How attitudes towards the welfare state are formed

6In both political science and sociology, the issue of national preferences in terms of political, economic, and social issues has been studied from two different perspectives. One line of research has concentrated on analysing the underlying social determinants of these preferences. At the same time, another avenue of study has instead focused on how national preferences are established, and how they are likely to evolve over time. The first section of this article describing the existing literature will engage with both of these perspectives, while the two subsequent sections will empirically test several conclusions which emerge from the literature.

7The social determinants of attitudes towards major political and economic issues are usually addressed using one of two very different theoretical approaches. In the neo-classical economic paradigm, individuals are exclusively guided by self-interest. They develop preferences that are congruent with their material interests, elaborating a cost-benefit analysis to decide on the best options available, given the situation. For example, individuals are thought to evaluate positively the aspects of the welfare state from which they can personally benefit, and rate negatively those aspects from which they cannot benefit. [11] The concepts of public interest or collective solidarity are, in this paradigm, ex post rationalisations that mask the reality of individual motivations.

8Social and political psychology has developed a competing theoretical corpus to account for the fact that individual preferences are also strongly influenced by values and beliefs. It has likewise been shown that deep-seated symbolic attitudes shape attitudes and behaviours by means of predispositions acquired during primary socialisation. [12] A number of studies have gone so far as to highlight that these symbolic attitudes can have a greater influence than material factors on individual preferences, [13] and that they can in fact neutralise the weight of self-interest under certain conditions. [14]

9Comparative literature on welfare attitudes has demonstrated that these two perspectives can help to explain the attitudes of Europeans towards the welfare state. On the one hand, individual preferences have been shown to be intricately linked to levels of socio-economic vulnerability: the most disadvantaged groups are those that most support a large welfare state, as well as social policies designed with their protection in mind. [15] A similar observation was recorded with regard to the middle class, generally viewed as a major consumer of public services. [16] But on the other hand, attitudes towards the welfare state are also strongly correlated with norms and values, in particular with regard to social justice, solidarity, and altruism. [17] In the same fashion, several studies have shown that individuals on the left of the political spectrum are more likely to support a strong state with regard to social and economic issues. [18] A significant portion of the existing literature likewise underscores the combined effects of interest and ideology on support for the welfare state. Nonetheless, the manner in which these two factors interact has hitherto been little studied in Europe. [19] The interaction between ideology and self-interest also remains to be analysed with longitudinal data over the long term.

10How attitudes towards the state develop is a much more controversial question. One of two different perspectives is usually adopted when discussing opinion formation. According to the “minimalist” paradigm, [20] ordinary citizens are not very interested in politics or the state of the economy. They do not have the skills to make informed judgements on economic and social issues, or to position themselves with regard to complex political options. Their attitudes are consequently unstable, weakly structured, and can be called “pseudo-opinions” or “non-attitudes”. [21]

11Countering this paradigm, some political scientists have argued that ordinary citizens do not necessarily have a single, “true” attitude towards a given subject. On the contrary, individuals are likely to adopt several possible, and sometimes contradictory, attitudes. Consequently, even though citizens are often “ambivalent” towards the welfare state, [22] the majority of individuals are nonetheless capable of thinking about their preferences, and in particular of observing the conflicting values underpinning their attitudes. [23] From this point of view, preferences are at least partly determined by individuals’ deep-seated symbolic attitudes. But they also depend on a number of different considerations that are in their mind at the point when they express an opinion. [24] Attitudes towards the welfare state can therefore be influenced by internal factors (such as political awareness), [25] more contextual considerations (such as the evolution of ideological cleavages), [26] or the extent to which the discourse of political leaders matches up with their experience of current social and economic issues. [27]

Why social groups should shift together (or not)

12The instability of individual responses to survey questions does not, however, prevent us from establishing the consistency of attitudes from a more global standpoint. Drawing on the conclusions reached in studies on opinion formation, Page and Shapiro put forth the following hypothesis: even if ordinary citizens are not very interested in a topic, and hold a number of ideas concerning it that are only partially compatible with each other, public opinion can be seen as rational at the aggregate level. Page and Shapiro have thus demonstrated that global opinion trends are coherently correlated with major political and economic events. But how do the preferences of individuals or groups come together to form aggregates that are capable of behaving in a rational manner? Even though their opinions vary widely on a certain number of topics, individuals do not differ fundamentally with regard to how they respond to the messages that they receive from their environment and how they update their attitudes and behaviours accordingly. Under such circumstances, all publics react in concert to external stimuli. As a result, it may seem that public opinion shifts as a homogenous whole. [28]

