Introduction

The overtime relative decline in the explanatory value of the socio-structural model of far-right vote has prompted scholars to examine policy and issue-related explanations of citizens’ voting decisions (e.g., van der Brug et al. 2005; Ivarsflaten 2008; Stockemer et al. 2018). The policy model of vote assumes that voters are rational consumers of policy platforms (Downs, 1957; Adams et al., 2005) and suggests that citizens tend to cast their vote for the far-right based on policy considerations, primarily expressed through the issue of immigration (Ivarsflaten 2008; Rydgren 2008; Stockemer et al. 2018; Halikiopoulou and Vlandas 2020). Far-right parties build their public profile around emphasizing extreme positions on the issue of immigration (Wagner 2012), which is consistent with their core ideology that seeks to preserve the unity, autonomy, and identity of the nation (Halikiopoulou and Vasilopoulou 2018). Voters spontaneously connect far-right parties to this policy issue and those who are motivated by immigration-related concerns opt for the far right in the polls on this basis (Ivarsflaten 2008; Stockemer et al. 2018).

Focusing on immigration helps the far right galvanize support among their core anti-immigrant constituency (Halikiopoulou and Vlandas 2020). Yet, electoral competition presents a dilemma to these parties: if they moderate their immigration policy, they are likely to be punished by their traditional anti-immigrant electorate. If they continue their focus on extreme anti-immigrant positions, they satisfy their core voting base, but risk alienating potential new voters. Nonetheless, the far-right party family has steadily improved its vote share in the past decades across Western Europe.

We argue that to understand the far right’s electoral fortunes, we need to consider the advantages and disadvantages these parties encounter in the multidimensional political issue space. We draw from theories of electoral competition and argue that changes in the public salience of policy issues offers far-right parties a comparative advantage which is more likely to be electorally beneficial compared to far-right parties’ positional changes. Yet, changing the salience of which issue is the most electorally beneficial for the far right? We furthermore posit that far-right parties enjoy an electoral advantage that is specific to the issue of European integration. The European Union (EU) has become a source of social division, creating a vast untapped electoral potential among the pool of Eurosceptic voters. Whereas immigration has already become politicized in West European party systems, the fact that far-right parties do not face serious competition over their position on EU integration creates strong expected electoral gains when the issue is salient among the voters.

Utilizing similar survey questions from the Chapel Hill Expert Survey (CHES) about parties’ positions (Polk et al. 2017; Bakker et al., 2020b) and the European Election Studies (EES) about voters’ policy preferences (Schmitt et al. 2016; 2019), we estimate multidimensional voting models in 12 West European countries. Then, we estimate the electoral consequences of changes in issues’ saliencies and parties’ policy shifts using simulation modeling techniques (Adams and Merrill 2000; 2005, Adams et al., 2005, Alvarez and Nagler 1995; 1998, Calvo and Hellwig 2011; Zur 2021a; b).

Our contribution is three-fold. First, we put forward an integrated theoretical model of far-right party success in the European multidimensional space. We do so through a rigorous research design, simultaneously testing for public issue salience and parties’ issue positions while also considering left-right ideology, and specific issue dimensions, including redistribution, immigration, and the EU. In doing so, we hope to set a novel research and methodological agenda in the fields of far-right parties and electoral competition in European politics.

Second, we examine parties’ potential payoffs given the distribution of policy preferences across the electorate and the policy positions of other parties. We then contrast these potential payoffs with the electoral effects of changes in the public salience of multiple issues, while holding parties’ positions and voters’ preferences constant. We demonstrate that by focusing on positions, the far-right policy model of vote overlooks public issue salience as a crucial contributor to far-right voting (e.g., van der Brug et al. 2005; Ivarsflaten 2008).

Third, we examine the far right’s expected vote share through the prism of opportunities and constraints posed by the multidimensional political issue space. Challenging the dominant view in the literature about the role of immigration in far-right party success (e.g., Stockemer et al. 2018), we posit that there is little room for vote gains on immigration given the far right’s extreme position relative to their more popular anti-EU position. The magnitude of the far right’s expected electoral gains due to increased public salience of European integration is larger than the expected electoral gains due to increased public salience of immigration. By demonstrating that EU integration and immigration are two distinct dimensions, we contend that future work on the electoral fortunes of this party family should integrate the EU issue as a core driver that is separate and separable from immigration; and that can account for the far right’s ability to extend its vote beyond its core electoral base.

