This is an open access article distributed under the terms of the Creative Commons
Attribution License (This article aims to understand when and why individuals sympathize with radical right-wing parties. The electoral success of radical right parties throughout Western Europe have constituted a major reconfiguration of the political landscape, and research show that mainstream parties can influence the success of radical right parties (Arzheimer & Carter, 2006; Bale, 2003; Meguid, 2005, 2008). For instance, studies have shown that mainstream party accommodation of radical right parties may legitimate these parties, which can, under some circumstances result in electoral success (Arzheimer & Carter, 2006; Bale, 2003; Dahlström & Sundell, 2012). The aim of this study is to investigate if radical right parties gain from mainstream party recognition, and to what extent individual factors moderate this effect.
Theoretically, we draw on the political psychological literature that suggests that when explaining the legitimization of radical right parties, we need to consider that many individuals in Western Europe are influenced by an anti-prejudice norm when forming preferences towards such parties and their policy stances on immigration (Ivarsflaten, Blinder, & Ford, 2010). Previous research shows that such norms can explain individuals’ political behavior, for example, discriminatory practices (Blinder, Ford, & Ivarsflaten, 2013, 2019).
We hypothesize that when mainstream parties signal that it is acceptable to associate with radical right parties, they challenge the anti-prejudice norm that dissuade voters from sympathizing with such parties, leading to a higher likelihood that voters sympathize with the radical right. By focusing on individual-level internal motivation to control prejudice (IMCP) we attempt to uncover the mechanism at play. We hypothesize that those with lower IMCP are more susceptible to be affected by mainstream party recognition of radical right parties as those with high IMCP have a stronger internalized anti-prejudice norm. Even though an alternative explanation for voter sympathy for the radical right could be that mainstream party recognition and increased support are consequences of real efforts on the part of the radical right party to clean up their agenda, the focus on IMCP as a mediator implies that such alternative explanations are less likely, than a norm change. While previous research has shown that a strengthening of the anti-prejudice norm can reduce discriminatory practices (Blinder et al., 2013), our study is the first to systematically investigate how a potential weakening of the anti-prejudice norm influences sympathy for a radical right-wing party.
We evaluate the effects of changes in the normative context in a survey experiment (N = 1,133) where we manipulate mainstream party recognition of a radical right party, the Sweden Democrats, before the Swedish parliamentary election in 2018. Our results show that when mainstream parties recognize a radical right party, individuals are more likely to sympathize with that party. Moreover, this effect is moderated by individual-level internal motivation to control prejudice – individuals who are less motivated to control prejudice are more strongly influenced by mainstream party recognition of the radical right. The fact that IMCP plays an important role is a hint that this is a case of a weakening of the anti-prejudice norm and not a case of a cleaned agenda on the part of the radical right party.
Theoretical Framework
The Literature on Mainstream Parties and the Radical Right
In the infancy of research on the electoral success of radical right parties, explanations regarding the demand from voters dominated the field. A prominent example is Ignazi's (1992) idea of how a demand stemming from the losers of modernization explained the success of the radical right. As the research field matured it became apparent that demand side explanations alone could not explain the variation of the success of this party family across Europe (Mudde, 2007), which led to the study of supply side factors. In broad terms, supply side factors can be divided into two categories; those that focus on how the radical right parties themselves shape their fortune (Art, 2011; Loxbo & Bolin, 2016; Lubbers, Gijsberts, & Scheepers, 2002), and those that focus on external factors, such as institutional arrangements (Golder, 2003) and the behavior of other political actors (Arzheimer & Carter, 2006; Bale, 2003; Meguid, 2005, 2008).
As argued by several scholars, radical right parties do not act in a vacuum, they operate in a party system occupied by other parties whose decisions affect the success or failure of them both (Immerzeel, Lubbers, & Coffé, 2016). It is, however, debated in what way the actions of the mainstream parties affect the success of radical right parties (Goodwin, 2011). In a seminal study, Kitschelt (1997) illustrated the importance of other political parties’ behavior in explaining the success of the radical right. He found that the convergence of mainstream parties granted radical right parties the political opportunity necessary to succeed electorally.
Further, Meguid (2005, 2008) argued that mainstream parties have the choice of three strategies when confronted with the threat of niche parties, including radical right parties; a dismissive-, adversarial- or accommodative strategy. These strategies signify the different positions the mainstream party take on the issue(s) most closely associated with the niche parties, which in the case of the radical right is assumed to be the immigration issue. The different strategies of mainstream parties may lead to either an increase or decrease in the salience of these issues. Meguid (2005, 2008) found that the success of the niche party is decided by the strategy employed by both mainstream parties, to the right and the left, in combination. An important implication is that accommodative strategies of mainstream parties reduce the electoral support of radical right parties, under certain conditions (see also Van Der Brug, Fennema, & Tillie, 2005). The empirical support for isolation, often referred to as a cordon sanitaire, as a strategy to weaken the support for radical right parties, is mixed as it is dependent on various institutional and contextual settings (Van Spanje & Van der Brug, 2009).
