Legitimize or Delegitimize? Mainstream Party Strategy toward (Former) Pariah Parties and How Voters Respond

Abstract

Mainstream parties have often shifted from initially portraying new competitors as undemocratic pariahs (i.e., a delegitimizing strategy) to portraying the same parties later as democratic (i.e., a legitimizing strategy). I argue that voters follow mainstream parties’ legitimizing strategies in their legitimacy evaluations of these parties. I investigate this argument with two independent survey experiments and a quantitative media content analysis in two countries that differ sharply in the nature of party competition—from mainstream parties delegitimizing a far-right party (i.e., Germany) to mainstream parties legitimizing it (i.e., Sweden). I find strong evidence that (a) mainstream parties can effectively legitimize pariah parties in the eyes of voters, (b) turning back to delegitimization has little effect, (c) legitimization is no less effective in the face of a third party’s delegitimizing strategy, and (d) legitimization resonates beyond co-partisans. The results suggest that mainstream party legitimization of pariah parties has far-reaching consequences.

INTRODUCTION

The rise of new parties is a recurring feature of multi-party democracies. Contemporary politics in Western Europe is characterized by the rise of far-right and far-left parties (Bale et al. 2009; Kriesi et al. 2006). Historically, Green parties and, even earlier, Social Democratic parties have disrupted and fundamentally transformed party systems throughout Western Europe (De Vries and Hobolt 2020).

In their initial stages, the legitimacy of many of these parties has been contested amongst voters and mainstream parties alike (Downs 2012; Van Spanje 2010). As a first reaction, mainstream parties have often chosen to portray their new competitors as undemocratic—for example, by ostracizing and rhetorically “demonizing” them (Schwörer and Fernández-García 2020; Van Heerden and Van der Brug 2017; Van Spanje 2010). I call this type of mainstream party behavior a delegitimizing strategy. Following extant literature, I refer to parties that have been delegitimized by one or more mainstream parties—currently or in the recent past—as pariah parties (Leander 2022; Moffitt 2021; Van Spanje and De Graaf 2018).

However, periods of (unanimous) delegitimization have often been limited in time and mainstream parties began to legitimize pariah parties, for example by forming coalitions with them (Akkerman, De Lange, and Rooduijn 2016; De Lange 2012) or by emphasizing rhetorically that they have been democratically elected by and represent the interests of voters. I call this type of mainstream party behavior a legitimizing strategy. A wide array of “actors as diverse as the far-left Rifondazione Comunista (RC—Communist Refoundation) in Italy, the populist right-wing Freiheitliche Partei Österreichs (FPÖ—Freedom Party) in Austria and Green and regionalist movements in various states” (McDonnell and Newell 2011, 444) underwent this process: initially delegitimized but later legitimized. In fact, very few consistently electorally successful parties have been subject to sustained delegitimization and were not legitimized at some point in their existence (Axelsen 2023, 8).

In this article, I ask whether voters follow mainstream party legitimization of pariah parties in their own evaluations of these parties’ legitimacy—which I define as voters’ perceptions of them as being democratic and not threatening democracy. Moreover, I examine whether (a) mainstream parties can effectively turn back to delegitimization after a period of legitimization, (b) mainstream party legitimization is still effective in the face of a competing, delegitimizing strategy by a third party, and (c) mainstream party legitimization resonates beyond co-partisans.

To date, we lack a theory linking mainstream parties’ legitimizing vis-à-vis delegitimizing strategies to voters’ evaluations of the targeted parties’ legitimacy. A few studies have examined how treating parties as pariahs affects these parties’ electoral fortunes (Van Spanje and Van der Brug 2009; Van Spanje and Weber 2017; Van Spanje and De Graaf 2018). However, so far, scholars have only alluded to the possibility that mainstream parties’ (de-)legitimizing strategies influence voters’ legitimacy evaluations of pariah parties (Art 2007; Ivarsflaten 2006; Moffitt 2021). More specifically, despite the pervasiveness of this phenomenon, we lack knowledge about voters’ reactions when mainstream parties abandon a delegitimizing strategy and legitimize a pariah party.

This is an important limitation of the current literature. Extensive research suggests that parties widely perceived as illegitimate face numerous challenges: they are strongly disliked (Bantel 2023; Wagner 2024), struggle to garner support for their policies (Blinder, Ford, and Ivarsflaten 2013), are unpopular coalition partners (Kelemen et al. 2023), spark countermobilization (Hager et al. 2021), and struggle to attract competent personnel (Art 2011) as well as broader segments of the electorate (Bos and van der Brug 2010). It is therefore not surprising that pariah parties themselves often aim to appear more legitimate (Moffitt 2021; Paxton and Peace 2020). Against this backdrop, mainstream parties may have a powerful tool at their disposal when competing against new parties if they can influence whether voters view these parties as legitimate by means of (de-)legitimizing strategies.

Advancing the literature, I present a theory that links (de-)legitimizing strategies to voters’ evaluations of pariah parties’ legitimacy. Drawing on the party cue literature, I first argue that voters follow legitimizing strategies in their own evaluations of the pariah’s legitimacy. I contend that (a) there is room for party cues because evaluating the legitimacy of controversial parties is a complex and salient issue for most voters and (b) mainstream parties are perceived as trustworthy cue-givers on this issue. Second, I hypothesize that the effectiveness of mainstream parties’ legitimizing strategies is mitigated by other parties’ delegitimizing strategies. Finally, I argue that co-partisans of the party pursuing these strategies are more influenced than all others.

I test my arguments with a quantitative media content analysis and two independent pre-registered survey experiments, ¹ focusing on the response strategies of the mainstream right vis-à-vis the far-right in Germany and Sweden. Both the Alternative for Germany (AfD) and the Sweden Democrats (SD) have been subject to strong delegitimization (i.e., they are pariahs). Yet, recently, the German Christian Democratic Union of Germany (CDU) and the Swedish Moderate Party (M) have taken diverging positions on the issue. As documented by the content analysis, CDU consistently delegitimized AfD after it entered parliament in 2017. In contrast, whereas M initially also delegitimized SD when it entered parliament in 2010, M later adopted a legitimizing strategy toward SD. In two independent large-scale, representative experiments in each country, I randomly exposed voters to either delegitimizing or legitimizing mainstream party cues. Additionally, in the Swedish experiment, I exposed some respondents to both types of cues—coming from different parties—simultaneously.

Overall, the results suggest that mainstream party legitimization of pariah parties has far-reaching consequences for public opinion. First, voters follow mainstream parties in their own evaluations of pariah parties’ legitimacy when mainstream parties legitimize. Second, turning back to delegitimization after a period of legitimization has little effect. Third, a legitimizing strategy is no less effective when countered by a delegitimizing strategy from a third party. Fourth, these effects are not limited to the supporters of the party pursuing a legitimizing strategy.

