Introduction

Individuals vote for political parties that represent them. This is at least what one can expect assuming that voting decisions are straightforward. However, the politics of representation is becoming more complex, at least in European multiparty systems where the diversity of party options and the variety of issue interests make individuals' voting choices increasingly difficult. The historical decline of party identification (Dalton & Wattenberg, Reference Dalton and Wattenberg2002; Mair, Reference Mair1989), together with the reconfiguration of the Rokkanian political cleavages and the emergence of radical populist parties (Mudde & Kaltwasser, Reference Mudde and Kaltwasser2012) contribute to make voting choices progressively less structured along the classic lines of political conflict (Hutter & Kriesi, Reference Hutter and Kriesi2019), putting traditional models of voting behaviour under strain. In this post-Rokkanian context, a voter who holds policy preferences across various issues can be pushed in different political directions because she can value some issues more than others when casting her ballot (Chou et al., Reference Chou, Dancygier, Egami and Jamal2021). At times in which European party systems are undergoing major transformations, how do cross-pressured voters weigh different, at times conflicting, issue preferences? What are the issues on which voters are likely to compromise? And are there differences between voters of radical and mainstream parties in terms of the issue trade-offs they are willing to make?

The relevance of issue-cross pressures in voting decisions has been acknowledged before (He, Reference He2016; Lefkofridi et al., Reference Lefkofridi, Wagner and Willmann2014) but empirical findings on how citizens navigate competing issue preferences remain sparse. In fact, despite a wealth of single-country or comparative studies on electoral behaviour (Arzheimer, Reference Arzheimer and Rydgren2018; Rooduijn, Reference Rooduijn2018; Steiner & Hillen, Reference Steiner and Hillen2021), the primarily survey-based evidence makes it hard to study issue-cross pressures, for at least two reasons. First, while available survey batteries offer insights on preferences concerning multiple issues independently, we cannot infer how respondents trade off their issue preferences against each other to satisfy their most important one. And second, surveys do not allow for drawing causal inferences about which issue preferences ultimately drive vote choices.

This paper employs an issue-centred research design that allows us to identify the conditions under which voters are (un)willing to trade off some preferences for others. We compare the behaviour of radical left and radical right voters who are known to have peculiar preferences on the issues of redistribution and immigration, respectively, with voters of mainstream parties. Specifically, we conducted conjoint experiments in France, Germany, Italy and Spain, where different types of radical parties have broken through. Participants were asked to choose between two hypothetical candidates in the next national parliamentary election who present diverging proposals on salient political issues and different stances on political representation. Randomizing the candidates' proposals on issues allowed us to identify the causal effect of each issue position on candidate choice. We avoided the confounding influence of party cues on issue trade-offs by not assigning party labels to candidates and taking information on party attachment from surveys run months before the experiment. As a validation of the results, we again implemented the same experimental setup almost 2 years later using a large German sample selection according to population margins.

The findings from all studies samples consistently show that voters of radical right parties are less likely to compromise on their most important issue. As a consequence, radical right voters are willing to accept even large trade-offs regarding undesirable proposals on climate change and European Union (EU) integration in order to achieve restrictive immigration policies. Differently, radical left voters have less skewed issue orientations: they have strong preferences on redistributive policies but are still less likely to accept candidates that additionally propose undesirable issue proposals, for example, climate denialist stances. Taken together, the study contributes to understanding long-term prospects for vote choices and the reconfiguration of politics in Europe, revealing complex patterns of issue (de)alignment of voters in contemporary democracies.

Political cleavages, issue-cross pressures and individuals' vote choices

The politics of representation is becoming increasingly multidimensional, at least in political systems where a variety of issue interests and political parties make individual vote choices more difficult. In this regard, an impressive body of research has examined the factors determining voting choices. Yet, despite recent advances (He, Reference He2016; Lefkofridi et al., Reference Lefkofridi, Wagner and Willmann2014; Steiner & Hillen, Reference Steiner and Hillen2021), only scarce attention has been paid to measuring and examining the consequences of issue-cross pressures that relate to ‘inconsistencies among individuals’ attitudes towards various political objects' (He, Reference He2016, p. 364). Our goal is to address this gap, shedding light on the impact of issue-cross pressures on vote choices and to examine the mechanisms of how voters of different party families trade off various issue preferences against each other.