Policy Positions, Public Issue Salience and the Far Right

Electoral competition consists of three core components: issue conflict, issue salience and parties’ valence image (e.g. Adams et al., 2005; Meguid, 2008; Basu 2020). The classical spatial theory postulates that parties strategically position themselves to maximize their electoral potential and voters support the party that announces the closest policy to their preferences (Downs, 1957). Issue salience theories assume that parties attempt to mobilize the electorate by emphasizing some issues more than others, and the weight voters attach to a given policy affects their decision at the ballot box (e.g. Budge & Farlie 1983; Dennison 2019). Valence models assume that electoral choice is based on party non-policy attributes, such as expertise, competence and integrity, that tend to be unanimously approved by the electorate (Stokes 1963).

How does this framework of competition apply to the far right? Voters tend to view most far-right parties as low-valence parties. A vast majority of them have not held office in Western Europe, and -even when they have- they tend to portray themselves as outsiders. This suggests that far-right parties are unlikely to win votes due to their non-policy characteristics. Far-right parties tend to be known, instead, for offering distinctive policy positions in a small range of issues; specifically, immigration (Basu 2020). This ‘product differentiation’ or ideological distinctiveness (Wagner 2012) tends to be rewarded in the polls. Ideological moderation, on the other hand, is likely to place far-right parties at a disadvantage vis-à-vis high-valence mainstream parties (Adams 2012; Adams, Merrill, and Zur 2020). Whereas moderation tends to positively affect citizens’ perceived party competence (Johns and Kölln 2020), when it comes to the far right, it is likely to be electorally costly. The core far-right electorate might interpret moderation as a betrayal, which might weaken their attachment. At the same time, moderate voters might consider it insincere, ‘cheap talk’ campaign promises (Adams 2012). In fact, when niche parties, including far-right parties, moderate their policy positions, they do not benefit electorally as much as mainstream parties (Ezrow, 2010). In short, extreme policy positions satisfy the expectations of far-right party activists and core supporters; while at the same time serve to foster an image of programmatic divergence from mainstream parties (Basu 2020).

If, therefore, positions are ‘sticky’ for far-right parties, how can we explain their broader support base and appeal to voters beyond their core voting group (see also Stockemer et al. 2021)? We posit that far-right parties benefit from changes in public issue salience.Footnote 1 Parties improve their electoral results by emphasizing the issues they ‘own’ (Dennison 2019: 441). Issue ownership can be electorally advantageous, especially when the issue is considered important at election time (Bélanger and Meguid 2008). Yet, issue ownership affects the voting decisions of ‘only those individuals who think that the issue is salient’ (Bélanger and Meguid 2008: 447). Whereas policy positions among the electorate tend to be stable over time, issue priorities vary across voters and across elections. If a given issue is salient, then voters will opt for the party that they consider the most credible proponent of it. This is because public issue salience tends to activate ‘latent and fairly stable attitudinal predispositions’ (Dennison 2020: 403). We hence hypothesize:

H1: Changes in the public salience of policy issues can increase the vote share of far-right parties more than changes in far-right parties’ policy positions.

Which Issues are more Electorally Beneficial for the Far Right?

We next study the opportunities and constraints posed by the multidimensional political issue space. Although, historically, political contestation in Western European politics centered around the left-right dimension, scholars increasingly agree that competition takes place in a multidimensional political issue space. In fact, the salience of non-distributive issues, such as immigration and the EU, has increased, disrupting European party systems (e.g. Hooghe and Marks 2018). These new dimensions of contestation tend to be distinctive from and cut across traditional ideologies (Bakker et al., 2020a).

The far right is conventionally considered to have ownership of the immigration issue (Ivarsflaten 2008; Rydgren 2008).Footnote 2 Immigration taps into the debate about entitlement to national membership, which is directly linked to the core tenet of far-right ideology, i.e., nationalism (Halikiopoulou and Vasilopoulou 2018). Immigrants are linked to a variety of socio-economic problems ranging from unemployment and crime to demographic change, posing a threat to the ‘national way of life’. There is plenty of evidence to suggest that individuals’ grievances over alleged out-groups constitute a powerful predictor of far-right vote (e.g., Stockemer 2016; Halikiopoulou and Vlandas 2020). However, there is a growing consensus that far-right parties are not single-issue parties (Immerzeel et al. 2016; Bergman and Flatt 2020). In fact, Stockemer et al. (2021) show that there are multiple routes to far-right voting, including a non-immigration route, as approximately a third of the far-right electorate have no immigration concerns.

We posit that European integration –much less discussed in the literature– is an additional issue that can benefit the far right’s electoral prospects. For years, the EU dimension was considered a ‘sleeping giant’, that if awoken, would transform European politics (van der Eijk & Franklin 2004). Over time, the EU has started to occupy a salient place in domestic European politics. Due to consecutive crises in recent European politics, the EU has been high on the agenda as a potential driver of or obstacle to policy solutions. Public contestation of the EU has increased with more actors -political or not- being involved in EU-related debates and expressing polarizing views. Citizens are increasingly becoming aware of and assign responsibility to the EU (Hobolt & Tilley, 2014). Public Euroscepticism has increased, with a significant drop in trust in the EU across its member states (Ejrnæs and Jensen 2019; Talving and Vasilopoulou 2021).