13The theory of parallel publics has increasingly been confirmed by American studies. The pioneering work of Page and Shapiro on the General Social Survey data highlighted opinion trends on economic issues, which evolved in similar fashion across different income groups from 1972 to 1988. Using the same data over a longer period of time (1973-1996), Erikson, MacKuen and Stimson [29] then showed that preferences regarding state intervention exhibited analogous shifts across segments of the American population with varying education levels. Enns and Kellstedt [30] probed these results further and discovered that both the most and the least educated individuals altered their economic opinions in the same manner. Looking more specifically at preferences regarding public spending and social issues, Ura and Ellis likewise observed similarities in how opinions shifted across different income groups. [31] Soroka and Wlezien [32] reached the same conclusion for the period 1973-2004: attitudes towards social spending evolved in parallel for different income groups, ideological segments, and education levels.

14More recently, Kelly and Enns [33] have examined the relationship between demand for state intervention and income inequality. They demonstrated that both the lower and higher ends of the income spectrum tended to respond in the same fashion to growing inequality, by lessening their support for the state. Replicating Stimson’s policy mood with disaggregated data at the individual level, Enns and Wlezien analysed opinion shifts from the perspective of policy responsiveness. They emphasised that changes in demand for state intervention were almost identical across income groups and education levels during 1956-2006; therefore, the attitudes of the groups in question did not become polarised. In general, Enns and Wlezien conclude that the concept of parallel publics “appears to be the norm” when analysing sub-group opinions in the United States. [34]

15Very few empirical studies in Europe have adopted this perspective, however. Although Svallfors has compared the evolution of attitudes towards the state in different social classes, [35] he focuses on proving that attitude cleavages are marked identically in all types of regimes, rather than on examining to what extent opinion shifts in concert across different social classes. Soroka and Wlezien identified similar opinion shifts regarding government spending in the United Kingdom, Canada, and the United States. [36] The validation of the parallel publics hypothesis remains geographically limited, however. Strictly speaking, it only applies to liberal, Western-style countries where attitudes towards the welfare state are highly specific. The question of whether publics shift in parallel thus remains unanswered for the rest of Europe.

16We can likewise imagine a series of counter-arguments in support of the opposite hypothesis that publics are not likely to move in parallel fashion. Research on partisan polarisation in the United States, for example, has draw attention to the growing influence and power of the most active political elites and the electoral hard core of the most mobilised voters. It has thus been observed that these transformations have led political parties to send clearer – and often more extreme – ideological signals since the 1970s. [37] And although the partisan conflicts that divide the elites are not necessarily replicated in an identical fashion in how ordinary citizens comprehend social and political issues, several American political scientists have observed very different shifts between ideological groups. For example, Enns and Wlezien [38] noted that the opinions of Democrats and Republicans on government intervention had in fact begun to be polarised during the Clinton presidency. Even more recently, Brooks and Manza [39] looked at how support for public action had been affected by the Great Recession. They showed that between 2007 and 2009, the opinions of Democratic supporters were very stable, while support for government intervention dropped off markedly among Republicans and Independents.

17To what extent can this tendency towards mass polarisation be applied to Europe? Some of the numerous comparative studies suggest that partisan differences regarding the welfare state are lessening in Europe, at the level of both general public opinion and the political elite. First of all, the major European political parties seem less likely to ideologically oppose one another over the role of the welfare state than to form wide government coalitions allowing for the creation of a “liberal consensus” on the reforms designed to implement the state’s divestiture of the welfare state. [40]

18Studies on new political cleavages have led to similar conclusions. We can indeed expect that public opinion will become less and less polarised regarding the “old politics”, as ideological cleavages begin to shift from socio-economic values to cultural values tied to immigration and European integration. [41] Neo-institutionalist studies have made an analogous observation: [42] as the welfare state enjoys widespread public support in general and among several special interest groups, it is likewise to be expected that publics as a whole will behave similarly and react in a consensual fashion to major changes in public policy. Overall, European public opinion may be much less ideologically polarised with regard to the welfare state than American public opinion – and it could even be becoming less polarised by the day.

