Introduction

Does a ‘drifting apart’ of ideological positions influence citizens’ political behaviour? The present study seeks to answer this question by investigating whether ideological polarisation of the citizenry drives political action. More specifically, I examine (1) whether ideological polarisation in individuals’ social environment causes them to become politically active, and (2) whether ideological extremists are particularly receptive to the influences of environmental ideological polarisation.

Voting as well as non-electoral modes of political participation allow people to express their views, make demands, or voice grievances (Verba et al. 1995; Oliver 2001). Thus, if ideological polarisation affects political participation and even unevenly mobilises different groups within the citizenship, a shift can occur in the representation of people’s preferences. For instance, if ideological extremists are mobilised by increasing polarisation while moderate citizens are not, the overall picture of people’s needs and desires may be distorted.

The research gap that I address is twofold. The potential consequences of ideological polarisation of the public have remained largely unexplored to date and constitute the first gap. The debate on public opinion polarisation (Baldassarri and Gelman 2008: 35), which has drawn increasing attention in recent years, is dominated by three questions. The first question is how to conceptualise polarisation (Iyengar et al. 2012; Mason 2015). The second question is whether public opinion has become increasingly polarised with regard to different ideological directions including, for example, liberalism/conservatism, economic orientations and distributional issues, attitudes towards other social groups, and moral beliefs.Footnote 1 The third question is whether people’s views have indeed become more extreme,Footnote 2 or whether the increasing elite polarisation merely better sorts individuals along ideological lines. Does this increase the gap between partisan subgroups, making mass polarisation an echo of elite polarisation but not an ideological movement per se?Footnote 3 However, the potential consequences of polarisation for the political and social system have not been generally considered. As such, we have limited knowledge about what happens when normative notions, ideological views or questions about how to organise social life become polarised. Does it politicise day-to-day life? Are people politically mobilised by polarisation in ideology in their social surroundings? To my knowledge, this question has not yet been answered (but see Kleiner 2018).

The second research gap pertains to the geographical focus. So far, research on mass polarisation has largely been limited to the US context, and the few studies related to Europe concentrate on the questions presented above (Jensen and Thomsen 2011; Down and Wilson 2010; Adams et al. 2012a, b; Schmitt and Freire 2012; Munzert and Bauer 2013; Kleiner 2016). The considerable challenges in Europe, however—autocratic tendencies, the Ukraine war, Islamic terrorism, the strengthening of populist parties and extreme right movements, growing immigrant populations, increasing numbers of refugees and increasing socio-economic inequality—necessitate further investigation of polarisation tendencies and their societal impacts in Europe.

The remainder of the paper proceeds as follows. After introducing the notion of polarisation, I argue that political behaviour is driven by anticipated normative deprivation due to mass polarisation. Next, I explain why the subnational context influences political behaviour. Subsequently, I describe the data, the operationalisation and the methodological design used for the empirical analyses. Macro-regressions as well as two-level regressions are conducted, and a new index to measure mass polarisation is introduced. I then examine whether mass polarisation with regard to political ideology on a subnational level has an impact on political participation. Empirical results indicate that ideological polarisation in the social environment motivates citizens to become politically active in a non-electoral way. Findings also show that ideological right-wing extremists are more mobilised by polarisation compared to non-rightists, while left-wing extremists are not mobilised by ideological polarisation.

