DEPARTMENT OF POLITICAL SCIENCE THE GENDER (DE)MAINSTREAMING OF THE POPULIST RADICAL RIGHT Quantitatively analyzing the effects of increasing female descriptive representation within populist radical right parties. Emma Olsson Master’s Thesis: 30 credits Programme: Master’s Programme in Political Science Date: 2024-05-20 Supervisor: Andrej Kokkonen Words: 19 640 Abstract All around the world, the visibility of women within populist radical right (PRR) parties has increased. Scholars typically explain the increase as a strategy for PRR parties to attract more female voters. Consequently, the goals are different from female descriptive representation within other parties. In light of such a strategic increase, there are reasons to investigate the effects of PRR female descriptive representation further. The aim of the following study is thus to analyze how increasing female descriptive representation affects the perception of PRR parties, and for whom such effects are strongest. So far, few studies have applied a party lens when analyzing female descriptive representation, and citizens’ perceptions of political parties by which this study fills a prominent gap in previous literature. Theoretically, the “Populist Radical Right Gender Mainstreaming Model” is used, suggesting that female descriptive representation has a mainstreaming effect on the perception of PRR parties. It is also expected that such an effect is strongest for women and far-right voters. To test these assumptions, OLS regressions including party-fixed effects are performed. The results from such regressions show that increasing female descriptive representation makes PRR parties appear more far- right and become more disliked among the public. In other words, the results show a de- mainstreaming effect instead of a mainstreaming. Also, the increasing dislike towards PRR parties appears strongest for far-left voters. These results thus invalidate all hypotheses and question the relevance of the Populist Radical Right Gender Mainstreaming Model. Keywords Populist radical right, gender mainstreaming, stereotypes, female descriptive representation Contents 1. Introduction ..................................................................................................................... 3 2. Previous research ............................................................................................................. 5 2.1 Citizens perceptions of political parties .......................................................................... 5 2.1.1 Ideological left-right position ................................................................................... 5 2.1.2 Party likeability ......................................................................................................... 6 2.2 The effects of female descriptive representation ............................................................. 7 2.3 Right-wing populism ........................................................................................................ 8 2.3.1 Defining right-wing populism .................................................................................. 9 2.3.2 Why analyze PRR parties? ..................................................................................... 10 2.4 The effects of strategic descriptive representation ........................................................ 14 2.5 Research gap and contributions .................................................................................... 15 3. Theoretical framework .................................................................................................. 17 3.1 Gender stereotypes ........................................................................................................ 17 3.2 The clash of the masculine and the feminine ................................................................. 19 3.2.1 PRR Gender mainstreaming ................................................................................... 19 3.3 Hypotheses ..................................................................................................................... 21 4. Research design .............................................................................................................. 23 4.1 Data and operationalizations ........................................................................................ 23 4.1.1 Independent variables ............................................................................................. 25 4.1.2 Dependent variables ............................................................................................... 27 4.1.3 Moderating variables .............................................................................................. 29 4.1.4 Control variables ..................................................................................................... 30 4.2 Method ........................................................................................................................... 32 5. Results ............................................................................................................................. 33 5.1 Diagnostics .................................................................................................................... 33 1 5.2 Descriptive statistics ...................................................................................................... 34 5.3 Regression results .......................................................................................................... 37 5.3.1 Hypothesis 1 ........................................................................................................... 37 5.3.2 Hypothesis 2 ........................................................................................................... 39 5.3.3 Hypotheses 3 and 4 ................................................................................................. 41 6. Discussion ....................................................................................................................... 45 6.1 The effects of PRR female descriptive representation ................................................... 45 6.2 The moderating effect of gender and left-right placements ........................................... 46 6.3 Implications and limitations .......................................................................................... 48 7. Conclusion ...................................................................................................................... 50 8. Bibliography ................................................................................................................... 52 Appendix ................................................................................................................................. 57 2 1. Introduction Politics is a field in which women have historically been, and still are today, underrepresented. A quick glance at the national parliaments worldwide reveals that only 26,5 percent of the parliamentarians are women (UN Women 2023). To improve the representation of women some parties and governments have taken action by implementing gender quotas through which the descriptive representation of women should increase (Krook & Zettererg 2014). Such actions have for the most part been adopted by left-wing parties for whom the issue of gender equality is close to heart (Santana & Aguilar 2019). Nevertheless, in recent years there has also been an increase in the descriptive representation of women within populist radical right (PRR) parties. The visibility of PRR women is shown through an increasing number of female PRR party leaders, for example, Marine Le Pen for “National Rally” in France and Alice Weidel for “Alternative Für Deutschland” in Germany (Chrisafis, Connolly & Giuffrida 2019), but also through an increase in the number of female PRR representatives. Many researchers describe this increase as a strategy for PRR parties to win further support, specifically targeting female voters who generally abstain from voting PRR (Weeks, Meguid, Kittilson & Coffé 2023). The gender gap in voting has been explained by referring to the aggressive and authoritarian political style of PRR parties rather than focusing on the difference between men and women in their attitudes and beliefs (Harteveld, Van der Burg, Dahlberg & Kokkonen 2015). Consequently, even though it has been established that men and women do not differ in their attitudes and beliefs, there still seems to be a gender-related difference in the way citizens perceive the political style of PRR parties. Furthermore, as this difference seems to go beyond political ideology it is interesting to analyze further how gender intersects with political ideology to affect party evaluations considering the increasing PRR female descriptive representation. The increasing PRR female descriptive representation is a puzzling phenomenon first and foremost with regards to the political agenda of such parties which by many is described as “[...] strongly conservative in the fields of family policies and the rights of women [...]” (Dona 2020:288). Additionally, analyzing its potential effects is of particular importance partly because the overall effects of female descriptive representation are contested within the literature but also because female descriptive representation within PRR parties seems to differ from the kind of descriptive representation that has previously been analyzed given the strategic aspect of it. 3 Previous research has only recently started to analyze the increasing PRR female descriptive representation, mainly because for long such representation was extremely low. Apart from studies explaining the gender gap and PRR methods for decreasing the gap, there is still much to be done in the area. Having established that many PRR parties use female descriptive representation strategically to win further support, it is interesting to further analyze whether such strategies also shape the public perception of these parties. In line with this, the aim of this study is to shed further light onto the effects of PRR female descriptive representation for the public perception of PRR parties. In so doing, this thesis contributes further to three fields of research, namely 1) right wing populism and its intersection with gender, 2) citizens perception of political parties and 3) the effects of female descriptive representation. More specifically, this thesis will be guided by the following research questions: ● Has the increasing PRR female descriptive representation affected the public perception of PRR parties, and if so, how? ● Among whom has the effect of PRR female descriptive representation on the perception of PRR parties been greatest? Following this introduction is a review of previous research in relation to the three earlier mentioned fields of research. The section on previous research is intended to serve as a background to the overall problem of this thesis and give the reader a deeper understanding of the overall research gaps. Additionally, previous research is also indicative of what factors will be mentioned in the theoretical framework where theories on gender stereotypes and PRR gender mainstreaming constitute the foundation. Following the theoretical framework is a section on methodology. To test the hypotheses and ultimately answer the research questions of this thesis, data from three different sources is used, combining individual-level data with data on party-level. In total, the dataset consists of 15 PRR parties from 15 different European countries covering the years of 1996 to 2021 making this the most comprehensive paper to date on the effects of female PRR descriptive representation. To test the effects, ordinary least squares (OLS) regressions including party fixed effect to measure within party change, are used. Two moderators are also included in the second party of the analysis to answer the second research question. In the final part of the thesis the results are discussed in relation to previous research, paving the way for future research. 4 2. Previous research The purpose of the following section is to review previous research relating to the topic of PRR female descriptive representation. The review will be organized in such a way that research on citizens´ perceptions of political parties will be discussed first, followed by a review of research on female descriptive representation. By first analyzing these two subjects separately, the reader is provided information on what has been done within each area of research, illustrating that they are rarely combined as done in this thesis. Following this separate discussion of the two main research areas will be a review of studies that have attempted to combine the two subjects. From such it should be evident that few, if any, have analyzed the relationship between female descriptive representation and citizens' perception of political parties within PRR parties, even though the visibility of women within these parties has increased. The purpose of the final section is thus to review research on right-wing populism and its intersection with gender, substantiating the decision to analyze this party family. 2.1 Citizens’ perceptions of political parties Numerous studies have been dedicated to the analysis of how citizens perceive political parties. When analyzing perceptions scholars want to capture citizens' feelings and thoughts rather than their actual behavior. To evaluate political parties’ citizens are often asked to place parties either on an ideological left-right scale or on a likeability scale. Seeing as these two points of evaluation are amongst the most used in previous research, the aim of the following section is to discuss them further. 2.1.1 Ideological left-right position The ideological left-right dimension is commonly referred to in political science studies where economy is the core issue (Caprara & Vecchione 2018). The left side of the dimension is said to promote more redistributive economic politics with an emphasis on the welfare state whereas the right-side advocates in favor of a free market-based capitalist economy with a limited government (ibid). Even though the relevance of the left-right dimension has been questioned today (ibid), it is still frequently used in individual-level surveys as an indication of citizens' perception of political parties. In most surveys, respondents are asked to place political parties on a scale from zero to ten, higher numbers for parties further right and lower numbers for parties to the left. 5 A variety of explanations have been brought up in the literature as to how citizens decide where to place political parties on the left-right scale. According to Russel J. Dalton and Ian McAllister (2015), citizens' perceptions of parties' left-right placements are durable over time, even though parties change their policy positions or try to appear in a different way. These results indicate that once citizens' decisions of parties' left-right placements are made there are not many factors that could potentially change such perceptions. On the other hand, Thomas M. Meyer, and Markus Wagner (2020), instead argue that citizens' perceptions of parties' left- right placements are affected by parties' positions on economic and cultural issues, and how salient such issues are to the party in question or to the party system as a whole. Using this knowledge, political parties can influence citizens' perceptions by emphasizing some issues more than others (ibid). Similarly, scholars James Adams, Simon Weschle, and Christopher Wlezien (2021), show that parties' cooperative public relations impact how citizens perceive their position on the left-right scale. In other words, cooperation between more extreme political parties and mainstream parties can make an extreme party seem more moderate, and a mainstream party seem more extreme (ibid). 2.1.2 Party likeability Another way of evaluating political parties is by letting citizens rate their likeableness. Whether a citizen likes or dislikes a party is subjective to their own stance on politics, however, such could also be influenced by other factors. In a similar manner as with the ideological left-right placement, factors in relation to electoral rules and the behavior of parties appear to be common in the literature. For example, it has been shown that when governing parties are in coalition with other parties, supporters of the governing party express feelings of approval and likeableness towards the other party (Gidron, Adams & Horne 2023). Consequently, not only could much disliked parties use their cooperation with more mainstream parties to appear more moderate, but they could also use such cooperation to become more liked in the public eye. Furthermore, electoral competition has also proven to be an important factor. In an experimental study conducted in Israel, Lotem Bassan-Nygate and Chagai M. Weiss (2022) demonstrate that the temporal proximity to an election date determines the level of affective polarization. Affective polarization is here understood as the feeling of dislike and hostility towards opposing parties; a feeling which is believed to increase when the salience of electoral competition increases (ibid). 6 All in all, even though parties’ position on the ideological left-right scale and parties’ likeability are two different ways in which to evaluate parties - ways which typically are analyzed separately, similar explanations for their development have been proposed in the literature. Predominantly, explanations in relation to electoral rules and the behavior of parties have been highlighted, however, more recently, a few scholars have also drawn attention to the gender of party MPs. Such studies relate to the effects of female descriptive representation which will be discussed further in the upcoming section. 