Despite the changing political landscape, mainstream parties have not caught up. Whereas the far-right immigration agenda has been mainstreamed, albeit with varying electoral success (Spoon and Klüver 2020), on the EU issue, mainstream parties have exhibited programmatic inflexibility (Hooghe and Marks 2018). The combination of reputational constraints and internal dissent suggests that they have refrained from politicizing the EU issue (Rohrschneider and Whitefield 2016; Hooghe and Marks 2018). Manifesto analyses show that mainstream parties attach much less importance to the EU issue compared to both the far-right and far-left party family; and that over-time changes in their EU issue salience have been minor (Braun et al. 2019).

The far right, on the other hand, has traditionally been one of the main opponents of European unification. Filling a vacuum on the supply-side of political competition, far-right parties have ‘rocked the boat’ by emphasizing their anti-EU credentials. Although there is variation within the party family, over time far-right parties have hardened their stance. The salience of the EU in their agendas has also increased over time, with some far-right parties emphasizing the EU more than the GAL-TAN dimension, which primarily relates to cultural issues (Vasilopoulou, 2018b).Footnote 3 Empirical analyses show that far-right parties can benefit electorally from the EU issue, above all in contexts of high EU polarization (Down and Han 2021). Eurosceptic voters tend to defect from mainstream to far-right parties (e.g. Hobolt and Tilley 2016), especially when mainstream parties advocate similar positions to each other and are further away from the positions of the far right (Spoon and Klüver 2019).

It is worth noting that the EU issue can act as a catalyst of key voter concerns across many other issues that are present across different electoral constituencies. It taps into questions related to culture and immigration, but it is equally related to the economy as well as the issues of sovereignty, democracy, and solidarity (Vasilopoulou, 2018b). Yet, we emphasize here, and show below, that both for voters’ preferences and parties’ policy positions the EU is statistically distinctive from the issues of immigration, as well as left-right ideology and economic redistribution (Bakker et al., 2020a, Zur 2021b). The data used in this article show a very low correlation (0.23 in 2014) between citizens’ positions on these two dimensions (Table 1). Therefore, we believe that immigration and EU integration should be analysed as separate policy dimensions.

These differences between European integration and all other issues are crucial to the electoral prospects of far-right parties. Whereas immigration has already become politicized in West European party systems, the fact that far-right parties do not face serious competition over their position on EU integration creates strong expected electoral gains when the issue is salient among the voters. We thus hypothesize:

H2: Changes in the public salience of European integration are more electorally beneficial for the far right compared to changes in the salience of other policy issues, including immigration.

Data and Research Design

To test our expectations, we combine two data sources. We use the Chapel Hill Expert Survey (CHES) for data about parties’ positions (Polk et al. 2017; Bakker et al., 2020b) and the European Election Studies (EES) for data about voters (Schmitt et al. 2016; 2019). These two sources provide a comparable set of positional issue placements for both parties and voters, allowing us to construct a multidimensional space of party competition in both the 2014 and 2019 EP elections. We construct a four-dimensional space of parties’ and voters’ positions, which includes the overarching left-right continuum and three additional issue dimensions, economic redistribution, immigration, and EU integration. As shown by Bakker et al., (2020a) these are individual and independent issue dimensions in West European politics. We also emphasize this point by providing respondents’ self-placement correlation matrix in Table 1. None of these correlations is high enough to be considered problematic for the analyses we present below.

Table 1 Respondents’ self-placement correlations on four dimensions

We provide empirical support for our theoretical arguments using EP election data because the EES is the only cross-national survey that asks voters to report their positions on both immigration and EU integration. This is crucial because our second hypothesis states that changes in the salience of EU integration are likely to be the most electorally beneficial for far-right parties, especially relative to the well-documented importance of immigration. We note here that national election surveys, including the German Longitudinal Election Study, the British Election Study, and the Dutch LISS panel, do not include all components of our research design. We also believe that the EP electoral context is a good starting point for examining the patterns we are interested in, not least because the ESS offers cross-country variation at two time-points which allows strong generalizability of our results. In addition, the literature suggests that EU issue voting takes place both in national and EU elections (e.g., De Vries 2010) and there are similarities in party strategies in both arenas (Zur 2021b). We also acknowledge the possibility that our theoretical arguments are more likely to be supported in the context of EP elections. Therefore, in the appendix we replicate all our analyses using the reported national election vote choice and the 'probability to vote’ questions. Our results are robust and consistent for all three types of vote choice models. We focus our discussion below on the reported EP vote choice due to the methodological concerns explained in the appendix.