These findings have important implications for our understanding of party competition in times of disruption by new parties with populist and radical profiles. Extant research has investigated how mainstream parties can influence the fate of these parties by adjusting their policy platform (Abou-Chadi 2014; Meguid 2005) or by emphasizing their competence (De Vries and Hobolt 2020). My findings show that mainstream parties may have another powerful tool at their disposal: (de-)legitimizing strategies. Mainstream parties can act as gatekeepers, having a strong impact on whether new parties enter the mainstream in the minds of voters or not by means of (de-)legitimizing strategies.

THEORETICAL BACKGROUND

Defining the Key Concepts: Legitimacy Evaluations, (De-)Legitimizing Strategies, Pariah and Mainstream Parties

Although many scholars argue that legitimacy in the eyes of voters is a crucial asset to parties (Art 2007; Ivarsflaten 2006), few provide a definition of what party legitimacy means.

I conceptualize legitimacy evaluations as voters’ perceptions of parties as democratic and not posing a threat to democracy. I derive my conceptualization from Tyler (2006, 376), who defines legitimacy from the perspective of individuals as “the belief that authorities, institutions, and social arrangements are appropriate, proper, and just.” Following others (e.g., Bos and van der Brug 2010; Van Spanje and Azrout 2019), I contend that for parties, as central institutions in modern democracies, voters’ perceptions of them as being “appropriate, proper, and just” is best captured by their voters’ perceptions of parties as democratic and not posing a threat to democracy. While voters may agree that a party is democratic simply because it was successful in democratic elections, they may still view a party’s presence as a threat to democracy (Downs 2012). Conversely, merely not being a threat to democracy might not imply that voters view a party as democratic—if voters think that democratic parties have more obligations than that. Hence, both aspects are part of my conceptualization of legitimacy evaluations.

Given that (self-reported) public support for democracy in Western Europe is high—even if many of them are “democrats in name only” (Wuttke, Gavras, and Schoen 2020, 416)—it seems unlikely that a party is perceived as undemocratic or a threat to democracy and simultaneously “appropriate, proper, and just”. Of course, one might conceive of legitimacy evaluations as consisting of more than voters’ perceptions. Yet this entails the risk of conflating the concept of legitimacy with other concepts such as affect (Wagner 2024) or tolerance (Petersen et al. 2010). Hence, I use this minimalist definition.

Following my conceptualization of legitimacy evaluations, I conceptualize mainstream parties’ delegitimizing strategies as portrayals of other parties as undemocratic and/or posing a threat to democracy. Such portrayals can take various forms, beyond simply calling a party explicitly undemocratic or threatening to democracy. For example, it can manifest itself as the systematic exclusion of a party—often called a ‘cordon sanitaire’ (Heinze 2018; Van Spanje and Van Der Brug 2007). Typically, such mainstream party behavior is accompanied by rhetorical demonization of a party (Schwörer and Fernández-García 2020; Van Heerden and Van der Brug 2017).

I conceptualize mainstream parties’ legitimizing strategies as portrayals of pariah parties as democratic and not posing a threat to democracy. Again, such portrayals can take various forms and rarely manifest themselves by explicitly calling a pariah party democratic or not threatening democracy. Perhaps the most explicit form of legitimization is (advocating for) inclusion in government (McDonnell and Newell 2011), as the arguably most powerful democratic institution a party can get access to. Yet I also conceive of other forms of collaboration or any rhetoric that suggests that a party is no different from other parties as legitimization.

Albeit closely related to perceptions/portrayals of a party as democratic, my concepts of legitimacy evaluations and (de-)legitimizing strategies explicitly acknowledge that a party can be democratic—in the sense that people voted for it in fair and free elections—yet still threaten democracy due to its conduct. As such, if, for instance, a mainstream party portrays a competitor not as outright undemocratic, but still portrays it as a threat to democracy, this qualifies as a delegitimizing strategy.

I conceive of parties that have been delegitimized by one or more mainstream parties—currently or in the recent past—as pariah parties. Importantly, this conceptualization does not restrict a priori the pool of parties that have been delegitimized to parties with certain ideological attributes (Downs 2012; Moffitt 2021)—such as radical, populist, anti-establishment, or anti-system parties. Many of these parties have been legitimized a long time ago. Hence, my conceptualization acknowledges that whether a party has been ascribed pariah status or not is ultimately the result of political debate (i.e., delegitimization) and not predetermined by its “objective” characteristics (Moffitt 2021, 5; Van Spanje 2010, 355). Moreover, it excludes parties whose legitimacy is not (or no longer) contentious, as they cannot be (further) legitimized. Finally, it is important to note that my conceptualization of pariah parties also includes parties that are delegitimized by some, but legitimized by other parties (Leander 2022).

Following Mair (2006), I define mainstream parties as those parties, which define the government alternatives. These parties constitute the core of party systems and have, in contrast to more peripheral parties, considerably more agency in (de-)legitimizing pariah parties. Most importantly, it is primarily up to these parties whether a competitor is granted or denied access to government and other important democratic institutions. Therefore, more than any other party, these parties decide if a competitor is delegitimized or legitimized.

To illustrate how a legitimizing strategy, following a period of delegitimization manifests itself empirically, consider two examples from Sweden. First, a delegitimizing statement by Frederik Reinfeldt, the then leader of M, from 2013: “SD is a party that leans toward racism and xenophobia. They should be isolated from political influence” (Sveriges Radio 2013). Second, a legitimizing statement justifying collaborations with SD by Ulf Kristersson, the current leader of M, from 2022: “They [SD] are certainly not alone in that in Swedish politics, but they have roots that are unpleasant, just like the Left Party. They came to terms with their history in the 90s” (SVT Nyheter 2022).

It is important to note that either strategy, delegitimization or legitimization, entails considerable potential costs and benefits to mainstream parties—which depend on voter reactions to these strategies as well as other factors. Mainstream parties’ choice to (de-)legitimize might be, for example, motivated by a trade-off between office- and vote-seeking motivations. On the one hand, if effective, delegitimization might be an attractive strategy for mainstream parties from a vote-seeking perspective (Blinder, Ford, and Ivarsflaten 2013; Bos and van der Brug 2010). On the other hand, when mainstream parties legitimize pariah parties, they win a potential coalition partner (De Lange 2012; Van Spanje 2010). Yet legitimization might also be costly for mainstream parties if voters do not follow this strategy in their own evaluations of pariah parties and accordingly punish mainstream parties. This tradeoff is particularly severe when mainstream parties choose to delegitimize or legitimize ideological neighbors, as they compete for similar voters and are potential coalition partners. Next to voters’ (anticipated) reactions, mainstream parties’ incentives to legitimize or delegitimize likely vary with the context and attributes of the targeted party. For example, maintaining a strongly delegitimizing strategy may become less tenable as a pariah party becomes electorally more successful and ideologically moderate. Conversely, legitimization might be a less viable strategy toward very extreme pariahs that openly oppose democracy and shun the establishment.

