Economic system justification reflects the belief that the system provides equal opportunities for success and that economic outcomes are determined by hard work and skills (Jost & Hunyady, 2005). People who endorse system-justifying beliefs often maintain high prospects of upward mobility, oppose progressive taxation and welfare policies, and vote for right-wing parties (García-Sánchez et al., 2022).

The status-legitimacy hypothesis of the system justification theory posits that low SES individuals (compared to high SES) are more inclined to justify the economic system as fair and legitimate in order to alleviate the strong sense of cognitive dissonance between system-justifying motivations and self/group-protection motivations (Brandt, 2013; Jost et al., 2004). However, the majority of the empirical evidence challenges this view (Andersen & Curtis, 2015; Brandt, 2013; Brandt et al., 2020; Brown-Iannuzzi et al., 2015; Vargas-Salfate et al., 2018) and is consistent with a self-interest perspective which posits that high SES individuals are more likely to endorse economic system justifying beliefs that legitimize their advantageous position. Consistent with this self-interest explanation, experimental studies revealed that when misperceptions about people’s economic position were corrected, and individuals were informed that they were richer than they had previously thought, they endorsed more system-justifying beliefs (Cruces et al., 2013; Mijs & Hoy, 2022).

It is worth noting, however, that the positive relationship between SES and economic system justification has primarily been documented in studies that operationalized SES as subjective social status or income (Brandt, 2013; Yang et al., 2019). In contrast, evidence regarding the link between education and system justification remains inconsistent. For instance, while some studies have reported a negative relationship between education and system justification (Li et al., 2020; Moya & Alcañiz-Colomer, 2023), others have found no significant association (Valdes et al., 2023). Consequently, we do not formulate a hypothesis about the indirect relationship between education and voting through economic system justification. Instead, we explore this relationship in a more exploratory manner.

Based on the strong evidence for the positive relationship between subjective SES and system justification, we hypothesize that subjective income (H1) will be positively associated with right-wing voting intentions, and this relationship will be mediated by increased endorsement of economic system-justifying beliefs (Figure 1).

According to social identity theory, low-status group members may distance themselves from their low-status ingroup and recategorize themselves at a higher level as a way to manage their negative social identity (Tajfel & Turner, 2001). In line with this idea, it has been suggested that low-SES individuals may exhibit reduced identification with their class identity and seek other positive superordinate identities, such as national identity (Jay et al., 2019). Indeed, low-income and low-educated individuals are less likely to define themselves in terms of their status position, and instead they are more likely to identify with their national group (Kuppens et al., 2015; Manstead, 2018; Shayo, 2009). Moreover, both income and education were found to be negatively correlated with national identification (Owuamalam et al., 2023) and low-education was additionally associated with ethnic nationalism and exclusionism (Coenders & Scheepers, 2003). Conversely, high SES individuals prioritize identities that emphasize their high-status position, such as social class (Easterbrook et al., 2020).

Evidence suggests that low SES individuals who strongly identify with the national group are less likely to support income and wealth redistribution (Shayo, 2009), are more likely to endorse conservative ideologies, have anti-immigrant sentiments (Pehrson & Green, 2010), and vote for right-wing parties (Oesch, 2008). Furthermore, right-wing voters and conservatives tend to be more committed to the nation, more nationalistic, and have stronger collective narcissist beliefs about the exceptionality of the national ingroup (Golec de Zavala et al., 2019).

Given that the negative link between SES and national identification has been reported by studies using income and education as SES indicators, we assume that education and income effects on national identification will converge. Based on these insights, we hypothesize that low education (H2a) and low income (H2b) are positively associated with intentions to vote for right-wing parties through increased national identification (Figure 1).

Another strand of literature has emphasized the crucial role of perceived threats in shaping political attitudes. Threats may pertain to loss of resources (i.e., economic threats), or to the integrity or validity of the ingroup’s meaning system and values (i.e., symbolic threats) (Rios et al., 2018; Stephan et al., 2009).

There is substantial evidence suggesting that symbolic threat is strongly associated with the cultural dimension of political preferences. For example, symbolic threat predicts conservative political identities and attitudes (Casteigne et al., 2023) as well as right-wing authoritarianism (Voca et al., 2024). Specifically, symbolic threats posed by immigrants have been linked to prejudice, intergroup hostility, and increased support for far-right populist parties advocating chauvinistic and exclusionist policies (Jay et al., 2019; Jetten et al., 2017).