Ideological polarisation

Mass polarisation extends beyond a simple variation in orientations. Rather, it can be interpreted as the intensification of orientation discrepancy dividing substantive parts of the society into opposing camps, while the moderates are losing ground (see Fiorina and Abrams 2011: 309; DiMaggio et al. 1996; Baldassarri and Gelman 2008).Footnote 4 Orientations and attitudes may systematically sort individuals along multiple lines of potential conflict and organise them into groups centred around exclusive identities (Baldassarri and Gelman 2008: 2), but such grouping does not necessarily lead to exclusive social identities. Rather, mass polarisation results in an increase in the ideological distance between oppositional factions within society—more precisely, the divide between the antagonistic camps becomes larger. The most influential literature (Fiorina and Abrams 2011; DiMaggio et al. 1996: 693; Baldassarri and Gelman 2008) identifies five characteristics of public opinion polarisation. First, the subject of disagreement is relevant not only to political elites, but also to the general population. Second, the public is divided, clustering at opposite extremes of opinion (Fiorina and Abrams 2011). Of course, the opinion structure can also be multimodal, but DiMaggio et al. (1996: 693) point out that bimodality is associated with the highest risk of social tensions. Third, tensions and polarisation become more likely as meeting the preferences and desires of both sides implies mutually exclusive social arrangements and societal goals (Jacoby 2014: 754; DiMaggio et al. 1996: 693). Fourth, although a population may occasionally be polarised on a single issue, it is more plausible that ‘people align along multiple […] issues’ (Baldassarri and Gelman 2008: 409). According to DiMaggio et al. (1996: 693), ‘[…] the more closely associated different social attitudes become […], the greater the likelihood of implacable conflict.’ That is, antagonisms about individual concerns rarely expand into serious problems, but those arising around a combination of social and political issues linked to a specific philosophy of life or world view have a greater tendency for expansion. Finally, these philosophies or world views underlie individuals’ identities, and the greater the degree to which ideologies or world views are bound to individual identities, the more likely they are to become the foci of affective social conflict (see Blau 1977, cited in DiMaggio et al. 1996: 693; Mason 2015).

Polarisation and political participation

Typically, the ordinary citizen concentrates on her or his private and working life and has little interest in becoming involved in politics or matters of public life (Hunter 1994: 10; Oliver 2001: 29), but social forces can influence an individual’s decision to behave in a political way (Oliver 2001: 110). Living in a social environment in which comparatively few fellow citizens hold moderate ideological beliefs, while the number of citizens with extreme positions—at both ends of the spectrum—is high, not only heightens the pressure to choose a side, but also leads to people to wonder if and to what extent politics takes their needs and interests into consideration. I argue that increasing ideological polarisation in the form of left–right-polarisation triggers worries about future social and political developments which could concern one’s values, convictions, and world views. Such worries may be accompanied by an anticipation of collective political deprivation, which in turn incites citizens to act. I derive my argument from the relative deprivation paradigm as well as theoretical considerations compiled by Chantal Mouffe and Émile Durkheim, outlined in the following section.

Relative deprivation

The concept of relative deprivation has had widespread influence in social sciences as an explanation for social protest (Dubé and Guimond 1986). It is centred on the sociological tradition of collective action research suggesting that citizenship is a product of structural aspects of the environment and that the social context explains political behaviour (Dalton and Wattenberg 1993; Ellemers 2002: 242). The starting point for relative deprivation is individuals with particular socio-economic characteristics and members of various social groups comparing and contrasting their life situation with that of comparable individuals or groups. They develop expectations about how economic, political, and social systems should operate in terms of equity–fairness and evaluate how fairly these systems treat them or their group. If these comparisons are unfavourable and people’s expectations about the goods and conditions of life to which they believe they are entitled are unmet, they experience a sense of grievance and frustration labelled relative deprivation (Merton and Kitt 1950; Taylor 2002). People may compare themselves to other individuals and feel personally deprived, or they may compare themselves as members of an important reference group to another group and feel group deprived (Runciman 1966; Smith and Ortiz 2002: 92). Studies have shown that personal deprivation leads to active attempts to change the social system (Vanneman and Pettigrew 1972; Smith and Ortiz 2002).

The sense of entitlement may stem from an uneven distribution with regard to socio-economic characteristics between groups (Taylor 2002: 14), and most portrayals of group deprivation indeed emphasise structural factors such as class, status, and power (Vanneman and Pettigrew 1972). It is likewise conceivable that groups compete for values, ideologies and convictions, and the feeling of belonging (Kaase 1976: 11). As moral concepts and convictions become politically relevant and citizens see that authorities can change outcome decisions, people may become active to improve their situation (ibid. 9; 13). The more central the dimensions on which deprivations are experienced and the clearer the perceived responsibility of political authorities for felt deprivation, the higher the potential for a reaction in the form of political behaviour (ibid. 17).