The relevance of different kinds of cross-pressures in voting decisions has been the focus of three major research schools. To begin with, the sociological model of voting behaviour proposed by Lazarsfeld et al. (Reference Lazarsfeld, Berelson and Gaudet1944) and colleagues at Columbia University contended that individuals' vote choices were heavily influenced by demographic characteristics and that belonging to a social group would be the predominant factor in determining voting decisions. In follow-up studies, they acknowledged the possibility of issue-cross pressures as they found some inconsistencies between policy preferences related to demographic indicators and those emerging from the characteristics associated with the groups to which individuals belong (Berelson et al., Reference Berelson, Lazarsfeld and McPhee1954). In their interpretation, these tensions would make individual voting decisions more difficult as they need to trade off some concerns over others.

A second major contribution to the study of vote choices came from the socio-psychological model of voting behaviour, initiated at the University of Michigan (Campbell et al., Reference Campbell, Converse, Miller and Stokes1960). These scholars assumed that party identification was central in informing voting decisions. Still, Campbell and colleagues also provided the theoretical tools to account for issue-cross pressures as they clarified that partisanship is not a factor that determines unambiguously how individuals will cast their ballots. Rather, it has to be understood as a ‘filter’ through which voters appreciate what is favourable to the orientation of the party and ignore (or value less) what is considered unfavourable. In other words, partisanship functions as an instrument to ‘decipher’ electoral campaigns and candidates' proposals.

Further theoretical elaboration came from the macro-sociological approach that understood Western European party systems as reflecting historical divisions originating in national revolutions (Lipset & Rokkan, Reference Lipset and Rokkan1967). Starting from the 1970s, the ‘defreezing’ of traditional political cleavages (Inglehart, Reference Inglehart1971) contributed to the decreasing role of party identification in voting decisions (Dalton & Welzel, Reference Dalton, Welzel, Dalton and Welzel2014). At the same time, these transformations of the political space have accompanied the emergence of novel parties in European party systems, notably Green parties and radical populist parties, increasing the diversity of available options and making voting choices even more complex (Ignazi, Reference Ignazi1992). Scholars suggest that in this post-Rokkanian political space, voting choices are increasingly determined by two main socio-political divides: an economic and a cultural one that may also create contradictory policy preferences (Hutter & Kriesi, Reference Hutter and Kriesi2019; Kriesi et al., Reference Kriesi, Grande, Dolezal, Helbling, Höglinger, Hutter and Wüest2012; Rovny & Polk, Reference Rovny and Polk2020).

While the specific issues of economic competition may vary from election to election, at a more abstract level, political competition on the economy tends to include debates about pro-state and pro-market positions (Castles et al., Reference Castles, Mair and Pedersen1997; Traber et al., Reference Traber, Giger and Häusermann2018). Specifically, this emergent economic divide opposes the advocates of a more interventionist state in regulating the economy, promoting social policies and setting taxation levels to those who call for a more limited role of governments (Dalton, Reference Dalton2018; Kitschelt, Reference Kitschelt1994). Another core political divide in contemporary European societies involves issues underlying a cultural cleavage. Central in these debates are newer issues associated with globalization, notably EU integration, immigration and, more recently, climate change (Hooghe & Marks, Reference Hooghe and Marks2018; Kriesi et al., Reference Kriesi, Grande, Dolezal, Helbling, Höglinger, Hutter and Wüest2012; Treib, Reference Treib2021). The transformation of the political space provides critical opportunities for both mainstream and radical parties that try to take up evolving preferences and values, (re)positioning themselves along these conflict lines. So far, cultural issues appear to be most central in explaining the vote for populist radical right parties but not for radical left or mainstream parties (Arzheimer, Reference Arzheimer and Rydgren2018; Rooduijn, Reference Rooduijn2018).

