Parliamentary Representation and the Normalization of Radical Right Support

Abstract

How do stigmatized political preferences become normalized? I argue that the parliamentary representation of the radical right normalizes radical right support. Radical right politicians breach established social norms. Hence their supporters have an incentive to conceal that support. When the radical right enters parliament, however, its voters are likely to perceive that their views have been legitimized, becoming more likely to display their private preferences. I use three studies to test this argument. Study 1 employs a regression discontinuity comparing the underreport of voting for radical right parties (RRPs) above and below thresholds of parliamentary representation. Study 2 compares how much individuals report liking RRPs in post-electoral surveys depending on interview mode. Study 3 employs a difference-in-differences that looks into the underreport of UKIP vote before and after entering parliament. The results support the argument and highlight the role of political institutions in defining the acceptability of behaviors in society.

Keywords

radical right social norms political backlash political behavior

In the Post War period, social norms were put in place in Western democracies against the expression of racism and prejudicial ideologies such as Fascism and Nazism (Billig, 1991; Rydgren, 2005; Sniderman et al., 1991). While certainly not disappearing, such views were increasingly regarded as socially undesirable, and individuals started concealing them or stating them in more symbolic, ambiguous ways (Billig et al., 1988; McConahay & Hough, 1976). Recent years, however, have provided us with highly mediatic examples of individuals displaying these kinds of stigmatized behavior in public. For example, a number of protesters during the Charlottesville “Unite the Right” were seen displaying Swastikas. Some participants in the Chemnitz demonstration that took place in Germany in August 2018 were seen doing the Nazi salute. These anecdotes suggest that the set of acceptable behaviors changes over time, and that previously stigmatized behavior can become more prevalent. What explains these changes?

This paper looks at the role of the parliamentary presence of radical right candidates and parties in normalizing previously sanctioned behavior. Radical right politicians openly defy established social norms (Mudde, 2004, p. 554). For example, Björn Höcke, a politician for the radical right party Alternative for Germany (AfD), recently referred to the Berlin Holocaust memorial as a monument of shame. Similarly, French radical right leader Jean-Marie Le Pen referred to the Holocaust as a “detail of history,” and often complained that the French soccer team did not reflect French society because it included non-white players. Politicians from Hungarian party Jobbik have made a number of similar Anti-Semitic claims, and associated the Roma people—the country’s largest minority—with parasitism.

Given their open defiance of norms against prejudice and racism, showing support for radical right politicians can come at the cost of social sanctions. Hence, their supporters have an incentive to falsify their preferences—that is, to misrepresent views that breach established norms (Kuran, 1995). The difficulties of accurately polling the electoral support for the radical right, which is often understated in polls, can be taken as an example of this preference falsification on the behalf of radical right voters.

Bringing together insights from political behavior and social psychology, I argue that this situation is likely to change if radical right parties (RRPs) acquire parliamentary representation. There are two reasons why parliamentary representation is likely to normalize radical right support. On the one hand, elections can work as information shocks. They provide summary information on the true preferences of individuals in a society, which can reduce uncertainty as to how many other individuals support the party. Such information signals that many others do support the radical right, even if that may not be what one ought to do. Moreover, electoral success can signal that radical-right support is more pervasive than one had previously thought, which means that others are less likely to punish the expression of radical right preferences. On the other hand, institutions can signal which set of behaviors are acceptable in a society. The fact that the radical right has parliamentary representation can make voters perceive that norms have shifted to make radical right views less stigmatized. As a consequence, they can become more confident in publicly demonstrating support for the radical right, in a way that they would not have done had the party not made it to parliament.

Testing this argument with resource to comparative observational data is a difficult task because it requires distinguishing between individuals who are unwilling to show their support for the radical right and individuals who simply do not support it. In other words, one needs to be able to distinguish between revealed and true preferences of individuals, an imperative problem in the study of public opinion and voting behavior. I rely on three studies that follow different strategies to overcome this problem. In the first and third, I develop a measure of the stigmatization of RRPs, which is based on the share of the official vote for each party that is reported in post-electoral surveys. The reasoning behind this measure is that one can take the official vote for a party as a measure of the true underlying support for the party, and the reported support for that party in surveys as a measure of its revealed support in conditions of social pressure not to report radical right support. Consequently, the share of the official vote that is reported in post-electoral surveys can be taken as a measure of the appropriateness of reporting support for the radical right.

In Study 1, I use comparative evidence to test how the parliamentary presence of an RRP affects this appropriateness. Using a regression discontinuity design (RDD) that takes advantage of legally established parliamentary thresholds as an exogenous source of variation, the analyses show that, for each ten individuals who voted for an RPP, four to five more (depending on model specification) were willing to tell the interviewer that they had done so if the party had narrowly made it to parliament than if it had narrowly failed to do so. I find no evidence that this effect is driven by any of four alternative explanations: a bandwagon effect; a difference in the characteristics of individuals who answer post-electoral surveys in the treatment and control group; a difference in the overall left-right placement of the party-system; or a difference in the level of extremism of RRPs in the control and treatment group.

Finally, when it comes to the mechanism, the analyses seem more compatible with the results being driven by the signaling role of political institutions than by an information shock. The effect seems to be stronger in elections with lower thresholds, unlike what one would expect following the information shock mechanism. Supporting the institutional signaling mechanism, the effect is stronger for individuals whose interview was more distant from the election date. This suggest that individuals learn about the new norm over time, instead of changing their behavior immediately after the election—as one would expect according to the information shock mechanism.

Study 2 complements these findings by looking deeper into individual-level data. It takes advantage of the fact that some elections in the Comparative Study of Electoral Systems (CSES) include respondents interviewed in ways that forced an interaction with the interviewer (public modes), while others were interviewed in ways that did not force such interaction (private modes). I show that when an RRP is represented in parliament, individuals report liking it more in public modes of interview vis a vis private modes. The analyses further show that the effect is much stronger for right-wing individuals (as proxied by the ideology of the party they had voted for) than for left-wing individuals.

Finally, Study 3 tests whether the same results can be replicated by looking into the same party before and after it enters national parliament for the first time. It draws upon the case of one RRP that ran for several elections before entering parliament: the United Kingdom Independence Party (UKIP). Doing so also provides the advantage of replicating the analyses of Study 1 and Study 2 in a majoritarian system. The results of difference-in-differences (DiD) models show that, for each ten individuals that had voted for the party, two more were willing to tell their interviewer that they had done so after it made it to parliament. These results are not being driven by changes in the sociodemographic characteristics of UKIP voters before and after the party made it to parliament or by the good electoral result of UKIP in the 2014 election for the European Parliament.

The findings of the paper highlight how the presence of norm-breaching political actors in democratic institutions can lead to normative change. They also show how social norms affect individuals’ decision of whether or not to publicly reveal their private political preferences.

Literature on Consequences of Radical Right Success

Previous research has shown radical right success to have a number of important consequences. In the first place, it can affect the strategy and ideology of the remaining parties and politicians. As a response to radical right success, the mainstream right can adopt more rightist positions (Abou-Chadi, 2016; Abou-Chadi & Krause, 2018; Bale, 2003; Han, 2015; Schain, 2006). There is also some evidence of a similar effect on the mainstream left (Abou-Chadi & Krause, 2018), although this finding is less consensual (Bale et al., 2010; Han, 2015; Schain, 2006).

Radical right success can also affect the set of issues that are discussed in the political debate. Extending an argument initially made in the US context (Carmines & Stimson, 1989; Riker, 1986), the literature on party entrepreneurship has argued that the radical right can politicize new dimensions of political conflict (Hobolt & De Vries, 2015). Entrepreneurship seems to be a profitable strategy, in that RRPs tend to reap electoral benefits from the politicization of new issues (De Vries & Hobolt, 2012).

Finally, RRPs can also influence the attitudes and issue priorities of voters. The case study by Schain (2006) looks at how the success of the French National Front made French voters more focused on issues such as security and immigration. Similarly, focusing on issues of economic integration, De Vries and Edwards (2009) find cueing effects of extreme parties at both ends of the ideological continuum. In turn, Bischof and Wagner (2019) find that the parliamentary entry of an RRP leads to increased polarization, which they argue is due to both a legitimization and a backlash effect.

This legitimization effect is one that deserves careful empirical examination, so as to assess the extent to which voters perceive stigmatized views to become normalized as a consequence of the parliamentary representation of RRPs. Doing so is complicated by the fact that one needs to distinguish between an individual’s true preferences and the preferences they display in public, so as to measure the extent to which the parliamentary entry of an RRP can make individuals more confident in displaying previously stigmatized views. Drawing on experimental evidence, Bursztyn et al. (2020) find that after being shown data suggesting that Trump’s election was given as certain in their state, individuals were more likely to donate to an anti-immigration organization, in a way that matched their private preferences. This very valuable insight, however, has the shortcoming of being confined to lab evidence from the US. Given the difficult task of studying this effect with resource to observational data, no study to my knowledge has thus far engaged in that endeavor. That is the goal of this paper.