Existing Research: Voters’ Responses to Mainstream Parties’ (De-)Legitimizing Strategies

So far, we know little about the consequences of (de-)legitimizing strategies for public opinion. Extant research has argued that ostracism (i.e., delegitimization) affects how effective voting for the targeted party is perceived by voters. Voters might deem a party irrelevant if it is systematically excluded by mainstream parties. However, such effects are conditional on the programmatic strategies pursued by mainstream parties (Van Spanje and De Graaf 2018) as well as on the institutional context. Under certain conditions, parties may even benefit electorally from mainstream parties’ ostracism (Van Spanje and Van der Brug 2009; Van Spanje and Weber 2017).

Delegitimizing and legitimizing strategies might have indirect effects on voters’ evaluations of the targeted parties. For example, scholars have argued that the inclusion of new parties in government affects their institutionalization and visibility (Bolleyer, Van Spanje, and Wilson 2012), which in turn might influence competence evaluations. Additionally, delegitimizing strategies might make populist parties’ anti-establishment messages more appealing to voters, while legitimizing strategies might have the opposite effect (Fallend and Heinisch 2015). Moreover, scholars have argued that parties become more radical when confronted with delegitimizing strategies (Van Spanje and Van Der Brug 2007), which could ultimately affect voters’ perceptions and evaluations of them.

Focusing on affect, Ekholm, Bäck, and Renström (2022) show that informing voters that Swedish mainstream party MPs are increasingly positive about cooperating with SD has a positive effect on voters’ likeability of SD. Yet it remains unclear whether such mainstream party behavior also shapes a perhaps more fundamental aspect of voters’ evaluation of parties: their legitimacy.

THE ARGUMENT

Although scholars have alluded to the possibility that (de-)legitimizing strategies directly shape voters’ legitimacy evaluations (Art 2007; Ivarsflaten 2006; Moffitt 2021), we lack a theory on this relationship. To develop my hypotheses on voters’ responses to mainstream party legitimization vis-à-vis delegitimization, I draw on the party cue literature. Party cues “link a party to a stand on an issue” (Bullock 2011, 497). I conceive of (de-)legitimizing strategies as party cues ² providing information to voters on where a mainstream party stands on the issue of a controversial competitor’s legitimacy. When mainstream parties legitimize, voters, who were initially exposed to mainstream party cues, suggesting a party is undemocratic (i.e., delegitimization), receive cues, suggesting the pariah is now democratic.

A large body of research shows that party cues shape public opinion. Party cues can influence voters’ opinions on political issues via framing—that is, selectively emphasizing certain relevant considerations or aspects of an issue (Nelson, Clawson, and Oxley 1997). Rather than changing the weight of existing considerations, party cues may also influence voters by providing new information on a given issue (Lenz 2009).

As evaluating the legitimacy of a controversial party is presumably a complex, yet salient issue to most voters, I expect that there is room for cueing effects (Bisgaard and Slothuus 2018; Slothuus and de Vreese 2010). Many competing considerations are potentially relevant for voters’ legitimacy evaluations of controversial parties. For example, on the one hand, these parties have been democratically elected into parliament. On the other hand, due to their often populist and radical platform, voters may perceive these parties as being at odds with democracy.

Furthermore, I expect that a majority of voters relies on mainstream parties as trustworthy cue-givers regarding the legitimacy of controversial parties. I contend that most voters view mainstream parties as central actors that represent, seek to preserve, and seek to safeguard democratic politics—at least in democracies with a history of stable and strong mainstream parties (Brader, Tucker, and Duell 2012). Thus, most voters should readily accept advice from mainstream parties because they are trusted sources on this issue.

Overall, I therefore hypothesize:

H1: Voters evaluate pariah parties as more legitimate when exposed to a legitimizing mainstream party strategy as compared to a delegitimizing mainstream party strategy.

Legitimizing strategies are rarely adopted in concert by all (mainstream) parties and seldom go unchallenged. Hence, I examine how mainstream parties’ legitimizing strategies influence voters in the face of a competing, delegitimizing strategy sponsored by a third party. In such a scenario, voters receive contrasting information on the issue of pariah party legitimacy from different parties. They receive not only a legitimizing mainstream party cue but also a delegitimizing cue sponsored by a third party. ³

There is only limited literature to draw on to develop a hypothesis about how voters react in such a scenario. The general literature on counter-framing suggests that messages become less effective if they are countered or opposed by another message in a competitive environment (Chong and Druckman 2007; 2013; Druckman 2004). Although it is less clear how competing messages affect voters when party sponsors are involved, I propose the following hypothesis:

H2: Mainstream parties’ legitimizing strategies are less effective when countered by third parties’ delegitimizing strategies.

It is an established finding in the literature that partisanship conditions voters’ reactions to party cues (Leeper and Slothuus 2014). I contend that legitimization is less effective for out-partisans and those who do not identify with any party than for partisans of the mainstream party pursuing those strategies. According to partisan-motivated reasoning theory, voters are more likely to follow a cue when it is sponsored by the party they identify with (Leeper and Slothuus 2014). This bias is particularly strong for salient and contentious issues (Druckman, Peterson, and Slothuus 2013; Slothuus and de Vreese 2010), such as the legitimacy of pariah parties.

I do not expect any effects of legitimizing strategies on voters who identify with the targeted party. First, these voters are presumably generally reluctant to take advice from mainstream parties, as they typically hold populist, anti-establishment views (Rooduijn, van der Brug, and De Lange 2016). Second, due to ceiling effects, there is arguably no room left for (further) legitimization effects on these voters. Those who identify with the targeted party already perceive it as legitimate.

Therefore, I propose the following hypothesis:

H3: Mainstream parties’ legitimizing strategies are most effective amongst voters who identify with the mainstream party pursuing this strategy.

RESEARCH DESIGN

Case Selection

Empirically, I focus on the mainstream right and its response to the far right. Not only is the legitimacy of the far-right arguably more contested than any other party family in contemporary politics, but also has the mainstream right recently often shifted toward legitimization of the far-right. I investigate my hypotheses in Sweden and Germany. These two cases are well-suited to test my hypotheses experimentally. While the German AfD and the Swedish SD have multiple similarities, the two countries differ sharply in the nature of party competition—yielding high comparability and generalizability.

Since AfD and SD entered the national parliament, they have become major players in their party systems. AfD became the third largest party after CDU and the mainstream left Social Democrats (SPD) when it entered the national parliament in 2017. Subsequently, AfD radicalized severely and extremist actors within the party organization became increasingly powerful (Arzheimer 2019). Yet the party has consistently been electorally successful since 2017.

