Swaying to the Extreme: Group Relative Deprivation Predicts Voting for an Extreme Right Party in the French Presidential Election

Why do people vote for the extreme right (ER)? Despite considerable evidence suggesting the role of group relative deprivation (GRD) in accounting for prejudice, collective action and support for protest movements, there is surprisingly little research that has tested the impact of various types of relative deprivation in explaining the support for the ER. Using a large and representative sample of the French population tested before the 2012 presidential election, we hypothesised and found that GRD is a better predictor of the intention to vote for Marine Le Pen, the ER candidate, than individual relative ­deprivation.­GRD­remained­a­significant­predictor­of­voting­for­the­ER­even­when­controlling­for­social­ dominance orientation and prejudice, while it did not predict self-placement on the left-right political continuum. Thus, the sense that the French as a group are unjustly treated compared to immigrants living in France underpins the vote for the ER but not, as we demonstrate, for any other populist party. We discuss how the rhetoric of the ER parties can appeal to voters and expand their base over and above the support coming from those who are overtly prejudiced. Data and supplementary materials [DOI: https:/

(IRD) when self is perceived to be worse off in relation to others, or group relative deprivation (GRD) when one's own group is perceived to be less fortunate than another group. Therefore, a typical measurement of IRD involves people thinking about their personal situation in comparison to that of other people like them, whereas for GRD, it involves comparing the ingroup to some other target outgroup and indicating the extent to which they judge their situation to be better or worse off (see Abrams & Grant, 2012;Pettigrew et al., 2008). Previous research has examined how various types of relative deprivation correlate with each other, and with relevant outcome variables (see Dambrun & Guimond, 2001). Both types of relative deprivation predict a range of personal and behavioural responses (Abrams & Grant, 2012;Hafer & Olson, 1993;Kawakami & Dion, 1993;Smith & Spears, 1996;Walker & Mann, 1987). Higher IRD is associated with poorer individual-level outcomes such as academic achievement or psychological stress (Smith at al., 2012). The GRD, on the other hand, is a particularly robust and relevant concept in the explanation of collective responses (Abrams & Grant, 2012;Grant, 2008;Kawakami & Dion, 1995;Walker & Mann, 1987) and intergroup attitudes (Dubé & Guimond, 1986;Guimond & Dambrun, 2002;Pettigrew et al., 2008). A study utilising three representative European data sets showed that GRD and not IRD predicts anti-immigrant prejudice, even when controlling for more distal predictors of prejudice such as education or age (Pettigrew et al., 2008). Thus, the finding that perceiving one's own group as being worse off than another group can motivate forms of prejudice is well established (for a recent review, see Anier, Guimond & Dambrun, 2016).
Linking GRD to political behaviours, Abrams and Grant (2012) found that GRD and not IRD predicted intention to vote for the separatist Scottish National Party by empowering people to believe that social change is possible (see also Grant, Bennett & Abrams, 2017). Specifically, it was the affective reaction to an unjust treatment of Scottish people in comparison to English people that was responsible for the desire to split from the UK. IRD, on the other hand, had no impact on political change beliefs. Similarly, Marchlewska et al. (2018) manipulated the long-term group-level disadvantage, assuming it evoked GRD, and found that it was responsible for higher likelihood of voting for Brexit in the UK sample (Study 2), whereas self-report GRD in the US was associated with a willingness to vote for Trump over Clinton in the most recent presidential elections (Study 3; see also Pettigrew, 2017). It appears then that feeling deprived in relation to some other groups may motivate individuals to support parties that address these feelings of deprivation in their political rhetoric. Thus, the major hypothesis put forward in the present research is that GRD better predicts voting for the extreme-right wing than IRD.
There is surprisingly little research that directly bears on this hypothesis. Of course, voting for Trump over Clinton or supporting the Brexit cannot be easily equated with voting for the ER. Trump was a candidate of the Republican party which has never been historically a party endorsing explicitly an extreme-right wing ideology such as the FN in France (Mayer & Perrineau, 1996). Similarly, although the campaign for the Brexit was supported by some ER leaders such as Nigel Farage, it was also strongly led by some members of the Conservative Party who are not usually considered to be a party of ER. Despite the increasing interest for the concept of populism and the tendency among some to equate populism with the ER, there is little evidence in support of such an equation (Van Assche et al., 2018). In fact, among 30 most salient populist parties in Europe, only 14 parties (equal to 42%) are further classed as ER (Polk et al., 2017). Populism, the idea that governments are elitist and not serving their citizens rightly, can be found on the ER but also on the extreme left (De Cleen, Glynos & Mondon, 2018). For example, in France, FN can be considered as both populist and ER while parties such as Parti de Gauche (PG), at the time led by Jean-Luc Mélenchon, can be classified as both populist but on also on the left of the political spectrum (see Figure S1 in the supplementary materials; https://osf.io/c3k9y/).