We examine voting models in 12 West European countries in which there is at least one far-right party.Footnote 4 We employ the term far right to define the group of parties -both radical and extreme- that put forward nationalist solutions to all socio-economic problems (Vasilopoulou & Halikiopoulou, 2015; Stockemer et al. 2021). All other parties are grouped by their size and left-right position. We analyze each party’s potential electoral rewards independently from other parties, but for convenience, we present parties grouped as: far left, greens, major left, centrist, major right, and our group of interest - the far right. In Germany, for example, Die Linke is defined as the far-left party, the Greens are in the green group, the Social Democratic Party (SPD) is the major left party, the Free Democratic Party (FDP) is the centrist party, the Christian Democratic Union/Christian Social Union (CDU/CSU) is the major right party, and our party of interest is the AfD on the far right.

In our voting models, the dependent variable is the respondent’s recalled vote choice. The independent variables - voter-party proximity on four dimensions - are measured as the linear absolute difference between the respondent’s self-placement and the experts’ mean placements of each party on each issue. Both voters and experts’ positions are measured on a 0–10 scale, where 0 indicates extreme left on the overarching left-right dimension, support for redistribution, less restrictive immigration policy, and support for European integration (see appendix for question wording, control variables and alternative model specifications).

Figure 1 shows parties’ positions (based on CHES) and the position of the median voter (based on EES) on each dimension. The gray and black symbols mark the 2014 and 2019 EP elections, respectively. This figure provides important information about party competition, pointing to temporal differences, and both within- and between-country variation. It shows that far-right parties take extreme positions on all dimensions except redistribution. On immigration, all far-right parties are perceived as extreme (over 8.5 on 0–10 scale), while on EU integration some of these parties are less extreme. The figure also indicates the distance of these parties from the median voter, where on average far-right parties are closer to the median voter on the EU dimension than on immigration. Additionally, Fig. 1 shows that the main competition of the far right on immigration are the major right-wing parties (usually high-valence parties). On EU integration, only far-left parties (low-valence parties) compete with the far right for the Eurosceptic votes. Thus, the ideological space creates larger vote-recruiting potential when EU integration salience is high.

Far-right parties also tend to take more extreme positions on immigration than on EU integration. On immigration, far-right parties consistently position themselves at the extreme end of the axis. Yet, on the EU they are comparatively less extreme and more diverse (see also Vasilopoulou 2018b). Far-right parties’ extreme positions, especially on immigration, limit their positional strategies. Considering that there is limited scope for becoming more extreme, moderation is their only feasible positional change. Yet, if they moderate their positions, far-right parties are not guaranteed (or even expected) to win the votes of moderate voters (Ezrow, 2010; Basu 2020). Importantly for our theoretical argument, the EU issue is unique in the sense that most far-right parties are the only parties to the right of the median voter. They do not encounter serious competition for Eurosceptic votes and when they do, competition comes from the weaker, low-valence, parties of the far left (Wagner 2022).

Fig. 1
figure 1

Party Positions on Four Dimensions. Note: Party positions (Chapel Hill Expert Survey) and the median voter’s positions (European Election Survey). The 2014 and 2019 elections are marked in gray and black, respectively. See appendix for parties’ names and groups.

In addition to voter-party proximity variables, we use party-specific coefficients to proxy voters’ evaluation of parties’ non-policy characteristics (i.e., valence image). These coefficients measure the (dis)advantage parties receive from their valence image, independently from their policy positions. The party-specific coefficients are equivalent to a series of dummy variables, where 1 represents the non-policy characteristics of the focal party and 0 otherwise. We use party-specific coefficients as control measures for the unobserved valence image of the parties. Thus, when the coefficients are (negative) positive the probability of voting for a focal party (decreases) increases. Including these controls in our voting models allows us to reduce possible biases in our proximity variables caused by party characteristics such as competence, integrity, party unity, etc. (see methodological discussion in Adams et al., 2005, appendix 2).

Using these data, we estimate country-specific conditional logit models of citizens’ voting decision rule in both the 2014 and 2019 EP elections. The conditional logit framework allows us to estimate two important features. First, our baseline model estimates the expected vote share of each party in each election. Second, it estimates the voters’ salience of each dimension. We interoperate the proximity coefficients similarly to the interpretation in previous works: as the size of the coefficients on each proximity variable increases (in negative terms) the importance of the corresponding dimension increases. This is because the coefficients measure the effect of voter-party proximity on vote choice (e.g., Adams and Merrill 2000; 2005, Adams et al., 2005). The conditional logit framework is especially relevant to our research, as we are interested in the probability of voting for each party conditioned on the probability of voting for other parties. In this framework, each respondent represents a number of observations equal to the number of parties in the system.