On the other hand, economic threat may be associated with both cultural and economic dimensions of political preferences. According to a motivated social cognition approach, economic threats undermine our sense of control and self-esteem (Fritsche & Jugert, 2017). This process elicits personal or collective attitudes and behaviors that can either be functional in that they try to eliminate the source of threat (e.g., collective action toward equality or support for redistributive policies) (Çakal et al., 2016) or are merely palliative without tackling the actual source of threat. These palliative responses aim at restoring a general sense of control or self-esteem by fueling outgroup prejudice and leading to ‘displaced’ intergroup conflict with minority outgroups who are perceived as competitors for limited resources (Becker et al., 2011; Fritsche & Jugert, 2017). Given that migration has recently come to the forefront of political debates across many European countries (Chouliaraki et al., 2017; Davis & Deole, 2017), this threat-driven outgroup derogation is often expressed as strong anti-immigrant sentiments (Davis & Deole, 2017; Mayer et al., 2020; Rydgren, 2012).

Therefore, in the current European context, both symbolic and economic threats are often attributed to immigrants, fueling anti-immigrant feelings which resonate with a growing right-wing populist discourse (Erisen & Vasilopoulou, 2022; Kende & Krekó, 2020; Messing & Ságvári, 2021). Although individuals across all SES levels may exhibit anti-immigrant attitudes in response to perceived threats, this relationship is stronger among low-income and low-educated individuals (Anier et al., 2016; Jetten et al., 2015). This is because these individuals experience chronic threats, are more susceptible to perceiving threats because they are at the mercy of higher status and more powerful groups, and are more likely to perceive competition with immigrants over jobs and social benefits (Corenblum & Stephan, 2001; Kraus et al., 2012; Manstead, 2018). Indeed, prejudice towards immigrants was independently predicted by lower levels of income and education (Carvacho et al., 2013).

Based on these insights, we hypothesize that low education and low income is associated with right-wing voting intentions through increased perception of economic (H3) and symbolic (H4) threats (Figure 1). Furthermore, we hypothesize that the negative link between SES and right-wing voting intentions through economic (H5) and symbolic (H6) threat are partly channeled through anti-immigrant attitudes (i.e., serial mediation). Given that both education and income have been associated with lower perceptions of threat (Kraus et al., 2012; Manstead, 2018), we test the aforementioned hypotheses for both education (H3a, H4a, H5a, and H6a) and income (H3b, H4b, H5b, and H6b).

On the one hand, some research suggests that low-status individuals are more likely to vote for right-wing parties (Napier & Jost, 2008; Rydgren, 2012; Stubager, 2010), while on the other hand, evidence also exists supporting the opposite pattern (Andersen & Curtis, 2015; Bartels, 2006; Brown-Iannuzzi et al., 2015). Given this inconsistency, we chose not to formulate a formal hypothesis regarding the relationship between SES and right-wing voting. Instead, our objective is to examine the explanatory power of the three previously discussed psychological mechanisms (Figure 1) in the relationship between SES and right-wing voting intentions. We aim to reconcile the contradictory findings in two key ways. First, by distinguishing between different indicators of SES—namely education and income—as composite measures of SES may obscure their divergent effects (Easterbrook et al., 2023; Yang et al., 2019). Second, by exploring different and potentially opposing pathways to right-wing voting, we seek to uncover the distinct and possibly conflicting motivations driving low- and high-SES individuals to support right-wing parties. These motivations reflect the separate economic and social dimensions of political preferences.

We use novel data that consists of large representative samples from nine European countries. Most existing studies used self-placement on a left-right political orientation scale as a proxy for right-wing voting preferences (Dvir-Gvirsman, 2015; Imhoff et al., 2022). However, previous research suggests that while SES shapes political attitudes, these attitudes are not always associated with complementary voting behavior (Brown-Iannuzzi et al., 2015). Therefore, we opted for a more covert measure of left-right ideology that reflects actual voting intentions. It is worth noting that this measure assesses voting intentions on a general left-right scale that does not distinguish between economic and cultural dimensions of political preferences. However, our study provides valuable insights into the different motives underlying people’s political preferences by examining distinct mediators that are linked either to the economic dimension (economic system justification and economic threat) or the cultural dimension (symbolic threat, national identification) of voting.