Deprivation always implies some kind of comparison, but the standard does not have to be current; it can be a past or future standard as well (Ellemers 2002: 243). Thereby, deprivation consists of a cognitive component, as the situation is judged and a comparison is drawn, as well as an affective component, like frustration or anger (Taylor 2002: 15). I argue that the fear of losing something valuable or not getting what one strongly desires triggers deprivation and can be a driver for action. Anticipated loss of the validity of one’s own values and ideology can be very threatening and energises people to act. To exemplify this argument, I additionally refer to Mouffe and Durkheim.

Values and identity

According to Chantal Mouffe, social systems are organised into groups with collective identities that contend for resources as well as sovereignty of definitions (Mouffe and Neumeier 2007: 12, 34–36). She argues that this struggle within the political arena is neither rational nor harmonic, but very passionate, especially if one’s way of life is at stake (ibid. 39). This passion can easily be understood through Émile Durkheim’s concept of mechanical solidarity. Durkheim searches for a mechanism that regulates social life and assures social cohesion, and he suggests that collectively shared moral values are the key to both. According to Durkheim, societies need common ideas about morality and concepts of obligations that are internalised and ‘shared by most average individuals in the same society […] inscribed upon everyone’s consciousness’ (Durkheim 2013: 60). Durkheim calls this moral intersection the collective conscience, which encompasses values, beliefs, ideas, and perceptions. Any violation of this collective conscience leads to feelings of morality infringement. If these values and beliefs are questioned or even threatened, the experience is not simply a violation of one’s personal values, but rather a violation of a universal principle. This experience of principle violation in turn typically triggers reactions and attempts to protect one’s own world view (Durkheim 2013: 76f.). To put it differently, ideologies and values are not restricted to the personal self but have universal claim and are experienced in this way (van Zomeren et al. 2012: 66). Thus, the threatening of the ‘social’ part of one’s personal identity is experienced as a violation of a universal principle which must be protected for the ‘greater good’ (Durkheim 2013: 76f.). In this way, ideological polarisation can lead to a defence of one’s own values as well as solidarity with those who share them. In line with this, Iyengar and Westwood (2015) showed for the US citizenry that hostile feelings for the opposing party are ingrained or automatic in voters’ minds, and that affective polarisation based on party exerts powerful effects on non-political judgments and behaviours.

Anticipated political deprivation and political action

Having reviewed these theoretical approaches, this paper now sets out the argument. Social groups vie with each other for resources as well as for prevailing norms. As long as the opinion structure is complex, the number of discussants is high and the more likely people are to discuss their views with persons who think differently (DiMaggio et al. 1996: 693). In the public arena, prestige bearers, leaders, and group representatives present their ideas in public meetings or conventions, and these ideas are published and discussed in the media. But as the opinion structure becomes bimodal, selection and evaluation of information as well as communication tend to reinforce attitudes and become circular (Leeper 2014).

As polarisation rises, the number of individuals posing a threat to the existence of one’s own world views increases, putting the ordinary citizen under social stress. Either there is a rather large group of people with a different world view, or one finds herself/himself in an environment of extreme ideological positions. Either way, the situation is competitive with regard to ideological views. People start to wonder how responsive politics are to differing interests and to what extent politics will consider their needs. The anticipation of increasing social and political influence of the opposing group (or two groups with opposing world views), accompanied by specific political policies that do not represent one’s own convictions, triggers political deprivation.

The process of becoming aware of such developments is a result of complex observations, interactions, and communication processes between individuals who hold the same views or between individuals with different views. It may even be more of a felt sensation than a conscious process. But certain social actors anticipate and articulate deprivation and push it over the threshold of individual awareness (Kaase 1976: 14). Once deprivation has crossed the threshold of awareness and people attribute responsibility to politics, the potential exists for them to become active to alleviate the situation (Kaase 1976: 13).