Building on these models, we use mathematical simulations to test our hypotheses. We follow the approach in works by Adams and Merrill (2000; 2005), Adams et al., (2005), Alvarez and Nagler (e.g., 1995; 1998), Calvo and Hellwig (2011), and Zur (2021a; b), among others. These simulations represent voters’ decision rules as probabilistic functions of voter-party proximity in the four-dimensional space. Then, we calculate the probability of each respondent voting for each party and compute changes to each party’s expected vote share under counterfactual conditions of interest. We simulate changes to parties’ vote share when: (1) parties alter their policy positions in the four-dimensional space, and (2) there is change in public salience of these four dimensions. In both counterfactuals, we simulate changes in the expected vote share when only one variable of interest changes and everything else related to the voters’ decision rule and parties’ strategic behavior remains unchanged. We further discuss these simulations below.

Results

We divide our results section into three parts. First, we discuss the 4-dimensional Downsian-proximity models. These models give us information about voters’ decision rules in the 2014 and 2019 EP elections and serve as the baseline for our simulations. From these models, we learn the country- and time-variation in the relative importance of the four policy dimensions and the parties’ positions on those dimensions. To better grasp our methodological approach, the second part of the results presents an example of our two sets of simulations. To do so, we take a close look at the 2019 EP elections in Germany as a clear case where a far-right party can gain from increased salience of the EU issue. The last part of our results section discusses the aggregation of the simulated changes in parties’ vote-shares and demonstrates that far-right parties can gain more votes through changes in the salience of policy issues rather than changes in their policy positions, and especially related to the EU issue.

Vote Choice Models in the 2014 and 2019 European Parliament Elections

We estimate country-election (Downsian) proximity models where the probability of voting for party j is conditional on voting for all other parties. These models establish the relative importance of parties’ positions and issue salience on the four dimensions. By estimating these models, we observe the relationship between parties’ positional strategies, the salience of different issues, and citizens’ decision rules when voting in EP elections.Footnote 5 The results of the models later become the baseline for our simulations. Moreover, the models are important to our argument because they show the independent (and relative) importance of each of the four dimensions, thus allowing us to differentiate between competition over the EU and immigration issues.

In Fig. 2, the conditional logit coefficients estimate the size of the punishment voters would inflict on parties for taking positions that are further away from their preference on each dimension. These coefficients imply that as the number gets closer to zero voters care less about the issue (i.e., smaller punishment). Because all four proximity variables are estimated simultaneously in the same model, the salience of each dimension can be interpreted as independent from the salience of the other dimensions. In addition, because these variables are measured using a 0–10 scale and are estimated simultaneously, they are comparable. The substantive meaning of these models is that when a party takes a popular position (i.e., close to a large section of the electorate and distinguishable from the positions of other parties) on an issue, high salience (large negative coefficient) of this issue is associated with a higher probability of voting for that party. Knowing from Fig. 1 that far-right parties tend to hold distinctive positions on the EU issue, they are expected to gain votes when this issue’s salience is high(er). Additionally, when an issue is salient, parties have a greater incentive to change their policy position and avoid the large punishment associated with incongruence (Basu 2020; Zur 2021b).

In the models below, all four dimensions show a statistically significant relationship with the probability of voting in all 12 countries in both the 2014 and 2019 EP elections. The coefficients presented in Fig. 2 can be compared within each cell to show how the salience of each issue changed in each country between 2014 and 2019. For example, the top-left cell shows that, in Austria, the left-right dimension was more salient in 2019 than in 2014, but these differences were statistically insignificant (because the confidence intervals overlap). The coefficients can also be compared across panels to show the variation in the relative importance of the issues. For example, the first row of coefficients shows that, in Austria, the left-right dimension was the most salient, followed by the issue of EU integration. The other two issues -immigration and redistribution- were less salient; yet had an independent and significant relationship with vote choice in the 2019 EP elections.

In all countries, the left-right dimension is the most salient, but the salience of the dimensions varies across countries and time. For example, in 2014 in the United Kingdom, there are statistically indistinguishable differences between the salience of the four dimensions, but EU integration is slightly more salient than the other three dimensions. This means that, in the 2014 EP elections, British citizens considered parties’ positions on all four dimensions to a similar extent when casting their vote. In Sweden 2014, on the other hand, the left-right dimension is significantly more salient than the other dimensions (both in statistical and substantive terms) and more salient than the left-right dimension in 2019. This means that Swedish voters considered parties’ positions on the overarching left-right dimension significantly more than they considered parties’ positions on each of the specific issues when casting their vote in EP elections. Lastly, in most countries, when controlling for the overarching left-right dimension and the immigration and EU issues, economic redistribution is the least salient issue.

