second hypothesis, these findings challenge the notion of a common social and electoral rootedness of voter dimensions across Europe. The comparative analysis instead underscores the role of national cleavage structures and the corresponding party system: new political issues arise from social transformations and are consequently organized along existing cleavages through national party competition (Borbáth et al. Reference Borbáth, Hutter and Leininger2023; Bornschier Reference Bornschier2010b; Elff Reference Elff2009). While social transformations have thus generated similar political potentials across Europe, national parties forge distinct programmatic linkages around different structural conflicts and cleavages.
Regarding this point, we can speculate that cleavage-based party competition produces the observed opposite patterns of structuring in Western and Eastern Europe. In Western Europe, where party competition traditionally ranges from left-libertarian to right-authoritarian poles, we find that most structural divides align with a similar ideological polarity. In Eastern Europe, however, economic and cultural dimensions are connected differently due to the distinct issue composition of party systems (Marks et al. Reference Marks, Hooghe, Nelson and Edwards2006; Rovny and Edwards Reference Rovny and Edwards2012). As a result, structural and political divides reflect different ideological conflicts in Poland and Hungary: the ‘left’, which consists of sociocultural professionals and the high-educated, is ideologically right-libertarian, while the ‘right’, comprising production workers and the low-educated, is ideologically left-authoritarian.
Discussion and conclusion
This article set out to compare the dimensionality and structuring of national electorates in Europe. Due to comparable socio-structural transformations in recent decades, it is often assumed that demand-side political spaces have converged into a ‘basic structure’ of political conflicts – in line with the common structure observed in European party systems. Focusing on eight dissimilar countries, two hypotheses were formulated. First, that the dimensionality of European voters does in fact vary across countries, because of differences in national demand, supply and institutional contexts. Second, that partisan and social groups are positioned differently in the national political space, due to country-specific variations in cleavage structures.
The above analyses have supported these hypotheses partially. On the one hand, this study shows that political attitudes continue to be framed along two dimensions – economic and cultural – which have been rearticulated to include new issues related to globalization and the de-bounding of the welfare state. In all countries, political spaces have indeed been redrawn towards heuristics of integration and demarcation, without giving rise to a third ideological dimension. On the other hand, the varying positions of electorates and social groups within the eight national political spaces indicate that the two dimensions divide distinct groups of voters in most countries. This second result points to specific divergences in the ideological structuring of national electorates, suggesting that European political spaces still carry the imprint of country-specific cleavage variations.
By considering both the dimensionality and the structuring of political spaces, this article has uncovered more by-country differences than extant research (cf. Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008; Wheatley and Mendez Reference Wheatley and Mendez2021). Although the empirical approach does not point to the exact source of these differences, they largely reflect established cleavage variations or their politicization by parties. First, it appears that the dimensionality of voters’ attitudes corresponds to the structure of national party systems. In post-communist societies, where environmentalist green parties have not developed out of postmaterialist social movements, climate attitudes are, for instance, (partially) detached from the cultural dimension (cf. Chaisty and Whitefield Reference Chaisty and Whitefield2015). Second, the structuring of political spaces remains defined by existing alignments between parties and social groups that were forged along national cleavages. These durable programmatic linkages are evident in, for example, the UK political space, where the dominant Conservative and Labour parties have channelled salient structural divides along a single left–right polarity.
An alternative reading of these results could suggest that the observed cross-national variations reflect different temporal stages in a common trajectory of political change – rather than diverging patterns of cleavage (re)structuring.Footnote 12 However, such a constellation of ‘laggards’ and ‘leaders’ in cleavage change risks downplaying the political choices of parties and political elites (Elff Reference Elff2007). Indeed, as Zsolt Enyedi (Reference Enyedi2005) contends, parties act as ‘combiners’ between social structure and attitudes: they can amalgamate group interests into ideological packages, consolidate and demarcate political identities and thus preserve the existing cleavage structure. Political elites may forge these linkages through programmatic appeals, policy achievements or even through clientelist networks – allowing specific groups of voters to (re)position themselves in the political space towards the party’s formal position (Bornschier Reference Bornschier2010b; Kitschelt Reference Kitschelt2000; Warncke Reference Warncke2025). Political agency may as such perpetuate different cleavages in different countries, so that the political space reflects both old and new conflicts. Overall, the main conclusion of a ‘common dimensionality but a different structuring’ is consistent with earlier comparative work in cleavage theory (e.g. Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006, Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008) – although, crucially, the current study has extended this finding to countries outside of Western Europe. Generalizing these results, the article’s key findings are corroborated by separate results for the 10 countries that were not presented in the analysis (see Appendices F and H of the Supplementary Material).
To conclude, it should be emphasized that the reconstructions of a political space may differ based on the selection of variables. As Rovny and Whitefield (Reference Rovny and Whitefield2019: 6) observe, ‘there is no population of issues to capture or sample from’ when reconstructing political spaces, and crucial issues on state intervention, populism, democratic organization or – in countries such as Spain and Belgium – regionalism and decentralization were left out of the current analysis. Furthermore, while this article found partial support for a demand-side basic structure, future research using more recent data should investigate whether potential critical junctures like the European refugee crisis, Brexit or the COVID pandemic have rearticulated political conflict differently across Europe (see, for example, Hooghe and Marks Reference Hooghe and Marks2018; Hutter and Kriesi Reference Hutter and Kriesi2019). Party systems in Germany, Spain and Italy have changed radically due to these events, yet remained relatively stable in countries such as Sweden, Belgium and Poland. It remains to be seen whether these transformations on the supply side will lead to a restructuring of the national demand side as well.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/gov.2025.10018.
Data availability
The data are available at the ESS Data Portal at https://ess.sikt.no/en/?tab=overview.
Acknowledgements
I would like to thank the reviewers and editor(s) of Government and Opposition for their valuable feedback on a previous version of this article. Special thanks go to Koen Abts, Simon Bornschier, Cecil Meeusen, Marc Swyngedouw, Cristiano Vezzoni and the participants of the NASP-POLS Seminar in Milan, whose comments all helped to improve the manuscript.
Financial support
Research was made possible with support of the KU Leuven Special Research Fund (C1l-19-00569).
Open access
