Psychological bases of support for radical right parties

Article history: Received 18 November 2015 Received in revised form 13 February 2016 Accepted 25 February 2016 Available online 10 March 2016 We examine the social–psychological and personality bases of support for radical right parties (RRPs), using cognitive–motivational approaches of ideology. Our comprehensive model includes core ideological variables which mediate personality traits (Big Five) or how individuals engage in social relationships and accommodate novel stimuli. Structural equation models were tested in an Austrian population sample to examine support for a RRP, the FPÖ. Our results suggest that a “perceived immigrant threat” and, in part, social dominance orientation are directly related to RRP support, whereas right-wing authoritarianism has consistent indirect (mediated) impact. Associations with lower Openness to Experience, lower Agreeableness, and to some extent also with Conscientiousness are mediated by the ideological variables. The conclusion discusses how RRPs' success and communication strategies can be linked to basic psychological motivations. © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
In recent years, Europe has witnessed growing electoral success of radical right parties (RRPs). As a "party family" RRPs are commonly characterized by their authoritarian beliefs, the return to traditional values, anti-immigrant and xenophobic stances, i.e., preference for an ethnically homogeneous population, as well as in-group/out-group thinking that portrays the existence of threats (e.g. Rydgren, 2007). Hence, the focus on grievances concerning immigration is considered a core feature of the RRP profile (Ennser, 2012;Ivarsflaten, 2008).
In addition, despite a large body of literature on basic personality traits as a factor in partisan orientation, few studies have attempted to link psychological traits (e.g. Big Five) to preference for RRPs (Zandonella & Zeglovits, 2012), extreme right-wing parties (Schoen & Schumann, 2007) or populist parties more generally (Bakker, Rooduijn, & Schumacher, in press). Furthermore, a coherent theoretical framework that links social-psychological factors of ideology and personality to core RRP stances is largely missing in the literature.
In the present study, we anticipate that voters gravitate toward RRPs when they: (a) exhibit authoritarian attitudes (right-wing authoritarianism, RWA), i.e., motivational goals to seek group security and stability in societal order (Altemeyer, 1981;Duckitt, 1989); (b) exhibit competitively driven motivations to maintain hierarchical or superior-inferior relations between social groups (social dominance orientation, SDO) (Pratto, Sidanius, Stallworth, & Malle, 1994); and (c) perceive social threats to identity and cohesion induced by immigration and, hence, exhibit motivations to reduce that uncertainty and threat (Jost, Glaser, Kruglanski, & Sulloway, 2003). Finally, we propose that (d) these attitudinal factors fully mediate basic personality traits (Big Five) that might predispose individuals to uphold stability in social relationships or make them less open to new social situations or stimuli (see DeYoung, Peterson, & Higgins, 2002).
After specifying our hypotheses, we analyze our theoretical model by using unique representative survey data from Austria. Our outcome variable is preferences for the Freedom Party of Austria (Freiheitliche Partei Österreichs), FPÖ, one of the most successful RRPs in Europe.
2. Radical right party support: its social-psychological and personality bases

Radical right party support and its relation to RWA and SDO
It is well established that basic cognitive-motivational goals drive our ideological orientations, namely advocating vs. resisting social change and rejecting vs. accepting inequality or RWA and SDO, respectively (see Duckitt & Sibley, 2009;Jost, Federico, & Napier, 2009, for an overview). Consistent with this framework, RWA and SDO are also believed to play a role in satisfying epistemic and existential motivations, namely reducing uncertainty and threat (Jost et al., 2003).
Following Altemeyer (1981), the main perceptional and behavioral consequences (or lower-level structure) of RWA are: (1) to accept and to adhere to authorities as well as to social norms ("submission"); (2) to approve of and demand the punishment of people who deny the legitimacy of these authorities or deviate from these norms ("aggression"); and (3) to be sensitive to threats to a given social order ("conventionalism"). We thus anticipate that motivational goals of RWA foster RRP support as these are vital characteristics of RRP stances.
In turn, SDO expresses competitively driven motivations to maintain or establish group dominance and superiority, i.e., people high on SDO support intergroup hierarchies and tend to arrange social groups in a superior-inferior order. Thus, SDO should predict a person's acceptance or rejection of ideologies and policies relevant to group relations (Pratto et al., 1994). We therefore expect SDO to be positively related to RRP preference.
Even though RWA and SDO can be moderately to strongly positively correlated (Roccato & Ricolfi, 2005), these factors represent distinct predictors of numerous sociopolitical and intergroup attitudes, especially political orientation and forms of prejudice (Duckitt & Sibley, 2009;Sibley & Duckitt, 2008). However, previous evidence suggests that SDO, rather than RWA, might be more important for party preferences or more directly related to them (Cornelis & Van Hiel, 2015).