The results presented in Fig. 2 confirm our knowledge about the importance of the EU integration and immigration issues in EP elections. Voters, as expected, punish parties for taking policy positions that dissent from their preferred positions. Based on these vote choice models, our main interest is to estimate whether far-right parties can gain more votes through changes in their policy positions or through changes in the public salience of policy issues.

Fig. 2
figure 2

Vote Choice Models and Public Issue Salience. Note: This Figure shows the coefficients and 95% confidence intervals from the conditional logit models described in the text. The dependent variable is the reported vote choice, and the proximity variables are linear (absolute) terms. Party-specific coefficients and model fit information are shown in the appendix. In the appendix, we analyze alternative model specifications that include respondents’ socio-demographics, exponential loss functions for the proximity variables, and directional voting variables.

Party Positional Shifts and Changes in Public Issue Salience– Simulated Results

To test our hypotheses, we simulate changes in parties’ expected vote shares under two counterfactual scenarios. We estimate parties’ positional changes in each of the four dimensions by up to one unit (in the direction that improves their expected vote share) on the 0–10 scale. In this scenario, parties are likely to increase their vote share by announcing a new set of policies in the four-dimensional space and can do so without backlash from the electorate (Feddersen, 2022), and changes to their valence image (Johns and Kölln 2020). Put simply, we keep everything but the positions of a focal party constant, while calculating the changes in the focal party’s expected vote share.

We limit parties’ positional changes to one unit on each dimension because we believe this is a realistic limit to parties’ policy shifts. Moreover, parties rarely sharply change their positions from one election to the next, and even when they try to communicate sharp changes, voters rarely update their perception of parties’ positions (Adams 2012). Below, we present the results pooled across all parties in both elections, but each party’s gains from the simulated scenarios are given in the appendix.

Simulations Example - Germany

For a better understanding of the simulation method, we first discuss an example of the counterfactual changes in parties’ expected vote share from the 2019 EP elections in Germany. Testing H1, Fig. 3A demonstrates how the expected vote share of the far-right party in Germany (AfD) changes as a function of its policy positions on the four dimensions. The x-axis in Fig. 3A represents the simulated position of the party on each dimension where 0 represents the observed position (as coded by CHES), -1 represents a leftward move by 1 unit, and 1 represents a one-unit move to the right. From these simulations, we learn that moving a whole unit to the right on the left-right dimension is associated with a significant decrease in the expected vote share of the AfD, but a very minor decrease if it moves to the right on any of the three issues dimensions. On the other hand, by moderating their left-right position the AfD can gain votes, but as we will show below, the AfD’s expected electoral gains are larger when the salience of the issue decreases.

Figure 3A shows the potential gains (losses) from positional shifts on each dimension independently Our multidimensional simulation results (full data shown in the appendix) show that the AfD could have increased its vote share by 5.7% by moderating its policy positions by 1 unit on all four dimensions. Note, however, that parties’ optimal strategy is not necessarily to moderate on all dimensions, nor that parties must change their position by a whole unit. Our simulations suggest that the CDU/CSU for example would be better off changing its left-right position from 5.86 to 5.10 (allowing it to gain an additional 1.56% of the vote) than by a whole unit to 4.86 (which will increase its vote-share only by 1.43%). Similarly, it would be better off taking a position slightly to the right of the center on immigration than to take the pure center. These results are concurrent with recent experimental (Johns and Kölln 2020) and observational (Zur 2021a; b) evidence that voters prefer moderate-left or moderate-right parties, but not purely centrist parties.

Fig. 3A
figure 3

Positional Shifts Simulations Example – The German AfD in the 2019 EP Elections. Note: Simulated changes to the expected vote-share of the AfD (y-axis) as a function of the AfD issue positions on the x-axis, where 0 represents the position of AfD on the focal issue.

In this second set of simulations (H2) we compare parties’ potential electoral gains due to altering their policy position with the potential changes of parties’ expected vote share through changes in the salience of each dimension. We do not change parties’ policy positions, nor do we change voters’ decision rules. The counterfactual simulated in this scenario intends to explain the electoral consequences parties face due to changes in the public salience of different policy dimensions, without changes to their perceived policy positions (Meyer and Wagner 2020). We limit public issue salience changes to a realistic range – from half (decrease to 50%) to double (increase to 200%) of the observed salience (100%) in the models above. This range only slightly exceeds the range observed in our dataset (34-188%) between 2014 and 2019.