2.2 The effects of female descriptive representation The effects of female descriptive representation have been extensively studied specifically focusing on outcomes related to the substantive and symbolic representation of women. However, for both these areas of research, results remain inconsistent. On the one hand, some studies have shown a positive relationship between female descriptive representation and the representation of women’s interests (Wängnerud 2009), as well as between female descriptive representation and women's political engagement (Barnes & Burchard 2013). On the other hand, several scholars have also pointed to the fact that descriptive representation is meaningless, rather substantive, and symbolic representation depends on the ideology and issue focus of the party. For example, according to Rosie Campbell and Oliver Heath (2017), female voters in the UK who value the importance of female descriptive representation also tend to prefer female candidates. A similar argument is proposed by Danielle Josten Martin (2019), who shows that women vote for female candidates if such candidates are also appealing ideologically. Both these studies are thus negative towards the actual importance of the candidate gender, and altogether, these studies reveal the inconsistency in research on the effects of female descriptive representation. In light of this, there is reason to evaluate how female descriptive representation has been measured, and what outcomes researchers have focused on. As mentioned earlier, scholars have recently started to acknowledge the effect of female descriptive representation on the perception of political parties. In such studies, gender has appeared to play a greater role than when analyzing other outcomes, and onwards, the discussion will thus revolve around the effects of female descriptive representation on citizens' perceptions of political parties. In a comparative analysis between 269 parties, Diana Z. O´Brien (2015) concludes that parties with a female party leader are perceived as more moderate than male-led parties (ibid). Theoretically, she describes the moderating effect of female party leadership as dependent upon 7 female stereotypes (ibid). In the public eye, women are often perceived as more left-leaning and warm which would explain why parties with a female leader are perceived as more moderate. In her analysis, she includes a variable for party family, thus distinguishing between different parties. In doing so she proves that there is a difference in the effects of female party leaders between parties to the left and parties to the right (ibid). Furthermore, in line with O´Brien, a few scholars have also started to analyze whether who serves in political office shapes the level of affective polarization. In an article about female descriptive representation and affective polarization, a group of scholars hypothesizes that out-parties with a higher proportion of women are evaluated more warmly in comparison to out-parties with fewer women MPs (Adams, Bracken, Gidron, Horne, O´Brien & Senk 2023). The authors argue that this is a result of both gender stereotypes and women´s participatory leadership styles (ibid). Being able to support their hypothesis, the authors prove that increasing female descriptive representatives can be used to ease cross-party hostility. These studies show the importance of combining the two fields of research, namely citizens' perceptions of political parties and female descriptive representation. Analyzing citizens' perceptions of political parties as dependent upon female descriptive representation might help clarify these otherwise inconclusive fields of research. 2.3 Right-wing populism In studies about the link between female descriptive representation and citizens’ perceptions of political parties, few if any, have distinguished between parties of different party families. In O´Brien´s study from 2015, party family was included as a control variable by which it was possible to interpret a difference between left-wing and right-wing parties, pointing to the importance of studying this phenomenon more in-depth. In other words, applying a party lens might help account for some of the previous inconclusiveness and provide more comprehensive results. The following section will thus highlight previous research about right-wing populism and simultaneously discuss reasons for analyzing this party family. 8 2.3.1 Defining right-wing populism According to political scientist Cas Mudde (2007:22) there are three major characteristics of radical-right wing populism: nativism, authoritarianism, and populism. Nativism is described as an ideology based on two main features: xenophobia and nationalism (ibid). Nativist parties propose that countries should only inhabit citizens of the native group and view other identities and ideas as threatening to the homogenous nation (ibid). The second characteristic, authoritarianism, is described as “[...] the belief in a strictly ordered society” (ibid:22) where citizens are subordinate to the authority and punished in the case of disobedience. Lastly, populism is also viewed as an ideology of its own where society is portrayed as being divided into two groups: the corrupt elite and the pure people (ibid). Dividing society into a corrupt elite and pure people resembles what psychologists refer to as the “us-vs-them mentality”, where the out-group is perceived as threatening to the in-group (ibid). These three characteristics are visible within other party families as well, for example, populist radical left (PRL) parties for which populism and authoritarianism are common features (ibid), and thus for a party to be considered PRR all three characteristics must be present. It is also this combination of characteristics that distinguishes the political style of PRR parties from the political style of other party families. The political style of PRR parties is often described as aggressive and polarizing (Immerzeel et.al. 2015) which stems from their nativist, authoritarian and populist features. Besides studying the actual conceptualization of right-wing populism, scholars have also attempted to explain who right-wing populists are, in terms of their representatives and their voters. One of the main findings from such studies reveals that PRR parties consist primarily of men (Mudde 2007:100). In politics, women have historically always been underrepresented regardless of what party is being analyzed, however, this trend remained for PRR parties even though most other parties started to incorporate more women (ibid). Thereto, in those few cases where women were appointed to PRR parties, their roles were irrelevant and non-influential (ibid). This domination of men within PRR parties also characterizes their electorate. In the literature, this skewed distribution between female and male PRR voters is often referred to as “the populist radical right gender gap”. Many researchers have investigated reasons for this gap; reasons which initially referred to a difference between men and women in their values and beliefs. Such an explanation implied that women are not as nativist, authoritarian, and populist as men are. However, this argument has proven insignificant in explaining the displayed gender gap (Immerzeel et.al. 2015; Harteveld et.al. 2015). Rather, Eelco Harteveld et.al (2015) show that men on average are somewhat less authoritarian and 9 nativist in comparison to women. In dismissing earlier assumptions about the gender gap, Harteveld et.al conclude that women refrain from voting for PRR parties because they attach less salience to such questions and because they tend to agree less with the political style of such parties (ibid). Considering the lack of female PRR representatives and voters, PRR parties are often referred to as “Männerparteien” (see men´s parties) (Mudde 2007:111). 2.3.2 Why analyze PRR parties? Following the definition of right-wing populism and their electorate, the next section is dedicated to a discussion on reasons for the focus of PRR parties in the study. This section will highlight three subjects that recently have gained scholarly attention, and together articulate the puzzle of this thesis, namely, 1) gender and right-wing populism, 2) strategic descriptive representation, and finally 3) citizens’ perceptions of PRR parties. Gender and right-wing populism With regards to the underrepresentation of women in PRR parties, as concerns both voters and political representatives, it is not surprising that the relationship between right-wing populism and gender has been largely understudied (Doná 2020). Besides studying the PRR gender gap, scholars have also attempted to analyze how gendered the PRR agenda is, by which the first reason for focusing explicitly on PRR parties is revealed. The question about how gendered the PRR agenda is, is really about the gender ideology of PRR parties. Doná (2020) shows in her article that PRR skepticism toward gender equality and feminism has since long been established in the literature. Even though skepticism towards gender equality-related issues and sometimes even anti-feminism is not one of the major characteristics of PRR parties, their traditional and xenophobic views explain their stance on such issues. PRR parties often advocate in favor of upholding traditional gender roles where men are portrayed as the breadwinners of the family and women are portrayed as the caretakers whose role in society is subordinate to men (Mudde 2007:92). Moreover, some scholars even point to the mobilizing against the concept of gender and the feminist movement, within PRR parties. This anti-feminist mobilization is believed to be a characteristic of the “new kind of right-wing populism” where female empowerment has led to a loss of hegemony and power for men; a loss PRR parties claim to be against (Sauer 2020). Following their traditional and conservative view on gender roles, it is further believed that PRR parties see feminism as a threat to the natural division of the sexes, where gender rather than being a cultural phenomenon is understood in biological terms (Löffner, Luyt & 10 Starck 2020). Analyzing female descriptive representation within PRR parties is thus especially interesting given the gender ideology of PRR parties. Moreover, in recent years scholars have noticed an increase in the descriptive representation of women within PRR parties making them even more interesting to analyze. Strategic descriptive representation In an article written by a group of political scientists, the strategic background of PRR female descriptive representation is analyzed. These scholars conclude that PRR parties elect female representatives to attract more female voters (Weeks et.al. 2023). The argument of this study is thus based on the previously discussed gender gap where PRR parties are believed to nominate more women hoping such measures will increase their support (ibid). This phenomenon is labeled “strategic descriptive representation” which appears to be most common among electorally struggling PRR parties and somewhat unique to these parties (ibid). Considering the increasing PRR female descriptive representation, the question has appeared of whether PRR parties are still to be considered “Männerparteien” or not, which according to Siliva Erzeel and Ekaterina R. Rashkova (2017) is not the case anymore. Similar to what Weeks et.al call strategic descriptive representation, Erzeel, and Rashkova refer to as a "standardization strategy", where PRR parties use the inclusion of women to appear more standardized and appealing in the eye of the public (ibid). Strategic descriptive representation is different from the kind of descriptive representation which has previously been studied. First and foremost, given that PRR parties historically have had very few female representatives, strategic descriptive representation has not been captured in previous studies. Additionally, in analyzing strategic descriptive representation, the aim is different than descriptive representation analyzed in other party families. For that reason, it is reasonable to assume the effects of such representation also differ. Altogether, from section 2.3.2 an interesting puzzle has appeared. On the one hand, the gender ideology of PRR parties explains the historically low level of female representatives, however as the visibility of female representatives has recently increased there is reason to wonder what effects to be expected from such an increase given the strategic aspect of the representation. These sections together validate the reason for focusing explicitly on PRR parties when analyzing female descriptive representation. Furthermore, in the following section, it will be made clear why the perception of these parties is of particular interest as well. 11 Citizens´ perceptions of PRR parties Given the aim of this thesis, it is important to establish what the general perception of right- wing populist parties is, to be able to say whether there has been a change in the perception of these parties or not. Considering studies on affective polarization, PRR parties are often mentioned. According to Eelco Harteveld, Philipp Mendoza and Matthijs Roodujin (2020), PRR parties are amongst the most disliked parties while simultaneously exacerbating levels of affective polarization. In other words, PRR parties receive high levels of dislike, and, especially from supporters of mainstream parties. However at the same time these parties are also responsible for expressing dislike towards other parties, yet again in particular against mainstream parties (ibid). This result is also confirmed by Gidron, Adams, and Horne (2023), who explain this trend with regard to the PRR political style. PRR parties are believed to generate negative emotions among their supporters due to their populist rhetoric and discourse of anger, rage, resentment, and fear (ibid). Harteveld et.al (2020) also show that this trend is not as common within populist radical left parties, even though they too are characterized by a populist rhetoric. Additionally, the strong “us-vs-them mentality”, which was mentioned in section 2.3.1 should also be a main contributor to the PRR affective polarization. Furthermore, PRR parties are not only responsible for receiving and expressing dislike but also for increasing toxic polarization, in other words, pernicious polarization. The term pernicious polarization is often used in studies on PRR parties and it is believed to arise when partisan loyalties created from the “us-vs-them-mentality” are prioritized before public interests such as equality and democracy. According to Murat Somer, Jennifer L. McCoy, and Russel E. Luke (2021), many PRR parties exploit existing societal cleavages to deepen polarization because such polarization is especially beneficial to these parties. Altogether, these studies reveal that PRR parties are amongst the most disliked parties because of their aggressive political style and because they are responsible for creating divisions between groups of people in society fostering feelings of hate and dislike. Moreover, the perceived ideological left-right position of PRR parties is not as clearly stated as their position on the likeability scale. This may stem from the critique of the ideological left-right scale in appropriately representing more newly developed parties. As a result of rising left-libertarian politics, with right-wing populism as a counter-movement, attention has been called to new cleavage structures in society based on attitudes towards sustainability, inclusion, and progressiveness (Bornschier 2010). In turn, new dimensions on which to place political parties have also appeared, for example, the so-called “green- 12 alternative-liberal vs traditional-authoritarian-nationalistic” (GAL-TAN) scale where PRR parties are described as being very traditional, authoritarian, and nationalistic (Hooghe, Marks & Wilson 2002). Given the rise of new political dimensions, it is often difficult to place newer parties on the traditional left-right scale, particularly PRR parties. On the one hand, many scholars point to the fact that PRR voters and representatives belong to the old “working class” with its roots on the left side of the left-right scale (Harteveld 2016). Consequently, when measuring PRR parties’ actual placements on the left-right scale using policy proposals, they are often described as more left-leaning than expected because they support some of the more traditional left-wing policies. On the other hand, it seems the old conceptualization of “left” and "right” as mainly dependent upon economic issues, has been replaced or at least expanded in the public eye. In line with Russel J. Dalton (2014), the conceptualization of left and right is contingent upon the contemporary political environment and, more specifically, reflects the current issues of salience to the public. Seeing as modernization and globalization have shed light on new political issues of cultural character, it is possible to argue that the understanding of left and right, at least in the public eye, has changed and will keep on changing to fit the current political environment. Provided that the understanding of left and right has expanded/changed, the PRR stance on cultural issues such as immigration, reveals their perceived position on the left-right scale. All in all, these results indicate that PRR parties are often perceived to be positioned far right on the ideological left-right scale and thus a more extreme form of right-wing ideology. This, even though their actual placement on the scale might be more to the middle. Consequently, not only are these parties often perceived as typical right-wing parties, but their political style and characteristics also make them appear more extreme in comparison to other parties, which is why they are often placed far right on the ideological left-right scale, speaking to both their ideology and level of extremism. In summary, given the perception of PRR parties, as concerns their level of likeability and placement on the left-right scale, it is interesting to analyze how such perceptions might develop in response to changes in the level of female descriptive representation. 