19Finally, in research on welfare regimes, several arguments have been put forth suggesting that social groups respond in different ways to economic and social changes. For advocates of power resource theory, different kinds of welfare states have given rise to different coalitions across social classes. [43] Today, several factors are threatening traditional coalitions. First of all, the globalisation of trade and the internationalisation of economies are challenging the redistributive capacities of European welfare states. While the effects of increasing international competition vary significantly from one country to another, they all reduce the amount of leeway that European governments have to satisfy citizens’ expectations in terms of employment, social security, and equality. [44] The supranational constraints imposed by the EU also limit the autonomy of member states. And even though the pressure to adapt to European norms can be sidestepped by individual nations, [45] increased EU intervention is transforming how social policies are elaborated and negotiated. [46]

20All of these factors have led to the reconsideration of how class interests structure political conflicts concerning the redistribution of wealth. [47] Certain scholars [48] have suggested that a conservative shift is taking place: they argue that the middle and upper classes display a growing aversion to the welfare state, as well as increased hostility towards its beneficiaries, when they no longer feel that they are benefiting directly from public intervention. Others [49] have, on the contrary, defended the hypothesis that the middle class remains the main consumer of public services: it should therefore continue to defend its self-interest and reject any reduction in government involvement.

21Empirical analyses contextualising reforms of the welfare state and the transformation of attitudinal cleavages between social groups have produced highly nuanced results. On one hand, some studies have shown that in times of economic crisis, [50] the lower classes and the least educated tend to show less solidarity with immigrants and other marginalised groups, whom they perceive as being undeserving of state support because they do not contribute sufficiently to the national economy. [51] On the other hand, comparisons between social classes do not reveal significantly different evolutions in attitudes towards the welfare state since the 1990s. As we have seen, support for social policies is stronger among the more disadvantaged segments of the population. At the same time, however, this support remains relatively stable over time for all social classes. [52]

22Overall, the question of whether parallel publics truly exist remains unanswered for Europe. Some of the research suggests that reductions in state intervention could lead to increased competition between social groups, and thus to the formation of divergent attitudes. However, other factors invite us to observe the presence of more permanent attitudes towards the welfare state, as well as more uniform dynamics of public opinion.

Europeans are increasingly statist, without becoming ideologues

23How is support for the welfare state evolving in Europe? Are different publics evolving in parallel? Before answering these questions, we must examine how expectations regarding the welfare state vary by country. My analysis is based on the data presented by the European Values Survey (EVS), conducted every nine years and in a growing number of countries since 1980. In order to cover a sufficient number of countries as well as a relatively homogenous political, cultural, and institutional context, I make comparisons using the 23 EU countries included in the last three waves of the survey. [53]

24Since the second EVS in 1990, Europeans have been asked to describe their socio-economic preferences by positioning themselves along a series of 10-point scales. These scales all combine a form of social orientation with a form of liberal (economic) orientation. For each question, Figure 1 presents a “liberal preference index”. This index corresponds to the ratio between the respondents who express a liberal (economic) preference and those who state a social preference. [54] If the index is equal to 1, liberal and social preferences are equally strong (or weak). When it is greater than 1, liberal preferences are more important than social preferences, and vice-versa. Figure 1 highlights the fact that public support for the main tenets of economic liberalism (competition, private ownership of the means of production, and individual effort) has significantly declined since the beginning of the 1990s. The same is true of the value of freedom, which was almost on par with the value of equality in 2008. The questions concerning individual responsibility and constraints to be imposed on those receiving unemployment assistance are more stable and relatively well supported. [55] Figure 1 thus gives us a somewhat paradoxical view of Europeans: public opinion seems to combine growing demand for state intervention with fairly strong preferences for certain aspects of economic liberalism.

25Figure 2 uses a statistical indicator – Cronbach’s alpha – to nuance this interpretation, illustrating the degree of correlation between the items presented in Figure 1. For the year 2008, the overall correlation was fairly weak across the 27 EU countries. [56] This result, which is consistent with existing research on the ambivalence of attitudes, reveals that Europeans express highly different, and at times opposing, socio-economic preferences. This result also supports the argument that public opinion as a whole is not made up of “ideologues” as defined by Converse, insomuch as individuals do not primarily structure their attitudes around a dichotomy between liberalism and socialism. Nonetheless, it should be noted that socio-economic preferences are more consistent in Scandinavian and Northern European countries, likely because the tension between the market and the state has taken on a more institutional character in those contexts.

26In order to evaluate to what extent political awareness influences the consistency of attitudes, an additional index was elaborated from the EVS questions concerning the respondents’ expressed interest in politics, the importance of politics in their daily lives, the frequency of political discussions with their friends, and the attention they paid to political events. [57] Figure 2 confirms existing research on attitudinal consistency by demonstrating that the most politically minded Europeans (who correspond here to the top quartile) combine their preferences in terms of economic and social issues in a much more systematic manner than their peers. These individuals are even more “ideological” when they are university graduates. Some Eastern European countries are exceptions to the trend, however. In Estonia, Slovakia, Poland and Hungary, higher than average education levels appear to mitigate the impact of interest in politics on the consistency of socio-economic preferences. It thus seems that in those countries, the tension between the market and the state was an issue of much greater concern for the most educated segments of the population.