Continuing with the German 2019 example, the coefficients on the left-right, redistribution, immigration, and EU integration proximity variables are − 0.48, -0.12, -0.07, and − 0.28, respectively. In these coefficients, a larger absolute value implies a lower probability of voting for a party as the distance between voters’ self-placement and parties’ positions increases. Therefore, these values are representative of the observed salience of each dimension. Then, we recalculate parties’ expected vote share when the public salience of the left-right dimension equals − 0.24 (half the observed value), then − 0.25, -0.26, and so on until it reaches − 0.96 (double the observed value). We then recalculate parties’ expected vote share when the public salience of the three other dimensions changes in a similar manner.

Figure 3B shows an example of our second set of simulations at the party level. Doubling the salience of the left-right dimension is associated with a substantial drop in the AfD’s expected vote share; while doubling the salience of the EU integration dimension is associated with a considerable increase. We then check which set of issue saliencies maximizes parties’ electoral returns. In this counterfactual scenario, if the salience of the issues changed (high EU integration salience and low salience for all other dimensions), the AfD could have increased its vote share by 9.1%.

Fig. 3B
figure 4

Changes in Public Issue Salience Simulations Example – The German AfD in the 2019 EP Elections. Note: Simulated changes to the expected vote-share of the AfD (on the y-axis) as a function of the public salience of the issue on the x-axis, where 100 represents the observed salience of the focal issue calculated from proximity model presented in Fig. 2.

Aggregated Simulations Results

The German example represents the overall trend in our data. To assess H1, Fig. 4 shows the electoral consequences for 6 groups of parties due to changes in the public salience of issue dimensions (in black) or due to shifting parties’ policy positions (in gray). While, overall, parties can make comparable gains in both scenarios,Footnote 6 for far-right parties there are statistically significant differences between the consequences of these scenarios. Specifically, across 29 party-election data points, far-right parties can gain (on average) an additional 4.67% of the vote due to changes in public issue salience, but only 3.34% by changing their own policy positions. On average, the electoral outcomes of far-right parties improve more than any other group of parties when the public salience of different issues changes. These differences imply that when choosing between campaign strategies, far-right parties face an easier choice than other parties. Changes in public issue salience are more beneficial for the far right than positional changes, which provides support for H1. Moreover, when comparing the expected electoral gains of far-right and major-right parties from moderating their policy positions by up to 1 unit on each of the scales (gray boxes), major-right parties are expected to win additional 4.26, but far-right parties are expected to win only 3.34%.

Fig. 4
figure 5

Electoral gains from hypothetical changes in parties’ positions and public issue salience. Note: Boxes show 25th to 75th percentiles of parties’ expected vote share increase under two counterfactual scenarios. Light-gray boxes show vote share increase for four-dimensional positional (as coded by the CHES) change within a one-unit interval. Black boxes show vote share increases for four-dimensional changes in public issue salience within half and twice of the observed salience interval.

Following these results, we ask: which issues are more electorally beneficial for the far right? To answer this question, we look at the changes in parties’ expected vote share when one dimension is twice as salient (relative to the estimated value shown in Fig. 2), while the salience of the other three dimensions remains in their observed value. These results emphasize the increasing importance of anti-EU rhetoric by far-right parties, especially relative to their anti-immigration rhetoric.

Figure 5 demonstrates changes in parties’ expected vote share when the salience of a focal dimension doubles (for simplicity we focus on the major parties on the left and the right, compared to far-right parties the results for all other parties are shown in the appendix). The estimations in this analysis are drawn from the second simulation described above, where we computed changes in parties’ expected vote share due to changes in each issue salience independently while holding everything else constant. Note that these vote-share changes are rather small relative to the increases in Fig. 4. This is because here we present changes due to public salience increase of each individual dimension, while in Fig. 4 we present the sum of all dimensions. Moreover, in Fig. 4 we present the ‘best case scenario’ of saliencies for each party, while in Fig. 5 we analyze public salience increase to 200% (doubled). As a result, parties’ vote share can decrease in the results presented in Fig. 5, but not in Fig. 4.

The results shown in Fig. 5 are consistent with H2 in two ways. First, doubling the salience of EU integration is associated with larger vote shares for far-right parties significantly more than doubling the salience of any other dimension. On average, doubling the public salience of EU integration increases far-right parties’ vote share by over 1.91% points. Doubling the public salience of immigration increases the average vote share of far-right parties only by 0.34%, while doubling the public salience of redistribution or the general left-right dimension decreases the vote share of far-right parties.