13 2.4 The effects of strategic descriptive representation As has been made clear from previous sections, there are not many studies analyzing the relationship between female descriptive representation and citizens’ perceptions of political parties. Yet, fewer have focused explicitly on PRR parties, even though an interesting puzzle has appeared in previous research on this subject. Following the study of Weeks et.al (2022), where the strategic aspect of PRR female descriptive representation was established, a couple of scholars have attempted to investigate whether such strategies for increasing PRR electoral support work. Whereas some scholars argue that the inclusion of women's interests and ideas, rather than the visibility of women, increases the support for these parties (Chueri & Damerow 2023), others have explained the increasing PRR support as a direct result of their increasing female descriptive representation (Ben-Shitrit, Elad-Strenger & Hirsch-Hoefler 2022). Following Lihi Ben-Shitrit, Julia Elad-Strenger and Sivan Hirsch-Hoefler (2022), female descriptive representation within PRR parties helps soften the image of these parties making them more acceptable in the public eye. The authors propose a “populist radical right gender mainstreaming” model whereby PRR parties are perceived as less authoritarian and aggressive when they include more women who, stereotypically speaking, are warmer and kinder (ibid). To test this effect, Ben-Shitrit et.al conducted two experimental studies in Israel, the first one focusing on the effects of political candidate gender on citizen support for radical right and radical left policy proposals, and the second one focusing on support for a radical right or a radical left movement where the movement activist is either a man or a woman (ibid). Policy proposals and movements were chosen based on their relation to the PRR and the PRL and each was tested before to ensure that they did represent the PRR and the PRL among the public (ibid). Having read policy proposals and statements from anonymous and fictitious political candidates and movement activists, respondents were then asked to rate proposals/statements on a scale ranging from 1 (= no support for policy/statement or associated party/movement) to 8 (= support for policy/statement and associated party/movement to a great extent) (ibid). In addition to the questions about support for policies/statements, respondents were also asked to rate candidates’/movement activists’ level of warmth and competence. Theoretically speaking, the focus of this study is thus on gender stereotypes, similar to the studies conducted by O´Brien (2019) and Adams et.al (2023), where women are proposed to “[...] soften the masculinist image of such parties and movements” (Ben-Shitrit et.al 2022:86). By conducting an experimental study, the authors were able to test their theoretical 14 mechanisms which is a major advantage of this study. In their analysis, the authors showed that respondents perceived female candidates as warmer than male candidates (ibid). Furthermore, the results also showed that respondents who were given a policy proposal written by a female candidate supported that proposal to a higher degree than respondents who were given a policy proposal written by a man (ibid). In line with this, the few studies that have focused on the effects of PRR female descriptive representation, have mainly focused on outcomes in terms of support for these parties or PRR policies. Even though support is one way of measuring whether a respondent likes or dislikes a candidate/movement, it does not necessarily translate into perceptions of an entire political party. Furthermore, even though proposals typical to the PRL and PRR were chosen in the study of Ben-Shitrit et.al, it is not guaranteed that the Israeli respondents were aware of what type of party or movement proposed a specific proposal. Consequently, and given that they asked about proposals written by single candidates at a single point in time, this study fails to test whether PRR female descriptive representation changes the image of PRR parties. Instead, they can show that proposals written by fictitious female PRR candidates are supported to a higher extent than proposals written by fictitious PRR men because Israeli respondents perceive women to be warmer and kinder. Nonetheless, the theoretical model proposed by Ben-Shitrit et.a (2022), is useful for this thesis and will thus be discussed more in the theoretical framework being one of few models in which the feminine stereotypes of female politicians and the masculine stereotypes of PRR parties are combined. 2.5 Research gap and contributions All in all, the overall effects of increasing female descriptive representation in PRR parties have been scarcely studied even though such an increase presents an interesting puzzle. First and foremost, it has become clear from previous research that few studies account for party ideology in their analysis which might explain why many results regarding, for example, the effects of female descriptive representation, are inconclusive. While extensive research exists regarding the possible effects of increasing female descriptive representation, there is little understanding of such effects within PRR parties and specifically regarding the effects on the perception of such parties. So far, the few studies that have attempted to analyze the effects of female descriptive representation within PRR parties, considering the strategic aspect, have focused on the level of support for these parties and not how they are perceived. 15 Analyzing female descriptive representation within PRR parties is important for several reasons, including the fact that such parties have a conservative view on women's roles in society whereby an interesting puzzle arises when such parties increase their female descriptive representation. Thereto, the strategic aspect of their female descriptive representation and the fact that the number of women within these parties has historically been low, further substantiates the reason for focusing explicitly on these parties. Finally, PRR parties are interesting to analyze in terms of how they are perceived by the public seeing as they are among the most disliked parties and often perceived as more right-wing than their actual policy proposals would suggest. Secondly, despite a wealth of research on the effects of female descriptive representation and citizens’ perceptions of political parties, few scholars have accounted for individual-level characteristics in their analysis, and whether such could change the strength or direction of the relationship. For those who have analyzed the perception of political parties as dependent upon the gender of representatives/party leaders, almost none have asked whether the effect of gender has a stronger impact on certain groups of people, which has also been highlighted as a potential way forward for future research. There are also research gaps of methodological character that will be addressed in this thesis. Not only is there a general demand for more quantitative research on female descriptive representation, but out of the few studies that have dealt with similar subjects, none have analyzed the effects of female descriptive representation within several PRR parties over time as done in this thesis. The dataset produced by Weeks et.al (2022), being the first to comparatively map out the percentage of women MPs in parties according to party family, speaks to the need for methodological improvements within this subject. To summarize, the contributions of this thesis are of both contextual, methodological, and theoretical character. In previous studies where there has been no distinction between party families, or where all party families have been included as control variables, the results have provided a more general overview of the problem at hand. Such studies have helped illustrate a difference between party families, however, this difference has not been analyzed further. Contextually, this study thus contributes with a more in-depth analysis of PRR parties which presents a new and interesting context to investigate. Also, by analyzing the puzzle at hand over time and across parties, and by expanding the original dataset produced by Weeks et.al, this thesis contributes methodologically and theoretically to existing research. The theoretical contribution refers to the application of theoretical models which has previously not been analyzed in such a large-scale study. 16 3. Theoretical framework The section above has been useful in articulating the research gaps, however previous studies on related topics are also useful in constructing a theoretical framework. Building on the mechanisms proposed by O´Brien (2015) and Adams et.al (2023), this theoretical framework will first highlight theories on gender stereotypes. Moreover, taking this theoretical framework one step further these factors and theories relating to gender stereotypes are contextualized and analyzed through a PRR lens. The PRR gender mainstreaming model proposed by Ben-Shitrit et.al. (2022) considers the context of PRR parties and will thus be used in this theoretical framework. In doing so, the proposed hypotheses are adjusted to the PRR context of this thesis. Providing such a lens will hopefully shed further light on how the effect of increasing female descriptive representation varies between groups of voters. Worth mentioning is that the mechanisms proposed will not be tested in this study with regards to the chosen method; instead, they are presented to make clear why a certain relationship is to be expected. 3.1 Gender stereotypes The definition of stereotypes used in this thesis follows in the steps of James L. Hilton and William von Hippel. They define stereotypes as “[...] beliefs about the characteristics, attributes, and behaviors of certain groups” (Hilton & Hippel 1996:240). Stereotypes are not necessarily negative, however when directed towards out-groups they tend to take on a more negative form (ibid). Gender-based stereotypes are very common and there is an extensive body of literature examining the beliefs about men and women. According to Leone Huddy and Nayda Terkildsen (1993), there are two approaches to gender-based stereotypes: the trait approach and the belief approach. Trait stereotypes concern the personality traits attributed to men and women. Men are typically attributed more masculine traits such as toughness, assertiveness, and aggressiveness whereas women are attributed feminine traits such as warmness, passiveness, and kindness (ibid). Belief stereotypes on the other hand consider the ideology and political issue focus attributed to men and women. Men are believed to focus more on conservative issues such as military and defense-related issues whereas women are suggested to focus on welfare-related questions such as school and healthcare (ibid). Additionally, women are also perceived as more democratic and liberal in comparison to men (ibid). In summary, gender stereotypes directly relate to the ideas about femininity and masculinity, or as expressed by Sven Kachel, Melanie C. Steffens and Claudia Niedlich 17 (2016:1), “[...] gender stereotype theory suggests that men are generally perceived as more masculine than women, whereas women are generally perceived as more feminine than men”. Gender stereotypes impact both men and women in their everyday lives and form the opportunities they are offered. Given that women and men are attributed different traits and beliefs, they are also perceived to be better fit for certain jobs and activities. Traditionally, jobs of caretaking have been set aside for women - jobs often associated with the political left where traits such as warmness and kindness are needed (Mudde 2007:111). Men, on the other hand, have been offered jobs of more power with greater opportunities (ibid). For positions of high power, traits such as toughness and assertiveness are often requested which are associated with masculinity. This is particularly evident within the political sphere and may explain why so few women historically have been included in politics. There is a strong link between political leadership and masculinity which not only complicates women's political inclusion but also makes it more difficult for those women who do enter the political sphere (Elad-Strender, Ben- Shitrit & Hirsch-Hoefler 2024). Female politicians who are perceived to lack the masculine traits associated with political leadership are generally less successful, however, this is also evident for female politicians who are perceived as more masculine (ibid). Yet again, this dilemma highlights the difficulties women face when trying to enter the political sphere and consequently, go against feminine stereotypes. Those women who have been successful are typically visible within ideologically left-wing parties where welfare-related issues are more regularly discussed (Krook 2010), which goes to show that belief stereotypes are also indicative of what jobs and activities women are better suited for. Besides the fact that gender stereotypes are expected to shape the opportunities given to men and women, gender stereotypes are also believed to influence female and male behavior. The differences between men and women in their political style relate to communication, issue focus, and attitudes toward other parties. So not only do the perceptions of men and women differ but these stereotypes are also indicative of the behavior of men and women. To exemplify, it appears women adopt less negative and aggressive styles of communication in comparison to their male colleagues (Halsemayer, Dingler & Jenny 2022) in line with the feminine stereotypes. In conclusion, gender stereotypes are not only ideas about the feminine and the masculine, but such stereotypes also influence the opportunity structure and behavior of men and women in their everyday lives. 18 3.2 The clash of the masculine and the feminine In previous studies, theories on gender stereotypes have been used to explain changes in citizens' perceptions of political parties. Increasing female descriptive representation has been hypothesized to have a standardization effect on political parties whereby they are perceived as less extreme and more likable due to feminine stereotypes. As indicated in section 3.1 female stereotypes have hindered women from entering the political sphere because it has been suggested that politicians must be assertive and strong. Nevertheless, within many party families such ideas have been forced out with the rise of the feminist movement and the recognition of gender equality. For the PRR though, the relationship with the feminist movement has been rather puzzling. In this final section of the theoretical framework, the focus will be on how increasing female descriptive representation and in turn female stereotypes clash with the strong sense of masculinity within PRR parties. More specifically, the PRR gender mainstreaming model proposed by Ben-Shitrit et.al (2022), will be analyzed, where the clash of feminine stereotypes and the masculinity of the PRR is expected to have a positive, mainstreaming effect on the image of PRR parties. This model has so far only been analyzed in an Israeli context at a single point in time and there is thus reason to evaluate its validity for cross-country studies. 3.2.1 PRR Gender mainstreaming So far only one study has attempted to understand the clash of stereotypes that occurs when women represent PRR parties, namely the experimental case study in Israel conducted by Ben- Shitrit et.al (2022). In their study, they propose a PRR gender mainstreaming model whereby PRR policy proposals and activists are perceived as more mainstream when they are presented by a woman. The PRR gender mainstreaming model builds on the traditional use of the term “gender mainstreaming”. Gender mainstreaming refers to the institutionalization of gender equality by integrating a gender perspective within all practices of public policy (Daly 2005). Nevertheless, PRR gender mainstreaming rather focuses on the institutionalization of right- wing populism whereby an increasing number of women is expected to mainstream PRR policies. The mechanism explaining the mainstreaming effect is related to the clash between feminine gender stereotypes and the PRR masculine image. When more women are included within these parties, their masculine image is expected to weaken by which they are perceived as warmer and less extreme. Following this theory, it would thus be expected that increasing 19 female descriptive representation within the PRR both make these parties seem less far right in the public eye and make them more liked. Additionally, there is reason to believe this mainstreaming effect is stronger for certain groups of people and we would thus expect a moderating effect from certain individual characteristics. First and foremost, given the PRR gender gap where women refrain from voting because they don't agree with the political style of PRR parties and given the strategic aspect of PRR female descriptive representation to appear less aggressive, it should be expected that the mainstreaming effect of PRR female descriptive representation is stronger for female voters. Moreover, seeing as it has been proven that women in general are not less PRR in their attitudes and beliefs (Harteveld et.al. 2015) it should also be expected that the mainstreaming effect mainly targets the likeability of these parties, rather than their perceived placement on the left-right scale. Secondly, it is also reasonable to assume that respondents' own ideological left-right placements have a moderating effect on the mainstreaming of PRR parties. As mentioned earlier, gender stereotypes typically go in line with the more traditional and conservative views on gender roles, and respondents who agree more with such an outlook on gender should thus also put more emphasis on its importance. Consequently, the more conservative and traditional a respondent is - in other words, the further right on the ideological left-right scale a respondent places themselves, the more they should believe the mainstreaming effect of gender stereotypes. Moreover, similar to the moderating effect of gender, the ideological left-right placement of respondents should also primarily affect the relationship between female descriptive representation and likeability, rather than the ideological left-right placement of PRR parties. As previously mentioned, for many voters, specifically female voters, there are other factors besides the ideology and issue focus of PRR parties that explain their choice to vote for non PRR parties. Among these voters who already to some extent share the views of PRR parties, their perception of these parties’ ideological left-right placement is not expected to change but rather the expressed likeability toward such parties. With this moderating variable I am able to target those right-wing, conservative respondents who share the attitudes and beliefs of PRR parties but dislike them for reasons other than their ideology - and whether an increasing presence of women in such parties has an impact on these voters. 