The view that a mainstream party shift can deplete a niche party is however contested. When mainstream parties co-opt the position of radical right parties, their success is rather facilitated (Arzheimer & Carter, 2006; Bale, 2003; Dahlström & Sundell, 2012). In a study of Swedish municipalities, Dahlström and Sundell (2012) find that when both left and right take a tougher stance on immigration, the radical right benefits electorally. Other studies have found that traditional right parties and their position play a central role in facilitating the radical right (Arzheimer & Carter, 2006; Bale, 2003). The assumed mechanism at play when explaining how mainstream parties facilitate the radical right involves a legitimization of the radical position and, by extension, their existence (Arzheimer & Carter, 2006). It should also be noted that radical right parties can have a contagion effect. Mainstream parties have been found to adapt their position when the radical right contender is electorally successful (van Spanje, 2010).
Given the electoral success of at least some members of the radical right party family, it is apparent that the task of appearing legitimate has been successful. In his influential work, Taggart (1995) made this argument in his comparison between what he referred to as “The New Populism” and “Neo-Fascist Parties”. Parties such as FPÖ (Freiheitliche Partei Österreichs in Austria), Lega (then Lega Nord in Italy) and DFP (Danske Folkeparti in Denmark) represent what was referred to as “The New Populism” by Taggart. These parties are still active and successful to this day, and Taggart concludes that this party family was more successful than their “Neo-Fascist” cousins because they tempered their rhetoric and agenda with regards to immigration. Being blatantly racist and holding anti-democratic views (Carter, 2013) has not been a winning formula for political parties, as showed by the Neo-Fascist parties. In fact, successful populist radical right parties have worked hard to be perceived as legitimate. This was observed by Ignazi (2005) in his study of Alleanza Nazionale and Lega Nord in Italy.
To protect themselves from racist claims, successful populist radical right parties use ‘reputational shields’. These shields are used to fend off accusations of racism and is constituted of a legacy or broader policy package (Ivarsflaten, 2006). These reputational shields are useful as arguments in debates with political opponents, and especially for voters to justify why it is acceptable to support or sympathize with a party that hold views that are at odds with the norms of mainstream society. Such reputational shields, and mainstream party recognition of the radical right party, are likely to legitimate the radical right as a viable political alternative. In essence, such legitimation lowers voters’ threshold for expressing sympathies with the radical right party. Due to such societal norms of what is accepted behavior or not, legitimation of the radical right party will influence individual voters’ sympathy for the radical right party. Hence, social norms are pivotal to understand why individuals sympathize with the radical right. We thus hypothesize that:
H1: When mainstream parties signal that it is socially acceptable to associate with a radical right party, individuals are more likely to sympathize with the radical right party.
Research Focusing on Norms, Prejudice and Social Influence
Scholars have long recognized that norms play a vital part in understanding behavior (Cialdini, Reno, & Kallgren, 1990). The study of how norms influence phenomena such as racism have a long history in the United States. Events and developments during the civil right movement created a norm of racial equality in this previously racially divided society. In other words, the normative context was changed. Such a norm shift has been used to explain why parties changed their campaign strategies to include less racial messages (Mendelberg, 2017).
While it is important to note that not everyone who supports radical right parties do so because of racist motivations (Rydgren, 2008), an “anti-racism” norm has been observed in Western Europe (Ivarsflaten et al., 2010). The presence of this norm has been used to explain why populist radical right parties are unable to attract their entire voter base. “Radical right potential” describes the proportion of the electorate that hold views and share attitudes similar to the populist radical right (Norris, 2005; Van Der Brug et al., 2005). Even the most electorally successful parties belonging to this family are not even close to reaching their potential in terms of how large their voter base is expected to be.
So, why are the populist radical right parties sometimes unable to reach their potential? Blinder and colleagues (2013) suggest that the anti-racism norm keep people from acting out on their prejudice, which could include voting for a radical right party. The process of making political decisions, such as voting for a radical right party, is often seen as a dual process composed of an automatic and a controlled part (Devine, 1989). While many individuals hold prejudiced views, such as negative attitudes towards immigrants, they avoid turning these attitudes into a discriminatory action because a normative context, symbolized by the anti-prejudice norm, prohibits such behavior. Blinder et al. (2013) suggest that the inclination to break the norm is determined by individuals’ motivation to control prejudice.
Motivation to Control Prejudice is a psychometric concept developed to study how willing people are to comply with social norms (Dunton & Fazio, 1997). A person’s motivation to act in accordance with the normative context can be either internally or externally motivated. We focus specifically on the Internal Motivation to Control Prejudice (IMCP). IMCP has been successfully used to understand how internalized ambitions to avoid prejudice affect political attitudes and behavior by moving beyond social desirability (Blinder et al., 2013, 2019; Plant & Devine, 1998). While prejudice is widespread, people who are internally motivated to control it are more likely to be able to avoid acting on their prejudice. Those who are motivated can avoid controlled prejudice due to an internalized willingness, which has been successfully measured as Internal Motivation to Control Prejudice (IMCP) (Plant & Devine, 1998).