Ideological extremism

In addition to a general effect, it is also plausible that people who are already politically sensitised may react differently from the ordinary citizen. People who are emotionally or cognitively involved in politics tend to be more attentive to developments in their sociopolitical environment and are aware that ‘doing something’ is important to the decision-making process (Rogowski 2014). One such type of citizen is the ideological extremist, who positions herself/himself at the fringe on one or the other end of the left–right spectrum.

Ideological extremists are politicised individuals; they typically hold high levels of political information and are more involved in political actions than ideological moderates (van der Meer et al. 2009). This involvement can partly be explained by their generally greater distance from the government’s ideological position (van der Meer et al. 2009), but ideological polarisation is also likely to influence the relationship between extremism and political action.

While very different in their beliefs, the extreme right and the extreme left are unhappy with the current sociopolitical order and strive for alternative social, economic, and political arrangements (Domhoff 2015: 9). They very much identify with the normative content of their ideology which not only helps them to envision a better world, but also assists in rationalising how they think things should be. Thus, as their vision differs from reality, ideological extremists experience political deprivation whether their environment is polarised or not. In the context of polarisation, however, their longstanding grievance is combined with newfound support; the group of supporters grows, and new resources become available. This situation heightens the internal efficacy and the expectancy for future success in terms of envisioned political outcomes. Hence, the relationship between ideological extremism and political participation is conditioned by ideological polarisation.

Subnational regions as context categories

Ideological polarisation is not only relevant on the national level, but on the subnational level as well. The social context in which individuals live, meet friends, and go to work has an impact on their behaviour (Kleiner 2016, 2018). A region can be interpreted as a spatial unity with its own specific constellation of historical, social, political, cultural, and intellectual elements (Hirschle and Kleiner 2014). The geographical proximity of the actors living within a region leads to comparatively high interconnectedness of the actors resulting in dense regional structures thinning out outwards.

Due to their specific linkage of power relations, religious references, aspects of the countryside, economic and social factors, density of relationships, and regionally endogenous innovation potential, subnational regions develop their own regional character including values, mentality, and lifestyle. This geographically limited culture is cultivated, having an impact on political and economic capacities and performance (Florida 2002; Putnam 1993; Charron and Lapuente 2013) and acting as reference point to individual behaviour.

Consequently, subnational regions have been gaining meaning over the last years and have more and more been used as categories for comparative analyses. And, in fact, empirical studies illustrate the existence of interregional differences in terms of cultural, social, and political aspects within countries (Charron and Lapuente 2013; Beugelsdijk et al. 2006; Minkov and Hofstede 2014; Hirschle and Kleiner 2014). Therefore, it is important to recognise subnational regions as analysis units and use them as context categories to explain social and political behaviour.

The left–right continuum and hypotheses

Following the theoretical considerations, I suggest that ideological polarisation has a mobilising effect on citizens’ political behaviour. This effect is especially true for ideological extremists.

To capture mass polarisation, the left–right heuristic seems quite functional. Although not devoid of ideological commitment, ideology exists and matters in most people’s everyday lives (Jost et al. 2008: 134)Footnote 5 and political beliefs are linked to a subset of core values and principles (Feldman and Zaller 1992; Jost et al. 2009: 316).Footnote 6 In Europe, people usually use this ideological pattern to summarise and locate issues of their greatest concern in the political realm (Downs 1957; Campbell et al. 1960; Dalton 2006). Left and right are labels for orientation in political debates as well as positioning of others and the self.

However, as a heuristic its content is not absolute, but to a certain extent context dependent (Freire 2006; Adams et al. 2012a, b). It has even been claimed that recent shifts in Europe’s ideological landscape challenge the conventional left–right divide (Azmanova 2011). But the left–right dimension is still the most relevant instrument for party competition (Mair 2007), and empirical studies point in the same direction concerning the population (Scholz and Zuell 2012: 1415), showing that within societies there is general agreement about how to understand left and right (Scholz and Zuell 2012).