The literature on the drivers of voting behaviour helps understanding that voting decisions are far from being straightforward. They are mediated by multiple, at times conflictual, individual issue preferences that cut across established and newer political cleavages. Still, available knowledge tends to overlook that as a response to partisan dealignment and changing lines of political conflict, cross-pressured individuals must and do trade off issues in formulating voting decisions. In line with the notions of ‘cross-cutting cleavages’ (Rokkan, Reference Rokkan, Lipset and Rokkan1967), ‘cross-cutting pressures’ (Berelson et al., Reference Berelson, Lazarsfeld and McPhee1954) or ‘issue cross-pressures’ (He, Reference He2016), a voter's congruence with a candidate's proposals on redistribution can be counteracted by a candidate's incongruent positions on immigration. Therefore, in most voting scenarios, the same voter has to trade off several more or less important preferences against each other to see her favourite one fulfilled.

More recently, researchers have started to use experimental designs to study the causal effects of voters' issue preferences on vote choice (Chou et al., Reference Chou, Dancygier, Egami and Jamal2021; Graham & Svolik, Reference Graham and Svolik2020; Hanretty et al., Reference Hanretty, Lauderdale and Vivyan2020; Neuner & Wratil, Reference Neuner and Wratil2022). Most closely related to our work are several conjoint studies with related, yet distinct, research questions. Chou et al. (Reference Chou, Dancygier, Egami and Jamal2021) designed candidate profiles with German party labels to investigate vote switching of radical right party supporters conditional on their issue preferences. The study found that AfD voters are willing to vote for mainstream parties that propose a complete stop of immigration. However, they also show that such an accommodation strategy alienates the core voters of mainstream parties. Another study of the Germany context investigated the electoral effects of ‘thin’ populist stances of candidates versus ‘thicker’ populist issue bundles. The authors identify differences in voters' issue priorities depending on individual levels of populist attitudes, but that anti-immigrant and pro-redistribution positions increase the appeal of German candidates, on average (Neuner & Wratil, Reference Neuner and Wratil2022). Finally, Franchino and Zucchini (Reference Franchino and Zucchini2015) used student samples to study the importance of valence issues in vote choices.

In sum, to the best of our knowledge, no existing study has experimentally compared the willingness to trade off different issue preferences among radical left, radical right and mainstream party voters across multiple countries. To address this gap, we designed a candidate conjoint experiment with respondents from France, Germany, Italy and Spain, including the most salient contemporary issues as well as an attribute capturing populist stances of candidates. In the next section, we formulate expectations about the drivers of issue trade-offs in vote choices for different groups of voters.

Hypotheses: How issue trade-offs shape vote choices

In our understanding, issue preferences cross-cut a bundle of distinct issues that may generate conflict in voting decisions. To resolve these conflicts, voters have to make issue trade-offs depending on the intensity of their specific preferences. In other words, when a candidate proposes the most desirable solution, voters will have to also accept less desirable proposals – at least to some extent. For example, voters who attach a high value to redistribution may be willing to compromise on EU integration and immigration as long as their redistribution preference is fulfilled. This theoretical framework has a high external validity, as a perfect party-voter congruence can only rarely be observed empirically (Costello et al., Reference Costello, Toshkov, Bos and Krouwel2020; Steiner & Hillen, Reference Steiner and Hillen2021; Traber et al., Reference Traber, Giger and Häusermann2018). The literature on political parties helps identifying relevant overlapping or contrasting issue preferences, informing how voters of different party families approach issue trade-offs.

Differently from voters of mainstream parties, which we understand as non-radical ‘traditional’ party families ranging from Green parties to conservatives, radical electorates have been shown to share dissatisfaction with the functioning of government, and express lower trust in political institutions. This more adversarial approach to conventional politics can be expected to yield major differences in issue trade-offs between voters of mainstream and radical parties (Van Hauwaert & Van Kessel, Reference Hauwaert and Kessel2018). Radical voters, in fact, may have stronger preferences on specific questions they regard as underrepresented in mainstream party politics and may be less ready than their non-radical counterparts to trade off these. Accordingly, we expect that

• H1: Voters of mainstream parties are willing to accept larger trade-offs regarding their most important issue preference than voters of radical left and radical right parties.