Figure 1

Evolution of the liberal preference index from 1990 to 2008 in the EU (23 countries)

Evolution of the liberal preference index from 1990 to 2008 in the EU (23 countries)

27In short, even though Europeans are often contrasted with Americans and presented as statists, egalitarian, and hostile to capitalism and the market economy, preliminary results suggest that their socio-economic preferences are not as homogenous as one might think. Even though social expectations and mistrust of the market are on the rise since the 1990s, Europeans appear to combine social economic values with more liberal economic values. These results are particularly interesting since they disprove the theories of both American and European exceptionalism. Consequently, they broadly support research demonstrating that Americans adhere to liberal ideals (“philosophical conservatism”), while in practice maintaining more social expectations (“operational liberalism”). [58]

Figure 2

Consistency of socio-economic preferences of Europeans in 2008 (27 countries)

Consistency of socio-economic preferences of Europeans in 2008 (27 countries)

Deconstructing socio-economic preferences and charting their evolution

28Preliminary results suggest that the socio-economic preferences of Europeans display composite logic, likely to evolve in different ways depending on the country in question. In fact, studies on welfare attitudes have recently emphasised the multi-dimensionality of attitudes towards the welfare state in Europe. [59] In order to test the theory of parallel publics, we must therefore first identify the main dimensions shaping European socio-economic preferences. [60] Different statistical techniques can be used to identify the latent structure of a set of quantitative variables. Here, I used principal component analysis (PCA) in order to rank the rationales behind European responses. Varimax orthogonal rotation enabled me to isolate two dimensions (Table 1). The first dimension is by far the strongest and the simplest to interpret. It accounts for 30 per cent of the total variance and encompasses questions on state control of the economy. [61] This dimension can therefore be seen as expressing the intensity of European social expectations and demand for state intervention. [62] This dimension will constitute the variable to be explained by the analysis which follows.

29In international comparisons, it is particularly important to ensure that the different items that make up an attitude scale have equivalent statistical association profiles in all the countries involved. [63] I controlled the validity of the attitude structure obtained by conducting different principal component analyses for each combination of countries and survey waves. Among the 23 countries examined here, the ranking of dimensions, the items correlated with them, and the percentage of variance that they represent barely changed between 1990, 1999, and 2008. The conclusion drawn from these analyses is therefore that the overall attitude structure is very stable over time and across space. [64] The demand for state intervention is therefore a clear and lasting dimension of European socio-economic preferences.

Table 1

Principal component analysis of economic preferences in the EU (23 countries, 1999-2008)

Demand for state Demand action for equality The state should take more responsibility for meeting everyone’s needs 0.70 0.05 Businesses and industries should be further nationalised 0.64 0.26 Competition is dangerous 0.64 0.19 Unemployed persons should have the right to refuse a job that does not satisfy them 0.60 -0.42 Incomes should be more equally distributed 0.02 0.76 Equality is more important than freedom 0.23 0.59 Percentage of variance explained 30 19

Principal component analysis of economic preferences in the EU (23 countries, 1999-2008)

The correlations of active variables with the factors are shown here. The KMO statistic (0.671) and the Bartlett test (χ² = 17244.42; p < 0.001) indicate an acceptable correlation between the items.

30In order to better account for variations in the demand for state involvement between European countries, and to analyse how these variations combine with individual characteristics, each respondent’s coordinates on the first dimension were isolated by the PCA and then extracted. For ease of interpretation, these factorial scores were adjusted to a scale of 0 to 1. Finally, different multi-level models were tested through the regression of the dependent variable thus established. Multi-level analysis is particularly useful here, as individuals are nested into different geographical and temporal units. The main goal of this kind of analysis is to explain variations of the dependent variable at the individual level, while controlling the average level variation of this same dependent variable between different aggregated units. [65]

31While there exist multi-level models specifically designed to analyse panellised data (growth models, for instance), their application to longitudinal datasets in particular is not yet completely stable. [66] The approach used here is one that is frequently employed for longitudinal analyses of international survey datasets. This approach consists of ranking variations across levels. Schematically, individuals, survey waves, and countries are separated, treating the first as level 1 units, the second as level 2 (meso) units, and the third as level 3 (macro) aggregate units. A three-level model is thus established, which has the advantage of being able to disentangle variations in the demand for state intervention that are attributable to evolutions over time and differences across countries.