It should be pointed out that this is not a measurement of social desirability (Ivarsflaten et al., 2010; Plant & Devine, 1998). Individuals with high IMCP have internalized the motivation to avoid prejudice to the point where they do not, for instance, act more prejudiced in private than in public (Plant, Devine, & Brazy, 2003). Furthermore, individuals with an Internal Motivation to Control Prejudice have not been found to act more prejudiced if they are subjected to an environment which allows them to act upon their prejudice (Butz & Plant, 2009). In previous studies, norms and individual Internal Motivation to Control Prejudice have been used to explain discriminatory practices such as giving asylum seekers’ access to employment and the welfare state, discrimination of non-natives and supporting the radical right (Blinder et al., 2013). Individuals with a low IMCP are more likely to support the radical right and act discriminatory towards Muslims (Blinder et al., 2013, 2019; Harteveld & Ivarsflaten, 2018). Following the findings of previous research, we hypothesize:
H2: Individuals with a lower internal motivation to control prejudice (IMCP) are more likely to sympathize with the radical right party.
Not all individuals react the same way to contextual changes; individual factors moderate how an individual reacts to the signal that it is acceptable to sympathize with a radical right party. As mentioned, IMCP captures an internalized avoidance of prejudice and should not be confused with standard concepts of social desirability (Ivarsflaten, Blinder, & Ford, 2010; Plant & Devine, 1998). In other words, those with high IMCP are not expected to simply follow changing elite cues if it goes against their own internalized motivations to control their prejudice. Those with low IMCP, on the other hand, have shown to exert more discriminatory behavior when mainstream actors have signaled a weakening of the anti-racism norm (Blinder et al., 2013, 2019). Some individuals should be more prone to use such a signal as an “excuse” to express anti-immigrant attitudes or to sympathize with a radical right party. Specifically, we expect that individuals who are low in IMCP will seize such an opportunity. Low IMCP can be seen as individuals having a low threshold for attending to and processing information that is consistent with the individual’s (anti-immigrant) attitude position. Alternatively expressed, we expect that low motivation to control prejudice, which is a cognitive effortful process, should be related to more automatic and spontaneous responses. Such automatic responses to out-group members are usually more negative than the controlled responses are (Dovidio & Gaertner, 2010). Because many Europeans have negative attitudes to immigrants, not appearing prejudiced requires cognitive resources. However, if one is less concerned with complying with an anti-immigrant norm, learning that mainstream parties affiliate with an anti-immigrant party is likely to affect whether a voter sympathizes with that party. In sum, individuals with low IMCP are more susceptible to changing normative contexts compared to individuals with high IMCP who have a stronger internalized anti-prejudice norm. We thus hypothesize:
H3: When mainstream parties signal that it is socially acceptable to associate with radical right parties, lower internal motivation to control prejudice (IMCP) will increase the likelihood that an individual sympathizes with the radical right party.
While the argument that mainstream parties run the risk of facilitating radical right parties by employing an accommodative strategy is commonly echoed in the literature (Arzheimer & Carter, 2006; Bale, 2003; Tyrberg & Dahlström, 2018), causal evidence is sorely missing. Our experimental study with an analytical focus on norms and individual susceptibility to these norms is able test this argument in a controlled setting. Furthermore, while the interaction effect of a strengthened anti-prejudice norm and IMCP on welfare chauvinistic attitudes have been found (Blinder et al., 2013), no previous study has looked at how mainstream party recognition affects sympathy towards radical right parties in an experimental setting.
Method and Data
To test our hypotheses, a survey experiment was carried out in Sweden prior to the 2018 election. Before we describe the experiment, along with data and methods, we briefly describe the case of the Sweden Democrats (SD) and the specific context we focus on.
The Case of the Sweden Democrats and the Swedish Context
The story of the Sweden Democrats is not only a story of electoral success but also the struggle of appearing legitimate. The core of the initial iteration of the party, founded in 1988, consisted of a splinter group from various national groups among them Keep Sweden Swedish (Bevara Sverige Svenskt, BSS) and the Sweden Party (Sverigepartiet, SP). This core had clear ties to the Nazi movement in Sweden. The party acquired their first seat at the local level in 1991. Since its infancy, the party has aimed to rid itself of the bad reputation they had largely due to their history. These attempts include excluding members who had clear ties to Nazi or other questionable movements, or those who made blatantly racist statements, and changing internal policies such as banning uniforms from official gatherings (Larsson & Ekman, 2001).