Beyond preferences about conventional politics, more specific campaigns associated with radical right and radical left parties can be expected to inform issue trade-offs. Various studies show that vote choices for radical right parties are primarily motivated by preferences on immigration (Arzheimer, Reference Arzheimer and Rydgren2018; Halikiopoulou & Vlandas, Reference Halikiopoulou and Vlandas2020; Van Hauwaert & Van Kessel, Reference Hauwaert and Kessel2018). This does not mean that a prototypical voter of the radical right is only concerned about immigration, but it means that this voter has stronger political preferences for the anti-immigrant claims that feature prominently in the campaigns of these parties (Halikiopoulou, Reference Halikiopoulou2019; Mudde, Reference Mudde2007). We expect that radical right voters' resolute preferences over immigration might thus make them regard other issue preferences as less desirable or important.

• H2: Voters of radical right parties are willing to accept large trade-offs regarding their other issue preferences as long as their preference for restrictive immigration policies is fulfilled.

It has also been shown that radical left electorates hold peculiar preferences over specific issues. While differences between mainstream left and right parties have become more blurred over time, radical left parties continue to campaign and appeal to voters who are primarily concerned about the skewed socioeconomic structure of contemporary capitalism (March, Reference March2012) and to call for a reduction of inequalities through redistribution, state subsidies and other major changes in economic and power structures (Ramiro & Gomez, Reference Ramiro and Gomez2017; Rooduijn, Reference Rooduijn2018). In sum, one may expect that voters of radical left parties hold stronger preferences for redistribution and are ready to trade off other issues considered as less important.

• H3: Voters of radical left parties are willing to accept large trade-offs regarding their other issue preferences as long as their preference for redistributive policies is fulfilled.

Assuming that radical left or right voters have rather peculiar tendencies when it comes to accepting trade-offs in their issue preferences, it is an intriguing question whether they differ in their desire to have their strongest preference fulfilled. While there is evidence that radical left and radical right electorates have different preferences about pluralism in society (Rooduijn et al., Reference Rooduijn, Burgoon, Elsas and Van de Werfhorst2017), such as minorities' rights, other studies show that both groups tend to have strong preferences for opposition to globalization and EU integration (Visser et al., Reference Visser, Lubbers, Kraaykamp and Jaspers2014), even if scholars identify different types of nationalism that inform these positions (Halikiopoulou et al., Reference Halikiopoulou, Nanou and Vasilopoulou2012). Another issue where there may be more common ground than assumed in the previous hypotheses is redistribution. While radical right parties display and appeal to voters who hold more blurred positions on redistribution (Rovny & Polk, Reference Rovny and Polk2020), they still promise consumptive social policies while deemphasizing social investment policies (Enggist & Pinggera, Reference Enggist and Pinggera2022). A substantial share of so-called ‘left authoritarians’ without proper party representation (Lefkofridi et al., Reference Lefkofridi, Wagner and Willmann2014; Steiner & Hillen, Reference Steiner and Hillen2021) also speaks for similarities, or at least smaller issue trade-offs that radical left and radical right voters might be willing to accept for one specific issue. The group of radical left voters should have become even more cross-pressured with the advent of post-materialist issues such as climate change that are salient in the platforms of left parties (Farstad, Reference Farstad2018).

In sum, previous literature has identified contradictory trends when it comes to the relative weight of different issues in the preference order of radical left and right voters. After all, it might be that these voter groups feel more cross-pressured beyond their primary orientation towards redistribution and immigration, respectively. To accommodate for unresolved puzzles in the literature, we formulate additional open questions for the analysis: What are the issues important to radical left and right voters besides redistribution and immigration, respectively? Are there differences between radical left and radical right voters in their willingness to compromise on their most important issue preference?

Data and methods

Attributes and levels

The substantive goal of the experiment was to provide a realistic representation of the most salient contemporary issues across all party families that are reconfiguring political conflict in Europe (Kriesi et al., Reference Kriesi, Grande, Dolezal, Helbling, Höglinger, Hutter and Wüest2012). To create a list of relevant issues (or attributes) on which candidates present a specific proposal (levels), we did extensive research incorporating information from the demand side (voters) and supply side (political parties). As the point of departure, we analysed the most important problem perceptions (‘MIP’, using the standard Eurobarometer issue battery) of radical right and radical left voters in our own surveys. We studied in detail the party manifestos for the 2019 EP election and recent national elections to empirically identify the most important issues and associated issue positions (Online Appendix Section A4.2). We also downloaded 121,108 Facebook posts posted by the biggest national parties in 2019 to verify that the chosen issues were also salient in party communication (Online Appendix Section 4.3). Finally, we consulted with 11 experts on populism and party politics at various stages of the design of the experiment to create levels (issue proposals) that were applicable across countries.