32Table 2 presents several models charting the influence of different individual level characteristics on the demand for state intervention. Model 1 corresponds to what is sometimes called an “empty” model, with no explanatory variable. Similar to a random effects analysis of variance (ANOVA) model, it allows us to compare the share of variance of the dependent variable that is explained by the different levels. This share of variance is provided by the intraclass correlation coefficient (ICC), whose value is 0.016 for the macro level and 0.056 for the meso level. The coefficient shows that 1.7% of the variance in demand for state action is explained by country differences and that 5.6% of it can be explained by differences between survey waves (the rest is explained by level 1 differences). Although they are weak, the variations attributable to differences in countries and survey waves are statistically significant. The most striking result is therefore that expectations of the state are more influenced by differences between individual Europeans than between national differences or survey waves. This kind of result, frequent in literature on welfare attitudes, [67] merely highlights that there are greater differences between individuals than there are between countries. It also demonstrates the more substantial value of contradicting culturalist interpretations, which often argue that national characteristics are homogenous and that countries differ more greatly than individuals. [68]

Material interests and political values simultaneously influence the demand for state intervention

33In order to understand how publics move in parallel (or not), we must determine, in a two-stage analysis, to what extent they are driven by self-interest and/or political orientation. First, in order to evaluate the influence of self-interest, employment status and household income are introduced into the analysis (model 2). Household income is divided into quartiles, with the individuals living in the wealthiest households serving as reference. A dummy variable combining the respondents who did not provide information regarding their income was added (DKNA income), in order to be able to compare this sizeable segment, which accounts for 17 per cent of the aggregated dataset.

Table 2

Multi-level analyses of the demand for state action in the EU (1990-2008)

Model 1 Model 2 Model 3 Coefficient t-value Coefficient t-value Coefficient t-value Woman 0.026*** 21.14 0.024*** 19.88 Age (std) -0.0003 -0.4 0.0006 0.96 Education level (std) -0.009*** -12.99 -0.009*** -13.11 Income --(ref. ++) 0.054*** 23.69 0.048*** 21.62 Income - 0.041*** 19.43 0.037*** 17.97 Income + 0.023*** 10.7 0.020*** 9.6 DKNA Income 0.02*** 8.44 0.017*** 7.48 Employed (ref. self-employed) 0.046*** 16.32 0.041*** 14.92 Unemployed 0.092*** 24.65 0.085*** 23.24 Retired or not working 0.044*** 14.94 0.040*** 13.91 DKNA employment status 0.051*** 8.2 0.046*** 7.66 Left (ref. right) 0.099*** 52.47 Centre 0.046*** 27.61 DKNA political orientation 0.058*** 28.56 Interest in politics -(ref. --) -0.003 -1.84 Interest in politics + -0.006** -3.22 Interest in politics ++ -0.005* -2.28 DKNA interest in politics 0.014*** 3.69 Constant 0.398*** 67.5 0.307*** 50.31 0.267*** 43.42 Variance of the constants (individual level) Regression variance (wave level) Regression variance (country level) Coefficient 0.0251 0.001 0.0004 Std. Err. 0.0001 0.0002 0.0002 Coefficient 0.0243 0.0012 0.0002 Std. Err. 0.0001 0.0003 0.0002 Coefficient 0.0232 0.0012 0.0002 Std. Err. 0.0001 0.0003 0.0002 N (individual level) 65,856 65,856 65,856 N (wave level) 69 69 69 N (country level) 23 23 23 ICC (wave level%) 0.056 0.056 0.055 ICC (country level%) 0.017 0.009 0.008 AIC -55,311.91 -57,629.62 -60,399.29 BIC -55,275.53 -57,493.19 -60,199.2

Multi-level analyses of the demand for state action in the EU (1990-2008)

std = standardised; * p < 0.05 ** p < 0.01 *** p < 0.001.

34Next the impact of ideology is plotted, on the one hand to appreciate the scope of European partisan polarisation, and on the other to see if ideology mitigates the influence of self-interest or not (model 3). Two canonical variables are used here. Political orientation is derived from subjective positioning on a standard 10-point scale. Four categories are specified: left (1 –4), centre (5 –6), right (7 –10), used as the reference point; plus a category termed DKNA political orientation for those respondents who did not position themselves on the scale (about 24% of the dataset). In order to understand the effect of political awareness and orientation, an additive index (presented in Figure 2) was likewise introduced. Those individuals least interested in politics provided the reference point. A dummy variable was added to include respondents for whom the index could not be calculated, due to insufficient observations (11.5% of the dataset). Age, sex, and education level (i.e., the age at which respondents left school) were introduced as control variables. Age and education level were standardised to obtain comparable parameters.