The transformation of the party can be tied to their success. The election after the appointing Jimmie Åkesson as chairman, in 2006, should be seen as the breakthrough election of the Sweden Democrats. While not gaining any seats in the Swedish parliament, the Riksdag, the party was awarded at least one seat in almost half of the municipal councils in Sweden. Since entering the Riksdag, SD have increased in size, and in the 2018 election, a total of 17.5% of the Swedish electorate voted for the party, which made it the third largest party in the Riksdag.
Despite the Sweden Democrats’ electoral success, no mainstream party has officially cooperated with them at the national level up until 2018, and they were treated as a “pariah” by all other parties during the government negotiations of 2010 and 2014. The arguments used by mainstream parties not to co-operate with them range from differences in specific issues to the fact that the party has “[…] deep roots in fascism and racism.” (Gabrielsson, 2019). Most of the mainstream parties have hence treated the Sweden Democrats as a pariah party, but during the election year of 2018, some parties (the Christian Democrats and the Moderates), seemed more open to the possibility of cooperating, at least in parliament, with the Sweden Democrats.
The fact that the pariah status of the Sweden Democrats was relatively unclear during the election year of 2018 makes this specific contextual setting suitable for studying the effect of mainstream legitimization of the radical right. Given how established most radical right parties are throughout Western Europe, Sweden, and possibly Germany was, at the time of the study, the only countries where an experiment including a manipulation of mainstream party facilitation of the radical right could credibly be conducted.
Experimental Design and Data Collection
The experiment was set up to manipulate the likelihood of a parliamentary cooperation between the Sweden Democrats and mainstream parties after the 2018 national election. In total, 1,133 participants took part in the experiment. The experiment consisted of two treatment groups and a control group. The treatment group that will be referred to as “Treatment Group – High (recognition from mainstream parties)” consisted of 405 respondents who read a news article stating that mainstream parties are increasingly open to parliamentary co-operation with the SD. The other treatment group referred to as “Treatment Group – Low (recognition from mainstream parties)” (n = 368) read an article stating the opposite - that mainstream parties had no interest to co-operate in any way with the SD. We also included a control group (n = 363) who did not read any article prior to answering the survey.
The treatment was a fictive article, detailing the possibility of mainstream party cooperation with SD or a continued dismissal of such a cooperation. Specifically, the article stated that the (fictive) Swedish Opinion Institute (SOI) conducted polls both among the population but also among MPs, and their latest survey among MPs indicated that many MPs were increasingly positive/continued negative to cooperating with the Sweden Democrats. A short text followed in which the CEO of the Institute said that this could be due to that the Sweden Democrats had cleared up their agenda and were now to be considered a party among all the others, or that the Sweden Democrats had not cleared up their agenda and were still seen as an extreme party. An important feature of our manipulation is that the possible cooperation is based on individual MPs’ opinions and do not (necessarily) represent the official party line. This was done to enhance credibility since it would be difficult for participants to actually control the truth in the message when framed in terms of individual opinions. Also, we did not state how many MPs were positive, only that “more and more MPs…”.1 See Appendix 1 in the Supplementary Materials for a detailed description of the experiment and the texts used as treatments.
The data were collected by the Laboratory of Opinion Research (LORE) at the University of Gothenburg (www.lore.gu.se) in April 2018. LORE is a service organization for researchers that collects data via web surveys using their own panels. Their largest panel, the Citizen panel, contains more than 60,000 active respondents all over Sweden. The Citizen panel is used in the present research. LORE continuously publishes methodological notes on their web site as well as general information about each data collection. All surveys conducted by LORE are approved by the regional Swedish Ethical Review Authority. Information about the present data collection and replication of all models can be found at https://doi.org/10.7910/DVN/O6YBBY.
Measurement of the Dependent Variable
In the present study, sympathy for (or liking of) the Sweden Democrats (SD) constitutes our dependent variable (DV). Hence, we use an evaluative measure of partisanship (Thomassen & Rosema, 2009). The reasons for looking at this measure instead of, for instance, vote intention is twofold. The first reason is theoretical - the main interest of this article is how sympathy towards radical right parties is affected when mainstream parties challenge the anti-prejudice norm. While vote intention undoubtedly is an important measurement of support, in the case of radical right parties being liked or favorably evaluated is a better indicator of whether the legitimacy of the radical right party changes. That is, one can deem a party legitimate or like a party without voting for it. Our second motivation for using this measurement of radical right sympathy is methodological. Voter surveys are consistently under-reporting vote intention for radical right parties compared to actual elections results (Hooghe & Reeskens, 2007). This is largely due to social desirability bias. While this may still affect the measurement we use, there is a lower threshold for liking a radical right party (compared to voting for them).
The question about sympathizing with or liking the Sweden Democrats read: “Use the scale below to indicate how much you like or dislike the different parties”. Participants were then presented with a matrix form question listing all the parties in the Riksdag, and rated each party on a scale from -3 = Strongly dislike to 3 = Strongly like, with 0 = Neither like nor dislike. The DV – Liking the Sweden Democrats (SD), ranged from -3 to +3. Hence, higher values indicate more liking. The distribution is presented in Figure 1.