Radical Right Politicians as Norm Defiers

I argue that the parliamentary presence of radical right politicians and parties can normalize public expressions of support for the radical right. This argument builds on the social psychological concept of social norm, which can be defined as a “behavioral rule that (1) is known to exist and apply to a class of situations; and (2) is followed by individuals in a population on condition that (a) it is believed that sufficiently many others follow it (empirical expectations), and (b) it is believed that sufficiently many others believe the rule should be followed, and/or may be willing to sanction deviations from it (normative expectations)” (Bicchieri, 2017, p. 66).

Previous empirical research has shown the impact of social norms on a number of behaviors. When provided with information about what others do or think should be done, individuals tend to change their actions to follow that norm. This finding is consistent across behaviors as disparate as reusing hotel room towels (Goldstein et al., 2008), choosing ways of transport (Kormos et al., 2015), or eating and drinking behavior (Pedersen et al., 2015; Rimal & Real, 2005).

The influence of social norms extends to political behavior as well, especially regarding the decision to vote or to abstain. Departing from the observation that voter turnout is often over-reported in surveys, Karp and Brockington (2005) show evidence suggesting that this is due to pressure to report the socially desirable answer. They find over-reports to be higher in contexts of higher turnout, where such social pressures should be stronger. Subsequent research has added to this finding by providing causal evidence of this relation. Using a field experiment, Gerber et al. (2008) find that sending messages to voters promising to advertise their turnout around their household or neighborhood significantly increases their propensity to vote. In digging deeper into the mechanism driving this effect, posterior studies have found that messages emphasizing expectations of high turnout seem to be more effective in increasing propensity to vote than messages emphasizing expectations of low turnout (Gerber & Rogers, 2009) and that emotional appeals can be very effective in making individuals behave in norm-conforming ways (Panagopoulos, 2010).

Voter turnout, however, is not the only sort of political behavior that is likely to be affected by social norms. The extent to which individuals publicly display their support for a given party or candidate is also likely to be affected by normative expectations. They may fear that, in expressing a political preference that is contrary to the most commonly held opinion, they may incur in social sanctions such as isolation or career repercussions. For this reason, they may choose to remain silent instead of revealing their preferences (Noelle-Neumann, 1974).

This paper draws upon a specific type of socially sanctioned political preference: support for radical right politicians and candidates. These parties are, by nature, norm defiers (Mudde, 2004, p. 554). Their xenophobic rhetoric openly opposes social norms against the derogatory treatment of minority groups. Following the de-legitimization of biological racism in the Post-War period (Ignazi, 1992; Rydgren, 2005), racism became increasingly regarded as socially undesirable in Western democracies of the 1970s and 1980s (Sniderman et al., 1991, p. 424). In other words, social norms emerged against the expression of those views (Billig, 1991; Billig et al., 1988). The existence of these norms means that expressing one’s support for the radical right may affect one’s reputation—a concern that may outweigh the utility one extracts from being true to their private preferences (Kuran, 1987, pp. 644−645). This gives radical right supporters an incentive to falsify their preferences to avoid the reputational costs of norm-breaching behavior.

I argue that parliamentary representation is one crucial channel through which radical right support can become normalized. There are two main mechanisms through which parliamentary representation could change social norms and thus bring about changes in the behavior of individuals. In the first place, elections can work as information shocks, providing an accurate depiction of how many others in one’s society privately support the radical right. This sort of information can play a crucial role in processes of norm change. In its absence, even if many individuals privately support the party, they can be locked in a situation of pluralistic ignorance, in which many disagree with the norm in place but still abide by it because they are not aware that others also oppose it (Bicchieri, 2017). The information provided by elections can help overcome this deadlock in two main ways. On the one hand, because voting is secret, individuals can freely express their true preferences (Kuran, 1995). Electoral outcomes can thus provide individuals with crucial information about the prevalence of norm-breaching preferences within a community. Hence, elections can signal that, even if supporting the radical right may not what ought to be done, many others do it. In situations such as these, individuals tend to follow what others do instead of what they think should be done (Bicchieri & Xiao, 2009; Pedersen et al., 2015).

On the other hand, the information shock provided by elections can affect behavior by changing individual perceptions of how costly it is to reveal their true preferences. Norms affect behavior to the extent that one believes their social referents are likely to punish them for transgressing the norm. Realizing that a lower-than-expected proportion of individuals supports the norm means that a smaller number of individuals are likely to punish such transgressions, which lowers the costs of breaching the norm (Bicchieri & Mercier, 2014).

Apart from working as an information shock, the parliamentary presence of the radical right can normalize radical right support because representative and governing institutions can signal what kind of behaviors are normatively desirable in a society (Tankard & Paluck, 2016, pp. 192−193). One particular way by which institutions can bring about normative change is by accommodating innovation in terms of the views or actors represented in them. In so doing, they can signal to individuals that social norms have moved in the direction of the innovation. Examples include the imposition of gender quotas for women in powerful political positions (Beaman et al., 2009) or quotas for stigmatized groups (Chauchard, 2014), both of which made the views of the population more positive toward these social groups.

One might expect an analogous mechanism to operate when radical right politicians are represented in parliament. Just like the majority of individuals regard underprivileged groups in a more positive light after they are represented in institutional bodies, the views of radical right candidates are likely to be perceived as less stigmatized once they acquire parliamentary representation. Importantly, the centrality of parliaments in democratic systems, and their salience in the media, means that the signal about normative change provided by parliamentary representation is highly public. In other words, each individual knows that most other members of society are also made aware of the normative signal—thus creating what Chwe (2013) has referred to as “common knowledge.” This is an important point because previous empirical research has shown that information transmitted in public is more likely to bring about behavioral change than information transmitted in private (Arias, 2019; Gottlieb, 2016; Gulzar & Khan, 2017).

Empirical Strategy

Testing the extent to which the parliamentary presence of RRPs normalizes radical right support is complicated by the need to distinguish between two types of individuals. On the one hand, individuals who support the radical right but who abide by social norms against the expression of such ideology, and hence falsify their preferences. On the other hand, voters who simply do not support the radical right. The obvious difficulty is how to know what someone’s true preferences are if they are consciously concealing them. The empirical sections of this paper rely on three complimentary studies that find ways of overcoming this problem.

The first and third study take advantage of the fact that the vote share of RRPs is consistently under-reported in post-electoral surveys. As Figure 1 shows, this difference is of around 1.6 percentage points, meaning that only around 80% of the official vote for RRPs is reported in the post-electoral surveys included in CSES. As Table A1 in the Supplemental Appendix shows, this difference is statistically significant ( $p < 0.001$ ). Figure 1 also shows that this under-report happens only for RRPs. Parties of other ideological placements are either not underreported or, if anything, over-reported in post-electoral surveys.

Figure 1.

Difference between parties’ official vote share and their vote share as reported in post-electoral surveys (CSES).

The under-report of voting for RRPs is likely to happen because a survey interview is still a social interaction, and social norms can influence the answers given by respondents. They are likely to feel judged by the interviewer and try to provide what they perceive as the socially desirable answer (Zaller, 1992). While the fact that the interviewer is aware of the respondents’ answer may seem like a rather weak treatment, a large body of literature has documented that others’ awareness of one’s actions can significantly alter behavior. For example, turnout is higher if individuals are aware that their neighbors or household members will be informed of whether they decided to vote (e.g., Gerber et al., 2008). Similarly, donations and public good spending increase when the donator can make others aware (Andreoni & Petrie, 2004; Ariely et al., 2009; Kessler, 2011). When it comes to survey answers, a large body of literature has expressed concerns that the answers individuals provide in surveys conform to social norms (Blair & Imai, 2012, pp. 47). Faced with an unknown interviewer, respondents have an incentive to try and manage the impression they convey by providing the norm-abiding answer (Kuklinski et al., 1997, p. 327). For this reason, some research has provided respondents with the information that the researcher will be made aware of their actions as a ways of manipulating extrinsic motivations stemming from possible social sanctions. In so doing, these studies have found such treatment to significantly alter behavior, making respondents more likely to act in norm-compliant ways (Bursztyn et al., 2020).