Radical right party support and perceived immigrant threat
To explain RRP support, we further consider a "perceived immigrant threat" (PIT), i.e., individuals' perception that immigration threatens their personal or the majority's societal value system, culture, social cohesion, or alleged ethnic homogeneity. Indeed, other authors have referred to this tendency as "cultural conflict" dimension (Kriesi et al., 2008), a "normative threat" (Stenner, 2009), or "symbolic threat" (Stephan & Stephan, 2000). Previous research suggests that this type of threat seems to matter most for RRP support (Lucassen & Lubbers, 2012;Oesch, 2008), even more so than economic or "material threats" (on this distinction see Lucassen & Lubbers, 2012;Stephan & Stephan, 2000), or that these types of threat by immigrants are not empirically distinguishable (Lucassen & Lubbers, 2012). According to a core component of RRPs' discourse, their supporters seek to mitigate perceived threats linked to immigration (PIT).

RWA, SDO, and social threat
In a nutshell, Duckitt and Sibley's (2009) model suggests that scoring high in RWA makes individuals more sensitive or responsive to types of social threat. Indeed, research has shown that authoritarians are more responsive to threatening messages (e.g. Lavine et al., 1999) or that (extreme) right-wing individuals show stronger psychological, but also physiological responses, to negative or threatening stimuli (e.g. Hibbing, Smith, & Alford, 2014). We thus anticipate that RWA is an important antecedent of PIT.
Another main hypothesis in Duckitt and Sibley's (2009) theoretical approach is that the relationship between RWA and political behavior (e.g. party preference) is, at least partly if not fully, mediated by perceived threats (i.e., PIT). As a consequence, RWA would only have an indirect effect on RRP support. SDO, on the other hand, is expected to be connected less strongly, if at all, to societal-level threats or normative threats (Onraet, Van Hiel, Dhont, & Pattyn, 2013). Instead, it will be related to threats explicitly activating competitiveness over relative superiority and dominance of groups. We nevertheless test, but do not expect mediation of SDO on voting preference via PIT.

Mediation of personality traits by ideological attitudes
Relationships between personality and political preferences are likely not to be direct, but rather indirect or mediated by ideological variables. In particular, RWA is assumed to have a unique foundation in personality, including social conformity traits or a combination of Conscientiousness and lower Openness to Experience (e.g. Brown, 1965;Duckitt & Sibley, 2009). In turn, Emotional Stability should be negatively related to feelings of threat and insecurity and could thus diminish RWA. SDO seems to be primarily related to lower Agreeableness (or lower trust, altruism, or compliance), higher Conscientiousness, and lower Openness to Experience (Sibley & Duckitt, 2008). PIT or prejudice might be rooted in traits that make people less capable of adapting to new stimuli and social situations, traits that entail lower levels of altruism or tolerance in social relationships, or traits that make them more anxious (see Brown, 1965). We anticipate that Openness to Experience, Agreeableness, and Emotional Stability are negatively associated with PIT, while Conscientiousness is positively associated with PIT (e.g. Gallego & Pardos-Prado, 2014;Sibley & Duckitt, 2008). Table 1 and Fig. 1 provide a summary of our hypotheses and the underlying model for our empirical analyses.