Of the surveyed issue categories, living costs, unemployment and the economy were salient in our sample and also in parallel Eurobarometer surveys of the general population (Online Appendix Section A4.1). Immigration was by avfar the most important perceived problem for radical right voters, whereas social security and the environment (‘The environment, climate and energy issues') stood out more clearly among radical left voters. In the survey, we also asked for free-text descriptions describing the chosen most important problem briefly so that we better understand how respondents interpret political issues. The issue descriptions provided by respondents indicate that their concerns boiled down to personal economic fears and not being able to cover the costs of daily life (see word clouds in Online Appendix Figure A6). Therefore, as a construct capturing preferences regarding redistribution, we tapped into a traditional survey question on the role of the state in the economy (interventionist vs. non-interventionist) but tailored it towards subsidies on staples and housing that would directly reduce living costs. Accordingly, the item should not just capture radical left voters’ well-documented preference for state intervention in social policies but also radical right voters' consumption-oriented demands towards the welfare state (Enggist & Pinggera, Reference Enggist and Pinggera2022). In addition, we included positions on the European Union that have become entangled in an ‘emerging centre-periphery cleavage’ (Treib, Reference Treib2021, p. 175) driven by Eurosceptic populist parties, and Climate change, an increasingly salient issue in European democracies (Farstad, Reference Farstad2018), most notably after the emergence of the movement Fridays for Future.Footnote ³ Finally, in order to capture preferences for a populist stance towards political representation, we included a Reason for running item, adjusted from the conjoint study of Chou et al. (Reference Chou, Dancygier, Egami and Jamal2021).Footnote ⁴

The final list of attributes and levels can be found in Table 1. Although we focus on radical right and radical left voters, we still wanted to compare their voting behaviour to voters of mainstream parties. Thus, we drew policy proposals from across the ideological spectrum (Online Appendix Section A4.2). We constructed attribute levels as follows: the first level is a right/authoritarian/nationalist policy proposal, the middle category is the status quo, and the third level is a left/liberal proposal, either in terms of economic or GAL-TAN issues. Choosing a similar number of levels for each attribute makes it possible to compare the relative importance of issue proposals across and within attributes.

Table 1. Attributes and levels (proposals) in the conjoint survey experiment

Results

For reporting the key findings of the conjoint analysis, we used marginal means (MMs) as our estimand (Leeper et al., Reference Leeper, Hobolt and Tilley2020), which has a straightforward interpretation as probabilities with binary outcome variables such as ours – the choice of a candidate. A MM of 0 means that respondents chose a particular profile feature with zero probability, a MM of 1 means the candidate profile with that feature was always chosen, ignoring all other features. Instead of a reference category, we chose a 0.5 probability to indicate that the feature was not significant for respondents in their choice of a candidate profile. We obtained MMs from simple ordinary least squares regression models. Because the features (levels) of candidate profiles were randomized, the effect of every feature on candidate choice in the conjoint experiment can be causally interpreted. We designed the levels in the conjoint experiment in a way that allowed for a complete randomization, meaning that a candidate profile could take any attribute combination and every feature had the same probability to appear in a profile. All analyses were performed in R, version 4.2.1. Replication materials including data and R scripts are available on OSF: https://doi.org/10.17605/OSF.IO/8P54D

Identifying issue preferences among different groups of voters

To test our hypotheses, we first needed to examine if radical right, radical left and mainstream voters have a dominant issue preference. We plotted the results in Figure 1 and also included the numerical outputs from the regressions in the Online Appendix Tables A7– A9. Consistent with the literature, our experimental evidence shows that radical right voters respond strongly to the immigration issue, radical left voters to the redistribution issue. Meanwhile, the unifying issue for the diverse group of mainstream voters is the EU.