Empirical research in political science has also found that perceptions of which answer is socially desirable influence how individuals report their behavior in surveys. For example, it has been consistently found that individuals tend to over-report voter turnout (Berent et al., 2016; Bernstein et al., 2001; McDonald, 2005). This over-report seems to be partly explained by the level of turnout in the previous election (Karp & Brockington, 2005), which suggests that the higher the social pressure to vote, the higher the likelihood that voters will report having voted when in fact they did not.

One can thus take the under-report of the vote for RRPs in post-electoral surveys as measure of the perceived strength of social norms against expressions of support for the radical right. Feeling that voting for the radical right is socially sanctioned, individuals have an incentive not to report it. Hence, the proportion of individuals who report voting for the radical right can be taken as a measure of how many individuals are likely to express that preference despite the social sanctions associated with it. These social pressures, however, should be much lower—or even inexistent—in the voting booth. As voting is secret, others will not know what one has voted for and are less likely to engage in sanctioning (Ewing, 2001; Kuran, 1995). Consequently, the vote share for RRPs can provide a measure of how many individuals in a population privately support the radical right, and the proportion of that vote that is reported in post-electoral surveys can be used as measure of the extent to which individuals in a population perceive the expression of radical right support as acceptable.

Taking advantage of this, Study 1 presents comparative evidence showing that individuals feel more confident in reporting having voted for an RRP when that party is in parliament. Following Dinas et al. (2015), this study employs an RDD that takes advantage of exogenous variation in legally fixed electoral thresholds to parliamentary entry and compares parties just above and just below those thresholds. The dependent variable is the proportion of the official vote share for each party that is reported in post-electoral surveys.¹ I rely on the CSES given the high quality of its data collection protocol and its comparability across different systems. For each country, I collected the vote share for each party in each election, as reported in CSES. Afterwards, I collected the official electoral results of the same parties in the same election. To calculate the final dependent variable (reported vote), I divided the vote share for the party in CSES by its official vote share:²

Reported vote = \frac{Vote share in CSES}{Official vote share}

In coding the parties as radical right, I follow two classical references on the comparative study of this party family: Mudde (2007) and Norris (2005). The coding of the parties was done such that any party considered to be radical right by at least one of these authors was coded as such. In other words, parties that were considered to be an RRP by both authors were coded as radical right; as were parties that were coded as such by one author but not by the other. Doing so avoids concerns of leaving some countries out by the sole reason they are not included in one of the references. Table A8 in the Supplemental Appendix shows that the results remain identical after removing parties regarding which there are discrepancies between these two references. I also included some parties that became successful after those references were published, and which the literature tends to consider as radical right.³ Table A5 in the Supplemental Appendix provides the full list of parties coded as radical right, along with the references that consider them to be so.

Study 2 complements this study by drawing upon individual-level data. It takes advantage of variation in the interview mode of CSES respondents. Some elections in the CSES simultaneously include respondents who were interviewed in a way that forced an interaction with an interviewer, such as telephone or face to face interviews (what I call public modes of interview); and respondents who were interviewed in a way that did not force such interaction, such as internet, self-administered, or mail-back interviews (what I call private modes of interview). Because social norms are less likely to influence private behavior, I take the group of respondents interviewed in private modes as a control group. Then I regress how much respondents report to like each party on their mode of interview (private or public), a dummy for whether the party was in parliament, a dummy for whether the party is an RRP, and the three-way interaction between these variables.

Finally, Study 3 tests the argument by looking at a single party across time. I draw upon the case of UKIP, an RRP that ran for several elections before making it to parliament. This makes it possible to study the effect of parliamentary entry on its normalization, in a way that would not be possible if the party had made it to parliament in the first election it ran for. Using data from the British Electoral Study (BES), I calculate the dependent variable of the Study 1 again (reported vote), and employ a DiD design that compares UKIP to the main parties in the country (Conservative, Labour and Liberal Democrat).

Study 1: Comparative Evidence

I start with testing the argument of the paper in comparative fashion. To do so, I rely on an RDD that takes advantage of the fact that a number of electoral systems have legally fixed thresholds that determine whether a party can enter parliament. The dependent variable is reported vote, which captures the proportion of the official vote share for the party that is reported in the CSES. As discussed above, a small number of post-electoral studies included in the CSES included different modes of interview, some of which do not include an interaction with the interviewer (e.g., mail back or self-administered surveys). Study 2 below takes advantage of this variation but, in this study, I use only respondents who were interviewed in modes of interview that imply an interaction with the interviewer: telephone or face-to-face.

Because different countries have different electoral thresholds, the forcing variable is not each party’s vote share, but rather the difference between a party’s official vote share and the fixed electoral threshold in the country. This decision follows previous research using a similar research design (Abou-Chadi & Krause, 2018). The cutoff point is the electoral threshold. The analyses include all elections included in the CSES that have a proportional or mixed electoral system and a fixed electoral threshold. I use a triangular kernel, which gives more weight to parties whose electoral results are closer to the cut-off (Fan, 2018). A sharp RDD would provide estimates for the parameters of interest that minimize the following equation:

\min_{α, β, γ, δ} \sum_{i = 1}^{N} 1 {c - h \leq X_{i} \leq c + h} (Y_{i} - α - β (X_{i} - c) - γ T_{i} - δ (X_{i} - c) T_{i})^{2}

In this equation, c is the electoral threshold in each party’s country; T is the treatment condition (0 meaning that the party failed to enter the parliament in that election, 1 meaning that the party did enter the parliament); $X_{i}$ is the vote share of the party in the election; and Y is the dependent variable, reported vote.

Some countries in the sample have mixed electoral systems, meaning that some seats are elected using majoritarian rules. Moreover, some countries with proportional electoral systems have rules that allow parties to win seats in the parliament even if their vote share is lower than the fixed threshold. For this reason, I run a fuzzy regression discontinuity. Instead of assuming that the cutoff point dictates whether or not each unit gets the treatment, as with sharp regression discontinuity, fuzzy regression discontinuity assumes that the cutoff point increases the probability that units will receive the treatment. In practical terms, this means that the forcing variable is used as an instrument of treatment status (Imbens & Lemieux, 2008). T is thus instrumented by Z—which takes the value of 0 for parties below the threshold and 1 for parties above it—and by $Z \times (X - c)$ . The first stage regression is given by the following equation:

T = α_{1} + β_{1} (X_{i} - c) + γ_{1} Z_{i} + δ_{1} (X_{i} - c) Z_{i} + μ_{i}

Afterwards, $\hat{T}$ is used to estimate the second-stage parameters, according to the following equation:

Y = α_{2} + β_{2} (X_{i} - c) + γ_{2} {\hat{T}}_{i} + δ_{2} (X_{i} - c) {\hat{T}}_{i} + v_{i}

The treatment effect is given by $γ_{2}$ . These analyses were run using the rdrobust package and make use of the optimal bandwidth (Calonico et al., 2017). The robust method provided by this package is robust to large bandwidths (Calonico et al., 2014), and represents the main specification used in robustness checks and additional analyses. For sake of transparency, in the main analyses I report the three approaches that are reported in the output of the rdrobust package: conventional, bias-corrected, and robust. As argued by Lee and Lemieux (2010), it is important to estimate RDDs using both this non-parametric approach and a parametric approach. As such, for each model I also estimate the global polynomial regression given by the following equation:

Y_{i} = α_{1} + τ D_{i} + β_{1} (X_{i} - c) + β_{2} (X_{i} - c) * D_{i} + μ_{i}

In this equation, $τ$ is the treatment effect, $D$ is the treatment status (coded 1 for parties in the parliament and 0 for parties outside the parliament), X represents each party’s vote share, and $c$ is the cutoff point (the electoral threshold). The specific number of polynomials to be used was determined by relying on the Akaike Information Criterion (AIC), as suggested by Lee and Lemieux (2010, p. 326). In all models reported in the main text, this meant using just one polynomial—only in Table A4 in the Supplemental Appendix did the AIC lead to the use of a different number of polynomials. Tables A18 through A30 in the Supplemental Appendix show these tests. Table A17 in the Supplemental Appendix shows that the main results hold if one uses a different number of polynomials.

How well does this design meet the crucial assumptions for causal identification using RDDs? One of the main assumptions of these designs is the as-if-random assumption, which postulates that the assignment of units to treatment or control group, within a given bandwidth around the threshold, is as good as random. This seems like a plausible assumption in this design. Electoral thresholds vary from country to country and RRPs cannot self-select into countries with lower electoral thresholds, nor can they manipulate their vote share to be just below or just above the threshold. RDDs also require that the continuity assumption is met. This assumption states that “the only change, which occurs at the point of discontinuity, is the shift in the treatment status” (Cuesta & Imai, 2016, p. 377). This assumption is more restrictive than the as-if-random assumption and, in this particular design, raises the concern that other changes apart from the shift in treatment status may drive the results. As discussed in more detail below, I find no evidence that the effects are being driven by jumps in other variables around the threshold.