SD entered parliament for the first time in 2010 and has been steadily gaining ground ever since. While SD barely made it into parliament in 2010, by the 2022 national elections it had become the second largest party with a vote share of more than 20 percentage points—behind the mainstream left Social Democrats (S) and ahead of M. With roots in neo-Nazi movements, SD has made great efforts to appear more moderate in the eyes of the public, not least by expelling many extremist party members (Widfeldt 2008).

However, as AfD and SD have become increasingly powerful, CDU and M have chosen fundamentally different response strategies. CDU has adopted a strict policy of non-cooperation with AfD (CDU 2020). Similar to CDU’s response to AfD, M initially ruled out any collaboration with SD after it entered parliament (Bolin, Dahlberg, and Blombäck 2022). However, this strategy led to a severe problem for M as SD grew stronger. Sweden has a tradition of bloc politics—that is, either a left bloc, led by S, or a right bloc, led by M, forms the government, while no governments are formed with parties of both blocs. As SD got stronger, M therefore had virtually no chance of leading a government without the support of SD (Aylott and Bolin 2019). Unlike CDU, M changed course toward legitimization. In 2019, M began negotiating possible collaborations with SD after the Liberals and Center party (C), formerly part of the right bloc, decided to support a government led by S after the election. After the 2022 general election—which took place six months after my experiment was fielded—M formed a minority government supported by SD (Aylott and Bolin 2023).

Several contextual features of Germany and Sweden secure high generalizability. First, both AfD and SD are hard cases for my theory. Due to Germany’s right-wing historical legacy and AfD’s radicalization, German voters are arguably particularly resistant to legitimizing efforts by political elites. While Sweden does not have a right-wing fascist history and SD has adopted a moderation strategy, it is one of the few successful far-right parties with clear roots in neo-Nazi movements and has been delegitimized much longer than most electorally successful parties (Axelsen 2023). Hence, persuading the Swedish public that SD is now a “normal”, democratic party might be difficult. However, legitimization might also be more effective in Sweden than in Germany due to a “habituation” effect. Second, albeit different, both are typical cases. CDU’s delegitimizing strategy constitutes a typical “first” mainstream party reaction to new parties with a (notionally) populist and radical platform, which could also be initially observed for M. Sweden constitutes a typical case of a legitimizing strategy after sustained electoral success despite delegitimization. Hence, the two cases are ideally suited to yield generalizable results.

The Context of the Experiments: A Media Content Analysis Mapping Mainstream Party Response Strategies toward AfD and SD

To illuminate the context of my experiments, I present the results of a quantitative media content analysis mapping CDU’s and M’s (de-)legitimizing strategies toward AfD and SD over time. This content analysis (a) illustrates the prevalence of (de-)legitimizing strategies and (b) uncovers potential methodological challenges regarding the experimental identification of possible public opinion effects of these strategies.

The observation period of the content analysis started when AfD/SD entered the national parliament for the first time: 2010 in Sweden and 2017 in Germany. It ends by the end of 2020 in Germany and by the end of 2021 in Sweden—just before the experiments were fielded (i.e., January 2021 in Germany and March 2022 in Sweden).

I manually coded every statement made by CDU/M representatives containing a reference to AfD/SD that appeared in the two major newspapers Dagens Nyheter (Sweden) and Süddeutsche Zeitung (Germany) during these time frames. In Germany, I coded 371 statements by CDU about AfD. In Sweden, I coded 255 statements by M about SD. The estimate of interest here is the proportion of (de-)legitimizing statements, which are, in line with my conceptualization, portraying AfD/SD as (un-)democratic/(not) posing a threat to democracy. Due to the lower number of coded statements in Sweden (see Appendix A4), I present this proportion per year in Germany, but per two years in Sweden. Further methodological details, the results for other parties than CDU/M, and typical statements for each coding category can be found in Appendix A.

Figure 1, which displays the proportion of delegitimizing statements, shows how CDU’s and M’s response strategies gradually diverged. CDU consistently pursued a strongly delegitimizing strategy toward AfD. In 2020, AfD was portrayed as undemocratic in 81% of all coded CDU statements. As shown in the right panel of Figure 1, like CDU, M strongly delegitimized SD when it entered parliament in 2010. However, in 2020/21, SD was portrayed as undemocratic in only 8% of all coded M statements.

Figure 1. Share of CDU’s and M’s Delegitimizing Statements toward AfD and SD over Time

Figure 2, which displays the proportion of legitimizing statements, shows the opposite pattern. In 2020, only 5% of all CDU statements toward AfD were legitimizing. Legitimizing statements were consistently rare, albeit slightly more frequent in 2017 than in more recent years. M, in contrast, clearly has switched to a legitimizing strategy. In 2020/21, 59% of its representatives’ statements toward SD portrayed it as a democratic party, while such statements were very rare in the years shortly after SD’s entrance into parliament.

Figure 2. Share of CDU’s and M’s Legitimizing Statements toward AfD and SD over Time

Overall, first, the content analysis shows that much of mainstream right parties’ communication about AfD/SD either portrays them as undemocratic or as democratic, indicating that (de-)legitimization is a salient dimension of mainstream right parties’ response to them. This motivates the question of whether voters are influenced by these strategies.

Second, it reveals how pretreatment might undermine the experimental identification of “real” effects. Respondents are pretreated if they already know from real-world exposure to political debate where the parties stand and what their arguments are before they are told in the experiment (Slothuus 2016). If respondents’ opinions were already influenced by a message akin to the message used as treatment in the experiment before the experiment, the difference between the control and the treatment group only captures the effect of additional exposure to this message (Druckman and Leeper 2012). Therefore, paradoxically, the more often (de-)legitimizing messages have occurred “in the real world” and were effective, the less likely are effects in the experiment (Slothuus 2016). This means that, because of pretreatment, the legitimizing condition is likely to have stronger effects in Germany than in Sweden. Conversely, the opposite holds for the delegitimizing condition.

Given these contextual differences, in conjunction, the German and the Swedish experiments allow for a comprehensive examination of the effectiveness of (de-)legitimizing strategies. The German experiment is ideally suited to examine voters’ reactions when a mainstream party breaks the line and legitimizes in a context of formerly unanimous delegitimization (in fact, all parties in the German national parliament pursue a delegitimizing strategy; see Appendix A4). The Swedish experiment, in contrast, constitutes a harder test of my theory as legitimization is pretreated. Furthermore, the Swedish experiment allows for a good test of voters’ reactions when mainstream parties return to delegitimization after a period of legitimization. In other words, the legitimizing condition in Germany and the delegitimizing condition in Sweden deviate from the strategy encountered by respondents before the experiments and hence simulate a change in the strategy taken by the mainstream right.

The Survey Experiments

I implemented the two survey experiments in collaboration with YouGov (Daur 2024). The German experiment was fielded in January 2021 and the Swedish experiment in March 2022. The sample of the Swedish experiment (N = 2,029) is representative of the Swedish adult population in terms of gender, age, region, and education. The sample of the German experiment (N = 2,079) is representative of the German adult population in terms of gender, age, region, and voting behavior in the national 2017 election. The sample demographics can be found in Appendix B3. The questionnaires with the exact wording of all questions as well as all vignettes can be found in Appendices B7 and B8.