Figure 1
Distribution of the DV "Liking SD" (as a Continuous Variable)
As seen in Figure 1 below, close to 60 percent of the respondents indicated that they strongly dislike SD. Given that the distribution is so heavily skewed towards not liking SD, it is problematic to treat the variable as continuous. Instead, we re-coded it into two groups where answers 2 and 3 were coded as liking SD and the rest (-3 to 1) were coded as not liking SD. Across the sample, 13.4 percent of the respondents sympathize with SD given this operationalization.2
Measurements of the Independent Variables
The main independent variable was Internal Motivation to Control Prejudice. This was measured with three items: “I feel guilty if I have negative thoughts/feelings towards immigrants”; “I get angry with myself when I have a thought or feeling that could be considered prejudicial”, and “It bothers me a lot when I think I have offended someone, so I am always careful to keep other people's feelings in mind”. Answers ranged from 1 = Do not agree at all to 7 = Completely agree, and were combined into a mean index (Cronbach’s α = 0.75).
We also included a range of control variables.
Left-right political self-positioning was measured with the item: Sometimes people describe political orientation on a left- to right scale. Where would you place yourself on such a scale? Answers ranged from 1 = Far to the left to 7 = Far to the right. This variable was included since the success of the radical right is often seen as a story driven by cultural and/or economic grievances. On an individual level a negative view on immigration is often seen as important factor in explaining why a voter is drawn towards a radical right party (Mudde, 2007). While the electoral success is often seen in the light of a two-dimensional (or multi-dimensional) political landscape the traditional left-right cleavage is still pertinent, warranting the inclusion of a subjective left-right measurement (Golder, 2016).
Employment situation was measured by asking the question: Which of the following groups do you belong to? The answers ranged from employed to unemployed but also contained pensioners and those on sick leave. Previous research has found that outsiders and vulnerable insiders on the labor market are more prone to support radical right parties (Dal Bó, Finan, Folke, Persson, & Rickne, in press). We treat unemployed, part-time workers and those on long-term sick leave as vulnerable insiders or outsiders. This variable is treated as dummy variable in our models, vulnerable insiders and outsiders are coded as 1.
Education level was measured using 9 categories, ranging from no elementary school to post graduate education. The question asked was: What education have you undergone/what education are you undergoing? Education was included as a continuous variable in our models.
Both employment and education were taken into account since they can signal who is an outsider or vulnerable in the labor market, which is connected to sympathy with a radical right party (Golder, 2016).
Political interest was measured on a 4-point scale from 1 = Very interested, 2 = Fairly interested, 3 = Not very interested, to 4 = Not at all interested. In general, political interest is an important social and psychological factor that has been linked to various political behavior and attitudes (van Deth, 2014). More specifically, since we try to manipulate the political landscape, individuals with different levels of political interest could be more or less susceptible to our treatment.
Gender was coded with a 0 for a woman and 1 for a man. Gender was included because previous research has shown that men are more prone than women to support radical right parties (Harteveld & Ivarsflaten, 2018).
Anti-Immigration attitude were measured with the items: “Immigration threatens the Swedish welfare”; “Immigration increases the risk of terrorism in Sweden” and “Immigration increases crime in Sweden”. Answers ranged from 1 = Do not agree at all to 7 = Completely agree. The items were combined to a mean index and used as a control variable (Cronbach’s α = 0.91).
A manipulation control variable was included in the end. Given that the experiment was intended to manipulate if the voter thought that mainstream parties was interested in co-operating with the SD, the manipulation variable asked “How likely do you think it is that some of the other parties in the Riksdag will cooperate with the Sweden Democrats after the 2018 elections?” Answers ranged on a Likert-scale from 1 = Very unlikely to 7 = Very likely.
The participants first responded to the demographic questions, and the anti-immigration attitudes measure. Then they were asked the IMCP items before the experimental manipulation. Then, we assessed the outcome variable, liking of the SD, and finally the manipulation control.
Empirical Analysis
We first present descriptive results for all variables included in the study. To test if the manipulation was successful, we ran an ANOVA. In the main analyses section, we first present how the proportions of individuals liking the SD differ in the two treatment groups, and how liking of the Sweden Democrats differs for individuals high and low in IMCP (split at the median). The main analyses were regression analyses. Given that the DV is binary, we chose to run a Linear Probability Model regression (LPM).3 The LPM models are presented both a strictly additive model and a multiplicative model with an interaction term consisting of Treatment Group and IMCP.