Figure 1. The effect of candidates' issue proposals on candidate choice in the conjoint survey experiment for radical left, radical right and mainstream party voters. Error bars represent 95 per cent confidence intervals. [Colour figure can be viewed at wileyonlinelibrary.com]

Three patterns stand out in Figure 1. First, in contrast to radical right voters who respond to the introduction of border controls (positively) and removing restrictions (negatively) with the same strength, the preferences of radical left and mainstream voters are more lopsided. They are more likely to reject proposals than to enthusiastically support proposals on their most important issue. Specifically, radical left voters only moderately supported candidates proposing to increase subsidies on staples and housing or keep the status quo but strongly oppose candidates who promote economic self-sufficiency. Similarly, mainstream voters moderately supported empowering the EU through weakening the veto rights of EU member states (a proposal found in party programmes of many mainstream parties, see Online Appendix Section A4.2) but strongly opposed candidates proposing to leave the Eurozone.

Second, there is barely a candidate for the second most important issue of radical right voters, while radical left and mainstream voters were also concerned about climate change besides income redistribution or the EU, respectively. Both groups were firmly against ignoring the problem of rising ${\mathrm{CO}}_2$$\rm CO_{2}$ emissions, a proposal even radical right voters rejected.Footnote ⁵ There were few similarities between the two radical poles in the electorate and no evidence for left authoritarians who might appreciate redistribute policies and anti-immigration rhetoric at the same time (Lefkofridi et al., Reference Lefkofridi, Wagner and Willmann2014).

Third, with ‘Reason for running’, we were aiming to signal to respondents a candidate's stance on political representation. All three groups of voters punished candidates running for a seat to continue to serve the government and ignored the neutral reason to participate in policy making. If a candidate was running for a seat to combat the corrupt elite and represent the real people, radical right voters' response to this anti-elitist candidate was more positive than the response of radical left or mainstream party voters. Taken together, the findings are only partially consistent with the literature (Van Hauwaert & Van Kessel, Reference Hauwaert and Kessel2018), where radical right and left voters prefer anti-elitist candidates. This suggests that, on average, voters of all three party families were not satisfied with the current government but that only radical right voters endorsed anti-elitist stances. Overall, however, our multidimensional survey experiment demonstrates that in the presence of salient issues like immigration or redistribution, radical left and radical right voters did not put much weight on candidates' populist features.

Cross-pressured voters and issue trade-offs

We use the results of Figure 1 to test our hypotheses on issue trade-offs among radical left, radical right and mainstream party voters. We specifically zoom in on the choices of each group under the condition that they were offered their most preferred issue proposal – restricting immigration for radical right voters, increasing subsidies for radical left voters and empowering the EU for mainstream party voters.Footnote ⁶ This empirical strategy allows us to observe to what extent voters with strong issue preferences are willing to compromise (trade-off) to have their most (desirable issue proposal fulfilled.

Radical right voters. Since the results of the pooled conjoint analysis showed that a restrictive immigration policy was most preferred by radical right voters (Figure 1), we reran the analysis for each issue position on immigration individually. In other words, we obtained the effects of issue proposals on respondents' vote choices holding immigration policy constant. Figure 2 shows that radical right voters were consistent in their behaviour. As all coefficients are clearly on the right-hand side of the 0.5 probability threshold, the approval of candidates with restrictive immigration policies increased regardless of their at times undesirable proposals on other issues such as climate change or the EU. Conversely, radical right voters punished candidates who wanted to remove restrictions on immigration despite the presence of other issue proposals that radical right voters embraced in the pooled analysis.

Figure 2. The effect of issue proposals on candidate choice among radical right voters holding candidates' proposals on immigration constant: border controls, status quo or removing restrictions. Error bars represent 95 per cent confidence intervals. [Colour figure can be viewed at wileyonlinelibrary.com]

Radical left voters. Figure 3 reports the results for radical left voters grouped by candidate proposals on redistribution. When candidates proposed an interventionist approach to subsidies, radical left voters mostly continued rewarding them. Yet the relative effect sizes were only moderate and almost identical to the coefficients of keeping the status quo. Some effects were even insignificant, meaning that some alternative issue proposals were of similar importance for radical left voters. More consistently, candidates who proposed non-redistributive policies were punished: all effects of other issues are on the left-hand (negative) side of the 0.5 probability threshold, with the sole exception being taxes on corporations for their ${\mathrm{CO}}_2$$\rm CO_{2}$ emissions. Overall, (dis)like of different redistribution policies played a less central role in vote choices of radical left voters compared to the consistently strong orientation of radical right voters towards immigration.