35The results produced by model 2 are entirely congruent with research on the determinants of welfare attitudes: that is, different publics seem to follow their respective self-interests. As a result, the impact of income is highly linear: the higher their household income, the fewer social expectations respondents express. The same holds true for employment status. Compared to the self-employed, all socio-professional groups are more likely to support an interventionist state. This is especially true for the unemployed, the retired, and those who are not engaged in active employment, who all depend more directly on social transfers. The same is true for women, which tends to support the argument that women are often more dependent on the state, but also that they are more attentive than men to issues of equality and redistribution. [69]

36The effects of education are less clear-cut. On the one hand, we might expect that higher education levels would make individuals more open, tolerant, and drawn to altruism and solidarity. Socialisation to democratic values, measured by number of years of study, would thus lead to a more marked preference for social rights. On the other hand, the sociology of school experience in relation to feelings concerning fairness has shown that high education levels often lead to increased belief in personal merit. We might therefore expect that the most educated individuals will reject the egalitarian norms enshrined by the welfare state; this tendency would only be exacerbated by the fact that higher education levels generally lead to higher income levels and consequently to less objective personal interest in social action. This second hypothesis is largely validated here, congruent with the results of several empirical studies. [70] As for age, it does not appear to be significantly linked to demand for state action, which is also a result supported by existing research on welfare attitudes. [71]

37Moreover, model 3 confirms the value of an approach based on symbolic politics. In fact, it can be observed that ideology has a very clear influence on attitudes towards the welfare state. If we compare the magnitude of the coefficients, left-right positioning on the political spectrum has a greater impact on support for state action than being unemployed or having a low household income. Interest in politics displays a negative correlation, albeit less markedly, with social expectations: the more an individual is politically aware, the less s/he is likely to support interventionist policies. Individuals who did not position themselves on this index (and who are perhaps the least politically active) were however significantly more statist in their responses than the most politicised respondents. Nonetheless, introducing ideological variables in model 3 left the socio-economic variables’ coefficients virtually unchanged. It would appear that self-interest and political values simultaneously influence the demand for state action. [72] Later on, we shall examine how these two elements interact.

Public cleavages evolving in parallel

38The multi-level models reveal that the demand for state intervention is on the whole significantly different from one public to another, and from one survey wave to another. However, the models do not reveal the direction in which the expectations of social groups are evolving. In order to have a more precise idea of these shifts, the values predicted by model 3 were calculated for different publics and for each of the survey waves. Technically, these are “adjusted” values, which add up the values predicted by the fixed effects part of the model with specific contributions based on the random effects predicted by the same model. Figure 3 gives a more intuitive visual picture of this situation, in the form of a series of regression lines illustrating the data point clusters formed by the individual predictions.

39The two panels of Figure 3 depict strikingly similar shifts for different publics. Between 1990 and 2008, all the social groups represented developed greater expectations in terms of state engagement. This was true for both men and women, across all age groups and education levels. Income and employment status were the most polarised, especially for the unemployed and the self-employed, whose respective expectations were separated by about 0.11 points on a scale ranging from 0 to 1. However, these groups still evolved in parallel. Ideological groups followed the same trend. Although Europeans close to the right side of the political spectrum are much less likely to support the welfare state than Europeans on the left (with a deviation of 0.1), they all shifted in unison and became more interventionist between 1990 and 2008. [73]

40The publics that make up European public opinion thus evolve in parallel, while remaining quite divided as to the extent of their social expectations. This strong result only partially confirms the existing research focusing on class conflict and growing competition surrounding state redistribution. Significant cleavages do exist between social groups, likely reflecting the fact that individuals have very different levels of expectation with regard to state action. At the same time, however, while a rise in anti-state preferences could have been expected among the wealthiest social groups, on the contrary these groups became more interventionist too. In short, all European citizens seem to be demanding more state intervention.

Who wants a bigger state?