Voting for the ER
Issues surrounding immigration and rejection of cultural diversity play a central role in the ER party programmes (Rydgren, 2007(Rydgren, , 2008, unlike left-wing populist movements. The French FN policy, for example, calls for 'national preference' in housing, employment, and state benefits to those who are of French citizenship. In this way, FN's policy encourages preference to the French citizens above the non-citizens. Within the French political landscape, the FN has a long history of being associated with and seen as a party of the extreme right (Mayer, 2017).
The link between anti-immigrant attitudes and voting for the ER parties is probably one of the most well-established findings in the literature. Decades of research in France have shown that authoritarian views, characterised by support for conventional values, social stability and submission to authority, and unfavourable attitudes towards immigrants predict support for FN (Lubbers & Scheepers, 2002;Mayer, 2017;Mayer & Perrineau, 1992). Similar findings were obtained in the United Kingdom (Cutts et al., 2011), Switzerland (Green et al., 2016) and Austria (Aichholzer & Zandonella, 2016). However, independently of the impact of authoritarian tendencies (Meloen, van der Linden & de Witte, 1996), Cornelis and Van Hiel (2015) have argued that support for the ER may also reflect a strong desire to dominate other groups or social dominance orientation (SDO; Sidanius & Pratto, 1999). Using the data from the European Social Survey conducted in seven European countries in 2002, they found support for a model suggesting that SDO has an indirect effect on voting for the ER, mediated by anti-immigrant prejudice (see also Van Hiel & Mervielde, 2002). This was a remarkable finding given the established role of authoritarianism in ER research. Since then, the relationship between SDO and voting for the ER has been further supported (Van Assche et al., 2018). Thus, ER parties may increase their vote by incorporating an anti-immigrant rhetoric which stems from wider systemic beliefs about the places of the group in the society, which in turn promotes the ER vote.
However, little is known about whether GRD represents an additional process potentially contributing the explain voting for the extreme right.
As intergroup relations (us versus the immigrants) are central to the politics of the ER parties, social comparisons are embedded in this context, in part, to make a positive distinction for own group (Tajfel et al., 1971). These comparisons may ultimately lead to feelings of relative deprivation, whereby people feel that they are worse off in contrast to some other individuals or groups. The rhetoric regarding the threat that immigrants pose to the economy is another thread running through the ER party promises, despite little evidence that indicators such as unemployment objectively contribute to the rise in the ER support (Lubbers & Scheepers, 2001. Thus, ER politicians portray immigrants as the cause of the declining economy at the cost of the national citizens. It is for this reason specifically that higher levels of GRD, a feeling that the immigrants may be better off than the national ingroup, may predict the likelihood of voting for the ER. In fact, higher levels of gross domestic product are often associated with a higher ER preference (Lucassen & Lubbers, 2012, see also Mols & Jetten, 2016). Thus, citizens of more prosperous countries and regions are more (and not less) likely to breed support for ER candidates, showing that objective economic perceptions cannot explain an appeal of the ER.
Indeed, across Europe, electoral support for ER appears to be higher in countries that are doing well economically whereas ER is still marginal in countries that were hit the most by the economic crisis (e.g., Spain and Portugal, Zick, Küpper & Hövermann, 2011). However, while ER voters tend to live in countries that are objectively doing fairly well economically, subjective perceptions of economy should also be considered. Following the first round of the presidential election in Austria, one voter of the ER Freedom Party of Austria said: 'Everything is given to them, whereas me, in the meantime, I have to struggle with €1300 a month. It has been 30 years that I work like crazy and get insulted in underground tunnels by foreigners' (Gauquelin, 2016). This voter's sentiment appears to point to not only relative deprivation, but it also captures both the pessimism and the economic struggle at individual level. Therefore, while IRD may not be expected to explain political behaviour, other negative perceptions of the economic prosperity could perhaps better explain why people may be attracted to vote for an ER candidate.