The sample consisted of 11,217 respondents (52% females, M_age = 48.95, SD_age = 16.02) who were 18+ years old and living in Belgium (1,215), Denmark (1,215), France (1,201), Greece (1,213), Hungary (1,215), Netherlands (1,202), Poland (1,529), Spain (1,215), or the UK (1,212). The sample was stratified using quotas based on gender, age, education, and region (NUTS-1). The distribution of these quota variables is based on Eurostat 2020 figures. Three attention checks were included in the survey. Respondents who failed more than one of these attention checks were screened out and not included in the data. The procedure was different in Poland, where all respondents who failed an attention check were screened out and excluded. The sample size, hypotheses, and statistical analyses were preregistered (https://osf.io/v3cdm). The exploratory analysis regarding the mediating role of immigration threats was not preregistered. Fieldwork was carried out from 8 December 2021 to 12 January 2022 using Ipsos online panels and Pollster Research Institute. Ipsos granted access to the data on 1 February 2022. The data and the survey items are available at the Open Science Framework repository.

The present research was part of an international project (UNDPOLAR-NORFACE) that focused on understanding and addressing political polarization in European societies (Kuppens et al., 2024). Using this dataset, other researchers identified psychological factors that predict affective polarization (Turner-Zwinkels et al., 2025). To ensure that our findings are not limited to specific socioeconomic and geopolitical conditions, the project included countries that are relatively affluent (Belgium, Denmark, France, Netherlands, UK) and less affluent (Greece, Hungary, Poland, Spain) and countries that are located in Western (Belgium, Denmark, France, Netherlands, UK,) Eastern (Hungary and Poland) or Southern (Greece, Spain) Europe. At the time of measurement, in most of these countries (the Netherlands, Belgium, France, the UK and Greece) center-right to right-wing governments were in power. In Hungary and Poland, right-wing populist and national-conservative political parties were in power. Lastly, in Spain and Denmark, center-left to left-wing coalition governments were in power.

The survey included around 100 items which covered a wide range of social and political attitudes and behavior. In the present study, we limited our analysis to the variables presented below.

We conducted regression analyses using the PROCESS macro in SPSS (Hayes, 2022). Hierarchical regression analyses of total effects (c), direct effects (c′), and bootstrapped bias-corrected 95% confidence intervals of the indirect effect (ab) were computed with 5000 bootstrapped samples (Preacher & Hayes, 2008).

To test our preregistered hypotheses, we fitted all hypothesized mediators in one model (Figure 1). Specifically, we conducted serial multiple mediation analyses (Model 80) to examine whether the influence of SES (i.e., level of education and subjective income) on right-wing voting intentions is mediated by economic system justification (Path 1), national identification (Path 2), economic threat, symbolic threat, and anti-immigrant attitudes (Path 3). In Path 3, we hypothesized a serial mediation whereby perceived threats mediate the relationship between SES and voting intentions through anti-immigrant attitudes. Therefore, the variable of anti-immigrant attitudes was used as a second mediator in the multiple serial mediation. Although we did not hypothesize a mediating effect of economic justification in the relationship between education and voting intentions, for reasons of symmetry, we included all aforementioned mediators when examining both education and subjective income effects on right-wing voting intentions.

To account for country-specific, gender, and age effects on the variables of interest, we included dummy variables for each country along with gender and age were included as covariates. We also tested our model separately in each country (see Lois et al., 2025S) to identify differences across countries. Using the PROCESS macro in SPSS, we directly compared the explanatory power of the different mediation paths by performing pairwise contrasts of the indirect effects (see Lois et al., 2025S). This contrast is based on the estimation of the 95% bootstrap confidence intervals of the difference between two indirect effects that have similar sign/direction (Hayes, 2022). Additionally, we conducted an exploratory multiple mediation analysis to address whether perceived immigration threat accounts for the indirect effect through anti-immigrant attitudes (see Lois et al., 2025S). In this exploratory analysis, immigration threat was included as an additional parallel mediator in the first step of the serial mediation model (Model 80). Lastly, we performed the same analysis using an economic scale of left-right voting intentions instead of a general left-right scale (see Lois et al., 2025S).