The sample includes all elections in the CSES that took place in non-majoritarian electoral systems and for which at least one RRP ran—a total of 80 observations from 58 elections in 21 countries. Table 1 shows the list of elections included in the sample and their electoral thresholds.⁴ A list of all RRPs included in the analyses can be found in Table A5 in the Supplemental Appendix. While this provides for a broad array of countries, many authors have argued that RRPs across Europe and other Western democracies—such as Israel, New Zealand, Canada, and Australia—have enough characteristics in common to be classified as a single party family (Mudde, 2007; Norris, 2005; Rydgren, 2007). One of the core characteristics of such party family is its xenophobia (Rydgren, 2007, p. 242). Given that such xenophobia is at the heart of why displays of support for RRPs are socially sanctioned, one should expect the stigma against such support to be broadly present across Western democracies. At the same time, because parliaments are central governing bodies in all democratic systems, and such bodies can have a crucial role in bringing about norm change (Tankard & Paluck, 2016), one should expect the effect of parliamentary entry in normalizing public expressions of support for the radical right to be fairly stable across the countries included in the sample. Similarly, while the study includes a time span of around twenty years, there are no reasons to believe that the effect of parliamentary entry would change during this period. To the extent that RRPs have not significantly changed their ideology and continue to breach social norms, there is no reason to expect that the stigma surrounding expressions of support for them—and the effect of parliamentary entry in lowering perceptions of such stigma—would be different across the time span of the study. Figure A2 in the Supplemental Appendix illustrates this point by showing that the dependent variable in this study (the proportion of the official vote for RRPs that is reported in post-electoral surveys) remains stable throughout the time span of the study.

Table 1.

List of Elections Included in the Sample and Their Electoral Thresholds.

Elections	Electoral threshold
Austria 2008, 2013, 2017	4
Bulgaria 2014	4
Croatia 2007	5
Czech Republic 1996, 2002, 2006	5
Denmark 1998, 2001, 2007	2
Estonia 2011	5
Germany 1998, 2002, 2005, 2009, 2013, 2017	5
Greece 2009, 2012, 2015 (Jan), 2015 (Sep)	3
Hungary 1998, 2002, 2018	5
Israel 1996, 2003	1.5
Israel 2006, 2013	2
Italy 2006	4
Italy 2018	3
Latvia 2010, 2011, 2014	5
Netherlands 1998, 2002, 2006, 2010	0.67
Norway 1997, 2001, 2005, 2009, 2013	4
Poland 1997, 2001, 2005, 2007	5
Romania 1996	3
Romania 2004	5
Serbia 2012	5
Slovakia 2010, 2016	5
Slovenia 1996	≈3
Slovenia 2004, 2008	4
Sweden 2006, 2014	4
Ukraine 1998	4

Source. Author’s elaboration, based on data from the European Election Database and official data from each country.

Figure 2 shows the results of the analyses. The left-hand side panel plots the share of the official vote for each RRP that is reported in CSES, with the electoral threshold as a cutoff point. The x-axis represents how many percentage points above or below that threshold each parties was. The figure suggests that there is, indeed, a discontinuity around the threshold of parliamentary representation, such that the share of the official vote that is reported in the CSES is lower for RRPs just below the threshold than for those just above it.

Figure 2.

Effect of parliamentary representation on the normalization of the radical right support (regression discontinuity design).

The panel on the right-hand side of the figure shows the coefficients from this regression discontinuity. I estimate parametric and non-parametric models that estimate the local average treatment effect (LATE) of parliamentary entry on the normalization of RRPs. As mentioned above, I replicate the analyses using the dependent variable weighted according to the demographic weights provided by CSES. To make sure that the effect is specific to RRPs, I also run a placebo where I replicate the same analyses for center right parties. These were coded based on the Chapel Hill Expert Survey (CHES) (Bakker et al., 2015). I considered any party to be center right whenever it was not coded as radical right and its ideological position in the CHES was between 5.5 (center) and 8.⁵ The parametric models include a global bandwidth, while the non-parametric models make use of the optimal bandwidth (Calonico et al., 2014).

This panel shows that the LATE is between 37 and 51 percentage points, depending on model specification. This is a remarkable effect: for each 10 individuals that voted for an RRP, four to five more were willing to tell the CSES interviewer that had done so when the party had narrowly made it to the parliament than when it had narrowly failed to do so. The results are very similar using the weighted and unweighted dependent variable, which suggests that they are not being driven by differences in the demographic characteristics of voters. Table A6 in the Supplemental Appendix provides further evidence of this point. In turn, this effect cannot be replicated when one draws upon center right. The LATEs for the analyses of these parties mostly fail to reach statistical significance and, for the most part, have negative signs. This suggests that the normalization effect of parliamentary entry is specific to the radical right. Figure A1 in the Supplemental Appendix further shows that this is because center right parties are not under-reported, regardless of their treatment status—which is consistent with the pattern shown in Figure 1, according to which RRPs are the only set of parties under-reported in post-electoral surveys.

This design makes the crucial assumption that the difference in the reported vote for RRPs in and out of parliament is due to a difference in the stigmatization of revealing one has voted for the radical right. Four main alternative interpretations might be advanced. In the first place, the results might be the product of a bandwagon effect, by which individuals are more likely to report having voted for a successful party than for an unsuccessful one. To the extent that parliamentary representation represents a proxy for success, this might explain the findings of Figure 2. If this were the case, however, one should find an effect when replicating the same analyses using parties with different ideological placements. Figure 2 already showed that this effect is not replicated when drawing upon center right parties. Table A4 in the Supplemental Appendix provides further evidence on this point, by replicating the main analyses for center left parties and for all parties left of center. Again, I find no evidence of a similar effect. Taken together, these analyses pay no support to the interpretation that the results can simply be explained by a bandwagon effect.

Another alternative explanation is that the results are not being driven by a normalization effect, but rather by a difference in the characteristics of individuals who are successfully contacted to answer post-electoral surveys when RRPs narrowly make it to parliament and narrowly fail to do so. Supplemental Table A6 tests for this possibility. It shows that there are no significant differences in the demographic characteristics of respondents successfully contacted to answer surveys when RRPs are just above and just below the threshold.

In the third place, the results might be driven by party systems of treated parties being generally more rightists, which might make respondents more willing to report having voted for an RRP. Supplemental Figure A7 shows no evidence supporting this alternative explanation, as there is no evidence of a relation between reported vote and the mean left-right ideology of the remaining parties.

Finally, the results might be driven by RRPs that narrowly cross the threshold being more moderate than those that narrowly fail to make it across the threshold. To address this possibility, Supplemental Table A7 replicates the main analyses shown in Figure 2 using the left-right position of RRPs as the dependent variable. I find no evidence that RRPs above the threshold are significantly less rightist than RRPs below the threshold.

This being said, these analyses present the caveat that they are unable to disentangle between two explanations. It may be that when RRPs are not in parliament, their supporters avoid telling the interviewer that they supported the party—a situation that changes once these parties make it to parliament. But the results may also be driven by voters of other parties who, once RRPs are represented in parliament, say that they voted for an RRP instead. It should be noted, however, that both explanations are compatible with the interpretation that reporting having voted for an RRP becomes less stigmatized when these parties make it to parliament.

Appendix A in the Supplemental Appendix provides a number of additional analyses related to the analyses of Study 1. In the first place, Supplemental Figure A3 shows evidence that suggests that the under-report of RRPs out of parliament does not overwhelmingly happen at the expense of revealing any other kind of behavior—if anything, it happens at the expense of more individuals reporting voting for center right parties. Supplemental Figure A4 shows that the effect seems to be stronger on RRPs that were in parliament in the previous election and fail to enter parliament in the current election.