Descriptive Results and Manipulation Check
Table 1 shows descriptive statistics for all variables included in the study. To test if the manipulation had worked as intended, a one-way univariate ANOVA with the manipulation control variable as DV and the experimental groups as between groups factor was run. The ANOVA test was significant, F(2, 1102) = 5.23, p = .006, η2 = .009. This indicates that at least two group means significantly differ and that the difference is likely not due to chance, but a consequence of a successful manipulation. Follow-up ANOVA tests between the groups showed that the treatment group high recognition differed significantly from both treatment low recognition, F(1, 750) = 8.72, p = .003, η2 = .011, and the control group, F(1, 743) = 7.07, p = .008, η2 = .009.4
Table 1
Descriptive Statistics
| Variable | Mean/ Proportions |
SD | n | Min | Max |
|---|---|---|---|---|---|
| DV: Liking SD | 0.13 | – | 1,133 | 0 | 1 |
| Internal Motivation to Control Prejudice | 3.88 | 1.59 | 1,108 | 1 | 7 |
| Anti-Immigration Attitude | 4.03 | 1.96 | 1,110 | 1 | 7 |
| Left-Right Placement | 3.87 | 1.55 | 1,127 | 1 | 7 |
| Gender | 0.58 | – | 1,130 | 0 | 1 |
| Political Interest | 1.67 | 0.64 | 1,132 | 1 | 4 |
| Employment Situation | 0.57 | – | 1,133 | 0 | 1 |
| Education Level | 6.38 | 2.02 | 1,128 | 1 | 9 |
Note. The mean is presented for continuous variables and proportions for binary variables.
The results show that in the treatment group high recognition, participants believed that some of the other parties in the Riksdag were more likely to cooperate with the Sweden Democrats (M = 4.54, SD = 1.76), compared to in the treatment group low recognition (M = 4.15, SD = 1.84), or in the control group (M = 4.18, SD = 1.93).
There is no significant difference between the control group and the low recognition treatment group (see Appendix 2, Supplementary Materials, for crosstab). In the low recognition treatment group, participants read an article stating that mainstream parties had no interest in cooperating with the SD. This situation in fact mirrored the real situation at the time of data collection, that is, no political party had officially co-operated with the SD at the national level. Given that there was no difference between the control group and the treatment group low recognition we combined these for the purpose of our multivariate statistical analysis.5
Main Analyses Predicting Sympathy With the Sweden Democrats
To test the hypotheses and gain an understanding of how mainstream party recognition affect individual sympathies with the Sweden Democrats, we start out by assessing the first two hypotheses (H1 and H2) descriptively. Our first hypothesis stated that in a situation where mainstream parties have signaled that they are willing to cooperate with the radical right party, individuals are more likely to sympathize with this former pariah party. Figure 2, shows that the proportion of individuals who likes SD is 16.3 percent in our Treatment Group – High in which we have manipulated mainstream party recognition of the SD. This should be compared to 11.8 percent when the SD is not recognized by mainstream parties. The difference between the groups is significant at the 95% level. Hence, this gives tentative support for H1.
Figure 2
Proportions of DV: "Liking SD" Across Treatment Groups
Note. CI = 95%. Two-sample proportion test, p = .032. Groups are Treatment Group Low/Control Group (n = 729) and Treatment Group – High (n = 404).
The second hypothesis was concerned with evaluating whether individual willingness to control prejudice affects the individual probability of sympathizing with a radical right party. In order to analyze this relationship descriptively we have split our sample at the median of the variable measuring Internal Motivation to Control Prejudice (the median is four in this seven-scale variable). While splitting the sample at the median of the variable comes with a loss of some nuance, it allows for a first descriptive look at the differences between groups, and here we follow a procedure that has been used previously (Blinder et al., 2013).
Figure 3 shows the differences in the proportion of the sample who like SD between individuals with low and high values in our variable Internal Motivation to Control Prejudice. The differences between these two groups are substantial. In the group of individuals with a low IMCP, 21.6% of the respondents like the SD while only 2.8% of those with high IMCP do that. In short, this descriptive look into the relationship between IMCP and sympathizing with radical right parties gives support for H2; individuals who are less inclined to control their prejudice are more prone to like radical right parties than those with a high IMCP.
Figure 3
Proportions of DV: "Liking SD" Across IMCP (Split at Median)
Note. CI = 95%. Variable IMCP split at the median (4). Two-sample proportion test, p < .0001). n = 1133.
In a next step, we evaluated our hypotheses in a Linear Probability Model regression using our dichotomous measure of Liking SD as outcome variable. Besides our main predictors, recognition of SD (treatment group), internal motivation to control prejudice (IMCP), and the interaction term treatment by IMCP, we included a set of control variables (gender, education, employment, left-right position and anti-immigration attitudes).
In Table 2 we present the results from our LPM models, both a strictly additive model (Model 1 in Table 2) and a multiplicative model (Model 2 in Table 2) with an interaction term consisting of Treatment Group and IMCP (logit model in Appendix 4, Supplementary Materials). As can be seen in Table 2, our treatment variable (Treatment Group – High, coded as 1) has a positive significant effect on the probability of liking SD. If mainstream parties signal that they are willing to cooperate with the SD in a parliamentary setting, individuals are 5.5% more likely to sympathize with the former pariah party. This result supports our first hypothesis (H1).