20 3.3 Hypotheses The puzzle of this thesis concerns the effects of PRR female descriptive representation in light of the strategic inclusion of these women and the gender ideology of PRR parties. Those studies that have focused on similar subjects have highlighted the importance of gender stereotype theory for the understanding of how women shape the perception of political parties, and applying a PRR lens, the PRR gender mainstreaming model gives further indication of what type of relationships one should expect. Important to keep in mind is that this model has not yet, to my knowledge, been applied in a study including multiple PRR parties over time and has not yet been used to analyze citizens' perceptions of PRR parties, but rather PRR support. Even though the authors who proposed the PRR gender mainstreaming model were able to confirm their hypotheses it is thus interesting to see whether this model is still relevant outside of Israel where it was initially tested, and for citizens' perceptions rather than their level of support. The first hypothesis concerns the relationship between female descriptive representation and PRR likeability. Since women are perceived to be warmer and kinder in comparison to men, increasing female descriptive representation should change the perception of PRR parties as being warmer and kinder resulting in the following hypothesis: H1: Increasing PRR female descriptive representation makes PRR parties more liked among the public. PRR female descriptive representation PRR likeability Similarly, the perception of women as more left-leaning as well as less extreme in comparison to men should also have a mainstreaming effect on the ideological left-right placement of PRR parties which results in the second hypothesis: 21 H2: Increasing PRR female descriptive representation makes PRR parties appear less extreme right in the public eye. PRR female descriptive PRR ideological left- representation right placement The two first hypotheses are concerned with the more direct relationship between female descriptive representation and PRR likeability, respectively PRR ideological left-right placement. In addition to these hypotheses, this thesis is also concerned with the moderating effect of gender and respondents' left-right placements which constitutes the content of the final two hypotheses. These moderating variables will however only be included for the relationship between female descriptive representation and PRR likeability since it was made clear in the theoretical framework that these should have no specific effect on the relationship between female descriptive representation and PRR ideological left-right placement. For the third hypothesis the moderating effect of gender is included where it will be expected that for women - being of target for PRR strategic descriptive representation, and for whom the political style of the PRR is most repelling, the relationship between female descriptive representation and PRR likeability is stronger. Thus, the following hypothesis is formulated: H3: The effect of PRR female descriptive representation on PRR likeability is stronger for women in comparison to men. PRR female descriptive representation PRR likeability Gender: female 22 Finally, it is also expected that for voters further to the right on the ideological left-right scale, the relationship between female descriptive representation and PRR likeability should be greatest. This is because such voters typically hold more conservative and traditional views on gender - in line with PRR parties. H4: The effect of PRR female descriptive representation on PRR likeability is stronger for far- right voters in comparison to other voters. PRR female descriptive representation PRR likeability Respondents left-right placement: far right 4. Research design 4.1 Data and operationalizations To answer the earlier presented hypotheses, data from three different datasets was collected and merged. All individual-level variables were collected from the Comparative Study of Electoral Systems (CSES). The CSES brings together research on elections from around the world facilitating comparative analysis (CSES 2024). Each participating country within the CSES is provided a module of questions to be sent out following a national election (ibid). These answers from around the world are then merged into public datasets. The questions asked are developed by a team of internationally leading scholars within political science and vary between different modules and contexts in which elections are held (ibid). Available to the public are datasets on individual modules over a few years, and datasets integrating all modules. More specifically, this study made use of the CSES Integrated Module Dataset (IMD) including elections from 59 countries in between the years of 1996 and 2021. In total, over 395,000 individual-level observations and evaluations of around 800 political parties are included. 23 Data on the party level was collected from two different sources: a dataset created by Weeks et.al (2022) on female descriptive representation and the Comparative Manifesto Project (CMP). The dataset by Weeks et.al (2022) is the first of its kind to measure the percentage of women MPs and female party leaders across 187 political parties within 30 European countries from 1985 to 2018. This makes their dataset the most comprehensive to date on female descriptive representation and thus appropriate for this thesis. In collecting variable information, the authors used parliamentary websites and electoral commissions. Finally, the CMP was also used for data on the party level. The CMP analyzes parties’ policy preferences by gathering information on parties’ election manifestos. In total, the CMP has collected information on over 1000 political parties from 1945 to the present. The CMP is a well-recognized source for data collection and is thus useful in adding party-level variables. When creating the new dataset for this study, single variables from the dataset by Weeks et.al and single variables from the CMP dataset were collected and then merged with the CSES IMD dataset. The choice to construct the new dataset on the individual level, rather than on the party level, relates to the purpose of also being able to analyze differences between various groups of voters. Furthermore, the selection of countries and years to include was first and foremost determined by the availability of PRR parties. Consequently, countries and years for which there were no PRR parties present in a national parliament, were excluded. Secondly, the selection of countries and years was also determined by the data availability in all three datasets. To exemplify, as the dataset from Weeks et.al only included European countries, the new dataset was restricted to countries/parties within Europe. More specifically, the following 15 countries were included in the new dataset: Austria, Switzerland, Germany, Denmark, Finland, France, Greece, Hungary, Latvia, Netherlands, Norway, Poland, Slovakia, Slovenia, and Sweden. Within all these countries, there has been one or more PRR parties present in the respective national parliament for at least two elections. PRR parties not represented within a national parliament after at least two elections, were excluded since it would not have been possible to measure change over time for such parties. Furthermore, the new dataset ranges between the years 1996 to 2021 and includes all elections after which the PRR parties of this thesis gained representation in a national parliament. Consequently, there is not information on all PRR parties for all years included in the dataset. The number of years per party was determined by the number of elections held during the investigated period and is shown in Table A1 in the Appendix. Important to mention here is also that the dataset from Weeks et.al only extends to the year 2018, and I thus had to add additional information on the independent variables myself. The reason for doing so was 24 twofold. First, I saw a need to expand the dataset to ensure the validity of the results. If elections after 2018 had not been added, several PRR parties would have had to be excluded. Secondly, seeing as much of the increase in PRR female descriptive representation has occurred in recent years, I found it reasonable to expand the dataset as much as possible. The expansion of the dataset, even though necessary, constitutes one of its major downsides. Given that data was collected at different points in time and by different people, the reliability could be questioned. Moreover, to ensure that the information on female descriptive representation after 2018, which was added, was compatible with the information collected by Weeks et.al, similar methods of data collection were used. Weeks et.al collected their information on female descriptive representation using information from parliamentary websites and electoral commissions which directed me in my way of collecting additional information as well. 4.1.1 Independent variables With this study, I wanted to speak to the effects of female descriptive representation within PRR parties and to do so, the operationalization of female descriptive representation had to match its definition. Descriptive representation, according to Pitkin (1972), refers to the representation of different groups within a population with respect to the different characteristics of the groups. Female descriptive representation thus refers to the presence of women within a political body who represent women as a group. Measuring this presence can be done in a number of ways; most commonly scholars either measure the number of female representatives within a party/parliament or measure the presence of a female party leader. Both these operationalizations are valid based on the definition of female descriptive representation, and both were reasonable to use in this thesis considering the PRR gender mainstreaming model. The independent variables were collected from the dataset produced by Weeks et.al (2022). The variable from Weeks et.al (2022) measuring the percentage of female PRR representatives, is the only variable to date that measures the percentage of women MPs within all parliamentary parties across 30 European countries. The value of this variable depends on the number of seats a specific party has within a parliament; smaller parties that have obtained a small number of seats might receive a higher value when electing one more female representative in comparison to bigger parties with more parliamentary seats to fill. Party size was however included as a control variable which will be discussed in upcoming sections. Furthermore, female party leadership is measured dichotomously. That is, PRR parties 25 with a female party leader after an election were given a value of one whereas parties without a female party leader following an election were given a value of zero, following Weeks et.al (2022). By testing both independent variables a more nuanced picture of the phenomenon in question was achieved. Notwithstanding that these are both sufficient ways of measuring female descriptive representation, there is reason to believe their relationship to the dependent variables differs. By thus having included both variables, no interesting results have risked being overlooked. The main difference between female representatives and female party leaders lies within the visibility of each post. Party leaders are more visible than party representatives and the effect on citizens' perceptions of PRR parties might thus be more apparent for such parties with a female party leader. Nonetheless, the effect is still expected to be the same for both independent variables given the PRR gender mainstreaming model. Another important matter to be discussed is what parties have been categorized as PRR and included in the dataset. First and foremost, it should be mentioned that only PRR parties that had obtained seats in a parliament after a national election were included in the sample, following the study by Weeks et.al (2022). Moreover, the categorization of PRR parties required more consideration. Going back to section 2.3.1, PRR parties are characterized by populism, nativism, and authoritarianism, however, in some cases, it is difficult to determine whether these characteristics are present or not within a party, specifically because PRR parties seldom express themselves as populist, nativist, and authoritarian. To categorize parties and enable the exclusion of all non-PRR parties, the so-called PopuliList was used. The PopuList is a widely used source among scholars for categorizing radical right, radical left, and populist parties in Europe (Roodujin, Froio, van Kessel, Lange, Mudde & Taggart 2023). The list clarifies whether a party is populist, radical left, radical right and/or Eurosceptic. Parties categorized as populist, far-right, and Eurosceptic according to the PopuList, were included in the sample for this thesis. The PopuList also identifies a couple of borderline cases for which there are disagreements among the experts on their classifications. For each such case, the PopuList has carefully explained the reasons for the borderline classification, and after closely investigating each case, I chose to include such parties in the sample. In total, 15 PRR parties were included in the sample, one from each country. 26 4.1.2 Dependent variables For this thesis, there were two dependent variables to be tested, namely PRR parties’ ideological left-right placements and PRR party likeability. These variables propose two different, however equally common ways, in which citizens evaluate political parties. By including both dependent variables the hope was to receive a more in-depth understanding of changes to the perception of PRR parties. Party likeability The first dependent variable for this thesis was the level of likeability expressed towards PRR parties. To measure citizens' expressed likeability towards PRR parties, data from the CSES IMD dataset was used. In the CSES surveys citizens were asked to rate the eight largest political parties in their country on a scale ranging from zero (= strongly dislike) to ten (= strongly like). The dataset thus contained eight variables for citizens' feelings towards political parties, one variable per party size. In other words, the first variable (feelings towards party A) illustrated citizens' feelings of like/dislike towards the largest party in their respective country. Consequently, before using these variables in the analysis, I had to identify all PRR parties from each country in the sample and their size following an election. For example, in Switzerland, the PRR Swiss People's Party was the largest political party after both the elections in 2003 and 2007. To capture citizens’ evaluations of that party I thus had to collect the values for Switzerland from the first likeability variable (feelings towards party A). In doing so for each country and PRR party, I was able to construct a new variable from the pre-existing eight that only contained citizens' feelings towards PRR parties. Whether a respondent likes or dislikes a party is individual and there is no information on the reasons behind feelings of like and dislike. Accordingly, in measuring parties’ likeability there is no single conceptualization of what is meant by liking or disliking a party. Using a 10-point scale like the one from CSES is one of the most common ways of measuring out-party hostility (Iyengar et.al 2019), that is, affective polarization, when combined with individual party choice. Even though affective polarization was not the focus of this thesis, the same thermometer was useful in measuring the overall level of party likeability. 27 Ideological left-right placement The second dependent variable for this study was citizens' perceptions of the ideological left- right placements of PRR parties. This variable was also collected from the CSES IMD dataset but changed to fit the structure of the new dataset. Like the variables measuring feelings towards political parties, the ideological left-right placements of parties were also measured in terms of party size and thus had to be compressed into a single variable measuring only the ideological left-right placement of PRR parties. That was done by using the same methods as for the construction of the likeability variable. Within the CSES survey respondents were asked where they would place the eight largest parties in their party system on a scale from zero (=far left) to ten (=far right). As mentioned earlier, some scholars have questioned the relevance of the ideological left-right scale specifically as regards PRR parties. Given that these parties arose in relation to new societal cleavages that go beyond the traditional economic-based cleavages, criticism has appeared towards using the ideological left-right scale on PRR parties. Following such criticism, one might question the use of the ideological left-right scale in this thesis. Nevertheless, seeing as the ideological left-right scale is one of, if not the most, used scale to capture citizens' perceptions of political parties, it was appropriate for this thesis as well. Additionally, in terms of political ideology and issue focus, there are few other political dimensions to be used, and more so to the point, most surveys only ask respondents about the ideological left-right scale and not other dimensions. Even though the conceptualization of left and right may differ between individuals and countries, changes in the perception of PRR parties’ placements on this scale should indicate whether the perception of these parties is changing due to increased female representation or not. Finally, the ideological left-right scale was also useful in that it speaks to more than just the ideology of political parties; it also speaks to how extreme a party is perceived. Parties that are placed furthest to the left and furthest to the right on the scale are perceived as extreme in their views, in comparison to the other parties of a party system. Consequently, including the ideological left-right scale as a dependent variable was appropriate for this thesis intending to analyze how these parties are perceived, as it speaks to both their perceived ideology and level of extremism. 