The present study
Collectively, the literature is in favour of the relevance of anti-immigrant prejudice in the ER vote. Moreover, GRD is known to be linked to anti-immigrant prejudice and to a range of politically oriented behaviours but has never been tested as a distinct predictor of voting intentions for the extreme right in national elections. Research investigating the specific voting intentions for nationalist and populist movements suggests that GRD is important in those contexts, but these are not always ER movements in their nature, which limits their comparability. Moreover, studies investigating the support for ER typically look at the party support or vote in the context of parties from the entire political continuum. The present study extends on this work by considering the role of GRD in the vote for an ER candidate, Marine Le Pen, in France. Like many other European countries, France employs a two-round system whereby in the second round, people are asked to vote for one of the two candidates who received the highest share of votes in the first round. This means that citizens may be asked to cast a vote for someone who may not be their first preference. It is important, therefore, to investigate what would encourage individuals to vote for the ER candidate over a representative from another party.
By operationalising the intention to vote for ER as relative to willingness to vote for another mainstream candidate in this setting, therefore, may be of higher validity than the traditional party support or the self-placement on the left-right political continuum measures. Although the choice between only two candidates is often studied in the US context, whereby most of voters support either a Republican or a Democrat candidate, it is rather different from the European politics where there is a wide range of political parties representing multiple positions from extreme left to extreme right to choose from. In European politics, research rarely focuses on the outcomes of a two-round system. The present research addresses this shortcoming.
In terms of our goals, the present research sought support for the hypothesis that GRD and not IRD predicts vote for the ER candidate. We also tested the extent to which GRD, but not IRD, may uniquely predict voting for the ER candidate independently of the effects of antiimmigrant prejudice and SDO (Cornelis & Van Hiel, 2015). Given that theoretically and empirically, GRD is geared toward the action mode (Abrams & Grant, 2012;van Zomeren, Postmes, & Spears, 2008), a direct link between GRD and the ER voting intentions would be expected. We hypothesised that a higher GRD would be related to a higher likelihood to vote for an ER candidate, even when controlling for other established predictors of ER support such as anti-immigrant prejudice and SDO. Moreover, we sought to determine the extent to which the role of GRD may be specific to a given political action, such as casting a vote for the ER, and may not necessarily extend to more general self-placement on the left-right political continuum. Finally, given the potential importance of an array of economic perceptions in motivating the ER vote, we aimed to explore the effects of a range of economic perceptions, involving comparisons with the present and the future, on the likelihood to vote for the ER. While based on the previous research we assumed that these economic perceptions may be related to the ER vote, we had no a priori hypotheses regarding these analyses.

Participants and procedures
The data in this study was collected by the French TNS Sofres polling agency using computer-assisted telephone interview method (see Bassili & Fletcher, 1991 for method description). 1 The survey was conducted in February 2011, several months before the 2012 French presidential elections. The total sample (N = 1001) consisted of 52.3% of females and 47.7% of males aged between 18 and 89 (M = 48.25, SD = 17.60) and was selected based on demographics to ensure the representativeness of French population. In terms of education, the sample consisted of 36.4% of people with no education or primary education only, 23.2% who have reached secondary level education, 15.4% with a university degree and 24.1% with a graduate or a professional degree (0.9% undeclared). Participants also declared their net monthly salary with 11.3% earning less than €1000, 14.5% earning between €1000 and €1500, 24.2% earning between €1500 and €2500, 32.1% earning between €2500 and €4000 and 11.7% declaring to earn more than €4000. The rest of the sample (6.2%) did not declare their salary. Missing data consisted of less than 3% of total cells. Participants with missing data were excluded in the relevant analyses. The survey was presented as being conducted by an independent research institute, not affiliated to any media or political organisation. It consisted of 53 questions divided across four categories (social issues, religion, politics and economy). The data in the study was collected in accordance with the provisions of the World Medical Association Declaration of Helsinki and all participants gave verbal consent prior to participating.