In sum, although there are differences between nations, there are strong reasons to suppose that within nations the political discourse is still organised in terms of left/right attitudes reducing the complexity of political reality for voters and making this scale a useful tool to measure ideological mass polarisation in European regions. Therefore, I assume the following associations:

H 1

The level of ideological polarisation on the regional level is positively associated with the citizens’ probability of political participation.

H 2

The effect of citizens’ ideological extremism on political participation is conditioned on the level of regional ideological polarisation.

Data and operationalisation

Data are drawn from the ESS cumulative data file conducted from 2002 to 2014.Footnote 7 To measure the dependent variables, two different forms of participation were used—voting behaviour and non-electoral participation. Voting behaviour is measured by the question whether the respondent had cast his/her vote in the last election. Respondents not eligible to vote—either because they were not old enough or not a citizen of the country in question—are marked as missing. Since I am not interested in the different forms of non-electoral participation, I generated a dummy variable for the seven different forms of participation obtained by the ESS. Respondents were asked whether they had contacted a politician, government or local government official, worked in a political party, action group or another association, worn or displayed a campaign badge or sticker, signed a petition, taken part in a lawful public demonstration or boycotted certain products within the last 12 months.

I use NUTS 2Footnote 8 level regions to model the context. Statistically, the subnational level offers the opportunity to broaden the analysis focus and to examine the relationship between polarisation and individual participation on a greater number of context units, increasing the reliability of the estimates. It also allows controlling for country-level effects (Kestilä-Kekkonen and Söderlund 2009; Beugelsdijk and Klasing 2016). Country dummies are included to control for unobservable country-specific characteristics.

In order to measure ideological polarisation, I created an indicator: First, the two categories furthest left on the ideological scale as well as the two categories furthest right are defined as left/right, while the more middle positions are summarised in a modest category.Footnote 9 Then, the regional mean for these categories was determined. Next, the level of ideological polarisation of a region was calculated by multiplying these means for extreme positions and dividing the result by the mean of the modest position.

$${\text{IPOL}}_{\text{Reg}} = {\text{MEAN}}_{\text{Left}} \times {\text{MEAN}}_{\text{Right}} / {\text{MEAN}}_{\text{Modest}}$$

The greater the index score, the greater the clustering of the respondents at the ‘ideological extremes’ and the greater the polarisation of the region in question.Footnote 10 I was able to obtain polarisation for 267 NUTS 2 regions in 31 European countries. Twenty countries were being considered in the analyses, since countries having less than three NUTS 2 regions and countries that are ‘not free’ or ‘partly free’ according to Freedom House were excluded for better comparability.Footnote 11

Ideological extremism is measured by the two outermost categories of the left–right dimension included in the ESS survey. Since it is unlikely that right- and left-wing preferences have the same impact on political action (van der Meer et al. 2009: 1430), I generated one dummy per political fringe.

Due to alternative explanations I control for several characteristics of individuals to exclude spurious effects: party attachment, ideological position, policy awareness through media consumption, political interest, age, gender, education, subjective household income, and left- as well as right-wing extremism (see Table 9 in the Appendix).

Analyses

In the first step, macro-regressions are conducted in order to consider the relationship between ideological polarisation and political participation on the NUTS 2 level. Table 1 shows the results regarding regional polarisation in the matter of voting (M01) and non-electoral participation (M02). As is evident, polarisation on the left–right dimension has a positive relationship with voting being non-significant at the 0.05 level (M01). In contrast, regional polarisation in ideology enters the regression positively with a statistically significant coefficient (M02), which holds true when controlling for points in time and countries.