High-educated individuals are less likely to vote for right-wing parties (b = -.129, SE = .016, t(1,8718) = -7.83, p < .001). On the other hand, higher subjective income is associated with intentions to vote for right-wing parties (b = .137, SE = .028, t(1,8718) = 4.96, p < .001). Moreover, anti-immigrant attitudes (b = .694, SE = .025, t(1,8713) = 27.77, p < .001) and economic system justification (b = .527, SE = .029, t(1,8713) = 18.01, p < .001) strongly correlated with right-wing voting intentions. Perceived symbolic threat (b = .226, SE = .023, t(1,8713) = 9.85, p < .001) and identification with the national group (b = .125, SE = .019, t(1,8713) = 6.50, p < .001) also displayed a positive, albeit weaker, correlation with right-wing voting intentions. In contrast, perceived economic threat negatively predicted right-wing voting intentions (b = -.133, SE = .026, t(1,8713) = -5.14, p < .001). In the next section, we test the hypothesized mediating paths separately for education and subjective income.

We performed a serial multiple mediation analysis (Model 80) to test and directly compare the mediating role of economic system justification (Mediator 1a), national identification (Mediator 1b), economic threat (Mediator 1c), symbolic threat (Mediator 1d), and anti-immigrant attitudes (Mediator 2) in the relationship between education and right-wing voting intentions. We observed a negative total effect of education on right-wing voting intentions. As depicted in Figure 2 and Table 1, all simple and serial mediating paths resulted in small negative indirect effects of education on right-wing voting intentions, except those through economic threat.

Consistent with H2a, the analysis revealed a negative, albeit small, indirect effect of education on right-wing voting intentions through national identification. Consistent with H4a and H6a, we observed a negative indirect effect of education on right-wing voting through perceived symbolic threat (simple mediation) and through symbolic threat and anti-immigrant attitudes (serial mediation). Contrary to H3a and H5a, we found no indirect effects (simple or serial mediation) through perceived economic threat. Lastly, although not initially hypothesized, the negative indirect effect of education on right-wing voting intentions through economic system justification was also significant.

Importantly, our analysis revealed a negative indirect education effect through anti-immigrant attitudes independent of all other mediators. Overall, we observed a negative indirect effect of education on right-wing voting intentions (b = -.093, SE = 0.007, 95% CI [-0.108, -0.078]). Controlling for all the mediators, the direct negative effect of education on right-wing voting was substantially smaller than the total effect, but it remained significant (b = -.036, SE = 0.015, t(1,8713) = -2.39, p = .017).

The pairwise contrast of indirect education effects revealed that the independent path through anti-immigrant attitudes is the largest negative indirect effect on right-wing voting. Specifically, the path through anti-immigrant attitudes was significantly larger than the indirect path through symbolic threat (ES = .033, SE = 0.006, 95% CI [0.022, 0.044]) or the path through economic system justification (ES = .033, SE = 0.006, 95% CI [0.021, 0.044]). In turn, the indirect path through symbolic threat (ES = .005, SE = 0.002, 95% CI [0.002, 0.009]) and the path through economic system justification (ES = .011, SE = 0.003, 95% CI [0.006, 0.016]) were larger than their respective serial indirect paths through anti-immigrant attitudes. The negative indirect effect through national identification is smaller than the path through symbolic threat (ES = .009, SE = 0.003, 95% CI [0.004, 0.015]) and economic system justification (ES = .009, SE = 0.003, 95% CI [0.003, 0.016]).

We performed a similar serial multiple mediation analysis (Model 80) to test the hypothesized mediating paths for the relationship between subjective income and right-wing voting intentions. We observed a positive total effect of subjective income on right-wing voting intentions. As depicted in Figure 3 and Table 2, all simple and serial mediating paths, except from those through symbolic threat and anti-immigrant attitudes, showed positive, albeit small, indirect effects.

Consistent with H1, we observed a strong positive indirect effect of subjective income on right-wing voting intentions through economic system justification. Contrary to H2b, we observed a positive indirect effect of subjective income on right-wing voting intentions through national identification. H4b and H6b are supported as we observed a negative indirect effect of subjective income on right-wing voting intentions through symbolic threat and a negative serial indirect effect through symbolic threat and anti-immigrant attitudes. It is worth noting, however that these negative indirect effects are cancelled out by the overall positive effect of subjective income on right-wing voting intentions. Contrary to H3b and H5b, the mediation analysis revealed a positive indirect effect of subjective income on right-wing voting intentions through economic threat and no serial indirect effect through economic threat and anti-immigrant attitudes. Similar to education effects, the analysis revealed a strong negative indirect effect of subjective income through anti-immigrant attitudes independent of all other mediators.