The Experimental Conditions

In the German experiment, one treatment condition received a delegitimizing CDU message, while the other condition received a legitimizing CDU message. The control condition read nothing before answering the outcome measures. The Swedish experiment had four conditions. Following the German experiment, one condition received a delegitimizing M message, another condition received a legitimizing M message, and, again, the control group read nothing.

A fourth condition in the Swedish experiment received a legitimizing M message, but next to this also a delegitimizing message by the small, ideologically centrist Center party (C). This condition serves to test my second hypothesis postulating that mainstream parties’ legitimizing strategies are less effective in the face of a third party’s delegitimizing strategy. I did not include this condition in the German experiment, as I conducted it first and used the insights gained from it to inform the design of the Swedish experiment.

I chose C as the sponsor of the delegitimizing message in this condition because C is arguably a credible and persuasive sponsor of this message. C, in fact, pursues a delegitimizing strategy—as my content analysis shows (see Appendix A4). Another plausible choice as the message sponsor would have been the mainstream left S. However, although using S as a sponsor of the delegitimizing message in this condition would have been equally credible, it might be less persuasive. First, for C, a delegitimizing strategy is more “costly” than for S, because C is ideologically closer to SD compared to S. C even left the right bloc after many years of collaboration in 2018 due to controversies sparked by M’s legitimizing strategy toward SD (Leander 2022, 338–9). Voters are thus likely to interpret a delegitimizing strategy pursued by C as more sincere than the same strategy pursued by the more ideologically distant S. Second, volatility among the alliance parties has been high, and many Swedish voters identify as being “left or right” rather than with a specific party. Hence, the fact that C is from the same, center-right camp as M (i.e., the Alliance), arguably makes the party a persuasive sponsor of the competing, delegitimizing message. Yet the rather small number of C partisans is potentially limiting the persuasiveness of its message. Moreover, pretreatment might impede the experimental identification of an effect. It is also important to note that C is not a mainstream party according to my definition of mainstream parties, whereas S is a mainstream party.

In each experimental condition, respondents read a series of four mock newspaper articles mimicking real articles published on the webpages of Frankfurter Allgemeine Zeitung and Dagens Nyheter. All mock newspaper articles, as seen by the respondents, can be found in Appendices B5 and B6.

The articles mimic real current debates in German and Swedish politics. This approach necessitates the use of different treatments in the two experiments. Given that the real-world mainstream right strategy toward AfD/SD differs sharply, the German treatment would not have been meaningful and credible in Sweden and vice versa.

In the German experiment, I manipulated (de-)legitimizing CDU messages in relation to a fictitious controversy about a possible boycott of a panel debate taking place at a high school due to AfD’s participation. The messages each contain CDU’s position—in favor or against a boycott—as well as an argument justifying this position by portraying AfD as (un-)democratic in the form of direct quotes. Below are examples of delegitimizing and legitimizing quotes from the German treatment vignettes. Manipulations are underlined.

German experiment, a quote from the delegitimizing condition (example): Since AfD “regularly disrespects democratic institutions”, as a democrat one cannot “treat the party like a democratic party”. Therefore, he pleads for “not sending a representative of our party to this event if a representative of the AfD will be present.”

German experiment, a quote from the legitimizing condition (example): Since AfD “is legitimized through its election into parliament by the electorate and is therefore democratic”, as a democrat one must not “discriminate against the party and treat them differently than all other democratic parties”. Therefore, he pleads for “sending a representative of our party to this event, even if a representative of AfD will be present.”

In the Swedish experiment, I manipulated M’s (de-)legitimizing messages toward SD in the context of a real debate about whether M should give ministerial posts to SD in a (potential) joint government after the September 2022 election. This issue was controversially debated when my experiment was fielded in March 2022 (Aylott and Bolin 2023). The official line was that SD should not be given ministerial posts, although this was justified by a lack of competence of SD (Polk 2022)—not a lack of legitimacy. As M advocated for increased collaboration—minority governments supported by SD were ruled out in previous elections—and SD’s exclusion from formally joining the government was not framed as a matter of principle, M’s real-world behavior toward SD qualifies as a legitimizing strategy according to my conceptualization. Yet this exclusion from formally joining the government was challenged by political pundits and voices within the party (Polk 2022), leaving room for credible manipulation. Following the German experiment, each of M’s messages contains a position—in favor or against giving ministerial posts to SD—as well as an argument justifying this position by portraying SD as (un-)democratic. Below are examples of delegitimizing and a legitimizing quote from the Swedish treatment vignettes.

Swedish experiment, a quote from the delegitimizing condition (example): “I do not think SD has genuinely changed its ideology. We must not be fooled by the façade that the party puts up to try to hide its undemocratic tendencies. For that reason, we must not give ministerial posts to SD.”

Swedish experiment, a quote from the legitimizing condition (example): “I think SD has genuinely changed its ideology. We have to admit that the party has become a serious, democratic competitor. For that reason, we should give ministerial posts to SD.”

Manipulation checks reassuringly showed that the treatment worked as intended (see Appendix B9). Respondents were fully debriefed about the fictitious nature of the mock newspaper articles. In Appendix B1, I elaborate extensively on ethical considerations regarding the experiments.

Survey Measures

After exposure to the four mock newspaper articles, all respondents in both experiments answered several outcome measures. To measure legitimacy evaluations and in line with my conceptualization, respondents reported to which extent they agreed with the following two statements: “AfD/SD is a democratic party” and “AfD/SD poses a threat to democracy”. Based on these items, I created an index (α = 0.85 in Sweden and α = 0.82 in Germany; labeled “Legitimacy” in the figures below).

As a harder test of my theory and to uncover a potential downstream consequence of mainstream party legitimization, I measure the political tolerance of the far right. Respondents were asked whether AfD/SD should have the right to (a) speak at public schools, (b) express themselves in public debate, and (c) hold public demonstrations, as well as (d) whether the police should have special permission to tap the phones of their representatives more easily. These items were taken from Petersen et al. (2010). Again, I created an index based on these four items (α = 0.82 in Sweden and α = 0.86 in Germany; labeled “Tolerance”).

Moreover, I examine whether respondents follow the position taken by CDU/M in the respective (fictitious) debates the treatment was embedded. In the German experiment, I measured respondents’ attitudes toward CDU’s boycott of the panel discussion with the item “CDU should not participate in the panel discussion in Holsterhausen ⁴ if AfD is represented” (“Against boycott” in the figures below). In the Swedish experiment, I asked respondents to indicate their agreement with the statement “M should be willing to give SD ministerial posts in a joint government” on a 5-point Likert scale (labeled “Ministers”).