This Appendix also provides a number of robustness checks. Supplemental Figure A5 plots the McCrary test for sorting. The p-value for the null of no-sorting is .90. Supplemental Figure A6 shows that there is a strong first stage. Supplemental Figure A8 replicates the RDD analyses using a number of different bandwidths. To make sure that the results are not being driven by influential country cases, Supplemental Figure A9 replicates the analyses after removing each of the countries. Supplemental Table A8 shows that the results remain similar if one removes parties that are coded as RRPs by Norris (2005), but not by Mudde (2007), and vice-versa. Supplemental Table A9 shows an alternative specification of the main model. Instead of using the proportion of the official vote that is reported in post-electoral surveys as the dependent variable, these analyses take the vote share of each party in the post-electoral surveys as the dependent variable. Because some elections include more than one RRP, Supplemental Table A10 replicates the main analyses using an alternative dependent variable, coded as the overall proportion of the vote for all RRPs running for a given election that is reported in post electoral surveys; Supplemental Table A11 replicates the analyses including a control for the overall vote share for RRPs in each election; and Supplemental Table A12 replicates the main analyses using only the most voted RRP in each election. Supplemental Table A13 replicates the analyses controlling for district magnitude. Supplemental Tables A14 and A15 replicate the main analyses including country fixed effects and survey wave fixed effects, respectively. Finally, Supplemental Table A17 shows that the findings hold using a parametric approach with different numbers of polynomials.

Information Shock or Institutions Signaling Norm Change? Investigating the Mechanism

These findings raise the question of what mechanism is driving them. In the theoretical section, I discuss two possibilities: that parliamentary representation works as an information shock about the true number of individuals who privately support the radical right; and that, in including the radical right in them, parliaments signal norms to have shifted toward a lower stigmatization of radical right support. While not being in the position to truly test which of these mechanisms is at play, the following analyses test observable implications of each of the two mechanisms.

If the results are being driven by elections revealing the pervasiveness of radical right support in one’s society, the effect should be stronger in elections with higher thresholds. The rationale is that, where thresholds are higher, parties need a higher vote share to enter parliament, making it a stronger signal of good electoral performance. Ideally, I would like to test this expectation by replicating the analyses shown in Figure 2 on two subsamples: the subsample of elections whose threshold is higher than four (the median threshold in the sample); and the subsample of the remaining elections. Unfortunately, the number of observations does not allow for this subsample analyses to be carried out. For this reason, I carry out descriptive analyses that show the proportion of the official vote for RRPs that is reported in post-electoral surveys as a function of the distance to the threshold, in the two subsamples.

The results, shown in Figure 3, do not seem to pay support to the information shock mechanism. The effects seem stronger in elections with low thresholds than in elections with high thresholds. This is the opposite of what one should expect according to the information shock mechanism.

Figure 3.

Proportion of the official vote for RRPs reported in post-electoral surveys as a function of the distance to the threshold, conditional on size of electoral threshold.

I then move on to testing an observable implication of the institutional signaling mechanism. If the effect is being driven by institutional presence signaling a norm shift that makes the radical right less stigmatized, individuals should need some time to learn about this normative change—unlike with the information shock mechanism, which should operate immediately after the election. I empirically assess this possibility by looking at how the effect changes, conditional on the distance between the date of the election and the date of the survey interview. To do so, I run linear probability models that regress a dummy coded 1 for individuals who reported having voted for an RRP and 0 for all others on a fully factorized variable indicating how many days had gone by since the election. I perform this exercise twice: once for RRPs that entered parliament and once for RRPs that did not enter parliament. To decrease concerns that the results may be driven by differences in the sociodemographic characteristics of individuals interviewed at different stages of the fieldwork period, the analyses make use of the demographic weights provided by CSES. The models also include election fixed effects.

As the results in Figure 4 show, the data seem to support this mechanism. This figure plots the predicted probability of reporting having voted for an RRP as a function of the distance between election day and interview day. While the probability of reporting having voted for an RRP out of the parliament is stable over time, the probability of reporting having voted for an RRP that entered parliament increases. This is what one would expect following the institutional signaling mechanism, but not following the information shock mechanism.

Figure 4.

Predicted probability of reporting having voted for a radical right party as a function of the distance between the date of the election and the date of the survey interview.

Study 2: Individual-Level Evidence

Study 1 relies on a measures of social norms at the party level. Study 2 provides a complement to it by drawing upon individual-level data. It takes advantage of the fact that, in some elections included in the CSES, different respondents were interviewed using different modes of interview. While interviews over the telephone or face-to-face made the answers of respondents public—in that they included an interaction with the interviewer—mail-back, online, and self-administered surveys kept answers private.⁶

The difference in interview mode is important because social norms should interfere less with private behavior such as survey answers using private modes of interview. The two remaining studies draw exclusively upon answers to surveys whose interviews were conducted in public modes. By contrast, this study takes advantage of variation in the interview modes and uses the group of respondents interviewed in private modes as a control group, who are more likely to report their true preferences.

I draw upon data from all elections in the CSES that include more than one mode of interview: Greece 2012 (June) and 2015 (January and September); Italy 2018; and Denmark 2007. I also included Germany 2002, which has two full studies in Round 2 of the CSES: one fully employing telephone interviews, and another fully employing mail-back surveys. As a dependent variable, I use the item asking how much each respondent reports to like each party, in a scale from 0 to 10. The main independent variables are a dummy for the mode of interview (coded 1 for public modes and 0 for private modes), a dummy for RRPs, and a dummy for whether each party had made it to parliament in the election previous to the survey. I also include an interaction term between these three variables, which is what I am substantively interested in. The extent to which respondent i reports liking party j is thus given by the following equation:

\begin{array}{l} {Like}_{i j} = α + β_{1} {Public}_{i} + β_{2} Radical {Right}_{j} + β_{3} {Parliament}_{j} \\ + β_{4} {Public}_{i} \times Radical {Right}_{j} + β_{5} {Parliament}_{j} \\ \times Radical {Right}_{j} + β_{6} {Parliament}_{j} \times {Public}_{i} + β_{7} {Public}_{i} \\ \times {Parliament}_{j} \times Radical {Right}_{j} + μ_{i j} \end{array}

Each individual was included in the sample several times: one for each party for which there is data concerning the dependent variable. To account for this, standard errors are clustered by individual.⁷

The results are shown in Figure 5. As the figure shows, all variables have the expected effect. Individuals tend to like RRPs less than the remaining parties; to like parties which are in parliament more; and to report liking parties out of parliament more when the mode of interview is private. The difference between how much individuals reporting liking a party in private and public modes of interview is reduced when the party makes it to the parliament, but this reduction is larger in the case of RRPs. The coefficient of this three-way interaction is 0.876. This represents around 25% of the mean dependent variable in the whole sample (3.52), making the coefficient of the three-way interaction far from negligible in substantive terms.

Figure 5.

Reported like/dislike for radical right and remaining parties, in and out of the parliament, by mode of interview.

While all the variables all have the expected direction, the large magnitude of the three-way interaction leads to the unexpected result that, when RRPs are in the parliament, individuals report liking them more in public modes of interview than in private ones. A possible interpretation of this finding is that the parliamentary entry of an RRP leads its supporters to feel some sense of euphoria that makes them extract utility from publicly expressing support for the party. Because the dependent variable in this study is a continuous variable that allows individuals to express the extent of their liking for a party—as opposed to just reporting having voted for it or not, as in Study 1—the feeling that the parliamentary entry of the party has legitimized its views can make individuals more eager to signal their support for the RRP, giving it a higher score. Since the utility extracted from such signaling is likely to affect public modes of interview more strongly than private ones, this could be driving the unexpected value.

The crucial assumption in this design is that individuals interviewed using private and public modes of interview are comparable, which may not necessarily be the case. To the extent that individuals may not be randomly assigned into different modes of interviews, the same characteristics that affect their assignment into a specific mode of interview can also affect the answers they provide. Supplemental Table B1 in Appendix B in the Supplemental Appendix addresses this concerns with two different strategies: by adding a number of control variables; and by estimating models that make use of the demographic weights provided by CSES. The models including control variables also control for the vote share of each party, which could be confounding the relation between parliamentary presence and the outcome variable. This table also replicates the analyses comparing RRPs only to center-right parties, instead of comparing them to all other parties. The results of these models are substantively similar to the ones presented in Figure 5. Finally, this Appendix provides some further detail to the analyses of this study. Supplemental Figure B1 shows that the effect of parliamentary entry on normalizing public expressions of support for the radical right is larger for individuals with a college degree.

One question that is raised by the findings shown in Figure 5 is which individuals become more likely to report liking an RRP when the party is in parliament. In this study I have data on the party each individual voted for, which allows me to provide further detail on this question. To do so, I focus on how individuals rate RRPs alone, excluding all other parties. Afterwards, I check how much individuals report liking RRPs in and out of parliament, conditional on their mode of interview and the ideological position of the party they had voted for in that election.⁸

As shown on Figure 6, the results are much stronger for individuals who voted for right-wing parties. However, I find no evidence of a backlash effect on voters of left-wing parties. The effect remains for left-wing voters, albeit much smaller. The fact that the effect is stronger for right-wing voters is in line with previous research in psychology, which has found that the effect of norm change tends to be stronger on individuals whose private preferences were already aligned with the new norm (Tankard & Paluck, 2016, pp. 198−199).