Measures
Intention to vote ER. As the study was conducted several months before the presidential elections, it was not clear how it would unfold exactly. Nicolas Sarkozy was finishing his five-year term as a president and was preparing his campaign, expecting to be a strong candidate. But,Marine Le Pen of the ER, galvanised by a series of electoral successes at the local level, was also a serious contender. In this context, participants in the survey were requested to respond how likely it is that they would vote for Nicolas Sarkozy if he was designated for the second round of the presidential election against the ER candidate, the FN Marine Le Pen on a scale from 1 (very likely) to 4 (very unlikely). Thus, higher score indicated higher intention to vote for the ER candidate. Details for this and all other measures of the present study can be found in the supplementary materials via the Open Science Framework (https://osf.io/c3k9y/). Self-placement on the left-right political continuum. This was measured simply by the self-report of placement on the political continuum from 0 (extreme left) to 10 (extreme right). Party preference was also noted whereby participants indicated which of the 13 French political parties they felt the closest to, with a full range of parties from the extreme left-wing Nouveau Parti Anticapitaliste to the ER party FN. This measure was recoded into seven groups whereby similar parties were categorised together (1 = extreme left, 2 = left-wing, 3 = socialist, 4 = green/eco-friendly, 5 = centre-right, 6 = rightwing, 7 = ER) for descriptive purposes.
Populist left versus populist right support. On the basis of data from Polk et al. (2017, see Figure S1 in the supplementary materials), party preference variable was further recoded to separate populist left-wing party supporters (PG; n = 51) and populist right-wing supporters (FN supporters; n = 40). SDO. A six-item adaptation of Sidanius and Pratto (1999) SDO scale was employed in the present study, a French version thoroughly validated in previous research (Duarte, Dambrun & Guimond, 2004). Participants responded to items such as 'Some groups are simply inferior to other groups' and 'Group equality should be our ideal' (reverse-coded) on a scale from 1 (do not agree at all) to 4 (completely agree). Higher score indicated higher antiegalitarian attitude (a = .68). 2 Anti-immigrant prejudice. Participants responded to eight items such as 'Today, there are too many immigrants in France' and 'I would have no problem moving into a neighbourhood where many immigrants live' (reversecoded) on a scale from 1 (do not agree at all) to 4 (completely agree). The scale was adapted from scales used in past research in France and in other countries (Berry & Kalin, 1995;Dambrun & Guimond, 2001;Zick et al., 2011). Higher score indicated higher anti-immigrant prejudice (a = .79).
IRD and GRD. To measure relative deprivation at individual and group level, single-item measures from Eurobarometer survey (used extensively in previous research utilising representative samples for its ease of administering; see Pettigrew et al., 2008) were adapted to the French context. The main advantage of such a strategy is that our results can be directly comparable to that of previous research linking relative deprivation with ethnic prejudice. To measure IRD, participants were asked 'If you were to compare your personal economic situation in comparison to that of most French people, would you say your economic situation is…?' 1 = much better to 5 = much worse. To measure GRD, participants responded on the same five-point scale to the following item: 'If you were to compare the economic situation of the French people to that of immigrants living in France, would you say the economic situation of the French people is…?' A higher score in both cases indicated higher feelings of deprivation.
Economic pessimism. For exploratory purposes, three items measuring perceptions of economic situation were also included. For two of them, participants rated France's general economic situation and personal economic situation on a scale from 1 (very good) to 5 (rather bad), whereas the third item related to the perceptions of one's future economic prospects with a response on a five-point scale (whether it will 1 = strongly improve to 5 = strongly degrade). Thus, for all items, a higher score indicated more pessimistic economic perception.

Results
The dataset is available via Open Science Framework: https://osf.io/c3k9y/. Mean and descriptive statistics broken down by party support are in the supplementary materials (Table S1) while zero-order correlations are presented in Table 1.