A second way the results in Fig. 5 support our argument is that doubling the public salience of EU integration increases far-right parties’ vote share at the expense of pro-EU parties. Both major-left and major-right parties are expected to lose votes when the salience of EU integration increases. These results have important implications for parties’ electoral strategies with regards to both position-taking and issue emphasis. Because previous research has demonstrated that center-right parties’ move towards the far right on the issue of EU integration increases its salience and decreases center-right parties’ vote share (Meijers and Williams 2020), we can conclude that both shifting right and emphasizing EU integration is not a recommended strategy for mainstream parties.

Fig. 5
figure 6

Changes in parties’ expected vote share when public issue salience is doubled. Note : Boxes show 25th to 75th percentiles of parties’ expected vote share change when the public salience of the focal dimensions is doubled.

Discussion

In this article, we have presented and tested a theoretical framework of how changes in parties’ positions and citizens’ issue salience influence the far right’s electoral success, taking into consideration the opportunities and constraints posed by the multidimensional political issue space. First, we demonstrate that changes in public issue salience are more likely to increase the far right’s electoral potential compared to the far right’s policy shifts. Second, we posit that not all issues are equally important in this respect. Because far-right parties hold a competitive advantage on the issue of European integration, they have higher expected electoral gains when this issue is discussed in electoral campaigns and becomes important in the eyes of the voters.

Our findings are significant in the context of the rising electoral success of the far right. They have important implications regarding how we think of and conceptualize far-right party competition, strategies, and agendas. First, they go some way to explaining how the far right has been able to expand its support beyond its core anti-immigrant voting base. While we do not dispute that these parties benefit electorally from the immigration issue, we have shown that there is ample room to compete based on an additional dimension, which optimizes their support. We thus suggest that the EU issue should be at the core of future work on far-right voting in Western Europe. For example, future surveys could break down the EU issue into economic, political and cultural grievances vis-à-vis the EU, and directly compare the importance of these sub-issues to immigration for the party family’s electoral success.

While past research shows the importance of immigration and the EU for the success of far-right parties independently, to the best of our knowledge, the literature has not juxtaposed simultaneously the effect of these two issues and their potential electoral returns, despite the fact that the question of Europe is core to far-right party competition (Vasilopoulou, 2018a). We test our argument using mathematical simulations because this method quantifies the empirical effects of changes in voters’ decision rule (due to changes in party positions and issue salience in our case) on the outcome of interest (parties’ vote shares). Future research could test our argument causally by employing experimental designs, such as for example conjoint experiments on the policy-valence-salience nexus. Research could also focus on small-N comparisons in order to identify patterns across different European regions, including Central and Eastern Europe, and explain variation.

More broadly, our findings allow us to contextualize far-right party campaign strategies that increasingly focus on criticizing the EU, without campaigning for their country’s EU withdrawal. Considering the reservoir of voters with Eurosceptic attitudes in each country and the rising salience of this dimension, a change of the status quo may lower the salience of the EU issue, which would be electorally harmful for the far right. In other words, the far right has a lot to gain electorally by making the election being more about the EU. We expect our theoretical arguments to hold across electoral contests. In the main text, we provide ample evidence that our hypotheses are supported in EP elections. In the appendix, we provide initial evidence that our arguments also hold in national elections. Yet we note that the EES surveys were fielded concurrent with EP -rather than national- elections.

Second, we have shown that changes in issue salience are more beneficial than positional changes; and that there is ample room for far-right parties to increase their vote share when in the public’s mind European integration is important. Far-right parties can engage in party competition without dramatically changing their policy positions, which may confuse or alienate voters. Instead, public salience of specific issues -especially if these are not already politicized in their respective party systems- may be enough to allow them to expand their voting base across the electorate. Note that, based on our findings, increasing the public salience of redistribution decreases the vote share of the far right. This is in line with Rovny and Polk (2020) who find that voter-party congruence on economic issues does not explain far-right voting. Although our focus has been on public [rather than elite] issue salience, our findings question the potential electoral uptake of the far-right’s change economic positions from market liberal to welfare chauvinism. Future research should comparatively consider the relative impact of changes in public versus party issue salience on the electoral fortunes of the far right.

Lastly, our findings have important implications for the campaign strategies of center-left and center-right parties and can serve as a warning sign against accommodating the positions of far-right parties. Similar to Meijers and Williams (2020) and Zur (2021b), we find that accommodating far-right parties on the EU mostly harms the electoral returns of centrist and center-right parties. By accommodating the positions of far-right parties, not only do mainstream parties legitimate these positions, but they also increase the public salience of European integration. These results imply that mainstream parties should not adopt more Eurosceptic positions, nor should they emphasize such positions, assuming their goal is vote-maximization.