28 4.1.3 Moderating variables Following hypotheses 3 and 4, two moderating variables were included, namely gender and respondents' placements on the ideological left-right scale. A moderating variable is included because it is expected to change the strength or direction of the focal relationship (Mehmetoglu & Jakobsen 2022). As stated in the theoretical framework, gender and individual left-right placements were included because these were expected to influence the strength of the relationship between female descriptive representation and citizens’ perceptions of PRR parties’ likeability. Gender is a dichotomous variable collected from the CSES IMD dataset where a value of zero has been given to individuals who identified as male, and a value of one has been given to individuals identifying as female. Respondents who identify as neither female nor male have therefore been excluded from the CSES dataset, leading to some non-responses. Nevertheless, seeing as the aim of the gender variable was to test the moderating effect of being a woman, I decided such a non-response would not influence the results. Furthermore, the second moderating variable, individual left-right placement, is a continuous variable ranging from 0 to 10 where higher values are provided by respondents who place themselves further right. This is the same scale used to place political parties on the ideological left-right scale. This variable was included to capture respondents with a more traditional and conservative view on gender and equality. However, in this regard, it should be pointed out that placements on the left-right scale do not translate into views on gender-related issues. Based on the conceptualization of left and right it is possible to assume far-right voters have a more conservative and traditional outlook on gender, although such cannot be known for sure. In the absence of a variable measuring attitudes toward gender equality and gender roles within the CSES dataset, the decision was made to use the individual left-right placement variable despite its shortcomings given the historical conceptualization of left and right. 29 4.1.4 Control variables To control the relationship between female descriptive representation and citizens’ perceptions, several control variables were included in the analysis. The choice of which control variables to include was based on what factors have been brought up in previous research as important for both the independent and the dependent variables. Control variables are those variables that are expected to influence both independent and dependent variables (Mehmetoglu & Jakobsen 2022). First, a variable measuring party size was included in the analysis. The size of a party is expected to influence the opportunity for women to become elected - O´Brien (2015) has shown the opportunity for women to become elected is larger for smaller, minority parties where positions entail less power than in more popular parties, and for citizens' perceptions of political parties. By including a variable for party size, relationships were controlled for changes in the popularity of parties as their sizes reflect how many people have voted for them. Seeing as more peripheral parties often are perceived as more radical and are more disliked (Adams et.al. 2021) it was reasonable to include the party size variable to account for its effect on the independent and dependent variables. The variable for party size was collected from the CSES IMD dataset based on information from the ideological left-right and likeability variables. As mentioned in section 4.1.2 about the CSES survey, respondents were asked where they would place the eight largest political parties in their country on a left-right scale and how much they liked/disliked such parties. From this, I was able to construct a new variable ranging from 1 (=largest party) to 8 (=smallest party) illustrating the size of all PRR parties included in the sample. Secondly, a dummy variable illustrating whether PRR parties were in government cabinet after an election, was also included. This variable was based on a similar variable from the dataset by Weeks et.al (2022). The reason for not using their exact variable was that they had lagged theirs by one year implying that PRR parties´ government position, as an explanatory variable for a specific year, was compared to the value of their outcome variable one year after. It is reasonable to include time lags when it is expected that the effect of an explanatory variable does not occur right away but rather one or more years later. Nevertheless, as there are a limited number of years for every party included in this dataset, using time lags was not an option for this study. Instead, I built on the variable from Weeks et.al theoretically and was able to collect information from parliamentary websites about PRR parties’ government position. Similar to party size, parties' position in government speaks to their popularity and success which may shape their incentives to elect female representatives and 30 female party leaders (O´Brien 2015). Furthermore, as shown by Adams et.al (2021) and Gidron et.al (2021) parties' cooperative public relations affect both how they are perceived in the public eye in terms of their ideological left-right placement, and with regards to their likeability among the electorate. Extreme parties in cooperation with more mainstream parties are both perceived as less extreme and often become more liked (Adams et.al 2021; Gidron et.al 2021). In this regard, it is important to note that being in government does not necessarily imply cooperation with other parties. The differences in the political systems of the countries included made a difference in the way the PRR party government cabinet variable was interpreted. In most European countries a proportional party system is applied where the government must be made up of more than one party. In such countries, it is highly unlikely that one party attains more than half of the votes given the range of party alternatives. Being in a government cabinet thus automatically implies cooperation with other parties. Nevertheless, in countries such as France, where a semi-presidential party system is applied, the winning party/president makes up the government and consequently, in those cases, the PRR party government position variable does not measure cooperation but rather success. Apart from party size and PRR party government position, a final variable measuring the actual ideological left-right placements of PRR parties was added, namely the “rile-index” variable. The rile index was retrieved from the CMP database and illustrates political parties’ position on the left-right scale using party manifestos. The CMP has identified 24 manifesto categories - 12 right-wing categories and the remaining 12 left-wing. To attain a score for an individual party, the CMP summarizes all manifestos proposed by that party and calculates how many left-wing manifestos have been mentioned and how many right-wing have been mentioned. The index ranges from -100, if a party has only mentioned left-wing issues, to +100, if a party has only mentioned right-wing issues. When including the rile index as a control variable the relationship between female descriptive representation and citizens’ perceptions was controlled for the fact that the actual placement on the ideological left-right scale might affect the individual perception. For example, if a respondent knew where a PRR party would be placed on the left-right scale it is possible that such knowledge would have affected their perception of the position of that party, even though they perhaps would have placed the party somewhere else had they not had that information. Finally, the rile score could also influence the independent variables given that previous research has shown the opportunity for women to become elected is higher within parties to the left (Krook 2010), so there seems to be an ideological component to the representation of women as well. 31 4.2 Method In choosing which method to use, the research question and aim of the thesis were the main determinants. In the interest of analyzing the relationships over time and within parties, rather than between parties, I had to find a method in which such comparisons were possible. By using ordinary least squares (OLS) regressions including party fixed effects and clustered standard errors, such comparisons were made possible. In an OLS regression, the linear relationship between one or more independent variables and dependent variables is calculated (Mehmetoglu & Jakobsen 2022). By including party-fixed effects, the relationships were controlled for unobserved differences that were constant over time across parties. This was done by the inclusion of dummy variables for all 15 parties. Change was thus measured within parties rather than between them. Similarly, the model also included year-fixed effects whereby the relationships were controlled for any time- specific factors that might have had an impact on the relationships. Time-fixed effects include all time-invariant factors that affect the dependent variables and are constant over time. Consequently, the relationships were not only controlled for party size, government position, and rile score as described in section 4.1.3, but also for anything that might have differed between parties and for any factors that influenced the relationship constantly over time. Alongside party fixed effects, clustered standard errors were also used. By including clustered standard errors, the regression was further adjusted for the clustered structure of the data. For the sake of analyzing within-party variations rather than between-party variations, clustered standard errors are useful because they analyze the level of correlation between observations within a cluster. In using clustered standard errors, one first has to identify the clusters which for this thesis were made up of PRR parties and there were thus 15 clusters to be accounted for. By means of clustering the regression between independent and dependent variables, the relationships took into consideration the correlation of observations within clusters. 32 5. Results The result part is divided into three sections, one for diagnostics which will be presented first, one for descriptive statistics and one for the regression results. In the diagnostics part, information on the various tests performed to ensure the results are valid and adequate, is provided. The descriptive part of the results aims to give a statistical background to the phenomenon at hand by illustrating developments and summary statistics for the independent and dependent variables. Following the descriptive part will be the actual testing of the earlier presented hypotheses using the two-way OLS model. 5.1 Diagnostics Diagnostics are important to perform on any given regression model to ensure the results are valid and reliable and that the model adequately explains the relationship between independent and dependent variables. There are several different tests to be performed that account for different potential regression problems. First, it should be mentioned that by including clustered standard errors, the regression accounted for several otherwise potential problems such as heteroscedasticity. Heteroscedasticity occurs when the spread of residuals changes depending on the values of the independent variable, in other words, the assumption of error independence is violated (Mehmetoglu, Jakobsen 2022). If heteroscedasticity is not considered, the predictions of the results can be misleading and inaccurate (ibid). The potential problem of heteroskedasticity was not fully resolved when including clustered standard errors. However, the correlation between observations within clusters, which violates the error independence assumption, was considered when including clustered standard errors. In addition to the clustered standard errors which helped account for the within-cluster heteroskedasticity, figures illustrating the plot of the predicted values in relation to the residuals are to be found in the Appendix. From these figures and together with the clustered standard errors, I was able to disregard heteroskedasticity as a problem for this thesis. Furthermore, in addition to heteroscedasticity, autocorrelation was another potential problem to be accounted for. Autocorrelation occurs when the observations within a dataset are not independent of each other but rather there seems to be a time trend present where the value of one observation at one point in time affects the value of another observation at the following time point (Mehmetoglu, Jakobsen 2022). As mentioned in part 4.2.1, time dummies were included in the regression to account for any potential time-fixed factors that might have affected the relationships constantly over time. By including time dummies, autocorrelation 33 was accounted for to some extent, since the regression was controlled for any time-related patterns that might have led to autocorrelation. Moreover, similar figures used to illustrate heteroskedasticity were also useful in visualizing autocorrelation, from where it was evident that no autocorrelation was present. Another final potential problem that was accounted for is that of multicollinearity. Multicollinearity occurs when two or more of the independent variables within a regression model are correlated with each other. To see whether multicollinearity was present in the models of this thesis a correlation matrix was created from which one could see how each of the variables in the model correlate with the other. The correlation matrix is displayed in the Appendix from which it was made clear that none of the independent variables correlate highly with each other. If two variables correlate perfectly with each other they will receive a value of 1 within a correlation matrix, however the highest observed value between two independent variables within the correlation matrix of this thesis was -0,3712 (between government cabinet position and party size) which indicated that there was no problem of multicollinearity present. 5.2 Descriptive statistics The very puzzle of this thesis relates to the observed increase in the descriptive representation of women within PRR parties. Such an increase has been observed for many years and documented by several scholars. Figures A5 and A6 in the Appendix illustrate the change in female descriptive representation over the years included in this dataset. Nevertheless, since the number of observations per year varies depending on which parties had elections during which years, the figures do not give an appropriate picture of change in female descriptive representation. Both figures do show that there is an increase in female descriptive representation as regards the percentage of female representatives and the number of female party leaders, however, this could also be because there were more PRR parties represented in parliaments in the late 2000s which most probably affect the illustration of the development. Even so, figures A5 and A6 in the Appendix are still useful in providing a broader picture and overlook of the data. Having descriptively analyzed the independent variables, the next step is to visualize the distribution of the dependent variables. 34 Figure 1. PRR likeability. Source: Comparative Study of Electoral Systems (2014), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: Values 0 to 4 indicate a low level of likeability whereas values 6 to 10 indicate a higher level of likeability. In Figure 1 the level of likeability for PRR parties is shown. More specifically, the figure shows where, on the likeability scale, respondents have placed the PRR party included from their country and the figure thus summarizes all PRR parties over all years included. The 15 PRR parties of this thesis are very disliked by the voters in their respective countries. Above 30 percent of the respondents have stated that they strongly dislike the PRR party from their country (alternative 0) whereas just over 5 percent have stated that they strongly like that party (alternative 10). When summarizing respondents who rated the PRR party of their country somewhere between 0 and 3 it is apparent that a majority of the respondents dislike PRR parties which goes in line with previous research on citizens’ perceptions of PRR parties. The results from Figure 2, where the ideological left-right placement of PRR parties is presented, also go in line with previous research on citizens' perceptions of PRR parties. 35 Figure 2. PRR ideological left-right placement. Source: Comparative Study of Electoral System, Comparative Manifesto Project & Weeks et.al (2022). Comment: Values 0 to 4 indicate a left-wing placement whereas values 6 to 10 indicate a right-wing placement. From Figure 2 one can see that almost 40 percent of all respondents placed the PRR party included from their respective countries furthest to the right and a majority placed them either furthest to the right or second furthest. Overall, there seem to be few respondents who place these parties anywhere on the left side of the scale (alternative 0 to 4), even though the actual left-right placement of these parties, according to the CMP, is more towards the middle. Consequently, this figure indicates that there is a general perception of PRR parties as being very far-right and extreme. Table A3 in the Appendix illustrates the summary statistics for all variables included in this thesis. There are also summary statistics for all parties in the dataset, illustrated in tables A4 to A18. 36 5.3 Regression results The regression results are organized so that each hypothesis and the accompanied results will be presented individually. The results will then be discussed further in relation to the research questions and previous research in the final part of this thesis. 