Figure 6.

Reported like/dislike for radical right parties, in and out of the parliament, by mode of interview and ideology of the voter.

Study 3: Case Study of UKIP

The two previous studies have shown the effect of parliamentary entry on the normalization of public expressions of support for RRPs by relying on party-level and individual-level data. These studies face the potential criticism that both compare different parties, and that they both focus on countries with proportional systems. Study 3 is designed to check if the results hold while analyzing the same party before and after it made it to parliament, and if they can be replicated in a majoritarian system. I look into the case of UKIP in the United Kingdom because it is a rare case of an RRP that ran for several elections before entering parliament, which allows one to look at differences before and after its parliamentary entry.

UKIP ran for the first time for the general election of 1997, failing to win any seat in the parliament and scoring a low 0.3% of the vote. The party remained extra-parliamentary until 2015, when its norm-defying, anti-immigration rhetoric proved successful. Its campaign relied heavily on the charisma of its leader Nigel Farage, who presented himself as someone not afraid to “tell it like it is” (Cowley & Kavanagh, 2016, pp. 175−176). UKIP gained $12.7 %$ of the vote and entered the lower house for the first time.⁹

To test the extent to which the parliamentary entry of UKIP normalized expressions of support to it, I draw on data from the BES and calculate the same dependent variable as in Study 1—reported vote, which captures the share of the official vote for UKIP that is reported in post-electoral surveys. The effect of UKIP’s parliamentary entry on this variable was estimated using DiD models, which calculate the average treatment effect on the treated (ATT) by comparing the difference between post- and pre-treatment values of treated units to those of units that do not receive the treatment. Assuming that, in the absence of the treatment, the two sets of units would follow parallel trends, the control group provides the missing potential outcome: the reported vote for UKIP, had it not made it to national parliament in the election of 2015.

To estimate these models, I create a binary treatment indicator coded 1 for RRPs that had made it to the parliament in the previous election, and zero otherwise. The analyses draw upon the elections of 2005, 2010, and 2015. The reported vote for UKIP is compared to that of the three main parties in the UK: Conservative, Labour, and Liberal Democrats. The ATT is then calculated on the basis of a fixed-effects model, given by the following equation:

Y_{i t} = α + γ_{i} + λ_{t} + τ D_{i t} + μ_{i t}

In this equation, $Y_{i t}$ is the dependent variable, reported vote, for party i at time t; $γ_{i}$ is a party-level fixed effect that controls for time-invariant unobserved heterogeneity; $λ_{t}$ is a period fixed effect that controls for similar trends; and $τ$ is the ATT.

Figure 7 shows the results. The left-hand side panel plots the evolution of the dependent variable in the control and treatment group. As the figure shows, units in the treated and control groups followed similar trends before the election of 2015. However, in that election, the share of the official vote for UKIP that was reported in BES increased substantially. The right-hand side panel shows the results of a number of DiD estimates for the effect of parliamentary entry of UKIP on the normalization of expressions of support for it. The first model represents the DiD estimate using period fixed effects for each of the three periods included in the analyses, while the second model simply compares the pre-treatment and post-treatment periods. The third model replicates the first model, with the difference that the dependent variable makes use of the demographic weights provided by the BES. As with the previous studies, this was done to address concerns that the results may be driven by differences in the demographic characteristics of individuals who reported having voted for UKIP before and after the treatment. Finally, the last model represents a placebo test that assumes the previous time point ( $t - 1$ ) as the moment of the administration of the treatment. In so doing, it also tests for the extent to which trends in the control and treatment groups were parallel before the treatment. Should that be the case, one should find no effect when running this placebo.

Figure 7.

Effect of the parliamentary entry of UKIP on its normalization (difference-in-differences models).

Regardless of the model specification, the results show an increase in the vote share for the party that was reported in the BES in 2015. The effect size is around 20 percentage points. This means that, for each 10 individuals who voted for UKIP, two more were willing to tell the BES interviewer that they had done so after the party made it to the parliament. This is a sizeable effect, even if smaller than the one found in the RDD in Study 1. The fact that the effect is slightly smaller than the one found in the RDD analysis of Study 1 seems to be driven by the vote for UKIP before it entered parliament being less underreported in post-electoral surveys (around 0.75 of its official vote) than that of RRPs that narrowly fail to make it to parliament in Study 1 (around 0.5 of their official vote). This suggests that UKIP was less stigmatized than the average RRP in other countries, leaving less room for its parliamentary entry to have a normalizing effect. Finally, it should be noted that the fourth model shown in Figure 7, using $t - 1$ as the moment of administration of the treatment, finds no effect. This means that there is no evidence of a violation of the parallel trends assumption.

Appendix C in the Supplemental Appendix presents some robustness check for these analyses. To check whether the results are sensitive to changes in the control group, Supplemental Figure C2 plots the effect sizes for models using all possible combinations of the three main parties in the control group. Their mean and median are very similar to the effect sizes reported in Figure 7. Supplemental Figure C3 shows the parallel trends plots comparing UKIP to each of the control parties individually: LibDem, Conservatives, and Labour. As with the two previous studies in this paper, a possible concern is that differences in the demographic characteristics of voters of UKIP in 2010 and 2015 may be driving the results. The fact that the results hold when the dependent variable is calculated using the demographic weights provided by BES lowers concerns that this might be the case. Supplemental Figure C1 tests for this alternative explanation more directly, showing no evidence of differences in the sociodemographic characteristics of voters between these two periods.

One possible criticism of this study is that, in 2015, UKIP not only made it to parliament, but its vote share grew substantially. This means that, unlike with the analyses of Study 1, one cannot identify the effect of parliamentary entry alone. This raises concerns of compound treatment effect, making it unclear whether the normalization effect is being driven by the UKIP’s parliamentary entry or its high vote share.

While not in position to fully dismiss this alternative mechanism, in the following I report some analyses that are more compatible with the parliamentary entry mechanism than with the increasing vote share mechanism. If the effect is being driven by the electoral success of UKIP, and not by its parliamentary entry, one should find a similar effect when replicating the analyses drawing upon elections for the European Parliament (EP). In the EP election of 2014, UKIP was the most voted party. Thus, if the effects shown in Figure 7 are due to the good electoral result of UKIP, instead of its parliamentary entry, one should expect that the 2014 EP election would have had a similar effect. I check whether this is the case by replicating the analyses shown in Figure 7 drawing upon EP elections instead of national elections, using data from the European Election Study (EES). The difference is that the treatment in these new analyses is not UKIP entering the UK parliament, but rather winning the EP election of 2014.

The results are shown in Figure 8. The figure shows no evidence of an effect similar to the one found for the analyses of national elections. There is no significant difference in the proportion of the official vote for UKIP that is reported is post-electoral surveys that follow the EP election won by UKIP as compared to previous ones. This evidence is more compatible with an interpretation that ascribes the effect shown in Figure 7 to the parliamentary entry of UKIP than to the increase in its vote share.

Figure 8.

Replication of the analyses in Figure 7 using data from elections for the European Parliament.

Conclusion

Once created, cultural equilibria significantly affect the preferences individuals are willing to display in public. In feeling that some preferences are likely to be punished by others, individuals have an incentive to falsify them. This paper has explored a causal mechanism by which stigmatized preferences can become normalized: the parliamentary representation of radical right actors. With resource to three different, complimentary studies, I have documented how such representation can make individuals more willing to express support for the radical right. Using comparative evidence, Study 1 showed that the vote share for RRPs is significantly less under-reported if these parties are represented in parliament. Study 2 looked into individual-level data, by comparing how much voters report liking RRPs in and out of parliament depending on their interview mode. Finally, Study 3 tested the argument drawing upon a single party throughout time, and in a country with a majoritarian (instead of proportional) electoral system. The three studies use different methods, variables, and case selections, but they come to the same conclusion: the parliamentary entry of an RRP makes individuals more confident in expressing support for the radical right.

These findings make two main contributions for existing literature. In the first place, they show how the presence of norm-breaching political actors in democratic institutions can bring about processes of norm change. Previous research has shown that institutional decisions can change individual perceptions as to which behaviors are deemed acceptable in a society (Tankard & Paluck, 2016). The findings of this paper highlight that not only the decisions that institutions take, but also the actors represented in them, can bring about such normative change.