Table 1 Regional ideological polarisation and regional turnout (M01) and non-electoral participation (M02), respectively
Full size table

Since coefficients from logistic regressions cannot be interpreted in an intuitively understandable way, I estimated marginal effects and plotted them across the full range of values of ideological polarisation (Fig. 1). Figure 1 plots the substantive effects of regional polarisation using the estimates of models M01 and M02 (Table 1). The black circles are the point estimates, and the grey area reflects the 95% confidence intervals associated with the probability estimates. Consistent with the regression model (M01), the graph on the left shows that polarisation concerning ideology is not connected to the probability of voting, since it does not rise or decline as polarisation rises. In contrast, as ideological polarisation increases, the respondents’ probability of non-electoral participation increases in a statistically substantive way, a finding that is consistent with the regression estimates.

Fig. 1
figure 1

Source: European Social Survey, 2002–2014 (www.europeansocialsurvey.org/. Accessed 15 Mar 2016)

Substantive effects of ideological polarisation and political participation at the NUTS2-level. Note The left graph depicts the marginal effects for voting, and the right graph depicts the marginal effects for non-electoral participation. The black circles are the point estimates, and the grey area reflects the 95% confidence intervals associated with the probability estimates.

Full size image

In sum, the impression is that ideological polarisation on the regional level is not connected to the probability of voting, but positively enhances the probability to participate in a non-electoral way.

Does this impression hold true for the individual level? In order to find out whether ideological polarisation affects the individual’s probability of participating two-level regressions with random intercept coefficients are conducted, whereby citizens (level 1) are nested in subnational regions (level 2). Again, inter-national differences are being controlled by country dummies.

Of the more than 150,000 respondents processed, approximately 80% reported that they had taken part in the election, and 43.2% stated they had participated in a non-electoral form within the last 12 months. About 12.1% of the interviewees positioned themselves at the margins of the left–right spectrum.

As a starting point, I analyse the relationship between the key variable measuring regional polarisation and the individual’s probability of voting (M03, Table 2) holding country differences and variations in time constant. To ensure an adequate number of individuals representing a region for the analyses conducted, only NUTS 2 regions with at least 50 respondents are used.

Table 2 Regional ideological polarisation and voting
Full size table

Since the dependent variable is dichotomous, logistic regressions are conducted. The intra-class correlation coefficient shows that approximately 11.8% of the total variance can be explained by introducing a second regional level of analysis. In a second step, I include the individual characteristics into the model (M04, Table 2).

Table 2 shows the results. As is evident, ideological polarisation has a negative relationship with voting, but the coefficient is non-significant at the 0.05 level (M03). However, when controlling for individual features the coefficient becomes statistically significant, but the effect is not robust. If I use the three most left and right categories for the polarisation index, there is no significant effect.

Even though the level-1 variables are not directly relevant to the question at hand, their influence on the dependent variable should be noted: being ideologically rather on the right, feeling close to a certain party, being ideologically extreme, being politically interested, educated, subjectively wealthy as well as being a female and of increased age raises the likelihood of casting his/her vote. The individual’s consumption of political programs in the media, however, negatively affects the probability of turning out to vote. In quite a few cases, the country in which the respondent lives also plays a role.

Table 3 shows the results concerning non-electoral participation. The intra-class correlation coefficient (ICC) indicates that including the regional level of analysis heightens the potential to explain the total variance of the dependent variable for an additional 16.3%.

Table 3 Regional ideological polarisation and non-electoral participation
Full size table

We see that polarisation has a positive relationship with non-electoral participation, and not only is the coefficient’s direction in line with our expectations, it is highly significant as well (M05) which holds true when controlling for individual features (M06).

Again, I estimated the marginal effects and plotted them across the full range of values of ideological polarisation (Fig. 2). Figure 2 plots the substantive effects of regional polarisation using the estimates of the models M03 and M04. The graph on the left shows that as polarisation raises the probability of voting declines, but to a rather small extent. For robustness, I rerun all analyses using 70 and 100 respondents per region as well as using the three outermost categories of the left–right dimension for measuring polarisation. All results remain stable (see Tables 6, 7, and 8 in the Appendix). In sum, I do find a weak but substantial connection between polarisation concerning ideology and the individuals’ decision to go to the polls.