Figure 3. The effect of issue proposals on candidate choice among radical left voters holding candidates' proposals on redistribution constant. Error bars represent 95 per cent confidence intervals. [Colour figure can be viewed at wileyonlinelibrary.com]

Mainstream party voters. Similar to radical left voters' balanced preferences over redistribution but unlike radical right voters skewed preferences, mainstream voters did not reward issue proposals for deeper European integration significantly more than keeping the status quo (Figure 4). However, mainstream voters consistently punished candidates proposing the anti-EU policy of leaving the Eurozone.

Figure 4. The effect of issue proposals on candidate choice among mainstream voters holding candidates' proposals on the EU constant. Error bars represent 95 per cent confidence intervals. [Colour figure can be viewed at wileyonlinelibrary.com]

Comparing differences in issue trade-offs

To more systematically test our hypotheses, we subtracted the effect sizes for every issue position in the pooled regression model (Figure 1) from the effects in Figures 2–4. As in the previous analysis, we again focus on the most preferred issue proposal for each group. Figure 5 presents a formal comparison of results, with the dashed vertical line representing the mean difference for radical right voters.

Figure 5. Difference in effect sizes for other issue proposals when the most desirable issue proposal among radical left, right and mainstream voters was shown. Error bars represent 95 per cent confidence intervals. The dashed vertical line shows the mean change for radical right voters. [Colour figure can be viewed at wileyonlinelibrary.com]

Overall, the bars demonstrate that in the scenarios when the most preferred issue position was fulfilled by a candidate, respondents were also willing to accept other issue positions. For instance, mainstream party voters were much more likely to choose the status quo for redistribution under the condition that the EU issue is addressed in line with their preference. However, the most striking finding pertains to radical right voters: their strong preference for restricting immigration made them bear even positions that fundamentally ran counter to their overall preferences (Figure 1). Importantly, in contrast to mainstream party and radical left voters, there is no noteworthy variation between issues, meaning that the preference order of radical right voters is much more lopsided towards a single-issue preference.

But what about the least preferred instead of the most preferred issue proposal? After all, Figure 1 revealed an asymmetric reaction of radical left and mainstream party voters: ceteris paribus, they tended to reject certain issue positions (e.g., the climate denialist proposal) more strongly than to enthusiastically support one of the available positions per issue. To account for these scenarios, Online Appendix Figure A19 replicates the analysis for the least preferred issue proposal per group. While the differences were less stark, radical right voters still stand out in terms of the overall strength of issue preferences and their unwillingness to differentiate further between their lesser preferred issue positions. An analysis of the average duration respondents needed for a conjoint task sheds more light on these patterns: radical right voters were significantly quicker in rejecting their least preferred and selecting their most preferred issue proposal than radical left and mainstream party voters (Online Appendix Figure A24).

Overall, there is mixed evidence regarding our hypotheses. As hypothesized, radical right voters were very much willing to accept issue trade-offs in order to fulfill their preference for restrictive immigration policies (H2). In fact, despite many different angles to look at the results, no clear second most important issue besides immigration emerged. Their radical left counterparts indeed had a preference for redistributive policies and an even more pronounced dislike of economic self-sufficiency. However, in contrast to the expectation in H3, radical left voters had a more variegated range of issue preferences, some of them so strong that they were not traded off for their redistribution preferences. Finally, H1 is only partially supported, since in contrast to radical right voters' preference for restrictive immigration policies, both, radical left voters and mainstream party voters were equally unlikely to accept large trade-offs in other issues (most importantly, climate change) to have their most preferred issues fulfilled.

Conclusion

Voting behaviour is becoming increasingly complex in the post-Rokkanian political space that is so characteristic of contemporary European democracies. One of the reasons is that voters' issue preferences can cut across established and emerging political cleavages, pressuring them to prioritize and trade off their concerns against each other. Accordingly, researchers increasingly acknowledge the importance of issue cross-pressures in vote choices (He, Reference He2016). Our goal was to examine the relative strength of issue preferences and to what extent voters of radical and mainstream parties are willing to make issue trade-offs.