41These results tend to confirm the hypothesis of parallel publics in Europe. One question remains, however: although all publics shift in the same direction, do they all shift to the same extent? In other words, if all social groups called for more state intervention in 2008 than they did in 1999, is this the result of certain groups developing greater expectations over time? In studies on opinion dynamics, asking the question “who shifts when public opinion shifts” is generally a way to examine the influence of elites versus that of ordinary citizens. We can thus wonder to what extent global opinion shifts are produced by minorities of well-informed individuals, who may be more sophisticated or educated than others. It is assumed that these individuals are more sensitive to changes in their social, political or economic environments, since they process more information and are thus in a position to adequately evaluate the transformations around them. [74]

Figure 3

Evolution of the demand for state action among different social groups

Evolution of the demand for state action among different social groups

42We can begin to answer this question by comparing shifts in the demand for state intervention for different social groups. Overall, social expectations grew about 10% [75] – a rather modest, albeit significant, growth rate for a period of almost twenty years, both statistically speaking and from the traditional perspective that emphasises the stability of European values and attitudes. [76] This rate largely conceals the varying shifts of different publics (Figure 4). These shifts are the most marked among Europeans living in the wealthiest households (respectively +17% and +13% for the two top quartiles), among Europeans on the right side of the political spectrum (+14%) and among the most educated Europeans (+13% for those who those who studied until they were at least 22 years old, and +19% for those who completed their studies between 19 and 21 years old). Similar trends have already been observed. [77] They attest to the fact that the most well-off individuals (both in terms of economic and educational capital) react more strongly to changes in their environment. On the other hand, the publics with the most stable social expectations are the most uninvolved with politics (+8% for those least interested in politics, as well as for those without any stated political orientation) and the most distant from the jobs market (+8% for individuals who are retired or not engaged in active employment), which might explain why they are less reactive than other Europeans.

Figure 4

Growth rate of the demand for state action across different social groups (1990-2008)

Growth rate of the demand for state action across different social groups (1990-2008)

43Even though there is a threefold percentage difference between the most labile publics and the least labile publics (5% versus 17%), most of the social groups exhibit variations that are close to the average (10%). Under these conditions, it is possible to argue that no public truly emerges as being solely responsible for the increased demand for state intervention in Europe, or even as contributing to this demand much more significantly than other groups. [78] These results confirm existing studies demonstrating that individuals are equally reactive across the board, as they assimilate the information to which they are exposed in the same fashion and similarly update their opinions based on that information. [79] In general, while the most well-off Europeans, both economically and educationally, are also the most reactive citizens, it is nonetheless also true that all citizens are reacting to transformations in their social environment by demanding greater state intervention.

Significant partisan cleavages, but no ideological polarisation

44We can nuance the analysis of partisan polarisation by looking at how the influence of ideology and that of self-interest interact over the long term. To do so, I estimated a model similar to model 3, to which were added the effects of interaction between political orientation and household income. The objective was to determine whether income had a different effect on the social expectations of Europeans on the left and right sides of the political spectrum.

45Figure 5 presents the values predicted for 1990, 1999, and 2008, respectively. The linear predictions were calculated for the fixed effects part of the model. In addition to visually confirming that ideological groups have shifted in parallel, the upper portion of Figure 5 illustrates that income has had a very different impact on the left and right sides of the political spectrum. While right-leaning Europeans are more likely to favour their self-interest and are less likely to support the welfare state, especially the richer they become, left-leaning Europeans are just as statist when they are poor as when they are rich. In other words, political values mitigate the influence of self-interest, but only for those on the left side of the political spectrum. As a result, it is among the right side of the spectrum that the attitude gap between the rich and the poor will be the most significant: poor right-wing Europeans are markedly less averse to the welfare state than their rich counterparts.

46This differential effect caused by income is very stable over time, and leads to similar observations in 1990, 1999, and 2008. It stands in stark opposition to the American theory of partisan polarisation, but also to the theory that the liberal consensus of political elites would have weakened ideological cleavages in Europe. When looking at income groups (which are particularly interesting because they are publics whose social expectations exhibit the greatest cleavages), it emerges that neither ideological polarisation nor convergence has occurred between the rich and the poor between 1990 and 2008. Rather, a kind of consensus of public opinion regarding growing state intervention seems to be emerging. However, as has been demonstrated for France specifically, the fact remains that this consensus appears to be more solid among left-leaning Europeans. [80] On the right side of the spectrum, support for state action appears to fluctuate much more in terms of wealth. Typically, the gap between the poor on the right and the rich on the right is almost as great as the difference between the poor on the right and the poor on the left.

47In order to determine whether political awareness influences expectations towards the state for different ideological groups, a final model was created with interaction effects allowing for interest in politics to vary in accordance with political orientation. The lower portion of Figure 5 shows that left- and right-leaning Europeans express increasingly different attitudes towards the state as they declare more interest in politics. In other words, the attitudinal distance between left- and right-wing Europeans is greater among the most politicised segments of the European population. And this distance narrows as interest in politics declines.