This result yields an interesting paradox: the institutional presence of norm-breaching political actors can normalize those actors, even though they are often critical of those very institutions. This finding is particularly interesting considering that there is generalized pattern of decreasing trust in institutions (Catterberg & Moreno, 2006), and that radical right voters are among the most distrustful of political institutions (Zhirkov, 2014). A possible interpretation of this seeming paradox is that, while radical right voters have low trust in political institutions, they still take the presence in these institutions as a cue of what is accepted by others—that is, the individuals who are likely to sanction their views. If the parliamentary presence of an RRP makes the potential sanctioners perceive that the party has been normalized, radical right voters become less likely to be sanctioned for expressing their support. This can explain why parliamentary representation affects their willingness to express radical right support regardless of their own views of political institutions.

The findings also highlight the role of social norms in determining the extent to which individuals are willing to publicly express political preferences that they hold in private. In the case of socially sanctioned preferences such as support for norm-defying politicians, individuals are likely to collect cues from their social and political environment before deciding whether to publicly express their private preferences.

The results of the paper open avenues for future research. In the first place, subsequent studies might try to gauge whether this normalization extends beyond the answers that individuals provide to survey answers—for example, in their willingness to participate in far-right demonstrations or rallies. Another potential avenue for future research is the long-term effect of this normalization process. Does the institutional presence of radical right politicians bring about a new, stable cultural equilibrium, or is this effect limited to the short-term? While going beyond the scope of this study, answering these question would provide important insights into the way political events affect culture.

Supplemental Material

sj-pdf-1-cps-10.1177_0010414021997159 – Supplemental material for Parliamentary Representation and the Normalization of Radical Right Support

Supplemental material, sj-pdf-1-cps-10.1177_0010414021997159 for Parliamentary Representation and the Normalization of Radical Right Support by Vicente Valentim in Comparative Political Studies

Footnotes

Acknowledgements

I thank Tarik Abou-Chadi, Amalia Álvarez-Benjumea, Daniel Bischof, Alex Coppock, Elias Dinas, Vicky Fouka, Marta Fraile, Julian Garritzmann, Federica Genovese, Sara Hobolt, Rob Johns, Krzysztof Krakowski, David Laitin, Elli Palaiologou, Bilyana Petrova, Ana Ruipérez Núñez, Julia Schulte-Cloos, and Zeynep Somer-Topcu for their very helpful comments on the manuscript.

Author’s Note

Replication materials and code can be found at Valentim (2020). Previous versions of the paper were presented at the 2019 Conference of the Midwest Political Science Association, the 2019 Conference of the Italian Political Science Association, the EUI Political Behavior Colloquium, the Max Planck Institute for Research on Collective Goods, and the Institute for Social Sciences in Lisbon.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the Portuguese Foundation for Science and Technology, Grant SFRH/BD/135089/2017.

ORCID iD

Vicente Valentim

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biography

Vicente Valentim is a PhD candidate at the European University Institute. His work focuses on comparative politics, political behavior, and political culture. In September 2021 Vicente will be joining Nuffield College, University of Oxford as a Postdoctoral Prize Research Fellow in Politics.

References

Abou-Chadi

(2016). Niche party success and mainstream party policy shifts–How green and radical right parties diﬀer in their impact. British Journal of Political Science, 46(2), 417–436.

Abou-Chadi

Krause

(2018). The causal eﬀect of radical right success on mainstream parties’ policy positions: A regression discontinuity approach. British Journal of Political Science, 50(3), 1–19. https://doi.org/10.1017/S0007123418000029

Andreoni

Petrie

(2004). Public goods experiments without conﬁdentiality: A glimpse into fund-raising. Journal of public Economics, 88(7–8), 1605–1623.

Arias

(2019). How does media inﬂuence social norms? Experimental evidence on the role of common knowledge. Political Science Research and Methods, 7(3), 561–578.

Ariely

Bracha

Meier

(2009). Doing good or doing well? Image motivation and monetary incentives in behaving prosocially. American Economic Review, 99(1), 544–555.

Arzheimer

(2015). The AfD: Finally a successful right-wing populist eurosceptic party for germany? West European Politics, 38(3), 535–556.

Aslanidis

Kaltwasser

C. R.

(2016). Dealing with populists in government: The SYRIZA-ANEL coalition in Greece. Democratization, 23(6), 1077–1091.

Auers

Kasekamp

(2013). Comparing radical-right populism in Estonia and Latvia. In Ruth

Majid

Brigitte

(Eds.), Right-wing populism in Europe: Politics and discourse (pp. 235–248). Bloomsbury.

Bakker

de Vries

Edwards

Hooghe

Jolly

Marks

Polk

Rovny

Steenbergen

Vachudova

M. A.

(2015). Measuring party positions in Europe: The Chapel Hill expert survey trend ﬁle, 1999–2010. Party Politics, 21(1), 143–152.

10.

Bale

(2003). Cinderella and her ugly sisters: The mainstream and extreme right in Europe’s bipolarising party systems. West European Politics, 26(3), 67–90.

11.

Bale

Green-Pedersen

Krouwel

Luther

K. R.

Sitter

(2010). If you can’t beat them, join them? Explaining social democratic responses to the challenge from the populist radical right in Western Europe. Political Studies, 58(3), 410–426.

12.

Beaman

Chattopadhyay

Duﬂo

Pande

Topalova

(2009). Powerful women: Does exposure reduce bias? (MIT Press). The Quarterly Journal of Economics, 124(4), 1497–1540.

13.

Berent

M. K.

Krosnick

J. A.

Lupia

(2016). Measuring voter registration and turnout in surveys do oﬃcial government records yield more accurate assessments? Public Opinion Quarterly, 80(3), 597–621.

14.

Berning

C. C.

(2017). Alternative für Deutschland (AfD) – Germany’s new radical right-wing populist party. https://www.econstor.eu/bitstream/10419/181255/1/dice-report-2017-4-50000000000855.pdf

15.

Bernstein

Chadha

Montjoy

(2001). Overreporting voting: Why it happens and why it matters. Public Opinion Quarterly, 65(1), 22–44.

16.

Bicchieri

(2017). Norms in the wild: How to diagnose, measure, and change social norms. Oxford University Press.

17.

Bicchieri

Mercier

(2014). Norms and beliefs: How change occurs. In Xenitidou

Edmonds

(Eds.), The complexity of social norms (pp. 37–54). Springer.

18.

Bicchieri

Xiao

(2009). Do the right thing: But only if others do so. Journal of Behavioral Decision Making, 22(2), 191–208.

19.

Billig

(1991). Ideology and opinions: Studies in rhetorical psychology. SAGE.

20.

Billig

Condor

Edwards

Gane

Middleton

Radley

(1988). Ideological dilemmas: A social psychology of everyday thinking. SAGE.

21.

Bischof

Wagner

(2019). Do voters polarize when radical parties enter parliament? American Journal of Political Science, 63(4), 888–904. https://onlinelibrary.wiley.com/doi/abs/10.1111/ajps.12449

22.

Blair

Imai

(2012). Statistical analysis of list experiments. Political Analysis, 20(1), 47–77.

23.

Bursztyn

Egorov

Fiorin

(2020). From extreme to mainstream: The erosion of social norms. American Economic Review, 11, 3522–3548.

24.

Byford

(2009). Denial and repression of anti-semitism. Central European University Press.

25.

Calonico

Cattaneo

M. D.

Farrell

M. H.

Titiunik

(2017). rdrobust: Software for regression discontinuity designs. Stata Journal, 17(2), 372–404.

26.

Calonico

Cattaneo

M. D.

Titiunik

(2014). Robust data-driven inference in the regression-discontinuity design. Stata Journal, 10(2), 1–36.

27.

Cantoni

Hagemeister

Westcott

(2017). Persistence and activation of right-wing political ideology. Volkswirtschaftliche Fakultät Ludwig-Maximilians-Universität München.

28.

Carmines

E. G.

Stimson

J. A.

(1989). Issue evolution: Race and the transformation of American politics. Princeton University Press.

29.

Catterberg

Moreno

(2006). The individual bases of political trust: Trends in new and established democracies. International Journal of Public Opinion Research, 18(1), 31–48. https://academic.oup.com/ijpor/article/18/1/31/797103

30.

Chauchard

(2014). Can descriptive representation change beliefs about a stigmatized group? Evidence from rural India (JSTOR). American Political Science Review, 108(2), 403–422.

31.

Chwe

M. S. -Y.

(2013). Rational ritual: Culture, coordination, and common knowledge. Princeton University Press.

32.

Cowley

Kavanagh

(2016). The British general election of 2015. Springer.

33.

Cuesta

B. D.

Imai

(2016). Misunderstandings about the regression discontinuity design in the study of close elections. Annual Review of Political Science, 19, 375–396.

34.

De Vries

C. E.

Edwards

E. E.

(2009). Taking Europe to its extremes: Extremist parties and public euroscepticism. Party Politics, 15(1), 5–28.