5.3.1 Hypothesis 1 In Table 1, the bivariate and multivariate relationships between both independent variables and PRR likeability, are illustrated. Having included party-fixed effects and clustered standard errors, the coefficient shows the within-party effect and variation. Table 1. Bivariate and multivariate results for the effect of female descriptive representation on PRR likeability. PRR likeability (1) (2) (3) Bivariate Bivariate Multivariate % women MPs 0.010 0.0038 (0.007) (0.008) Female party leader 0.695 -0.559* (0.550) (0.251) Party size 0.019 (0.070) In government 1.116*** (0.097) Rile score -0.007 (0.005) Constant 2.454*** 1.869† 3.253** (0.402) (1.010) (0.850) Adjusted R2 0.099 0.106 0.103 Observations 75 997 81 192 75 997 Parties 15 15 15 Party FE Yes Yes Yes Year FE Yes Yes Yes Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: Standard errors in parentheses. † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Clustered standard errors included. 37 Column one presents the bivariate relationship between the percentage of female PRR representatives, and the level of likeability expressed towards these parties. The coefficient is not statistically valid, and it is thus not possible to say anything about that relationship. The same goes for the model in column two where the bivariate relationship between female party leaders and PRR likeability is tested. Yet again, the insignificant beta coefficient makes it impossible to draw statistically valid conclusions. Finally, in column three the multivariate relationship is tested where both independent and control variables are included. For this model, the coefficient for the percentage of female PRR representatives is yet again insignificant, however, the coefficient for female party leaders is significantly negative. A significantly negative relationship indicates that the presence of a female party leader decreases the likeability of PRR parties, opposite to what was stated in the first hypothesis. More specifically, having a female party leader decreases the likeability of PRR parties with 0,559 units on the likeability scale. Such results point to the difference between the two independent variables mentioned in section 4.1.1 where the significant result for female party leaders and insignificant result for a percentage of female PRR representatives might be explained by the greater visibility of female party leaders. The “in government” control variable is the only one with a significant effect on the dependent variable indicating that being in government increases the likeability of PRR parties, as expected from previous research. Additionally, the value for the adjusted r square reveals the explanatory power of the model at hand. The multivariate model in column three has an adjusted r-squared value of 0,103 indicating that the model explains 10,3 percent of the variation in the dependent variable. From Table 1 it is thus not possible to confirm the first hypothesis about the positive relationship between female descriptive representation and PRR likeability. Rather, the only significant result showed a negative relationship between female descriptive representation and PRR likeability. 38 5.3.2 Hypothesis 2 Table 2 is similar to Table 1 however with a change of dependent variable from likeability to ideological left-right placement. Table 2. Bivariate and Multivariate results for the effect of female descriptive representation on PRR ideological left-right placement. PRR Ideological left-right placement (1) (2) (3) Bivariate Bivariate Multivariate % women MPs 0.019** 0.021*** (0.005) (0.001) Female party leader 1.370*** 1.241*** (0.241) (0.188) Party size -0.077 (0.053) In government -0.047 (0.094) Rile score -0.013*** (0.002) Constant 6.991*** 5.489*** 5.146*** (0.233) (0.440) (0.443) Adjusted R2 0.167 0.160 0.172 Observations 75 297 80 063 75 297 Parties 15 15 15 Party FE Yes Yes Yes Year FE Yes Yes Yes Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: Standard errors in parentheses. † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Clustered standard errors included. The significantly positive coefficient for column one shows that an increase in the percentage of female PRR representatives within PRR parties results in these parties being perceived as more far right. This leads one to believe that increasing female descriptive representation has a de-mainstreaming effect rather than the mainstreaming effect which was proposed in the second hypothesis. In column 2, the de-mainstreaming effect of increasing female descriptive representation is further substantiated seeing as the coefficient for female party leaders is significantly positive as well. The adjusted r-squared values for models 1 and 2 are 0,167 and 39 0,160. The explanatory power is thus slightly stronger for the first model including the percentage of female PRR representatives, which explains 16,7 percent of the variation in the dependent variable. In the final model, the multivariate relationship is illustrated where both independent variables are significantly positive which further substantiates the relationship found in the bivariate models. When including control variables, relationships are controlled for factors that could have an impact. These results are thus more reliable as they stay significant even when controls are included. The coefficient with a value of 0,021 indicates that an increase in the percentage of PRR female representatives leads to a shift on the ideological left-right scale towards the right with 0,021 scale steps. Furthermore, the coefficient with a value of 1,241 shows that the presence of a female party leader within PRR parties leads to a radical right shift of 1,241 scale steps. The fact that both coefficients for the independent variables are significantly positive further proves the validity of the results and demonstrates the usefulness of including both independent variables. By including both independent variables, the relationship between female descriptive representation and citizens' perceptions of PRR parties undergoes a much tougher test than if only one of the variables had been included. As in Table 1, only one of the control variables is significant, namely the rile score. The rile score variable appears to have a significantly negative effect on the dependent variable indicating that the more right-wing a PRR party is according to their party manifestos, the more left-wing they are perceived by the public. Such a result is interesting because one would not have expected that outcome. However, since it does not speak to the main relationship any more than clarifying that the rile score does have an impact, it should not be discussed further in this thesis. Finally, the adjusted r-squared value of 0,172 demonstrates that the model including all control variables and both independent variables has the most explanatory power by which it is possible to explain 17,2 percent of the variation in the dependent variable. All in all, from these results it is not possible to confirm the second hypothesis, but rather than there being no significant effect, these results imply a negative relationship instead of a positive one. The results from both Tables 1 and 2 thus question the usefulness of PRR gender mainstreaming in explaining citizens´ perceptions of PRR parties. Such will be addressed further in the discussion part of the thesis. 40 5.3.3 Hypotheses 3 and 4 The results from Table 3 account for the effect of the moderating variables, gender, and respondents' left-right placements. All models are multivariate, thus including control variables, however, these have been excluded from the table to facilitate the reading. Table 3. Bivariate and multivariate results for the effect of female descriptive representation on PRR likeability when including moderators. PRR likeability (1) (2) (3) (4) Model 1 Model 2 Model 3 Modell 4 % women MPs 0.002 0.003 0.005 0.002 (0.008) (0.009) (0.027) (0.009) Gender: female -0.317* -0.276* -0.293** -0.288** (0.119) (0.096) (0.088) (0.088) Female # women MPs 0.001 (0.005) Female party leader -0.555 -0.516 -0,550 -1.631* (0.396) (0.433) (0.394) (0.594) Female # party leader -0.082 (0.114) Individual left-right 0,456*** 0.455*** 0.462* 0.418** (0.093) (0.093) (0.161) (0.404) Left-right # women MPs 0.000 (0.004) Left-right # party leader 0.199† (0.112) Constant 0.537 0.519 0,476 0.759 (1,438) (1.459) (1,665) (1.416) Adjusted R2 0.219 0.220 0.219 0.233 Observations 70 103 70 103 70 103 70 103 Parties 15 15 15 15 Party FE Yes Yes Yes Yes Year FE Yes Yes Yes Yes Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: Standard errors in parentheses. † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Clustered standard errors included. The following control variables are included: party size, government coalition and rile score. 41 In model 1 the multivariate relationship between the percentage of female PRR representatives and PRR likeability including gender as a moderator, is visualized. When including a moderator, the interpretation of the beta coefficient must be adjusted. The value of the coefficient reveals the effect of the independent variable on the dependent variable when the moderating variable is held constant. Additionally, the interaction effect between independent and moderating variables explains if, and how, the effect of the independent variable on the dependent variable changes based on the value of the moderating variable. From the results in model 1, it is evident that gender does not have a moderating effect on the relationship between the percentage of female PRR representatives and PRR likeability. The interaction effect is insignificant and so is the coefficient for the independent variable indicating that there is no statistically valid effect of the percentage of female PRR representatives on likeability when gender is held constant. The effect of the moderating variable is however significant, which shows that gender influences the level of likeability expressed towards PRR parties when the percentage of female PRR representatives is held constant. More so, changing from a male respondent to a female respondent decreases PRR likeability with 0,317 units. Model 2 presents a similar relationship as in model 1 but with a change of independent variables. In model 2 female party leadership is included as the independent variable and the interaction effect is thus between female party leadership and gender. Still, there is an insignificant interaction effect between female party leadership and gender implying that gender does not change the direction or strength of the relationship between independent and dependent variables. However, the moderating variable gender is still significant when the female party leader variable is held constant. In comparing the two first models there are small but negligible differences. Gender as a moderating variable does not seem to have an impact on either the relationship between the percentage of female PRR representatives and likeability or female party leadership and likeability. Moving on to models 3 and 4 where the second moderating variable is tested. In model 3 the relationship between the percentage of female PRR representatives and likeability when including respondents’ left-right placement as a moderating variable, is visualized. Neither the interaction term, nor the independent variable is significant, and it is thus not possible to draw any statistically valid conclusions regarding the effect of individual left-right placement on the relationship between independent and dependent variables. Moreover, in model 4 there is an observable effect of individuals’ left-right placement on the relationship between female party leadership and PRR likeability. First, one can observe that the coefficient is significantly negative with a value of -1,631 which indicates that when individual left-right placement is 42 held constant, there is a negative effect of female descriptive representation on the level of PRR likeability. Furthermore, the interaction term in model 4 has a statistically significant value of 0,199 which implies that the effect of female descriptive representation on PRR likeability changes with 0,199 scale steps resulting from a one-step change towards the right in the moderating variable. The fact that the interaction effect is positive rather than negative also indicates that an individual change towards the right has a positive effect on likeability rather than a negative effect which was observed when the moderating variable was held constant. To receive a more intuitive picture of the moderating effect, a margins plot is included in the analysis, and presented below. Figure 3. Margins plot, Female party leadership, and respondents’ left-right placements. Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: The slope shows the average marginal effect of female party leadership, that is, when the dichotomous variable for female party leader is 1. In Figure 5 the results from Model 4 in Table 3 are further illustrated. Figure 5 shows the average marginal effects of having a female party leader on PRR likeability depending on different values of the moderating variable. The confidence intervals shown in the graph help evaluate the significance of the results. If a confidence interval for a value of the moderating variable includes zero, the effect is not significant. However, if a confidence interval for a value of the moderating variable does not include zero, the effect of having a female party leader on PRR likeability is significantly different from zero for such values of the moderating variable. 43 In the Figure, one can thus observe that an individual left-right placement of 0, 1, 2, 3 or 4 has a significant effect on the relationship between female descriptive representation and PRR likeability. For all these values of the moderating variable, the effect of female descriptive representation on PRR likeability is negative however the dislike towards PRR parties seems to increase for individuals further to the left. Consequently, for respondents who place themselves on the left side of the left-right scale the negative relationship between female descriptive representation and PRR likeability which has previously been established, is strengthened. For respondents who place themselves on the right side of the ideological left-right scale (placement 5 to 10), the results are not significant seeing as the confidence intervals include zero. The effect of the moderating variable is thus only valid for left-wing respondents whereas the presence of a female party leader does not seem to affect the likeability of PRR among right-wing voters. Altogether, these results from Table 3 and Figure 5, do not confirm either of the final hypotheses. As regards the moderating effect of gender, the results showed no sign of gender moderating the relationships between independent and dependent variables leading to a rejection of the third hypothesis. Individual left-right placements as a moderator did however prove to have a significant effect on the relationships, although not the same effect as expected from the fourth and final hypothesis. The fourth hypothesis stated that the effect of female descriptive representation on the expressed likeability towards PRR parties should be stronger for far-right voters in comparison to other voters. Nonetheless, the effect of the moderating variable was only significant for left-wing voters thus finding no support for the fourth hypothesis. 44 6. Discussion From the results, neither of the four hypotheses can be confirmed. This is either because of insignificant results or the results showing a negative relationship where the hypotheses propose a positive. The results indicate that increasing female descriptive representation within PRR parties makes such parties seem more radical right and become less liked, especially among far-left voters. Consequently, these results show a different picture than the one proposed in previous research. In this final part of the thesis, the results will be discussed further in relation to previous research and theory. The discussion will be divided into three sections representing the two research questions followed by a discussion of the implications and limitations. 6.1 The effects of PRR female descriptive representation In answering the first research question, the results show an effect of increasing female descriptive representation on the perception of PRR parties. Having found such results there is reason to believe citizens respond to changes in the level of PRR female descriptive representation by altering their perception of PRR parties. Nevertheless, rather than having a mainstreaming effect as proposed by the PRR gender mainstreaming model, female descriptive representation appears to make PRR parties seem even more radical right and become more disliked. Having found results that go against previous research, it is further necessary to elaborate on possible reasons for the rejection of the first two hypotheses. First and foremost, it should be mentioned that up until the conduct of this thesis, the PRR gender mainstreaming model had only been applied in an Israeli context. One of the major downsides with single case studies is the problem of generalization - the results from one case are not necessarily relevant in other cases given the difference in contexts. The case of Israel is in many ways different from the countries included in this thesis given their current involvement in a protracted conflict that affects not only the political agenda but also the attitudes toward the PRR (Ben-Shitrit et.al. 2022). Periods of conflict and polarization are particularly favorable to the PRR and consequently, the observed mainstreaming of the PRR in Israel might partially be explained by their involvement in a protracted conflict (ibid). Furthermore, it would also be possible to explain these results with regard to the dilemma women face when being appointed to political positions of power. In section 3.1 about gender stereotypes, this dilemma was partially highlighted. It appears women face difficulties 45 when entering politics because of the historical masculinization of politics. On the one hand, female politicians who are perceived as being “too feminine” often face penalties, however on the other hand those female politicians who instead counteract feminine stereotypes also risk facing penalties for being “too masculine” (Elad-Strender et.al. 2024). Since the masculinization of politics is particularly evident within PRR parties, the dilemma highlighted above should be even more apparent for female politicians within PRR parties where gender norms are more nuanced and where masculinity is more widespread. On the one hand, female PRR representatives who are perceived as “too feminine” should face more penalties than female politicians in other political parties because of the PRR masculinity. On the other hand, PRR female politicians should feel a greater need to counter feminine stereotypes because masculinity is so strong in right-wing populist parties, and in turn, face more penalties in comparison to other female politicians. Additionally, the fact that the PRR political style is generally more aggressive than the political style of other parties (Gidron et.al 2023), further substantiates these assumptions. This argument thus highlights a situation where the clash between the masculine and feminine results in a de-mainstreaming rather than a mainstreaming. Finally, the rejection of the PRR gender mainstreaming model could also be explained by the fact that citizens may see these PRR gender mainstreaming strategies as for what they are - strategies rather than actual gender equality improvements. For such respondents, I would expect female descriptive representation to have a de-mainstreaming effect on PRR parties. In other words, it is reasonable to assume PRR parties become more disliked among respondents who see the increasing female descriptive representation as a support strategy. 