Table 2
Linear Probability Model (LPM) – DV: Liking the Sweden Democrats
| Variable | 1
|
2
|
||
|---|---|---|---|---|
| B | SE | B | SE | |
| Treatment Group – High | 0.055** | 0.018 | 0.193*** | 0.049 |
| Internal Motivation to Control Prejudice (IMCP) | -0.034*** | 0.006 | -0.021** | 0.008 |
| Anti-Immigration Attitude | 0.064*** | 0.006 | 0.064*** | 0.006 |
| Left-Right Placement | 0.027*** | 0.007 | 0.027*** | 0.007 |
| Gender | -0.012 | 0.019 | -0.016 | 0.019 |
| Political Interest | -0.047** | 0.014 | -0.045** | 0.014 |
| Employment Situation | 0.014 | 0.018 | 0.012 | 0.018 |
| Education Level | -0.019*** | 0.005 | -0.019*** | 0.005 |
| Treatment Group – High × IMCP | -0.035** | 0.012 | ||
| Constant | 0.086 | 0.061 | 0.038 | 0.063 |
| Observations | 1094 | 1094 | ||
Note. Treatment Group – Low & Control Group is the reference category.
*p < .05. **p < .01. ***p < .001.
In terms of H2, that states that individuals with a low IMCP are more likely to sympathize with radical right parties compared to those with high IMCP, the results presented in Table 2 support this hypothesis. As can be seen, there is a significant and negative effect of IMCP on Liking SD, such that a one-unit increase in IMCP lowers the probability of liking SD by 3.8%.
The effects of our main predictor variables are significant even when controlling for other potential explanatory variables. With regards to these control variables, anti-immigration attitudes and left-right placement have positive effects meaning that being critical of immigration and being right-leaning increases the probability of liking the SD. Being interested in politics and having higher education lowers the probability of sympathizing with SD, while gender and employment situation has no significant effects.
To test the conditional hypothesis (H3), which states that as the anti-prejudice norm is challenged by mainstream political actors, individuals with differing IMCP are affected in different ways when it comes to liking radical right parties, we have to look at the multiplicative model (Model 2 in Table 2). The interaction term (Treatment Group – High × IMCP) has a negative and significant effect. However, to interpret interaction effects, we need to turn to figures illustrating these (Brambor et al., 2006). Figure 4 shows the predicted probabilities of sympathizing with the SD based on the model presented in Model 2 in Table 2.
Figure 4
Predicted Probabilities: DV: Liking SD = Treatment Group – High × IMCP
Note. See Table 2 for regression model and coefficients. (4.1) shows the predicted probability of Liking SD for groups Treatment Group – High (dashed line) and Treatment Group – Low/Control Group (solid line). (4.2) show the difference in the predicted probability of Liking SD for groups Treatment Group – High and Treatment Group – Low/Control Group. Both figures show histogram of Internal Motivation to Control Prejudice and 95% confidence intervals, n = 1133.
Figure 4.1 shows the predicted probabilities at all the different values of IMCP for the groups, Treatment Group – High (dashed line) and Treatment Group – Low/Control Group (solid line), while Figure 4.2 shows the differences between the two groups in the probability of liking the SD. For an individual with a really low IMCP (value of one) the effect of a weakening the antiprejudice norm, by mainstream parties signaling that they are willing to cooperate with the radical right party, increases the probability of liking SD by 15.8% (from 18.1 to 33.8% probability). The difference between our treatment group and the reference group decreases as IMCP increases but there is a significant difference of our treatment on all individuals with an IMCP value lower than 4.4 (which is 64 percent of all individuals in our sample, and all below the median, which was 4). While out manipulation by itself is unable to distinguish if the main effect can be attributed to a changing normative context or cleaner radical right agenda, the fact that IMCP works as a mediator strongly indicates that the at work mechanism is a challenging of the anti-prejudice norm. In sum, our results support H3 and show that as mainstream parties challenge the anti-prejudice norm, individuals with lower IMCP are more likely to sympathize with radical right parties, compared to individuals with higher IMCP.6
General Discussion
We set out to answer the question if voter sympathy for a populist radical right party would be influenced by mainstream party recognition. To answer this, we ran an online experiment manipulating mainstream facilitation of the radical right. Specifically, we led participants to believe that the mainstream parties were increasingly acknowledging the radical right party the Sweden Democrats (SD) as a potential collaborator, or that they were negative to such a cooperation. By this manipulation we sought to weaken the anti-prejudice norm, which should lead people to increasingly sympathize with the radical right (Blinder et al., 2013).