Figure 5

Demand for state action across ideological groups according to income and interest in politics (values predicted with confidence intervals at 95%)

Demand for state action across ideological groups according to income and interest in politics (values predicted with confidence intervals at 95%)

48While political competence shapes expectations towards the state by helping individuals to translate their symbolic predispositions and political values into more or less statist attitudes, ideological groups are nevertheless no more polarised in 2008 than they were in 1990. Regardless of the survey wave, the most politicised Europeans seem, both on the left and the right, to align themselves equally with pro- and anti-statist positions, while the least politicised citizens are developing expectations that edge closer to the average. This final result reminds us that even if ordinary citizens all react to transformations in their environment, they are not all equally susceptible to the latent ideological cleavages behind their social preferences.

49* * *

50A large corpus of empirical studies has examined the socio-economic preferences of Americans since the 1970s. This body of work, with no equivalent in Europe, has identified several key factors to account for attitudes towards the welfare state and their evolution over time. Drawing on longitudinal survey data covering 23 EU countries, the results presented here highlight several opinion trends that are very similar to those identified by American researchers.

51First of all, my research confirms studies on the parallelism of changes in public opinion. Between 1990 and 2008, Europeans as a whole became more favourable to state intervention in economic and social matters. Moreover, their social expectations globally evolved in similar proportions. Nonetheless, there are important cleavages in attitudes towards the state. One of the most striking cleavages relates to political orientation, pitting the statist preferences of left-wing Europeans against those of right-wing Europeans. This fault line is exacerbated by other factors: left- and right-wing Europeans develop increasingly different social expectations as they earn more money or become more interested in politics.

52Ultimately, however, these cleavages seem stable over time and can be observed for each survey wave, merely moving upwards from 1990 to 2008. Contrary to the common sense argument that ideological oppositions are gradually disappearing in Old Europe, one of the conclusions drawn here is that ideology remains a very divisive factor with regard to European socio-economic preferences, and in particular with attitudes towards the welfare state. Running counter to the theory of partisan polarisation, these results show that political orientation does not divide Europeans with regard to state intervention more in 2008 than it did in 1990. In short, “old politics” continue to divide Europeans, but no more so today than yesterday.

53It is tempting to view Europeans’ growing demand for state intervention as the response of ordinary citizens to what has sometimes been identified as a turning point or a liberal consensus among political elites. While social expectations have grown significantly since the 1990s, we must not forget that European public opinion remains very ambivalent on the subject. Its statist preferences are combined with much more conservative economic preferences. Parallel publics demanding a larger welfare state are thus far from constituting a homogenous majority. As a result, the goal for policy makers is perhaps not merely to satisfy the expectations of public opinion, but more specifically to satisfy them without also frustrating more economically liberal expectations.

54The theory of parallel publics stems from the premise that different social groups similarly interpret the political and economic signs and stimuli that they receive from the outside. Consequently, these publics can react analogously and shift in unison. The data provided by the EVS tends to support this theory in Europe. However, we should not misconstrue the nature of this theory’s empirical value. The fact that different publics can shift in the same direction in Europe and the fact that parallel shifts can also be observed elsewhere does not necessarily mean that these shifts are guided by the same reasons. The empirical validity of the mechanism underpinning the parallel publics theory is one thing: opinions shift in the same direction because they respond to the same contextual information. The more substantial content accompanying this mechanism is another: opinions can in fact shift in the same direction, but for very different reasons.

55Consequently, although Europeans seem to react identically to changes in their environment by calling for greater state intervention, their social expectations are no less likely to stem from very different attitudes towards public action from one public to another, or one country to another. Typically, left-wing Europeans and right-wing Europeans may not envision the same forms of state action to mitigate social inequality. In the same fashion, the growing demand for state action expressed by the most economically fortunate Europeans is quite likely of a different nature than that expressed by less well-off households. A more qualitative approach could help to clarify how publics view state responsibility and its forms of intervention in times of economic crisis. In-depth interviews could also help to understand which social policies and forms of redistribution Europeans mean when they demand greater state intervention.

56While, in the absence of qualitative data, it is difficult to imagine the content of these social expectations, they likely contain both potential for legitimacy and potential for dissent that should be of interest to policy makers. The potential to legitimise public action: Europeans’ growing demand for state involvement could represent a counterweight to what is often presented as a crisis of faith in public institutions or of the legitimacy of the welfare state in general. But they may also provide potential for dissent: there is nothing to guarantee that left- and right-wing Europeans would agree on how to implement new forms of state intervention. Parallel opinions do not, therefore, preclude social conflict.