In total, the mediation model yielded a positive, albeit relatively small, indirect effect of subjective income on right-wing voting intentions (b = .063, SE = .014, 95% CI [.035, .091]). Controlling for all the mediating paths led to a reduced, albeit still highly significant, positive effect of subjective income on right-wing voting (b = .074, SE = .026, t(1,8713) = 2.81, p = .005).

The pairwise contrast of indirect income effects revealed that economic system justification is the largest positive indirect effect of subjective income on right-wing voting intentions. Specifically, the indirect effect through economic system justification is significantly larger than the indirect effect through economic threat (ES = .059, SE = 0.010, 95% CI [0.040, 0.079]). In turn, the positive indirect effect through economic threat is significantly larger than the indirect effect through national identification (ES = .020, SE = 0.007, 95% CI [0.031, 0.043]). All the aforementioned positive indirect effects are significantly larger than their respective serial indirect effects through anti-immigrant attitudes. The negative indirect effect through symbolic threat does not differ in size from the independent negative indirect effect through anti-immigrant attitudes but is significantly larger than the serial indirect effect through symbolic threat and anti-immigrant attitudes (ES = .014, SE = 0.005, 95% CI [0.005, 0.023]).

We tested the revised serial multiple mediation model (model 80 with immigration threat) separately in each country to identify countries that deviate from the dominant pattern reported in the entire sample. All countries displayed a total negative indirect effect of education on right-wing voting intentions which was mainly mediated by immigration threats and anti-immigrant attitudes in most countries, except Poland and Spain. Nonetheless, national identification (for Spain, Poland, and Greece), economic threat (for Hungary), symbolic threat (for Spain), and economic system justification (for the UK) also contributed to the negative total indirect effect. These differences may reflect the distinct meaning that people ascribe to left and right political parties or more tangible differences in the political agendas of left and right political parties across these countries.

In all nine countries, we observed a robust positive indirect effect of subjective income on right-wing voting through economic system justification. However, there were substantial differences across countries in the other pathways. In Greece, Hungary, and Poland, the indirect effect of subjective income on right-wing voting through economic threat was strongly positive suggesting that lower subjective income in these countries leads to left-wing voting through perceived economic threat. On the other hand, rich countries in the present sample displayed negative indirect effects through immigration threat (Belgium, the Netherlands, and France) and symbolic threat (Denmark). These differences resulted in a positive total indirect effect for the first cluster of countries and a negative or no total indirect effect for the second cluster (except from France).

Despite the growing economic inequality, redistributive policies such as progressive taxation and welfare programs are not very popular among the public and, even more surprisingly, among low SES individuals (Mijs, 2021; Piff et al., 2018). At the same time, far-right populism is on the rise and issues regarding national borders and the refugee crisis have come to the forefront of the political debate (Kende & Krekó, 2020; Messing & Ságvári, 2021). In light of these developments, a crucial contribution of the present study is the identification of immigration threat as the most quantitatively relevant explanation of why low SES individuals vote for right-wing parties.

The uniqueness of the immigration issue may pertain to recent changes in the supply side of European politics since the refugee crisis of 2015 (Chouliaraki et al., 2017). Far-right political parties responded to the recent migration waves by either cultivating or capitalizing on the growing anti-immigrant sentiment which depicted immigration as the primary threat for the country’s economy and culture (Kende & Krekó, 2020; Messing & Ságvári, 2021). Even mainstream political parties on the right adjusted their agendas by becoming more skeptical about the increased migration flows (Davis & Deole, 2017).

Our results suggest that unspecific economic and symbolic threats are not major explanatory factors of the relationship between SES and voting probably because these threats can be attributed to different actors (e.g., politicians, the economic system, globalization, or immigrants) leading to different outcomes (e.g., collective action, political cynicism, left or right voting) (Becker et al., 2011). In this respect, a strong narrative in which perceived economic and symbolic threats are clearly attributed to immigrants may fuel anti-immigrant sentiments and increase right-wing voting. However, this anti-immigrant narrative from mainstream media and leading political actors is hugely based on insufficient or unsubstantiated evidence. In this respect, a major challenge is to restrict these false or misleading claims and to promote factual information about how social welfare works by debunking zero-sum beliefs about economic and cultural resources (Komendantova et al., 2023).