Fig. 2
figure 2

Source: European Social Survey, 2002–2014 (www.europeansocialsurvey.org/. Accessed 15 Mar 2016)

Substantive effects of ideological polarisation at the NUTS2-level and individual political participation. Note The left graph depicts the marginal effects for voting, and the right graph depicts the marginal effects for non-electoral participation. The black circles are the point estimates, and the grey area reflects the 95% confidence intervals associated with the probability estimates.

Full size image

Does regional polarisation on ideology have an impact on non-electoral participation? Consistent with the results of M05 and M06, the right graph in Fig. 2 shows that participation increases substantially as regional polarisation grows. Thus, empirical results indicate that the probability of participating in a non-electoral way is enhanced as ideological polarisation increases. I therefore summarise that ideological polarisation on the subnational level does indeed have a substantial impact on non-electoral participation, but not on the probability of voting.

Next, I examine whether ideological extremism makes individuals more susceptible to environmental polarisation on ideology, and whether polarisation shows conditional effects on the individual-level relationship between ideological extremism and political participation. Since interesting differences may exist between ideological groups, I employ cross-level interactive terms by which left-wing extremism and right-wing extremism are considered separately, and I rerun the regression models. Tables 4 and 5 show the results obtained with these modified models.

Table 4 Regional ideological polarisation, extremism and voting
Full size table
Table 5 Regional ideological polarisation, extremism and non-electoral participation.
Full size table

With regard to polarisation in ideology, I find no robust interaction effect between polarisation and extremism. The rightest-by-polarisation interaction does not meet statistical significance (M08). In contrast, the leftist-by-polarisation interaction term shows a positive and statistically significant effect after controlling for left-wing extremism, polarisation and individual-level features (M07), which holds true when using an index with the three most extreme positions on the left–right-scale.

The plotted effects of polarisation in terms of ideology in Fig. 3 show that neither the extreme leftists, nor the rightists change their voting behaviour as ideological polarisation increases. Moreover, the probability of voting for non-leftists and non-rightists decreases as regional polarisation enhances.

Fig. 3
figure 3

Source: European Social Survey, 2002–2014 (www.europeansocialsurvey.org/. Accessed 15 Mar 2016)

Substantive effects of ideological polarisation at the NUTS2-level and individual political participation. Note The graphs depict the marginal effects for voting. The circles and diamonds are the point estimates, and the grey area reflects the 95% confidence intervals associated with the probability estimates.

Full size image

Finally, the results of Table 5 show that left-wing extremists (M09) and right-wing extremists (M10) have a higher probability of participating in a non-electoral form than non-leftists and non-rightists, respectively, do.

The leftist-by-polarisation interaction shows a negative and a highly significant effect (M09), and the corresponding marginal effects plotted on the left side of Fig. 4 show that again not the left-wing extremists, but the non-leftists are mobilised by increasing ideological polarisation. In contrast, the rightist-by-polarisation interaction term shows a positive and statistically significant effect after controlling for right-wing extremism, polarisation and individual-level features (M10), and the marginal effects plotted on the right side in Fig. 4 show that both the extreme rightists and the non-rightists are substantially mobilised by polarisation, but the impact on the rightists is much stronger.

Fig. 4
figure 4

Source: European Social Survey, 2002–2014 (www.europeansocialsurvey.org/. Accessed 15 Mar 2016)

Substantive effects of ideological polarisation at the NUTS2-level and individual political participation. Note The graphs depict the marginal effects for non-electoral participation. The circles and diamonds are the point estimates, and the grey area reflects the 95% confidence intervals associated with the probability estimates.

Full size image

In sum, firstly, the impression that ideological polarisation at the regional level does not mobilise for voting, but indeed does mobilise for non-electoral participation is confirmed. Secondly, we learn that especially right-wing extremists are mobilised by ideological polarisation, while left-wing extremists are not. These findings are confirmed if we use the three outermost categories of the left–right dimension to measure polarisation.