Validity of the ER vote measure
As expected, self-placement on the left-right political continuum was strongly related to party preference, r(868) = .62, p < .001. To test the validity of the measure of the intention to vote for Marine Le Pen, we examined the relations between this measure and both self-placement on the left-right political continuum and party preference. In terms of party preference, a one-way ANOVA supported the validity of the measure, yielding a large effect size, F(6, 839) = 36.05, p < .001, η p 2 = .21. Participants affiliated with the FN display the strongest intention to vote for Le Pen (M = 3.55, SD = .85, see supplementary materials for descriptive statistics for other party groupings). This confirms that responses to the measure do indicate the relative likelihood of supporting the ER candidate. The measure of voting intention was also significantly correlated with left-right placement r(901) = -.21, p < .001, confirming that it relates to the political orientation of the respondents. However, the correlation is relatively small and negative, meaning that placing oneself further on the right was related to less likelihood to vote for Le Pen, demonstrating that placing oneself more on left did not automatically mean a vote for Sarkozy just because he is posited closer to the left than Le Pen. The modest size of this correlation indicates that voting for the ER in a presidential election is clearly distinct from more general ideological preferences.

Validity of the measures of IRD and GRD
Previous research has consistently shown that GRD is a better predictor of anti-immigrant prejudice than IRD. If our measures are valid, we should find the same. Simple correlational analyses indicate that GRD is related with IRD, r(984) = .23, p < .001 and that both GRD and IRD predict prejudice, r(892) = .25, p < .001,and r(896) = .11, p = .001 respectively. Partial correlation analyses indicate that GRD remained correlated with prejudice when IRD was controlled, r(888) = .23, p < .001 but the reverse was not the case, r(888) = .05, p = .11). These findings are in line with previous research.

Swaying to the ER
Our central hypothesis was that a vote for Marine Le Pen and the ER is motivated by GRD, not IRD. To initially test this idea, mean levels of reported GRD were compared across political party supporters (see Figure 1). On scale from 1 to 5, French citizens reported a mean level of GRD of 2.48 (SD = .81). One-way between groups ANOVA showed that there was a significant effect of a political party on GRD levels, F(6, 873) = 7.91, p < .001, η p 2 = .05. As indicated by the Bonferroni-corrected post hoc tests, FN supporters were the only group that displayed significantly higher levels of GRD in comparison to all other groups (all ps < .002, 95% CIs [from .14, 1.07 to .48, 1.31]. Thus, the French FN voters felt more collective deprivation than any other political party supporters, providing initial support for the importance of GRD. Intention to vote ER. To further evaluate our central hypothesis, we conducted a multiple regression analysis with voting ER intention as the outcome variable. In Step 1 of the analysis, we entered SDO and anti-immigrant prejudice as the predictor variables. While higher SDO was associated with higher intention to vote ER, r(866) = .11, p = .002, this effect was supressed when anti-immigrant prejudice was entered into the model (see Table 2), meaning that the positive relationship between SDO and the ER vote was mediated by higher feelings of anti-immigrant prejudice. 3 This finding replicates the results of Cornelis and Van Hiel (2015), providing further evidence for the validity of our measure of ER voting.
Next, GRD and IRD were added as predictors of the ER voting intention in Step 2. As shown in the first column of Table 2 (Vote ER), higher GRD predicted the likelihood of voting for the ER candidate, Marine Le Pen, b = .14, t = 3.78, p < .001, but not IRD, b = .03, t = .74, p = .463. Importantly, this effect of GRD occurred independently of participant's levels of anti-immigrant prejudice and SDO. In other words, the findings indicate that there is a direct effect of GRD on voting intentions, which further supports our hypothesis.
For exploratory purposes, three additional variables, namely the perceived economic situation in France, the perception of one's own current economic standing, and the perception of one's future economic standing, were included in Step 3 (see bottom of Table 2). In the ER vote model, the effect of GRD as predictor prevailed after  adding the new variables, b = -.11, t = -3.78, p < .001, but perceptions that the French and own personal economic situation is rather bad predicted more likelihood to vote ER, b = .16, t = 4.59, p < .001 and b = .12, t = 2.84, p = .005, respectively. Moreover, thinking pessimistically about one's future economy (i.e., that it will degrade in the next few years) also further predicted a vote for the ER candidate, b = .14, t = 4.01, p < .001. Furthermore, in Step 3, the role of prejudice in the ER vote diminished even more, b = .09, t = -2.14, p = .032, compared to Step 1. This demonstrates that feelings of being deprived and negative perceptions of current and future economy were as important, if not more so, than anti-immigrant prejudice itself in influencing the ER vote. Overall, the model predicting the ER vote explained 12.6% of the variance F(7, 800) = 16.51, p < .001. We conducted relative weight analysis using relaimpo R package (Grömping, 2006) to compare which of the predictors were statistically the most important for voting ER. Anti-immigrant prejudice, SDO and GRD were all equally important predictors of ER vote (see supplementary materials for R code and output). 4 Self-placement on the left-right political continuum. To understand which variables uniquely underpin the motivation to vote for the ER, as opposed to being generally supportive of an extreme-right ideology, the same regression analysis model was conducted this time using the self-placement on the left-right political continuum as the outcome variable. As shown in the second column of Table 2, once SDO and prejudice were controlled for, GRD did not predict the left-right political continuum, b = -.03, t = -.76, p = .450. Therefore, GRD uniquely predicted vote for the ER candidate over a mainstream right-wing candidate and not the left-right self-placement.  In terms of IRD, it predicted self-placement on left-right continuum, b = -.13, t = -3.78, p < .001, whereby identifying as more right-wing was associated with less IRD. In sum, Step 2 analyses confirm that GRD and not IRD predicts an ER vote. 5 Similar to the previous model predicting ER vote, in Step 3, we added three additional exploratory variables. Negative perceptions of France's general economy, b = -.15, t = -4.57, p < .001, personal economy b = -.10, t = -2.42, p = .015 and pessimistic thinking about one's future economy, b = -.10, t = -2.95, p = .003 were all associated with lesser likelihood to self-identify as right wing. In other words, it was the self-identifying left-wing supporters who were more likely to perceive the economy in more negative terms. Overall, the model predicted 19.6% of the variance, F(7, 823) = 29.59, p < .001. Relative importance analysis revealed prejudice to be the strongest predictor of self-placement on the left-right political continuum (see supplementary materials).