All dependent variables were measured using a 5-point Likert scale, which I transformed into a 0–1 scale. Higher values indicate that voters evaluate AfD/SD as more legitimate. When necessary, scales were reversed accordingly. Respondents who answered “Don’t know” were excluded from the analyses.

In my third hypothesis, I postulated that partisanship conditions voters’ responses to (de-)legitimizing strategies. To measure partisanship in the German experiment, I asked the following question before exposure to the treatment: “In Germany, many people tend to vote for a particular political party for a long time although they also vote for another party from time to time. What about you? Do you—in general—lean toward a particular party? And if so, which one?” Subsequently, respondents picked either one of the parties represented in the national parliament, “Other party,” “I do not lean toward any party” or “Don’t know.” In the Swedish experiment, survey participants first answered the question “Is there a party that is closer to you than any other party?” ⁵ If they answered the “Yes” or “Don’t know,” they received the following question: “Which party do you feel closest to?” and chose one of the eight parties represented in national parliament or “Other party.” Next, I created three groups: first, only CDU/M partisans; second, partisans of other parties than CDU/M and AfD/SD as well as voters who do not identify with any party or responded “Don’t know”; third, partisans of AfD/SD.

RESULTS

Do Voters Follow When Mainstream Parties Legitimize Pariah Parties?

Before discussing the experimental findings, I begin by comparing voters’ legitimacy evaluations of AfD and SD in the control conditions (“No cue”). ⁶ Without the interference of my treatments, SD is perceived as considerably more legitimate than AfD (left panel of Figure 3). The average on the legitimacy evaluations index is 0.35 for AfD, but 0.51 for SD in the respective control conditions. Moreover, Swedish voters are much more tolerant toward SD (0.68) than German voters toward AfD (0.51). These differences might be due to M’s legitimizing strategy in recent years, as shown by the content analysis. However, the divergent mainstream party strategies in the two countries could also merely follow public opinion—instead of the reverse relationship that I hypothesized. Moreover, this association could be solely due to confounding factors, such as the actual behavior of SD and AfD. Hence, the experiment.

Figure 3. Effects of Treatment—“CDU/M Legitimizing” versus “CDU/M Delegitimizing” Conditions—on Legitimacy Evaluations of AfD and SD

Note: Point estimates represent means. Error bars represent 95% confidence intervals. The numbers corresponding to the horizontal braces represent the OLS coefficients. Indicators of significance: ***p < 0.01, **p < 0.05, *p < 0.1. Full regression output is in the Appendix, Tables SI9 and SI10.

My first hypothesis is that voters evaluate pariah parties as more legitimate when exposed to a legitimizing strategy compared to a delegitimizing mainstream party strategy. To test this hypothesis, I compare respondents’ legitimacy evaluation in the legitimizing condition (“Leg” in Figure 3) and the delegitimizing condition (“Deleg”) in both experiments. I begin by presenting the treatment effects on legitimacy evaluations. Thereafter, I elaborate on the results for tolerance judgments and respondents’ attitudes toward a boycott of AfD in Holsterhausen (Germany) and giving SD ministerial posts (Sweden).

Supporting my first hypothesis, the German experiment shows that when mainstream parties legitimize, voters perceive the pariah as more legitimate. As shown in the very left panel of Figure 3, respondents in the legitimizing condition perceived AfD as significantly more legitimate than respondents in the delegitimizing condition (β ≈ 0.09; p < 0.01). However, there is almost no difference between the delegitimizing condition and the control condition in Germany. This might be the case because strong delegitimization by CDU is the status quo. Respondents were likely pretreated, and an additional delegitimizing message did not further move voters’ legitimacy evaluations of AfD. Yet I cannot rule out that delegitimization simply has no effect.

How did respondents react to (de-)legitimizing mainstream right messages in Sweden, a context in which the mainstream right legitimizes the far-right after initial delegitimization? Strikingly, the findings look very similar. Like in the German experiment, respondents in the legitimizing condition perceived SD as more legitimate than respondents in the delegitimizing condition. However, this difference is smaller than in Germany, at only three percentage points on the scale. Again, although much less pretreated than in the German experiment, the delegitimizing message did not persuade respondents. The difference between the estimate of the control condition and the delegitimizing condition is negligibly small, not significant and in the opposite direction than hypothesized.

This suggests that mainstream parties cannot easily return to a delegitimizing strategy after a period of legitimization. There are at least four plausible reasons for this. First, respondents may view M’s turnaround to delegitimization as disingenuous, strategic, and opportunistic. If so, they are likely to be less persuadable. Second, competing legitimizing frames are likely still available in voters’ minds, which might mitigate the effect of the delegitimizing party message. Third, those who are receptive to delegitimizing party messages might have already been influenced by the delegitimizing messages of other parties that have maintained their delegitimizing strategy (e.g., C). Thus, their opinions are not further swayed by M turning back to a delegitimizing strategy. Fourth, it could simply indicate that delegitimization is per se ineffective, independently of whether mainstream parties have legitimized before or not.

As an additional, harder test of my theoretical propositions, I also present effects on tolerance judgments (second panel of Figure 3). In the German experiment, there is a positive effect of the legitimizing CDU message on tolerance judgments toward AfD. With an increase of eight percentage points compared to the delegitimizing group (p < 0.01), this effect is only marginally smaller in size than the effect on legitimacy evaluations. Again, this effect is primarily driven by the legitimizing condition, while the delegitimizing condition does not differ significantly from the control condition—arguably due to pretreatment. However, this finding provides further evidence that legitimization has far-reaching consequences beyond “mere” perceptions.

In Sweden, the differences between the legitimizing and the delegitimizing groups are negligibly small. While the lack of an effect of the legitimizing message might be due to pretreatment, the delegitimizing condition is likely not pretreated. Nevertheless, there is virtually no effect. This again suggests that turning back to a delegitimizing strategy after a period of legitimization is difficult—if not impossible.

My third dependent variable captures whether respondents followed the positions taken by CDU/M in relation to the “boycott of panel discussion” (third panel) and “ministerial posts” (fourth panel) debates. In the German experiment, higher values indicate that survey participants were more against a boycott and therefore evaluated AfD as more legitimate. In the Swedish experiment, higher values indicate that survey participants were more supportive of giving SD ministerial posts and therefore evaluated SD as more legitimate.

In both experiments, the legitimizing conditions differ significantly from the delegitimizing conditions in the expected direction. In Germany, respondents in the legitimizing condition were on average 14 percentage points more opposed to a boycott than respondents in the delegitimizing condition. In Sweden, respondents in the legitimizing condition are on average eight percentage points more in favor of giving SD ministerial posts in a potential joint government than respondents in the delegitimizing condition. Both treatment conditions in both experiments differ significantly (p < 0.01) from the no cue conditions in the expected directions. Especially in Sweden, the effect of the delegitimizing treatment condition is remarkable, given that M’s official line was not to give ministerial posts to SD.