Our study innovated by conducting a conjoint experiment featuring salient contemporary issues across four major European democracies. Setting a similar number of levels (proposals) per issue and avoiding party labels allowed us to identify the strength of issue preferences of different groups of voters. The results show that radical right voters were willing to make large issue trade-offs as long as their most important issue preference of restricting immigration is fulfilled. In contrast, radical left and mainstream party voters were more willing to compromise on their most important issue position in favour of issues ranked lower in their order of preference.

Our experimental findings have implications for several streams of research. The striking absence of a pronounced second most important issue preference among radical right voters adds to ongoing debates about the broader shifts in party systems due to the advent of radical right parties. Much of this research has centred on the (re)positioning of conservative and mainstream left parties on the issue of immigration and the (lack of) success of such accommodation strategies (Chou et al., Reference Chou, Dancygier, Egami and Jamal2021; Spoon & Klüver, Reference Spoon and Klüver2020). Our research adds another perspective to these debates. If their voters are barely considering any issue proposals other than restrictive immigration policies, attempts of radical right parties to broaden their platform (e.g., Marine Le Pen's flirtations with ecological issues) seem futile. In many electoral scenarios, the chances of radical right parties will therefore hinge on the salience of the immigration issue.

While bundles of cultural explanations were identified as the core drivers of the radical right vote, voting behaviour of radical left voters most often tends to be reduced to economic concerns and traditional political cleavage structures. Yet in our study, radical left voters were not willing to accept climate denialist stances as a trade-off for their preferences over redistribution. These results were confirmed in the well-powered replication study where respondents were sampled systematically according to German population margins. Taken together with more EU-friendly preferences than found in previous studies, these results tentatively point towards an increased sorting of radical left voters along a new post-materialist cultural axis. However, further issue-specific research is needed. More action-oriented proposals, for instance, to ‘tax the rich’ might trigger stronger reactions among radical left voters (Neuner & Wratil, Reference Neuner and Wratil2022).

The unifying issue preference among the heterogeneous group of mainstream voters was in the field of EU politics. While they were equally likely to choose candidates who proposed to keep the status quo or who wanted to institutionally strengthen the EU, they despised leaving the common currency Euro. Voters of mainstream parties also clearly preferred climate policies aiming to curb ${\mathrm{CO}}_2$$\rm CO_{2}$ emissions while rejecting climate denialist positions. These experimental insights contribute to emerging research on the role of climate politics in European party systems (Farstad, Reference Farstad2018).

Beyond the need for additional replication studies with bigger samples for all four countries, our study design comes with additional limitations. The survey experiment took place while the COVID-19 pandemic was intensifying in spring 2020. Accordingly, the ongoing discussions about closing borders to curb the spread of the pandemic might have increased the appeal of border controls to restrict immigration. Reassuringly, there is first evidence that the pandemic did not drastically influence behaviour in experiments (Peyton et al., Reference Peyton, Huber and Coppock2020). The replication study that we conducted in early 2022 also confirmed the results. We further acknowledge that the operationalization of populist stances might suffer from a limited external validity. While in line with other experimental studies (Chou et al., Reference Chou, Dancygier, Egami and Jamal2021; Neuner & Wratil, Reference Neuner and Wratil2022), one possible explanation of why an anti-elitist stance of candidates did not have large effects might be that such strategies only appeal to radical electorates in combination with specific party cues. While being a necessary design choice to separate issue preferences from partisan identity, one specificity of the research design is the lack of party labels of candidates. Similarly, forced-choice conjoint experiments reveal preferences of participants but do not allow for abstention, a viable option when voters are feeling cross-pressured. Finally, while our process of identifying issues was informed by survey responses, party manifestos, parties' social media communication and consultations with experts, the issues and associated positions were chosen in a way that they are applicable across the four democracies under study during one specific research period. As a consequence, country experts perhaps regard other issue areas as more important than some of the ones chosen in this period.

Despite these caveats, the paper has offered an important step towards uncovering the similar and distinct issue considerations underlying vote choices of radical left, radical right and mainstream party voters. Revealing these mechanisms holds implications not only for specialists of populism and political behaviour but also for scholars interested in how issue dynamics are re-configuring political conflict in Europe.