Populism or ER?
Although the data set used in the present study is limited in terms of participants who typically support PG or FN, the two populist parties on the opposite poles of the left-right political continuum, we further decided to compare these supporters on the key outcome variables. The logic guiding this analysis is that if the outcome variable (e.g., GRD or prejudice) is a key element of populism, but not of ER, the participants will not significantly differ on these dimensions. A series of multiple independent t-tests were run to compare mean scores on the key outcome variables between PG and FN supporters. There were large differences between PG and FN among three variables. FN supporters, i.e., the right-wing populists, scored higher on anti-immigrant prejudice, SDO and GRD than the leftwing PG supporters (see Table 3 for statistics). This lends further support to the thesis that anti-immigrant prejudice, SDO and GRD are predictors of the support for the right-wing populists and not populist parties in general. Furthermore, there were also medium-to-large differences between PG and FN supporters in IRD, with FN scoring on average higher than PG supporters. Economic percep-tions, on the other hand, appear to be less important for the right-wing populist voters as FN supporters did not differ as much from PG supporters on the perceptions of own economical situation or their pessimism towards the future of the country's economy. There was, however, a small difference in perceptions of the country's current state of economy, with FN supporters perceiving it slightly more negatively than PG supporters. As such, populist voters, independently of their leaning to the left or to the right of political spectrum, may share more concerns to do with the economy.