Case
To test our hypotheses we use data from Austrian voters and support of the FPÖ. In terms of its policy profile, the FPÖ is considered comparable to other RRPs in Europe (Ennser, 2012) and voting patterns are considered similar to other RRPs (see Aichholzer et al., 2014;Lubbers et al., 2002). In the last Austrian federal election 2013 the FPÖ gained 20.5% of the valid votes.

Data
The data used in this study were collected by the Austrian National Election Study (AUTNES) Pre-and Post-Election Survey (panel) in 2013 (Kritzinger et al., 2013. The pre-election interviews were conducted face-to-face (CAPI) and the post-election interviews by telephone (CATI). Respondents were sampled from the Austrian population eligible to vote (i.e., aged ≥16), using an address-based multistage stratified clustered sample and selection of a random respondent within each household (total n = 3266, pre-election response rate: 61.8%; post-election n = 1504 or 46.1% re-interviewed). The sample composition was: 49.1% male; age: M = 45.3 years, SD = 19.5 years; education: 22.1% compulsory schooling, 47.3% lower secondary/vocational training, 18.4% admission to tertiary education, 12.3% college/university degree. Data were weighted for post-stratification and sample design adjustment.

Measures
All measures except actual vote choice were administered in the preelection wave (see Appendix A for the exact question wording). For our dependent variable, we rely on two different operationalizations: first, we use the respondents' "propensity to vote" (PTV) for the FPÖ (preelection wave, see Appendix A for exact question wording), which is measured on a quasi-metric 11-point scale (0 = very unlikely, 10 = very likely, M = 2.75, SD = 3.17). This measure refers to the general affiliation with a party beyond the vote in a specific election. Secondly, we employ a binary measure of respondents' "actual vote" (post-election), which is coded 1 if respondents voted for the FPÖ and 0 for all other parties, no party/invalid-answers, and excluding voters of the FPÖ splinter group BZÖ (n = 28) to avoid any overlap (16.2% FPÖ). The Pearson correlation between PTV and actual vote was r = .52.
Big Five personality traits were measured using the German BFI-10 ( Rammstedt & John, 2007). Our analyses support the hypothesized five-factor structure of the items, although it is not regarded as perfectly "clean" (i.e., items have non-ignorable cross-loadings) (see Table A1 in the Appendix A).
We measure PIT based on five items that tap into the respondents' attitudes and emotions regarding immigrant and cultural threat (vs. enrichment), since emotional reactions as well as factual attitudes explain perceptions about immigration (Brader, Valentino, & Suhay, 2008). These are indicated by questions on concerns about (Muslim) immigrants as well as on respondents' specific emotions towards immigration, namely anger and anxiety. We also adjust for common method variance in the two questions capturing emotions (see Table A1).
Finally, we measure SDO with two items that strongly resemble the wording of the original SDO scale by Pratto et al. (1994) (see Appendix A). All Likert-type items use a fully labeled 5-point rating scale (1 = agree completely to 5 = disagree completely).

Analysis
We use structural equation models (SEM) to take into account unreliability or unsystematic measurement error in survey measures when analyzing the structural relationships between variables. Secondly, we take into account systematic acquiescence bias in agree-disagree items, using a response style factor as a control (Aichholzer, 2014). Thirdly, we allow for a "complex" (unrestricted) item-factor structure in the BFI-10 scale, applying ESEM (Asparouhov & Muthén, 2009), which is then held constant across dependent variables (see Table A1 for the measurement models).
Further, we intend to control for socio-demographic heterogeneity and add age, education (1 = admission to tertiary education), and gender (1 = female) as controls for all latent factors and RRP support (see Fig. 1, detailed results not presented).
All analyses were conducted in Mplus (Version 7) (Muthén & Muthén, 1998-2012, using linear SEM with full information MLR or WLSMV estimation for missing data. In order to evaluate the models' fit we review the goodness-of-fit indices CFI and RMSEA [90% CI] and interpret them jointly. A combination of cutoff values CFI N .90 and RMSEA b .08 is often regarded as acceptable and CFI N .95 and RMSEA b .05 is considered to be an excellent fit.