Overall, the findings provide strong evidence for my first hypothesis. Given the considerable contextual differences, it is striking that the findings of the German and Swedish experiments are very similar. Voters evaluate pariah parties as more legitimate when exposed to a legitimizing mainstream party strategy. In particular, the results suggest that initial legitimization in a context of (formerly) unanimous delegitimization is highly effective, as revealed by the German experiment. Furthermore, the Swedish experiment suggests that mainstream party delegitimization has only little effect after a period of legitimization of a pariah party. Moreover, even in this context, a legitimizing message further moved voters’ perceptions toward perceiving the pariah as more legitimate. The legitimization of pariah parties is thus a consequential choice by mainstream parties.

Can a Legitimizing Strategy Be Countered by a Third Party’s Delegitimizing Strategy?

My second hypothesis proposed that mainstream parties’ legitimizing strategies are less effective when countered by third parties’ delegitimizing strategies. To test this hypothesis, I compare the condition containing only a legitimizing M message with the condition containing a legitimizing M message and a delegitimizing C message. As described before, there was no comparable condition in the German experiment, as it was conducted before and informed the Swedish experiment.

Figure 4 shows that M’s legitimizing message is no less effective in the face of a competing delegitimizing message sponsored by C. The differences between the “M legitimizing” and the “M legitimizing & C delegitimizing” conditions are negligibly small and insignificant for all three dependent variables. If anything, those who have also received a delegitimizing C message next to the delegitimizing M message evaluate SD as more legitimate. Against this backdrop, it seems unlikely that the small number of C partisans dilutes aggregate effects, although it cannot be ruled out that this null finding is due to pretreatment.

Figure 4. Effects of Treatment—“M legitimizing” versus “M Legitimizing & C Delegitimizing” Conditions—on Legitimacy Evaluations of SD

Note: Point estimates represent means. Error bars represent 95% confidence intervals. The numbers corresponding to the horizontal braces represent the OLS coefficients. Indicators of significance: ***p < 0.01, **p < 0.05, *p < 0.1. Full regression output is in the Appendix, Table SI11.

Overall, the evidence presented here suggests, that mainstream party legitimization is therefore not only far-reaching because it is effective and cannot be reversed, but also since it might not be feasible for third parties to counter it effectively.

Are Legitimizing Strategies Effective beyond Co-Partisans?

My third hypothesis postulated that partisanship conditions voters’ reactions to legitimizing strategies. Before I discuss the experimental results regarding this hypothesis, I elaborate on how mainstream right supporters in the control conditions evaluate AfD’s/SD’s legitimacy as compared to the rest of the electorate. In the absence of any messages received during the survey experiment, this comparison reveals that the different nature of party competition in Germany and Sweden is also mirrored in the level of voters (see Figure 5). ⁷ In Germany, there is a broad consensus amongst all voters, except AfD partisans, that AfD is not a legitimate party. CDU supporters view AfD as slightly less legitimate than the rest of the electorate. In Sweden, in contrast, M partisans view SD as much more legitimate than the rest of the electorate, with the obvious exception of SD partisans.

Figure 5. Effects of Treatment—“CDU/M Legitimizing” versus “CDU/M Delegitimizing” Conditions—on Legitimacy Evaluations of AfD/SD by Partisanship

Notes: Point estimates represent means. Error bars represent 95% confidence intervals. Full regression output is in the Appendix, Tables SI12 and SI13.

Hence, the Swedish case exemplifies, how party elites as well as voters can polarize on the question of whether a pariah party should be considered legitimate. This suggests that the issue of how to deal with a pariah can become an important political issue that leads to a restructuring of party competition and perhaps even results in a realignment of the electorate. Yet again it is unclear whether mainstream parties are just responsive to the demands of their constituencies or whether partisans follow the messages disseminated by their party.

I hypothesized that the effects of legitimization are strongest for partisans of CDU/M. Figure 5 shows that this hypothesis is only partially supported. In line with my theoretical expectations, I begin by comparing co-partisans with other partisans as well as non-partisans (labeled “Others” in the figures) and subsequently proceed by comparing co-partisans with partisans of the targeted pariah parties.

Legitimization resonates beyond the electorates of the mainstream right. Comparing the persuasiveness of the legitimizing message vis-à-vis the delegitimizing message for CDU supporters versus voters who support neither CDU nor AfD, reveals that the effect on legitimacy evaluations is 0.13 (p < 0.01) for the former and 0.07 (p ≈ 0.01) for the latter. Supporting my hypothesis, the interaction of treatment with CDU partisanship (versus all others) yields a moderate, significant effect (β ≈ 0.07; p ≈ 0.09) (see Appendix C4).

In the Swedish experiment, the differences in treatment effects between M partisans (β ≈ 0.05, p ≈ 0.3) and voters who support neither M nor SD (β ≈ 0.04, p ≈ 0.12) are minor. Contradicting my hypothesis, the interaction of treatment with M partisanship (versus all others) is insignificant, and the coefficient is negligibly small (β ≈ 0.01; p ≈ 0.86) (see Appendix C4). However, as mentioned before, Swedish voters might identify rather by bloc than by party, and voters of L and KD would be affected by SD’s formal participation in government as well since M had formed a pre-electoral alliance with both parties (Aylott and Bolin 2023).

Hence, I next examine the treatment effects by identifiers of parties of the right versus left bloc, rather than M versus all others. The regression table for this analysis can be found in Appendix C5. Perhaps surprisingly, the treatment effect for supporters of all right bloc parties, in comparison to M voters only, is slightly bigger (β ≈ 0.08; p ≈ 0.04). In comparison to everyone but SD and M voters, only the treatment effect for supporters of left bloc parties is smaller and remains insignificant (β ≈ 0.03; p ≈ 0.331). However, the interaction of right versus left bloc partisanship is also insignificant (β ≈ 0.06; p ≈ 0.23).

Turning to tolerance judgments of AfD, the results are in line with the results for legitimacy evaluations. There is no effect on tolerance judgments in the Swedish experiment, regardless of respondents’ partisan identification. The results for attitudes toward a boycott of AfD and giving SD ministerial posts also echo the results on legitimacy evaluations.

Finally, I look at whether mainstream parties’ (de-)legitimizing strategies are effective amongst partisans of the targeted parties. I theorized that they are not. The Swedish experiment supports this, while the German experiment indicates that mainstream parties might indeed be able to persuade partisans of AfD that their own party is illegitimate. This difference between the two experiments could be due to the actual conduct of AfD and SD. While the former has radicalized severely, the latter has become more moderate. However, considering the small number of AfD voters in my samples, the reliability of these estimates is limited and the difference in legitimacy evaluations between the legitimizing and delegitimizing conditions is insignificant for AfD voters (p ≈ 0.154). Therefore, I refrain from drawing firm conclusions from this finding and leave it to further studies to investigate whether mainstream parties can effectively delegitimize pariah parties in the eyes of a pariah’s own partisans.