Discussion
Why do people vote for the ER candidates? Given the increasing electoral success of ER political movements in various countries, this question is perhaps more important now than ever before. With a representative sample of the French population, the present research showed that in addition to prejudice and antiegalitarian attitudes, GRD is a significant factor to consider in ER voting. Indeed, considering all seven major political party categories in France, our findings showed first that the FN supporters, that is supporters of the long-standing ER party in France, stand out as being more likely to vote for Le Pen and reporting higher levels of GRD than supporters of any other party. This suggests that GRD and voting for the ER are linked. Subsequent analyses strongly confirmed that higher GRD predicted higher likelihood of voting for Le Pen over the mainstream right-wing candidate, Nicolas Sarkozy, even when controlling for other powerful predictors such as anti-immigrant prejudice or SDO. This indicates, consistent with past research on GRD in other domains (e.g., Abrams & Grant, 2012), that a deep feeling of collective injustice involving a comparison between the national ingroup and immigrants is a strong motive behind voting for the ER in France. On the other hand, however, our analysis showed that there were large differences between populist left-wing and populist right-wing supporters in their mean levels of GRD, suggesting that feelings of deprivation may not necessarily underpin populist movements as such, as suggested by Marchlewska et al. (2018), but rather that it may be unique to those who sway towards ER candidates. Together, GRD is a distinct predictor of the ER vote alongside anti-immigrant prejudice, but it does not predict left-right political placement. This highlights the importance of feeling group-level deprivation in pushing voters to consider an ER candidate, even if they may not normally identify as right-wing.
Furthermore, in line with Cornelis and Van Hiel (2015), we found that SDO indirectly influences vote for the ER. While the results of Cornelis and Van Hiel (2015) for France were retrospective, based on the party that people voted for in the past, the voting intentions in the present research were prospective, indicating the probability of voting for the ER candidate as oppose to the outgoing president, Sarkozy. Thus, our findings using the novel measure of the ER vote considerably reinforce those reported previously. Nevertheless, some caution is needed when considering these results because of their correlational nature and because some additional variables that we did not consider could alter the findings. For example, we did not measure authoritarian attitudes in the present research. Thus, while Cornelis and Van Hiel (2015) found that the effect of SDO is independent from authoritarianism, we are unable to conclude whether the effects observed here would be sustained even after controlling for authoritarian attitudes. Finally, we provided some exploratory evidence that negative perceptions related to the personal and country's economic situation as well as pessimistic outlook for the future economy further predicted the ER vote. In other words, feeling that things are not going the right way economically nor that they will improve in the future can motivate voters to turn towards the ER. However, economic concerns were also salient among left-wing populist voters meaning that thinking pessimistically about the economic situation may be an important factor that may persuade citizens to vote for a populist party, independently of their left or right leaning. This is in line with arguments presented by Inglehart and Norris (2016), who advocated for the role of economic insecurity in the rising popularity of populist parties. Indeed, economic insecurities may best be characterised by the quadratic model whereby these fears are higher among the extreme left and the ER identifiers as shown by van Prooijen, Krouwel, Boiten and Eendebak (2015). Admittedly, however, this line of research requires further investigation to understand whether there are aspects of economic insecurities that are specific to left-versus right-wing voters.
The present study also makes a distinct contribution by contrasting predictors of general political self-placement on the left-right continuum with those of voting intentions in the context of a specific election. While our measure of the intention to vote for an ER candidate in contrast to another popular candidate is not a typically used way of tapping into political voting behaviour, we found that it is strongly related to party preference, confirming its face validity, but that it is more modestly linked to the selfplacement on the left-right political continuum. The data showed that the variables underpinning the vote for the ER differ from those predicting the self-placement on the political spectrum. Indeed, political self-placement and the actual vote in the election may not completely correspond; the ER candidates may swing some voters that have previously not identified with the ER ideology. This raises the critical issue of how ER movements may expand their base and eventually seize power through democratic means. The current findings highlight that a typical rightwing self-identifier tends to be more prejudiced against immigrants and less likely to think negatively about the economy in comparison to a left-wing self-identifier. However, these are not necessarily the factors that drive the intention to vote for an ER candidate in a national election. Marine Le Pen, the ER candidate, was more likely than Nicolas Sarkozy, a right-wing candidate, to receive support from French people who felt collectively deprived in comparison to the immigrants living in France and from those who felt negatively about the present and future economy, independently of their level of anti-immigrant prejudice.
Therefore, the ER parties may expand their base and appeal to voters through the political rhetoric utilising GRD and by instilling negative perceptions of the economy, not only by blatant anti-immigrant prejudice. This demonstrates why the ER parties may gain some votes from people who otherwise would not support them. The ER candidates appeal to voters by constructing the reality of disadvantage, promising that no citizen will longer feel left behind in comparison to other groups such as immigrants. While this statement may not be factually correct, once the belief is instilled, it encourages people to seek out cases whereby a French citizen gets a 'raw deal' in comparison to an immigrant to confirm their views (i.e., it is prone to the confirmation bias; Nickerson, 1998) and eventually becomes sufficient to produce a strong image of group deprivation. In that way, ordinary citizens who normally are not prejudiced against immigrants can be attracted by the ER candidate.
These results provide novel insights into issues that should be of concern to social and political psychologists such as the question of how to tackle the current trends in increasing political polarisation (e.g., see Pew Research, 2014). As political ideologies become more divisive, there is a pressing need to develop ways in which people who do not share political views can have a productive discussion regarding the way political systems can serve their societies as a whole. The present research demonstrates that those inclined to vote for the ER candidates cannot be dismissed as doing so just on the grounds of having negative attitudes towards immigration. The ER party programmes may well attract some individuals who do not usually consider themselves as right-wing. Therefore, some of the concerns faced by the ER voters can be shared with those on the left of the political spectrum who also feel more negatively about the country's economy. This offers opportunities for opening a dialogue and decreasing partisanship.