Results
Overall, each structural equation model displays at least an acceptable or a good fit to the data. Furthermore, all models exhibit an equally good fit when allowing for direct effects of the Big Five traits on RRP support. Thus, it is reasonable to assume full mediation of effects from these variables. In Table 2 we present direct and total (direct + indirect) effects with fully standardized coefficients (β) separately for the two dependent variables: PTV and actual vote for the FPÖ.
Our first empirical test examines how basic social-psychological factors of ideology are related to RRP support. With regard to direct effects, PIT strongly (β = .51 and .42) and SDO moderately (β = .22) influences the PTV for the FPÖ (no mediation), whereas the impact of SDO disappears when the respondents' actual vote choice is considered (β = .02, n.s.). One reason for this pattern could be panel attrition effects, since preliminary analyses suggest that the probability to remain in the panel increases with higher SDO but decreases with higher RWA, whereas PTV and PIT seem to have no impact. On the other hand, RWA has virtually no significant direct impact on RRP preference when including SDO and PIT, consistent with previous evidence suggesting negligible direct effects in the presence of the other variables. Nevertheless, it is evident that the effect of RWA is mediated by a strong relationship with PIT (β = .63 and .67), resulting in a highly significant positive total effect with RRP support (total β = .32 and .50), which is higher for actual voting. Finally, 36% or 39% of the variance in our dependent variables (R 2 ) can be explained by the explanatory variables in our model.
Our second empirical test investigates the relationship of personality traits with RRP support. Since our results corroborate full mediation via the ideological variables, we provide indirect (=total) effect estimates of the Big Five on RRP support. We find that the most consistent and moderate (negative) relationships with RRP support can be found for Openness to Experience (total β = −.17 and −.18) and Agreeableness (total β = −.12 and −.14), whereas Conscientiousness displays weak and inconsistent positive relationships with RRP support (total β = .05 (n.s., p = 0.061) and .09). Extraversion did not reveal significant or consistent relationships with the ideological constructs (apart from SDO: β = .08 and −.14), neither with RRP support. In turn, (lower) Emotional Stability only plays a moderate role for explaining RWA (β = −.15 and −.14), but is irrelevant for FPÖ vote preference.
Finally, our results confirm most of the existing evidence on Big Five associations with RWA, SDO, and PIT as well as on the relation between RWA and SDO (see Table 2 for detailed results). On the one hand, this lends support to the criterion validity of our measures. On the other hand, it helps us to clarify how personality traits are related to RRP support. To sum up, we find that scoring low in Openness to Experience is directly, and most clearly, related to RWA, to SDO, and indirectly to PIT, whereas Agreeableness negatively correlates with PIT and in part also with RWA, resulting in their indirect negative association with RRP support. Surprisingly, however, we do not find the hypothesized relationship between SDO and lower Agreeableness. In turn, Conscientiousness is substantially related to RWA, but not to PIT or SDO, thus being only partly related with RRP support.
With regard to socio-demographics and FPÖ vote preference, education was significantly negatively, though indirectly, related to FPÖ vote preference (total β = −.21 and −.28), because of consistent negative effects on RWA, SDO, and PIT (total βs = −.21 to −.33). In turn, higher age displays a moderate negative direct effect (β = −.20 and −.18), positive relationships with RWA, SDO, and PIT (total βs = .18 to .30), but virtually no total effect. Lower preference for the FPÖ among women only plays a direct role for PTV (total β = −.07), but not for actual vote.