Moreover, we sought to explore what role Internal Motivation to Control Prejudice have when the normative context changes. Following the previous literature, we expected that individuals with a low internal motivation to control their prejudice would express a higher sympathy towards the radical right as compared to those with a higher motivation. In addition, we expected that especially individuals with a low motivation to control their prejudice would be affected by a mainstream party recognition cue. Since individuals with a low Internal Motivation to Control Prejudice should clasp any cue that satisfactorily can function as a legitimization of the radical right, these individuals should be more prone to sympathize with the radical right when provided with such a cue (compared to individuals with a higher IMCP).
Our results support all three of our hypotheses. First, when informed that the mainstream parties consider SD a realistic political partner, sympathy for SD increases. Second, individuals with a low internal motivation to control their prejudice are more in favor of SD than those with a higher internal motivation to control prejudice. Finally, we found the expected interaction effect that individuals with a low internal motivation to control their prejudice were especially prone to liking the SD when the mainstream parties recognized SD as a collaborator. By itself, our manipulation is unable to isolate the mechanism as being a changing normative context or a cleaner radical right agenda. However, the fact that IMCP is an important mediator indicates that this is a case of a changing norms. Hence, when the normative context is weakened and the anti-prejudice norm is challenged, sympathy for the radical right increases. Importantly, the effects hold when controlling for relevant other factors such as attitudes towards immigrants and political self-placement.
This article makes a significant contribution to the literature. There is an ongoing debate in the literature regarding how mainstream parties facilitate radical right parties (Arzheimer & Carter, 2006; Bale, 2003; Meguid, 2005, 2008). Causal evidence is sorely missing in this literature. While some scholars have argued that mainstream party accommodation of the radical right reduces their success (e. g. Meguid, 2005), others have argued the opposite – that mainstream party accommodation increase radical right party success (e. g. Arzheimer & Carter, 2006; Bale, 2003; Dahlström & Sundell, 2012). With the use of experimental methods, we have demonstrated that there is a causal effect of mainstream party legitimization of a radical right party on the likelihood that a voter sympathizes with such a party. We therefore add to the literature that show that mainstream party accommodation of radical right parties may facilitate their success (Arzheimer & Carter, 2006; Bale, 2003; Dahlström & Sundell, 2012).
While previous research has shown that a strengthening of the anti-prejudice norm can explain why people with a high motivation to control their prejudice avoid making discriminatory political choices (Blinder et al., 2013), our article is the first to show how a weakening of this norm at the hands of mainstream parties increases sympathies with radical right parties. In addition, we show that this effect is strongest among those people in the electorate that are the least motivated to control their prejudice, showing only some individuals will be affected by mainstream party recognition of a radical right party.
One potential limitation of this study is that some respondents may have viewed the experimental manipulation as a testament to the increased legitimacy of SD achieved by the party’s own actions rather than as a result of mainstream party recognition. The manipulation read “One explanation could be that the Sweden Democrats are not seen as extreme anymore. They have cleaned up their agenda in the last few years and it seems to have provided results.” This latter statement was included as a proposed rationale for the co-operation between mainstream parties and the SD in our manipulation to increase external validity. While our intent was to manipulate the anti-prejudice norm by signaling that mainstream parties are willing to co-operate with them, there is a possibility that some respondents viewed the co-operation as a consequence of the now legitimate, less extreme, or “cleaner” radical right party. Indeed, this is in line with research arguing that the electoral success of the SD is in part due to their increased legitimacy (Rydgren & van der Meiden, 2019). While our treatment yields an effect on sympathy towards the radical right, we are unable determine to what extent this is due to a weakened anti-prejudice norm or a more legitimate radical right party, and we thus suggest that future studies attempt to disentangle this complex relationship.
Our results suggest that part of the explanation to the electoral success of radical right parties throughout Western Europe is related to a changing normative context. One reason for a normative context to change is that mainstream political parties change their attitudes and behavior towards radical right parties, legitimizing these parties when signaling that they are willing to cooperate with them. However, there may also be other reasons for a changing normative context which may weaken an anti-prejudice norm, for example related to how the news media, and social groups that an individual belongs to, portray and discuss issues of immigration and the policy stances of the radical right. Future research should investigate other mechanisms that may change a normative context influencing radical right party support.
With regards to individual traits, we focused on IMCP as it has shown to play pivotal role in explaining the occurrence of individual acts of prejudice (Blinder et al., 2013, 2019). Future research should look at the intricate relationship between those who are externally motivated to control their prejudice as they should be drawn between acting in accordance to the anti-racism norm and following elite cues. In addition, personality traits such as OCEAN (the Big Five) have shown to play a key role in the research on both prejudice (Bergh, Akrami, & Ekehammar, 2012) and support for the radical right (Bakker, Schumacher, & Rooduijn, 2021). Such personality features may also interact with norms in influencing support for the radical right, as our research suggests, showing that a focus on personality features and contextual features related to social norms is a fruitful avenue for research on radical right.