Overall, the moderating effect of partisanship on the persuasiveness of (de-)legitimizing messages, postulated in my third hypothesis, is surprisingly weak. Legitimization resonates beyond co-partisans of the party pursuing this strategy.

CONCLUSION

Party competition over policy is a defining feature of modern democracy. Accordingly, generations of political scientists have investigated how parties compete over policy. Parties compete against each other by taking positions on political issues (Downs 1957; Stokes 1963) and selectively emphasize issues on which they hold popular positions and are perceived competent by voters (Green-Pedersen 2016; Petrocik 1996). When parties adjust their own policy platforms, voters respond and update their evaluations of parties (Stubager and Seeberg 2016). In line with this perspective, existing research has mostly investigated how mainstream parties adjust their policy platforms in response to new parties challenging them and how voters react to these adjustments (Abou-Chadi 2014; Chou et al. 2021; Meguid 2005).

My findings suggest that next to policy-based response strategies, there is an additional dimension of mainstream parties’ response to such parties, centering on legitimacy. Mainstream parties have typically attacked the legitimacy of these parties in their early stages by means of delegitimizing strategies, making them pariahs. Yet such strategies were often short-lived and mainstream parties have adopted legitimizing strategies. These response strategies matter.

I provided strong causal evidence that voters evaluate pariah parties as more legitimate when mainstream parties pursue a legitimizing strategy. Furthermore, my findings suggest that (a) it is difficult for mainstream parties to turn back to delegitimization after a period of legitimization, (b) legitimization is no less effective in the face of a third party’s delegitimizing strategy, and (c) legitimization resonates beyond co-partisans. Therefore, the results suggest that mainstream party legitimization of pariah parties has far-reaching consequences.

Against this backdrop, I contend that scholars should recognize competition over legitimacy as a dimension of party competition in times of disruption by parties challenging the mainstream. Extant scholarship has argued that populist parties portray themselves as the only legitimate party (Mudde 2007) and attack the legitimacy of mainstream parties by means of fierce anti-elite rhetoric (De Vries and Hobolt 2020). I argue mainstream parties may also have a powerful tool at their disposal to engage in a competition over legitimacy with pariah parties: (de-)legitimizing strategies.

Moreover, my results contribute to the literature on the “normalization” or “mainstreaming” of the far right in the eyes of voters by highlighting the agency of mainstream parties in this process. Scholars have primarily theorized the legitimization of the far right in voters’ eyes as an unintended byproduct of accommodative issue-based mainstream party strategies (Arzheimer and Carter 2006; Dahlström and Sundell 2012), as the result of institutional representation (Valentim 2021), or as a consequence of pariah parties’ own conduct and behavior (Paxton and Peace 2020). I have shown that challenger party legitimacy is susceptible to influence by mainstream parties’ legitimizing strategies.

Future studies should apply my theory to other cases to gain more insights into when mainstream party legitimization is effective. I have argued that studying the mainstream right-far-right party dyad in Germany and Sweden yields highly generalizable findings, as the two contexts, albeit similar in many respects, differ sharply in terms of (a) strategies chosen by the mainstream right (M: legitimization; CDU: delegitimization), and (b) the conduct of the pariah (AfD: radicalization; SD: moderation). As such, it is likely that voters respond to mainstream party legitimization toward far-right parties like the Portuguese Chega or the Spanish Vox in a similar vein. On the left of the political spectrum, legitimization has likely been effective for the Greens and many radical left parties—such as the German Die Linke, which has participated in several state government with the mainstream left. There are fewer examples of mainstream parties turning back to delegitimization after a period of legitimization, which I have argued is likely to be ineffective. The case of the Austrian FPÖ seems to be in line with this reasoning. It was one of the first legitimized far-right parties of the post-World War II era and any efforts to delegitimize the party subsequently, in light of various scandals and sustained radicalism, seem to have fizzled out without effect.

However, it takes further empirical tests to confirm that my theory holds for legitimization (a) by the mainstream left toward the far left and (b) across the ideological aisle—that is, by the mainstream right toward the far left or by the mainstream left toward the far right. Arguably, the latter type of legitimization might be even more effective as it might be seen as more sincere and less driven by opportunistic motivations (i.e., office or vote). In a similar vein, regarding the legitimization of far-right pariah parties, future studies should examine whether competing, delegitimizing messages from the mainstream left are equally ineffective as those from smaller parties with a center-right platform (i.e., C in my study). In addition, the duration a pariah has been in parliament might condition the effectiveness of legitimization.

Given my finding that mainstream party legitimization has far-reaching consequences for public opinion toward the targeted parties, investigating the predictors of legitimization and the attitudinal and behavioral downstream consequences of legitimacy evaluations appear ever more important.

Against this backdrop, future studies should scrutinize the predictors of legitimizing vis-à-vis delegitimizing strategies in terms of the constraints and incentives mainstream party elites face or perceive. Extant literature has argued that office motivates mainstream right parties to form coalitions with far-right parties (Backlund 2022; De Lange 2012; Van Spanje 2010). But which role do other factors, such as vote, policy, ethical considerations, or (historical) context play in that regard? To better understand the (perceived) agency of mainstream parties when dealing with pariah parties, future research might study mainstream party elites’ expectations about voters’ reactions to (de-)legitimizing strategies.

Finally, scholars should investigate when and how legitimacy evaluations are linked to other behavioral and attitudinal outcomes. Studies have suggested that parties widely perceived as illegitimate, inter alia, have a harder time to garner support for their policy proposals (Blinder, Ford, and Ivarsflaten 2013), are unpopular coalition partners (Kelemen et al. 2023), and struggle to attract competent personnel (Art 2011). Yet often we lack causal evidence that this is because of a lack of legitimacy in the eyes of voters and it remains unclear when legitimacy evaluations are translated into attitudinal or behavioral consequences. Investigating the effects of legitimacy evaluations on vote choice is perhaps the most pressing. While scholars have often argued that voters refrain from voting for parties they deem illegitimate (Bos and van der Brug 2010; Van Heerden and Van der Brug 2017), this effect might be conditional. To what extent do legitimacy evaluations matter for vote choice vis-à-vis competence evaluations or policy proximity? Moreover, (de-)legitimizing strategies might influence vote choice not only through persuading voters but also through priming considerations related to legitimacy. If so, even a legitimizing strategy might have detrimental consequences for pariah parties if it leads voters who agree with the pariah’s policies, but firmly deny its legitimacy, to refrain from voting for it.

By showing that (de-)legitimizing strategies shape voters’ legitimacy evaluations of pariah parties, my study may fruitfully inspire future studies on party competition in times of party system disruption and fragmentation.