Another important implication of this work pertains to the negative link between education and right-wing voting, which can create a negative feedback loop between growing economic inequality, limited opportunities for education of low SES, and reduced support for redistribution policies expressed through voting. In highly unequal countries, equal access to public education can break this negative cycle by acting as a buffer to the deleterious effects of growing economic inequality (Manstead, 2018). In a similar vein, the opposite effects of income and education on right-wing voting intentions, as well as their distinct psychological mechanisms, suggest that activating income or education identities can crucially shape political attitudes and actual voting behavior. In this respect, a political discourse that emphasizes the unequal distribution of resources and renders disadvantaged identities more salient can impede the tendency of low SES individuals to vote against their material interests (García-Castro et al., 2023).

In light of the growing inequality and the recent migration flows, we propose a research agenda that goes beyond individual-level psychological explanations by investigating the joint impact of country-level variables, such as the degree of inequality and net migration rates, and individuals’ SES on political preferences. Such a scientific endeavor combined with the use of longitudinal data can provide valuable insights into how contextual factors interact with subjective perceptions to shape political preferences over longer time periods.

Our large and representative samples from multiple European countries provide us enough statistical power to test complex mediation models. Furthermore, given the declining impact of the traditional left-right divide on voting (Evans & Mellon, 2016), we opted for a measure of left-right ideology that reflects actual voting intentions compared to self-placement in a political orientation scale. In doing so, we hope to avoid issues related to how coherent is the meaning of the political “left” and “right” across individuals and across Western and Eastern European countries (Piurko et al., 2011). However, we encountered problems with omissions, as a substantial portion of the sample (17.6%) indicated no intention to vote or an intention to cast a blank vote.

We acknowledge that mediation analysis on cross-sectional data cannot establish causal links or distinguish between alternative causal models (Bullock & Green, 2021; Fiedler et al., 2011; Preacher, 2015). We also recognize that we cannot rule out the effect of third confounding variables or a reverse causality issue. For example, supporting a far-right populist party may increase susceptibility to its anti-immigrant discourse, thereby reinforcing anti-immigrant sentiments (Kende & Krekó, 2020). Experimental mediation approaches (Pirlott & MacKinnon, 2016) and longitudinal studies are more suitable methodologies to address these issues and provide more robust evidence of the underlying psychological mechanisms. Nevertheless, our model remains valuable for understanding why low-educated and high-income individuals support right-wing parties given that the proposed pathways are grounded on previous theoretical knowledge of how threats, identities, and beliefs influence political behavior. Although the effects of income and education are difficult to test through experimental or longitudinal designs, future studies could manipulate threats, identities, and beliefs to examine the causal effects of our suggested mediators on right-wing voting.

The comparison of the main SES effect on voting with the indirect effects clearly suggests that the proposed mediators contribute substantially to the main effects. Specifically, the indirect education effects account for 72% of the main education effect on right-wing voting account, while the indirect income effects account for 46% of the main income effect on right-wing voting. However, it is important to note that the total indirect effects of income and education were small and we observed significant direct effects, especially for subjective income indicating that other important psychological mechanisms were not considered. Future work should incorporate additional factors such as relative deprivation, right-wing authoritarianism, and social dominance orientation, which have been previously linked with left-right political preferences and may differ across socioeconomic strata (Kunst & Obaidi, 2020; Napier & Jost, 2008; Urbanska & Guimond, 2018). Moreover, subjective and objective measures of SES are often weakly correlated and reflect different psychological mechanisms. In our study, income was assessed using a subjective measure, while education was assessed through an objective measure. In this respect, our study is limited by the absence of an objective measure of household income, which could have ruled out the possibility that the differing patterns observed with income and education are due to the more objective or subjective nature of their assessment.

Another limitation, common in large cross-country studies, is that some key variables in our model (i.e., national identification, perceived threats, anti-immigrant attitudes and subjective income) are composed of single-item measures, which exposes the results to potential biases regarding the validity and reliability of the instruments. Nonetheless, we preregistered all the selected variables which have been widely used in the literature and are deemed as representative of the constructs of interest.

We provide evidence that immigration threat and anti-immigrant attitudes are the most quantitatively relevant mediators of the negative link between education and right-wing voting and this mechanism is independent of unspecific economic and symbolic threats. Furthermore, economic system justification is the most important mechanism explaining the positive relationship between subjective income and right-wing voting. Taken together, our study constitutes the first systematic quantitative assessment of the psychological mechanisms that explain how people of different educational level and subjective income cast their vote.