Discussion

The present article concentrates on political culture as a contextual factor of political participation. More specifically, it addresses the question of whether ideological polarisation of the social environment influences individuals’ decision to go to the polls or participate in non-electoral form. My argument comprises two parts. In the first part, I argue that mass polarisation with regard to central concepts puts citizens on the defensive with regard to their values and beliefs. In an environment of polarisation, people become insecure about whether their norms, values, and standards will be superseded by other values and standards accompanied by corresponding policies in the future. Hence, they experience deprivation which in turn energises them for political action. I expected the level of contextual polarisation to be positively associated with citizens’ probability of participating, and I conducted macro-regressions as well as two-level regressions to examine whether regional polarisation with regard to the left–right dimension has an impact on political behaviour. My empirical results show that while the average citizen is not motivated to vote over ideological polarisation, s/he is motivated to become active in a non-electoral way. In the second step, I specified my considerations by assuming that the mobilising effect of polarisation particularly affects individuals who are emotionally or cognitively fairly involved in politics and thus more attentive to developments in their sociopolitical environment. My findings show that while left-wing extremists are not really mobilised by ideological polarisation, right-wing extremists are likely to get active in a non-electoral way.

What are the implications of these findings? First, our findings provide a step towards understanding how polarisation can affect democratic quality. The democratic ideal of equal responsiveness to the interests of the citizenship requires equality of the voices of citizens in politics. Although public officials have different ways of learning about what citizens want and need, systematic studies have shown that what policymakers hear from citizens influences what they do (Verba et al. 1995: 526). Individuals who express their preferences in an active and perhaps louder way have a better chance of being recognised by the political system (Oliver 2001: 19). This means, unequal participation bears the potential for representational imbalance. Why? It is well known that participation is fostered by resources, generalised political engagements, and recruitment (Verba et al. 1995: 513). Likewise, citizens’ ideological preferences have been shown to be an important determinant for political action (van der Meer et al. 2009). This study reveals that ideological polarisation is a source of participatory inequality as well. Right-wing extremists differ from members of the public in terms of the extent of their activity. The issues that motivate their participation may deviate from the priorities of those who are less active, which may in turn jeopardise equal protection of interests. If right-wing extremists are more likely than moderates or left-wing extremists to become active, their voice is likely to be heard more strongly and more often by policy makers. In this manner, ideological polarisation can stimulate participative inequality and lead to inequality in policy output.

In a more general way, this may be also true for the mode of participation. If certain social groups are inclined to go to the polls to express their needs and preferences, while others prefer more direct and less limited options, ideological polarisation can stimulate participative inequality and lead to inequality in policy output. Research hints to the direction that those who can, want to, or have been mobilised to be active have the capacity not only for voting, but for the participatory repertoire as a whole (see, e.g. Novy 2014). The fraction of the population engaged in non-electoral forms of participation is rather small and typically younger and not representative of the population at large (Finkel et al. 1989; McCaffrie and Marsh 2013).

The second contribution of this study is a strengthening of political culture as an essential context feature for political participation. Political culture is defined as ‘[…] the particular distribution of patterns of orientation’ (Almond and Verba 1963: 14), and this paper shows that a bimodal distribution of orientations has to be taken into account when explaining political behaviour.

I certainly recognise the limits of the study. The main limitation relates to my use of cross-sectional data. Mass polarisation is fundamentally a question of dynamics over time, so panel data are needed to understand its causes and effects. The correlational nature of my study does not enable identification of exactly which causal direction underlies my findings. Polarisation may lead to action, but the opposite is also possible, for example, political activism may bring activists closer together but alienate friends and family (Jost et al. 2017). Likewise, the selection and evaluation of information may take place in an attitude-reinforcing fashion which makes the development of even more extreme viewpoints more likely (Leeper 2014: 30). I expect polarisation and action to provide mutual reinforcement, but a need for a spark to ignite motivation and polarisation is evident, and polarisation seems plausible. Nevertheless, further studies are needed to disentangle this puzzle; accordingly, my analysis is one step towards future research.