Limitations
Despite several contributions, the present research has many limitations. Firstly, the problem with researching the ER is that they are sometimes also populist in their political agenda (see Polk et al., 2017). This makes it difficult to distinguish whether the variables investigated truly predict ER vote or whether populism predicts a part of the variance, as the FN's political rhetoric consists of both populist and right-wing elements. Future research should devise ways in which these can be distinguished, for example, by contrasting these variables to the likelihood to vote for left-wing populist parties.
Secondly, the final model predicting the ER vote accounted for 12.6% of the variance, meaning that there are still other factors that can contribute to our understanding of the vote for the ER candidates. In contrast, models explaining the self-placement on the left-right political continuum was stronger, with around 20% of the variance explained. One of the reasons for that is because the vote for Front National's Marine Le Pen was compared in relation to the outgoing president, Nicolas Sarkozy, who is also considered to represent right-of-centre politics. On the one hand, framing the political choice in this manner is useful to identify factors that may be specific to the ER. On the other hand, there is a need for more research to replicate these findings in context whereby the second candidate is more centre or left, something that was not addressed in the present study. In addition, we did not consider whether participants would vote at all. If given a choice of two candidates that do not fit within one's political preferences, it is possible that one may abstain from voting. Our findings are not able to account for this.
Thirdly, the explanations relating to the importance of negative perceptions of the economy need to be further tested and developed as the results presented here are exploratory. There is a clear scope for political voting theory development as these perceptions of the economy predicted voting behaviour as well as self-placement on the left-right political spectrum when IRD could not account for it. Clearly, not only group-level perceptions motivate political behaviour, and these should be better integrated with the individual-level factors. It is a further limitation that we only employ single-item measures to capture these economic perceptions. While the same shortcoming regarding single items could be argued to apply to IRD and GRD items, the main difference is that there are both theory and research supporting the validity of the measures of relative deprivation. For example, in the present study, anti-immigrant prejudice was related to GRD, but less so to IRD in line with Pettigrew et al. (2008). Nevertheless, further research should improve the measures by using scales such as those developed by Dambrun, Taylor, McDonald, Crush and Méot (2006). Better measurements may also allow for uncovering effects of relative gratification, the reverse of relative deprivation, something that we failed to address in the present research (see Anier et al., 2016). Indeed, some research suggests that relative gratification also may fuel anti-immigrant prejudice (Mols & Jetten, 2016), but it is not yet clear how it may affect voting behaviour for the ER candidates. Finally, correlational analyses conducted with cross-sectional data are not sufficient to establish cause and effect and thus, future research should experimentally confirm the role of GRD in voting for ER candidates in context of elections or employ a longitudinal design.

Conclusion
With the rising popularity of the ER parties across Europe, it is necessary to continue to develop models which can robustly explain why the ER candidates are so appealing, even to those who previously would not consider themselves as right-wing. This study is the first to demonstrate that, in addition to anti-immigrant prejudice, group feelings of deprivation in comparison to immigrants and concerns about the economy can directly motivate the ER vote. As such, more research is needed to study the role of GRD in explaining why people support the ER.

Notes
1 Parts of the findings from this survey were also previously published by Guimond, Streith and Roebroeck (2015).