35.

De Vries

C. E.

Hobolt

S. B.

(2012). When dimensions collide: The electoral success of issue entrepreneurs. European Union Politics, 13(2), 246–268.

36.

Dinas

Georgiadou

Konstantinidis

Rori

(2016). From dusk to dawn: Local party organization and party success of right-wing extremism. Party Politics, 22(1), 80–92.

37.

Dinas

Riera

Roussias

(2015). Staying in the ﬁrst league: Parliamentary representation and the electoral success of small parties. Political Science Research and Methods, 3(2), 187–204.

38.

de Lange

S. L.

Art

(2011). Fortuyn versus wilders: An agency-based approach to radical right party building. West European Politics, 34(6), 1229–1249.

39.

Ellinas

A. A.

(2013). The rise of Golden Dawn: The new face of the far right in Greece. South European Society and Politics, 18(4), 543–565.

40.

Ewing

(2001). Altruistic, egoistic, and normative eﬀects on curbside recycling. Environment and Behavior, 33(6), 733–764.

41.

Fan

(2018). Local polynomial modelling and its applications: Monographs on statistics and applied probability 66. Routledge.

42.

Fernández-Vázquez

Dinas

(2012). Projection bias in the survey placement of Spanish political parties: Diﬀerences between national and regional parties. South European Society and Politics, 17(3), 519–531.

43.

Gattinara

P. C.

(2017). Framing exclusion in the public sphere: Far-right mobilisation and the debate on Charlie Hebdo in Italy. South European Society and Politics, 22(3), 345–364. https://doi.org/10.1080/13608746.2017.1374323

44.

Gerber

A. S.

Green

D. P.

Larimer

C. W.

(2008). Social pressure and voter turnout: Evidence from a large-scale ﬁeld experiment. American Political Science Review, 102(1), 33–48.

45.

Gerber

A. S.

Rogers

(2009). Descriptive social norms and motivation to vote: Everybody’s voting and so should you. The Journal of Politics, 71(1), 178–191.

46.

Global Health Observatory. (2018). WHO|global health observatory (GHO) data. http://www.who.int/gho/en/

47.

Goldstein

N. J.

Cialdini

R. B.

Griskevicius

(2008). A room with a viewpoint: Using social norms to motivate environmental conservation in hotels. Journal of Consumer Research, 35(3), 472–482.

48.

Gottlieb

(2016). Common knowledge and voter coordination: Experimental evidence from Mali. In Blais

Laslier

J. -F.

Van der Straeten

(Eds.), Voting experiments (pp. 89–113). Springer International Publishing. https://doi.org/10.1007/978-3-319-40573-5_5

49.

Gulzar

Khan

M. Y.

(2017). Why do citizens become politicians? Experimental evidence on the social dimensions of candidacy (Technical report working paper).

50.

Han

K. J.

(2015). The impact of radical right-wing parties on the positions of mainstream parties regarding multiculturalism. West European Politics, 38(3), 557–576.

51.

Hobolt

S. B.

De Vries

C. E.

(2015). Issue entrepreneurship and multiparty competition. Comparative Political Studies, 48(9), 1159–1185.

52.

Ignazi

(1992). The silent counter-revolution. European Journal of Political Research, 22(1), 3–34.

53.

Imbens

Kalyanaraman

(2012). Optimal bandwidth choice for the regression discontinuity estimator. The Review of Economic Studies, 79(3), 933–959.

54.

Imbens

G. W.

Lemieux

(2008). Regression discontinuity designs: A guide to practice. Journal of Econometrics, 142(2), 615–635.

55.

Karp

J. A.

Brockington

(2005). Social desirability and response validity: A comparative analysis of overreporting voter turnout in ﬁve countries. Journal of Politics, 67(3), 825–840.

56.

Kessler

J. B.

(2011). Signals of support and public good provision. Pennsylvania. http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.188.4392

57.

Kluknavská

Smolík

(2016). We hate them all? Issue adaptation of extreme right parties in Slovakia 1993–2016. Communist and Post-Communist Studies (The transformations of far right and far left in Europe), 49(4), 335–344.

58.

Kormos

Giﬀord

Brown

(2015). The inﬂuence of descriptive social norm information on sustainable transportation behavior: A field experiment. Environment and Behavior, 47(5), 479–501.

59.

Kuklinski

J. H.

Cobb

M. D.

Gilens

(1997). Racial attitudes and the “New South”. The Journal of Politics, 59(2), 323–349.

60.

Kuran

(1987). Preference falsiﬁcation, policy continuity and collective conservatism. The Economic Journal, 97(387), 642–665.

61.

Kuran

(1995). Private truths, public lies: The social consequences of preference falsiﬁcation. Harvard University Press.

62.

Lee

D. S.

Lemieux

(2010). Regression discontinuity designs in economics. Journal of Economic Literature, 48(2), 281–355. https://www.aeaweb.org/articles?id=10.1257/jel.48.2.281

63.

March

(2011). Radical left parties in Europe (1st ed.). Routledge.

64.

Mareš

Stojar

(2016). Extreme right perpetrators. In Fredholm

(Ed.), Understanding lone actor terrorism (pp. 82–102). Routledge.

65.

McConahay

J. B.

Hough

J. C.

(1976). Symbolic racism. Journal of Social Issues, 32(2), 23–45.

66.

McCrary

(2008). Manipulation of the running variable in the regression discontinuity design: A density test. Journal of Econometrics (The regression discontinuity design: Theory and applications), 142(2), 698–714.

67.

McDonald

(2005). Reporting bias. In Radcliﬀ

Best

(Eds.), Polling in America: An encyclopedia of public opinion. Greenwood Press.

68.

Mudde

(2004). The populist zeitgeist. Government and Opposition, 39(4), 542–563.

69.

Mudde

(2007). Populist radical right parties in Europe. Cambridge University Press.

70.

Mudde

(2013). Three decades of populist radical right parties in Western Europe: So what? European Journal of Political Research, 52(1), 1–19.

71.

Mudde

(2014). Fighting the system? Populist radical right parties and party system change. Party Politics, 20(2), 217–226.

72.

Noelle-Neumann

(1974). The spiral of silence a theory of public opinion (Oxford Academic). Journal of Communication, 24(2), 43–51. https://academic.oup.com/joc/article/24/2/43/4553587

73.

Norris

(2005). Radical right: Voters and parties in the electoral market. Cambridge University Press.

74.

Panagopoulos

(2010). Aﬀect, social pressure and prosocial motivation: Field experimental evidence of the mobilizing eﬀects of pride, shame and publicizing voting behavior. Political Behavior, 32(3), 369–386.

75.

Pedersen

Grønhøj

Thøgersen

(2015). Following family or friends. Social norms in adolescent healthy eating. Appetite (Social inﬂuences on eating), 86, 54–60.

76.

Pytlas

(2015). Radical right parties in Central and Eastern Europe: Mainstream party competition and electoral fortune. Routledge.

77.

Riker

W. H.

(1986). The art of political manipulation. Yale University Press.

78.

Rimal

R. N.

Real

(2005). How behaviors are inﬂuenced by perceived norms. Communication Research, 32(3), 389–414.

79.

Rydgren

(2005). Is extreme right-wing populism contagious? Explaining the emergence of a new party family. European Journal of Political Research, 44, 413–427.

80.

Rydgren

(2007). The sociology of the radical right. Annual Review of Sociology, 33(1), 241–262. https://doi.org/10.1146/annurev.soc.33.040406.131752

81.

Schain

M. A.

(2006). The extreme-right and immigration policy-making: Measuring direct and indirect eﬀects. West European Politics, 29(2), 270–289.

82.

Sniderman

P. M.

Piazza

Tetlock

P. E.

Kendrick

(1991). The new racism. American Journal of Political Science, 35(2), 423–447.

83.

Tankard

M. E.

Paluck

E. L.

(2016). Norm perception as a vehicle for social change. Social Issues and Policy Review, 10(1), 181–211. https://spssi.onlinelibrary.wiley.com/doi/abs/10.1111/sipr.12022

84.

Valentim

(2020). Replication data for: Parliamentary representation and the normalization of radical right support. Harvard Dataverse Type: Dataset. https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/Z6S26H

85.

World Bank. (2018). World bank open data. https://data.worldbank.org/

86.

Zaller

(1992). The nature and origins of mass opinion. Cambridge University Press.

87.

Zhirkov

(2014). Nativist but not alienated: A comparative perspective on the radical right vote in Western Europe. Party Politics, 20(2), 286–296. https://doi.org/10.1177/1354068813511379

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

2.46 MB