6.2 The moderating effect of gender and left-right placements The rejection of the first two hypotheses can be explained further by analyzing the second research question. In hypotheses 3 and 4, the moderating effect of gender and individual left- right placement is tested whereby both have to be rejected. In answering the second research question, these results indicate that individual left-right placement does influence the relationship between female descriptive representation and PRR likeability where the effect is strongest for far-left respondents who express more dislike towards PRR parties following an increase in PRR female descriptive representation. Evidently, female descriptive representation does not have a significant effect on PRR likeability among far-right voters. This might be explained by the fact that these respondents already have a somewhat positive image of PRR parties, given their political ideology. One 46 could thus suggest that the PRR ideology weighs heavier on the perception of PRR parties than their representatives. Going back to previous research, similar arguments have been suggested by other scholars as well, for example, Campbell and Heath (2017) as well as Martin (2019). The observed negative effect of female descriptive representation thus only applies to left-wing voters. On that note, in the previous discussion, I considered whether respondents who see the increasing female descriptive representation as a support strategy become more skeptical of PRR parties. Having established that the increasing dislike towards PRR parties is strongest for far-left respondents, one could also assume that these respondents see through the PRR gender mainstreaming strategy. This is a reasonable assumption to make, seeing as far-left voters should be most skeptical towards PRR parties in general, regardless of PRR gender mainstreaming. Furthermore, the observed increasing dislike among far-left voters could also be explained with regard to the dilemma of female politicians which was mentioned in section 6.1. Given the strong sense of masculinity within PRR parties, PRR female politicians might feel the need to appear not only more masculine but also more far-right in comparison to their male colleagues. Such behavior would in turn repel far-left voters even more as these voters should be more sensitive towards radicalization of the PRR in comparison to far-right voters. The dilemma of female politicians could also explain the rejection of the third hypothesis, about the moderating effect of gender. PRR female politicians might counteract their feminine stereotypes to the extent that they are perceived as less warm and kind which would repel female voters even more. Moreover, the fact that the moderating effect of gender did not prove to be significant further questions the effects of female stereotypes. With regards to the PRR gender gap where women are expected to vote less for PRR parties because they disagree with their political style and because PRR issues are less salient to women (Harteveld et.al. 2015), one would expect that had female stereotypes been of importance to the perception of PRR parties, the likeability would increase amongst women. In sections 6.1 and 6.2 the results and hypotheses concerning both research questions have been discussed, providing possible explanations for the rejection of all four hypotheses. In the final section of the discussion, I aim to understand the results within a broader context examining its implications for the research fields at large. 47 6.3 Implications and limitations Seeing as these results invalidate all four hypotheses, the overall relevance of the PRR gender mainstreaming model is to be questioned. Going back to the theoretical framework, the PRR gender mainstreaming model assumes a positive relationship between female descriptive representation and PRR support because of the clash between the feminine and masculine (Ben- shitrit et.al. 2022). This thesis has, rather than further substantiating the use of the PRR gender mainstreaming model, questioned its relevance, at least in analyzing the perception of PRR parties. In relation to those studies that have attempted to investigate the effects of female descriptive representation on citizens´ perceptions of political parties using stereotypes, it has from this study become evident that applying a party lens is important because political ideology plays a role in determining what type of effects should be expected and the direction of the relationships. Additionally, political ideology plays another important role in this context seeing as the PRR political ideology appears to weigh heavier on the perception of these parties in comparison to the gender of their representatives. For far-right voters, the PRR ideology surpasses the effect of female descriptive representation in such a way that the perception of PRR parties does not change. For far-left voters, the PRR ideology surpasses the effect of female descriptive representation in such a way that they become even more skeptical of these parties. This contributes to the overall field of female descriptive representation by showing that applying a party lens might help understand under what circumstances political ideology versus representatives matters. Some researchers have shown that when female descriptive representation is not accompanied by substantive representation, there is no real improvement in the overall representation of women - in terms of how female voters feel (Campbell & Heath 2017; Martin 2019). The results from this study are thus aligned with previous research but also show that such is true in analyzing party perceptions, particularly regarding the perceptions of PRR parties. Under the circumstances that the visibility of women is not accompanied by gender-related improvement to the political agenda (as is often the case with PRR female descriptive representation), one might expect representatives to matter less. Furthermore, these results in relation to both the first and the second research question, are also relevant in commenting on the successfulness of strategic descriptive representation. Going back to previous research, strategic descriptive representation is used by PRR parties to attract more voters, but primarily female voters as there is a significant gender gap in voting. Whether such strategies do attract female voters is however still up for debate. Even though this thesis has focused on citizens’ perceptions rather than PRR support, the results offer 48 another gateway into the understanding of strategic descriptive representation and its effectiveness. First and foremost, since the results show female descriptive representation, in general having more of a de-mainstreaming effect on citizens' perceptions of PRR parties, the effectiveness of strategic descriptive representation could be questioned. Strategic descriptive representation relies on similar mechanisms as proposed by the PRR gender mainstreaming model whereby female descriptive representation should attract female voters because PRR parties are perceived as warmer and less extreme. Nevertheless, as the perception did not appear to change in the expected direction, not only is the PRR gender mainstreaming model partially to be questioned, but the effectiveness of strategic descriptive representation too. Secondly, the fact that gender did not have a significant moderating effect further questions its effectiveness in attracting female voters. Finally, it is also important to highlight some of the thesis limitations. First, it is worth mentioning that individual left-right placement as a moderator does not directly measure attitudes toward gender-related issues. In the theoretical framework, it was suggested that the more conservative and traditional a respondent is, the more they should be affected by female stereotypes. Far-right voters are typically more conservative and traditional and individual left- right placement was thus included as a moderating variable. Nevertheless, since respondents were not asked about their attitudes towards gender-related issues there is no way of knowing whether all far-right respondents share the same conservative and traditional views on gender roles. Additionally, it should be mentioned that the significant relationships show correlation rather than causality. The difference between correlation and causality is that with causality one can show that the occurrence of one event is a direct result of another event. Even though, the results indicated a de-mainstreaming effect of female descriptive representation, there is no way of knowing, for sure, whether such de-mainstreaming is the result of increasing female descriptive representation. Causality is commonly achieved in experimental studies where the mechanisms are tested, which was not the case in this study. Consequently, the only conclusion to be drawn regarding the PRR gender mainstreaming model, is that it failed to explain the tested relationships. Although, there is no way of knowing whether stereotypes played a role in the observed de-mainstreaming of PRR parties. Finally, I should also mention that since there were only a limited number of years available per party - even though the dataset was expanded, there is a need for more research on this topic in the future when more data is available. 49 7. Conclusion The aim of this thesis has been to highlight the effects of PRR female descriptive representation on the public perception of such parties. More specifically, the analysis has been guided by two research questions about the effects of increasing female descriptive representation for the perception of PRR parties, and for whom such effect is strongest. PRR female descriptive representation is an important area of research given the historical underrepresentation of women within these parties, their traditional view on gender issues, and their strategic appointment of female representatives/candidates. These are also the underlying factors of the puzzle at hand. Nevertheless, in both areas of research, that is citizens’ perceptions of political parties and female descriptive representation, few studies have applied a PRR party lens. For those who have, the focus has either been on explaining PRR support or PRR strategic descriptive representation. Going beyond previous research, the following thesis has applied a PRR party lens making contributions of contextual, theoretical, and methodological character. Building on previous research, theories on gender stereotypes and more specifically, the PRR gender mainstreaming model was used. From such, I expected PRR female descriptive representation to have a mainstreaming effect on PRR parties by which they would be perceived as less extreme right and become more liked. These assumptions were represented by the first two hypotheses and joined by two final hypotheses about the moderating effect of gender and individual left-right placement. To test these four hypotheses, data from three different sources were used combining individual-level data with data on party-level in-between the years of 1996 and 2021. The method of choice was a two-way OLS regression model including party fixed effects and clustered standard errors. From the results neither of the four hypotheses were confirmed; rather the results showed a de-mainstreaming effect opposite to what was expected from the PRR gender mainstreaming model. Increasing female descriptive representation within PRR parties appeared to have made such parties seem even more extreme right and become more disliked amongst the public. Also, this negative effect was strongest for far-left voters whereas there was no effect for far-right voters and no difference between female and male voters. This rejection of the hypotheses has implications for the reliability of the PRR gender mainstreaming model and questions the successfulness of PRR strategic descriptive representation. From this study, it has become clear that applying a party lens helps bring further clarity into an otherwise inconclusive field of research. The clash between feminine and masculine stereotypes in PRR 50 female descriptive representation should be analyzed more in future studies focusing on possible explanations for the observed de-mainstreaming effect. By conducting more experimental studies in the future with the aim of explaining the de-mainstreaming effect, causality is also more easily achieved. Possible explanations for the de-mainstreaming effect were brought about in the discussion part of this thesis and have thus paved the way for future research. Additionally, seeing as this study is the first of its kind with data on PRR female descriptive representation up until 2021 it is important to analyze the de-mainstreaming effect further including more PRR parties and more elections. In summary, this thesis has brought to attention the necessity of analyzing PRR female descriptive representation - a subject of continued importance for future studies. 51 8. Bibliography Adams, Bracken, D., Gidron, N., Horne, W., O’brien, D. Z., & Senk, K. (2023). Can’t We All Just Get Along? How Women MPs Can Ameliorate Affective Polarization in Western Publics. The American Political Science Review, 117(1), 318–324. https://doi.org/10.1017/S0003055422000491 Adams, Weschle, S., & Wlezien, C. (2021). Elite Interactions and Voters' Perceptions of Parties' Policy Positions. American Journal of Political Science, 65(1), 101–114. https://doi.org/10.1111/ajps.12510 Barnes, & Burchard, S. M. (2013). “Engendering” Politics: The Impact of Descriptive Representation on Women’s Political Engagement in Sub-Saharan Africa. Comparative Political Studies, 46(7), 767–790. https://doi.org/10.1177/0010414012463884 Bassan-Nygate, & Weiss, C. M. (2022). Party Competition and Cooperation Shape Affective Polarization: Evidence from Natural and Survey Experiments in Israel. Comparative Political Studies, 55(2), 287–318. https://doi.org/10.1177/00104140211024283 Ben‐Shitrit, Elad‐Strenger, J., & Hirsch‐Hoefler, S. (2022). ‘Pinkwashing’ the radical‐right: Gender and the mainstreaming of radical‐right policies and actions. European Journal of Political Research, 61(1), 86–110. https://doi.org/10.1111/1475-6765.12442 Bornschier. (2010). The New Cultural Divide and the Two-Dimensional Political Space in Western Europe. West European Politics 33(3): 419-444. Campbell, & Heath, O. (2017). Do Women Vote for Women Candidates? Attitudes toward Descriptive Representation and Voting Behavior in the 2010 British Election. Politics & Gender, 13(2), 209–231. https://doi.org/10.1017/S1743923X16000672 Caprara, G. V. & Vecchione. (2018). On the Left and Right Ideological Divide: Historical Accounts and Contemporary Perspectives: On the Left and Right Ideological Divide. Political Psychology, 39, 49–83. https://doi.org/10.1111/pops.12476 Chrisafis, A., Connolly, K., & Giuffrida, A. (2019). From Le Pen to Alice Weidel: how the European far-right sets its sight on women. The Guradian. https://www.theguardian.com/lifeandstyle/2019/jan/29/from-le-pen-to-alice-weidel- how-the-european-far-right-set-its-sights-on-women Chueri, & Damerow, A. (2023). Closing the gap: how descriptive and substantive representation affect women's vote for populist radical right parties. West European Politics, 46(5), 928–946. https://doi.org/10.1080/01402382.2022.2113219 52 Dalton, R. J., & McAllister. (2015). Random Walk Or Planned Excursion? Continuity and Change in the Left–Right Positions of Political Parties. Comparative Political Studies 48 (6): 759–787. Dalton, R. J. (2013). Citizen politics: Public opinion and political parties in advanced industrial democracies. Cq Press. Daly. (2005). Gender Mainstreaming in Theory and Practice. Social Politics, 12(3), 433–450. https://doi.org/10.1093/sp/jxi023 Danmarks Statistik (2020). Folketingsvalget 5. juni 2019. https://www.dst.dk/Site/Dst/Udgivelser/GetPubFile.aspx?id=34694&sid=ftvalg2019 (Accessed: 2024/02/07) Danmarks Statistik (2001). Denmark in figures 2001. https://www.dst.dk/Site/Dst/Udgivelser/GetPubFile.aspx?id=3151&sid=entire_public ation (Accessed: 2024/02/07) Danmarks Statistik (2001). Befolkning og Valg. https://www.dst.dk/Site/Dst/Udgivelser/GetPubFile.aspx?id=20330&sid=valg2001 (Accessed: 2024/02/07) Dona. (2020). What’s gender got to do with populism? The European Journal of Women's Studies, 27(3), 285–292. https://doi.org/10.1177/1350506820929222 Elad-Strenger, Ben-Shitrit, L., & Hirsch-Hoefler, S. (2024). Mainstreaming democratic backsliding: The role of gender stereotypes. European Journal of Political Research. https://doi.org/10.1111/1475-6765.12667 Erzeel, & Rashkova, E. R. (2017). Still men's parties? Gender and the radical right in comparative perspective. West European Politics, 40(4), 812–820. https://doi.org/10.1080/01402382.2017.1286181 