Discussion
In this study, we provide a comprehensive picture of the interplay between psychological aspects in voters' preference for radical right parties (RRPs), including personality (Big Five), authoritarianism (RWA), social dominance orientation (SDO), and a "perceived immigrant threat" (PIT). Putting the scattered pieces of the puzzle in order, we were able to comprehensively address the individual differences and social-psychological mechanisms that drive RRP support. Here, we investigated the case of Austria with an established RRP in its party landscape, the FPÖ, using representative survey data.
To summarize, most of our initial expectations on the associations between the Big Five and RRP support were in line with the empirical evidence. Most importantly, we established indirect relationships (mediation hypothesis) which help us to understand why and how personality explains RRP support. Among basic social-psychological factors of ideology, PIT most clearly increases the likelihood of RRP support, SDO has a moderate direct positive impact for PTV, whereas the association with RWA seems to be indirect. There is reason to believe, however, that different conceptualizations and measurements of "authoritarianism" explain diverging evidence on its role in RRP support.
In theoretical terms, our findings largely support the notion of cognitive-motivational goals in individuals, such as managing uncertainty or threat as well as maintaining stability in societal order and intergroup hierarchies, which manifest themselves in political ideology and voting behavior (Jost et al., 2003(Jost et al., , 2009. In other words, individuals seek to support parties or politicians that match these goals. Furthermore, our results could be interpreted in the light of basic functions of personality traits, such as engaging in social relationships through trust and compliance, on the one hand, as well as the (in)ability of enjoying and processing novel stimuli and situations, on the other (see DeYoung et al., 2002). The exact personality origins of RRP support beyond the Big Five or its relation to their more narrow facets nevertheless deserve to be studied further.

Conclusion
What have we learned about the nature and the success of RRPs? With our proposed model we cannot explain the electoral success of specific RRPs at specific elections. However, learning about deeply rooted patterns in voters' political preferences adds new insights to electoral research, particularly in the light of insufficient explanatory power of more classical theoretical models (see Aichholzer et al., 2014). Furthermore, this also contributes to our understanding of how appealing political elites' and parties' political communication is to voters. Our results suggest that a successful RRP offers policy stances that deliberately address basic psychological motivations and cognitions that are rooted in voters' personalities and core ideological attitudes. For instance, triggering social threat and negative emotions associated with immigrants among voters seem to be effective in RRPs' voter mobilization (see also Brader et al., 2008).
Some limitations of this research must also be considered: first, our results ideally require replication. Austria may be special due to its particular cultural and historical context (e.g., WWII) when it comes to Note: Socio-demographic controls age, education, and gender included, but not shown. MLR or WLSMV estimates with standardized coefficients β (using yx-standardization), *p b .05, **p b .01, ***p b .001. R 2 values for binary variables refer to the latent response variable (y*). RRP = radical right party, PTV = propensity to vote, RWA = right-wing authoritarianism, SDO = social dominance orientation, PIT = perceived immigrant threat.
political ideology and RRPs. That said, the alignment of RWA and SDO in the ideological spectrum and with their antecedents can be contingent on the ideological contrast among parties in a country (see Roccato & Ricolfi, 2005). Second, a common limitation in large-scale representative surveys are measurement instruments. Specifically, short measures of complex traits are inherently restricted in their breadth and, hence, more extensive and cross-culturally invariant scales would be desirable. Third, future research may want to investigate factors that activate and mediate SDO, such as actual experiences of intergroup competition (e.g. labor market disadvantage) which, in turn, may foster RRP support. We also expect future research to further investigate the RWA lower-level structure, because its sub-dimensions may differ in how they relate to aspects of social threats and, hence, to vote choice. Further theoretical work and empirical tests are required to fully understand the role of psychological aspects in supporting RRPs.  Note: Results for Geomin rotation. RWA = right-wing authoritarianism, SDO = social dominance orientation, PIT = perceived immigrant threat. Full information MLR estimates, n = 3099, Highest factor loadings per row presented in bold numbers. A residual correlation (correlated uniqueness) between PIT4 and PIT5 adjusts for common question wording effects (r = .62). In order to capture common variance in RWA sub-dimensions, RWA2, RWA5, and RWA6 have correlated uniquenesses (CUs) as well as RWA3, RWA4, and RWA7.