Gidron, Adams, J., & Horne, W. (2023). Who Dislikes Whom? Affective Polarization between Pairs of Parties in Western Democracies. British Journal of Political Science, 53(3), 997–1015. https://doi.org/10.1017/S0007123422000394 Giger, Traber, D., Gilardi, F., & Bütikofer, S. (2022). The surge in women's representation in the 2019 Swiss federal elections. Swiss Political Science Review, 28(2), 361–376. https://doi.org/10.1111/spsr.12506 Harteveld, Mendoza, P., & Rooduijn, M. (2022). Affective Polarization and the Populist Radical Right: Creating the Hating? Government and Opposition (London), 57(4), 703– 727. https://doi.org/10.1017/gov.2021.31 53 Harteveld. (2016). Winning the ‘losers’ but losing the ‘winners’? The electoral consequences of the radical right moving to the economic left. Electoral Studies 44: 225-234. Harteveld, E., Van Der Brug, Dahlberg & Kokkonen. (2015). The gender gap in populist radical-right voting: examining the demand side in Western and Eastern Europe, Patterns of Prejudice, 49:1-2, 103-134, DOI: 10.1080/0031322X.2015.1024399 Haselmayer, Dingler, S. C., & Jenny, M. (2022). How women shape negativity in parliamentary speeches - a sentiment analysis of debates in the Austrian parliament. Parliamentary Affairs, 75(4), 867–886. https://doi.org/10.1093/pa/gsab045 Hilton, & von Hippel, W. (1996). Stereotypes. Annual Review of Psychology, 47(1), 237–271. https://doi.org/10.1146/annurev.psych.47.1.237 Hooghe, Marks, G., & Wilson, C. J. (2002). Does Left/Right Structure Party Positions on European Integration? Comparative Political Studies, 35(8), 965–989. https://doi.org/10.1177/001041402236310 Huddy, & Terkildsen, N. (1993). Gender Stereotypes and the Perception of Male and Female Candidates. American Journal of Political Science, 37(1), 119–147. https://doi.org/10.2307/2111526 Immerzeel, Coffé, H., & van der Lippe, T. (2015). Explaining the gender gap in radical right voting: A cross-national investigation in 12 Western European countries. Comparative European Politics (Houndmills, Basingstoke, England), 13(2), 263–286. https://doi.org/10.1057/cep.2013.20 Iyengar, Lelkes, Y., Levendusky, M., Malhotra, N., & Westwood, S. J. (2019). The Origins and Consequences of Affective Polarization in the United States. Annual Review of Political Science, 22(1), 129–146. https://doi.org/10.1146/annurev-polisci-051117- 073034 Kachel, Steffens, M. C., & Niedlich, C. (2016). Traditional masculinity and femininity: Validation of a new scale assessing gender roles. Frontiers in Psychology, 7, 956–956. https://doi.org/10.3389/fpsyg.2016.00956 Krook, M. L. & Zetterberg. (2014). Introduction: Gender Quotas and Women's Representation - New Direction in Research, Representation, 50:3, 287-294, DOI: 10.1080/00344893.2014.951168 Krook, Mona Lena. (2010). Why Are Fewer Women Than Men Elected? Gender and the Dynamics of Candidate Selection. Political Studies Review 8 (2): 155–68. https://doi.org/10.1111/j.1478-9302.2009.00185.x. 54 Löffler, Luyt, R., & Starck, K. (2020). Political masculinities and populism. Norma: International Journal for Masculinity Studies, 15(1), 1–9. https://doi.org/10.1080/18902138.2020.1721154 Martin. (2019). Playing the Women’s Card: How Women Respond to Female Candidates’ Descriptive Versus Substantive Representation. American Politics Research, 47(3), 549–581. https://doi.org/10.1177/1532673X18776622 Mehmetoglu, M., & Jakobsen, T. G. (2022). Applied statistics using Stata: a guide for the social sciences (Second edition). SAGE. Meyer, & Wagner, M. (2020). Perceptions of parties’ left-right positions: The impact of salience strategies. Party Politics, 26(5), 664–674. https://doi.org/10.1177/1354068818806679 Mudde. (2007). Populist Radical Right Parties in Europe. O'Brien. (2019). Female leaders and citizens' perceptions of political parties. Journal of Elections, Public Opinion and Parties, 29(4), 465–489. https://doi.org/10.1080/17457289.2019.1669612 O Brien. (2015). Rising to the top: Gender Political Performance, and Party Leadership in Parliamentary Democracies. American Journal of Political Science. 59(4): 1022-1039 Palonen. (2020). Finland: Political Developments and Data in 2019. European Journal of Political Research. Political Data Yearbook, 59(1), 130–141. https://doi.org/10.1111/2047-8852.12297 Pitkin. (1972). The concept of representation. Univ. of Calif. P. Republic of Latvia. (2011). 11th Saeima. https://titania.saeima.lv/Personal/Deputati/Saeima11_DepWeb_Public.nsf/deputies?O penView&lang=EN&count=1000 (Accessed: 2024/03/25) Republic of Latvia. (2010). 10th Saeima. https://titania.saeima.lv/Personal/Deputati/Saeima10_DepWeb_Public.nsf/deputies?O penView&lang=EN&count=1000 (Accessed: 2024/03/25) Rooduijn, Pirro, L.P., Halikiopoulou, Froio, van Kessel, de Lange, L., Mudde, and Taggart. (2023). The PopuList: A Database of Populist, Far-Left, and Far-Right Parties Using Expert-Informed Qualitative Comparative Classification (EiQCC). British Journal of Political Science, 1–10. https://doi.org/10.1017/S0007123423000431 Santana, & Aguilar, S. (2019). Bringing Party Ideology Back In: Do Left-Wing Parties Enhance the Share of Women MPs? Politics & Gender, 15(3), 547–571. https://doi.org/10.1017/S1743923X1800048X 55 Somer, McCoy, J. L., & Luke, R. E. (2021). Pernicious polarization, autocratization and opposition strategies. Democratization, 28(5), 929–948. https://doi.org/10.1080/13510347.2020.1865316 The Comparative Study of Electoral Systems (www.cses.org). CSES Integrated Module Dataset Phase 4 release [dataset and documentation]. February 27, 2024 version. doi:10.7804/cses.imd. (Accessed: 2024/02/27). Tyska Förbundsdagen. (2023). Andel kvinnor i den 20:e tyska förbundsdagen efter parlamentarisk grupp 2023. I Statista. https://de.statista.com/statistics/daten/studie/1063172/umfrage/frauenteil-im- bundestag-nach-fraktionen-in-deutschland/ (Accessed: 2024/03/12) UN Women. (2023). Facts and figures: Women's leaderships and political participation. https://www.unwomen.org/en/what-we-do/leadership-and-political-participation/facts- and-figures (Accessed: 2024/03/01) Várnagy. (2019). Hungary: Political developments and data in 2018. European Journal of Political Research. Political Data Yearbook, 58(1), 123–131. https://doi.org/10.1111/2047-8852.12257 Weeks, Meguid, B. M., Kittilson, M. C., & Coffé, H. (2023). When Do Männerparteien Elect Women? Radical Right Populist Parties and Strategic Descriptive Representation. The American Political Science Review, 117(2), 421–438. https://doi.org/10.1017/S0003055422000107 Wängnerud. (2009). Women in Parliaments. Descriptive and Substantive Representation. Annual Review Of Political Science, 2009, Vol. 12, Pp. 51-.69, 12, 51–69 56 Appendix Table A1. Illustration of which PRR parties are represented during which years. Party Module 1 Module 2 Module 3 Module 4 Module 5 Austrian Freedom Party 2008 2013 2017 Swiss People's Party 1999 2003 2007 2011 2019 Alternative for Germany 2017, 2021 Danish People's Party 1998 2001 2007 2019 True Finns 2007, 2011 2015 2019 French National Front 2007 2012 2017 Greece Golden Dawn 2012, 2015 Fidesz Hungary 2002 2018 “All for Latvia” 2010, 2011 2014 2018 Party of Freedom Netherlands 2006 2010 2017, 2021 Progress Party Norway 1997 2001, 2005 2009 2013 2017 Law and Justice Party Poland 2001 2005 2007 2011 2019 Slovak National Party 2010 2016 Slovenian National Party 1996 2004 2008 Sweden Democrats 2014 2018 Source: Comparative Study of Electoral Systems (2024). Comment: The categorization of years into five modules is based on the CSES modules. Module 1 includes the years 1996 to 2001; Module 2 includes the years 2001 to 2006; Module 3 includes the years 2006 to 2011; Module 4 includes the years 2011 to 2016, and finally, module 5 includes the years 2016 to 2021. Figure A1. Scatterplot of residuals vs predicted values for multivariate regression with PRR likeability variable. Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: The figure shows no signs of heteroscedasticity since the spread of residuals does not appear to change depending on the predicted values. If heteroscedasticity is present within a regression the plot of residuals often resembles a cone where the spread increases/decreases depending on the predicted values. 57 Figure A2. Scatterplot of residuals vs predicted values for multivariate regression with PRR ideological left-right placement variable. Source: Comparative Study of Electoral System (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: The figure shows no signs of heteroscedasticity since the spread of residuals does not appear to change depending on the predicted values. If heteroscedasticity is present within a regression the plot of residuals often resembles a cone where the spread increases/decreases depending on the predicted values. Figure A3. Scatterplot of residuals over time for multivariate regression with PRR likeability variable. Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: The figure shows no signs of autocorrelation since the residuals seem to be randomly distributed over the years. Had autocorrelation been present there would have been signs of a trend in the distribution of the residuals. 58 Figure A4. Scatterplot of residuals over time for multivariate regression with PRR ideological left-right placement variable. Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: The figure shows no signs of autocorrelation since the residuals seem to be randomly distributed over the years. Had autocorrelation been present there would have been signs of a trend in the distribution of the residuals. Table A2. Correlation Matrix 1 2 3 4 5 6 7 8 9 1. Left-right 1,0000 2. Likeability -0,1134 1,000 3. % women -0,0430 0,0077 1,000 MPs 4. Female party 0,1537 -0,1151 0,0857 1,000 leader 5. Party size -0,1489 -0,1128 0,1114 0,2425 1,000 6. Government 0,0015 0,11755 0,0326 0,0064 -0,3712 1,000 cabinet 7. Rile score 0,0830 -0,0055 -0,2853 0,1100 -0,0747 0,1333 1,000 8. Respondent -0,0828 0,3813 0,0538 -0,0383 -0,0435 0,0881 -0,0058 1,000 left-right 9. Gender -0,0264 -0,0594 0,0109 -0,0199 -0,0237 0,0120 0,0125 -0,0546 1,000 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: 1,000 implies perfect correlation. The correlation matrix shows no signs of multicollinearity. 59 Figure A5. Change in the percentage of PRR party female representatives. Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: Each dot represents an election with a percentage for the number of PRR female representatives. Figure A6. Change in the number of female PRR party leaders. Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: Each of the dots with a value of 0 represents an election where a woman has not been elected party leader, and each of the dots with a value of 1 represents an election where a woman has been elected party leader. 60 Table A3. Variable description. Observations Mean Std. dev. Min Max Left-right placement 80 063 7,82 2,61 0 10 Likeability 81 192 3,35 3,22 0 10 % PRR party women MPs 88 881 20,54 10,93 0,16 67,00 Female party leader 94 364 0,17 0,37 0 1 Party size 94 364 2,89 1,54 1 7 Government cabinet 94 364 0,28 0,45 0 1 Rile score 94 364 15,17 15,69 -13,81 48,66 Gender 94 364 0,52 0,49 0 1 Party choice 67 863 0,11 0,31 0 1 Year 94 364 2010,81 7,07 1996 2001 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Table A4. Descriptive statistics for the Austrian Freedom Party. Observations Mean Std. dev. Min Max Left-right placement 3 038 7,86 2,39 0 10 Likeability 3 293 3,31 3,07 0 10 % PRR party women MPs 3 368 20,94 4,58 15,00 25,80 Female party leader 3 368 0 0 0 0 Party size 3 368 0 0 0 0 Government cabinet 3 368 0,36 0,48 0 1 Rile score 3 368 6,48 5,78 -1,29 12,17 Year 3 368 2012,69 3,78 2008 2017 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Austria. Table A5. Descriptive statistics for the Swiss People's Party. Observations Mean Std. dev. Min Max Left-right placement 13 624 8,34 2,35 0 10 Likeability 6 938 4,21 3,58 0 10 % PRR party women MPs 15 666 14,37 6,97 5,45 25,50 Female party leader 15 666 0 0 0 0 Party size 15 666 1 0 1 1 Government cabinet 15 666 0,72 0,45 0 1 Rile score 15 666 28,32 17,87 0,83 43,11 Year 15 666 2010,27 6,80 1999 2019 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Switzerland. 61 Table A6. Descriptive statistics for the Alternative for Germany Party. Observations Mean Std. dev. Min Max Left-right placement 4 738 8,87 2,24 0 10 Likeability 5 037 1,55 2,73 0 10 % PRR party women MPs 5 153 11,32 0,27 10,98 11,54 Female party leader 5 153 1 0 1 1 Party size 5 153 4,21 0,98 3 5 Government cabinet 5 153 0 0 0 0 Rile score 5 153 22,65 4,21 17,43 26,05 Year 5 153 2019,42 1,96 2017 2021 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party for Germany. Table A7. Descriptive statistics for the Danish People's Party. Observations Mean Std. dev. Min Max Left-right placement 6 149 8,09 2,39 0 10 Likeability 6 596 3,02 3,0 0 10 % PRR party women MPs 6 851 32,89 7,70 20,00 40,91 Female party leader 6 851 0,79 0,40 0 1 Party size 6 851 3,58 0,91 3 5 Government cabinet 6 851 0 0 0 0 Rile score 6 851 23,01 11,32 9,72 35,00 Year 6 851 2005,04 7,73 1998 2019 Source: Comparative Study of Electoral Systems (2024) , Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Denmark. Table A8. Descriptive statistics for the True Finns Party. Observations Mean Std. dev. Min Max Left-right placement 5 061 6,07 2,23 0 10 Likeability 5 402 4,29 2,94 0 10 % PRR party women MPs 5 764 27,26 4,16 20,00 31,60 Female party leader 5 764 0 0 0 0 Party size 5 764 3,61 1,86 2 7 Government cabinet 5 764 0,28 0,45 0 1 Rile score 5 764 -6,53 5,14 -13,81 0,47 Year 5 764 2013,43 4,45 2007 2019 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Finland. 62 Table A9. Descriptive statistics for the French National Front. Observations Mean Std. dev. Min Max Left-right placement 3 872 8,42 2,74 0 10 Likeability 5 705 1,99 2,89 0 10 % PRR party women MPs 3 844 39,81 10,69 28,60 50,00 Female party leader 5 844 0,31 0,46 0 1 Party size 5 844 3,03 0,81 2 4 Government cabinet 5 844 0 0 0 0 Rile score 5 844 13,06 11,43 1,67 28,49 Year 5 844 2011,86 4,05 2007 2017 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from France. Table A10. Descriptive statistics for the Greece Golden Dawn Party. Observations Mean Std. dev. Min Max Left-right placement 2 751 9,36 1,91 0 10 Likeability 2 944 1,09 2,31 0 10 % PRR party women MPs 3 052 8,95 3,02 4,70 11,10 Female party leader 3 052 0 0 0 0 Party size 3 052 3,67 0,94 3 5 Government cabinet 3 052 0 0 0 0 Rile score 3 052 14,35 16,93 2,33 38,17 Year 3 052 2013,99 1,42 2012 2015 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Greece. Table A11. Descriptive statistics for the Fidesz Party. Observations Mean Std. dev. Min Max Left-right placement 2 131 8,07 2,21 0 10 Likeability 2 284 4,54 3,78 0 10 % PRR party women MPs 2 407 6,89 1,39 5,50 8,27 Female party leader 2 407 0 0 0 0 Party size 2 407 1,49 0,50 1 2 Government cabinet 2 407 0,50 0,50 0 1 Rile score 2 407 28,73 20,01 8,65 48,66 Year 2 407 2010,03 8,00 2002 2018 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Hungary. 63 Table A12. Descriptive statistics for the National Alliance “All for Latvia”Party. Observations Mean Std. dev. Min Max Left-right placement 2 879 6,72 2,89 0 10 Likeability 3 553 3,89 2,95 0 10 % PRR party women MPs 4 056 34,50 19,93 14,29 67,00 Female party leader 4 056 0 0 0 0 Party size 4 056 4,25 0,43 4 5 Government cabinet 4 056 0,75 0,43 0 1 Rile score 4 056 17,89 18,44 -1,67 41,18 Year 4 056 2013,26 3,10 2010 2018 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Latvia. Table A13. Descriptive statistics for the Party of Freedom Netherlands. Observations Mean Std. dev. Min Max Left-right placement 8 411 7,96 2,68 0 10 Likeability 9 266 2,88 2,75 0 10 % PRR party women MPs 6 415 18,58 7,77 1,11 30,00 Female party leader 9 898 0 0 0 0 Party size 9 898 3,28 1,03 2 5 Government cabinet 9 898 0 0 0 0 Rile score 9 898 19,15 11,59 9,09 38,69 Year 9 898 2014,26 6,14 2006 2021 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Netherlands. Table A14. Descriptive statistics for the Progress Party of Norway. Observations Mean Std. dev. Min Max Left-right placement 10 865 8,39 1,99 0 10 Likeability 11 252 4,01 3,03 0 10 % PRR party women MPs 11 420 18,11 6,32 8,00 26,92 Female party leader 11 420 0,31 0,46 0 1 Party size 11 420 2,49 0,49 2 3 Government cabinet 11 420 0,31 0,46 0 1 Rile score 11 420 14,49 9,05 -1,95 23,96 Year 11 420 2006,56 6,84 1997 2017 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Norway. 64 Table A15. Descriptive statistics for the Law and Justice Party Poland. Observations Mean Std. dev. Min Max Left-right placement 8 223 7,20 2,67 0 10 Likeability 9 068 4,39 3,38 0 10 % PRR party women MPs 9 932 19,79 2,27 17,50 23,83 Female party leader 9 932 0 0 0 0 Party size 9 932 1,92 1,08 1 4 Government cabinet 9 932 0,44 0,49 0 1 Rile score 9 932 7,88 6,65 -1,59 17,47 Year 9 932 2008,62 6,08 2001 2019 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Poland. Table A16. Descriptive statistics for the Slovak National Party. Observations Mean Std. dev. Min Max Left-right placement 1 924 4,65 2,67 0 10 Likeability 2 215 4,01 2,92 0 10 % PRR party women MPs 2 353 15,46 4,44 1,11 20,00 Female party leader 2 353 0 0 0 0 Party size 2 353 5,02 0,99 4 6 Government cabinet 2 353 0 0 0 0 Rile score 2 353 13,31 0,36 12,95 13,66 Year 2 353 2012,93 2,99 2010 2016 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Slovakia. Table A17. Descriptive statistics for the Slovenian National Party. Observations Mean Std. dev. Min Max Left-right placement 2 313 5,16 2,76 0 10 Likeability 3 266 3,36 2,78 0 10 % PRR party women MPs 4 088 21,08 4,39 14,29 25,00 Female party leader 4 088 0 0 0 0 Party size 4 088 6,24 0,84 5 7 Government cabinet 4 088 0 0 0 0 Rile score 4 088 10,23 4,20 6,25 16,22 Year 4 088 2001,06 5,22 1996 2008 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Slovenia. 65 Table A18. Descriptive statistics for the Sweden Democrats. Observations Mean Std. dev. Min Max Left-right placement 4 084 8,17 2,41 0 10 Likeability 4 373 2,39 3,31 0 10 % PRR party women MPs 4 512 23,68 11,18 0,16 29,00 Female party leader 4 512 0 0 0 0 Party size 4 512 3 0 3 3 Government cabinet 4 512 0 0 0 0 Rile score 4 512 -8,70 0,82 -9,09 -6,98 Year 4 512 2017,26 1,55 2014 2018 Source: Comparative Study of Electoral Systems (2024), Comparative Manifesto Project (2024) & Weeks et.al (2022). Comment: PRR party from Sweden. 66