• 1 Introduction

  Some women come close. In the Dutch parliamentary elections of 2021, Sigrid Kaag managed to get a record number of votes for D66, but her party trailed the VVD led by Mark Rutte. Some women do make it all the way to the top; Angela Merkel, the German federal chancellor for over a decade, and Jacinda Ardern, president of New Zealand, are probably the most well-known. Overall, however, women are underrepresented in Western liberal democracies: at the beginning of 2021, 26% of all parliamentarians elected at the national level were female (Inter-Parliamentary Union, 2021b).
  Scholars refer to various factors to explain this underrepresentation. Some look at the institutional setting, of which the electoral context is crucial: an election can be considered a marketplace with parties supplying candidates and voters picking their preferred candidate(s). Here we focus on the demand side: do voters perceive female and male politicians differently? Specifically, are women considered less fit for office because of a gender bias? We test whether citizens hold gender stereotypes and their impact on politicians’ evaluations, and do so via an experimental survey design with vignettes. Our findings show that male and female candidates are not perceived differently. Although gender stereotypes exist, citizens do not seem to use these stereotypes in evaluating candidates.
  This article downplays the ‘public wisdom’ or common explanation that female candidates are negatively evaluated by voters by showing that gender stereotypes do not explain the evaluation of political candidates. Alternatively, we suggest that such stereotypes play a role mainly within the context of political parties. Another possible explanation is that citizens are ‘impressed’ by female politicians because they assume that they must be extremely qualified since women have to overcome extra barriers to become active in men-dominated politics.
  We study gender bias in a party-centred parliamentary system. Most similar research is conducted in the United States, that is, a candidate-centred system with weak parties where individual candidate characteristics are highly relevant, with findings that cannot simply be generalised to the European context with party-centred systems. It is thus essential to study the impact of gender on candidate evaluations outside the United States. Furthermore, a meta-analysis shows that for European proportional systems the effect of gender on voting is inconsistent: both in Norway and Switzerland there were significant positive effects of being a woman, but in Denmark and Romania there was no effect (Schwarz & Coppock, 2019).
  Our study focuses on the Dutch case. Studying such a least-likely case is important: if women are underrepresented in countries where we would expect it least, it is even harder to fight underrepresentation in countries where we would expect it. More generally, it is crucial to understand contextual mechanisms at play in a least-likely case; studying other countries than the ‘usual suspect’ is relevant since contextual variables impact the effect of gender stereotypes. Exposure theory (Jennings, 2006), for example, poses that when voters become more familiar with female politicians, they are more likely to develop a gender-neutral attitude towards them. Taylor-Robinson et al. (2016) indeed shows this effect: in Costa Rica, a country with experience with women in government, voters have a relatively neutral attitude towards female politicians, while in Israel, with less experience with women in politics, female candidates are evaluated less positively than male candidates. For Europe, findings are mixed. Matland (1994) shows that the presence of women in Norwegian politics is not associated with gender-neutral evaluations. However, for Denmark (Dahl & Nyrup, 2021) and Belgium (Devroe & Wauters, 2018), where historically women have been present in politics, there is no difference between the evaluation of male and female politicians. Our study thus contributes to these nuanced findings by including the Dutch case.
  This article starts with a short overview of why it is important to study the representation of women in politics, followed by a theoretical framework on why gender stereotypes disadvantage women in politics. However, our findings show differently: the evaluation of women is relatively positive, and gender stereotypes do not explain these findings. In the discussion and conclusion, we suggest alternative explanations for our main finding by pointing at different stereotypes and the understudied role of a crucial actor in party-centred systems: political parties.

• 2 Why Representation of Women Is Important

  Various forms of representation exist, as authoritatively argued by Pitkin (1967). The concept of standing for is related to descriptive representation and refers to the resemblance of voters/citizens and their representatives. Parliament as a representative body should mirror society (Pitkin, 1967). This form of descriptive representation is essentially about who the representative is; arguments in favour may be based on ideas of justice, fairness, and equity (e.g. Celis & Meier, 2006; Phillips, 1994). This reasoning entails that when half of the population consists of women, it is logical and from a normative perspective imperative that women are present in politics in proportional numbers.
  Descriptive representation shows that (sub)groups of citizens are equal and equally present in the main democratic body, signalling that all groups have equal opportunities to govern (Mansbridge, 1999). Consequently, descriptive representation counters the conventional ‘wisdom’ and often implicit but dominant norm that politics is for (white, middle-aged) men; if every second Member of Parliament (MP) is a woman, politics obviously is not an exclusively male affair. Moreover, descriptive representation impacts positively on the legitimacy of policies and politics since multiple societal (sub)groups are involved in deliberation and participate in law- and policy-making processes (Celis & Meier, 2006; Mansbridge, 1999).
  It can be argued – and empirically substantiated – that descriptive representation is related to and enables substantive representation, essentially meaning that politicians act for the (sub)group of citizens they represent: acting for (Pitkin, 1972). The basic idea is that with women present in politics, they are likely to act in favour of women and women’s issues (Celis & Childs, 2008). In electoral democracies dominated by political parties (at least in Europe), representatives are elected on a party programme. However, when policy is considered that is not (clearly) identified in this programme the position and judgement of the individual MP is very important. In such circumstances, background and identity characteristics (e.g. gender) are relevant or even crucial (Blais, 2011; Celis & Meier, 2006; Mansbridge, 1999). The underlying logic or mechanism is that identity characteristics reflect particular life experiences (Celis & Childs, 2014; Mansbridge, 1999). Arguably, the life experience of women is not universal (Celis & Childs, 2008; Celis & Meier, 2006), but women as a group do share a common core identity and have essentially different interests, needs and life experiences compared to men. These can be gendered (Celis & Meier, 2006), and because of such differing interests, descriptive representation of women is important on its own and also due to its impact on substantive representation. Descriptive representation allows for different, gendered interests to be present in politics: the Politics of Presence (Phillips, 1995).

• 3 Evaluating Politicians and Gender Stereotypes

  To explain the political underrepresentation of women, the four-stage recruitment model (Norris & Lovenduski, 1993) is often used. This model is a chain containing four links: from being a citizen to being an eligible citizen (phase 1); being an eligible citizen aspiring to be a politician (phase 2); being a candidate for office (phase 3); and, ultimately, becoming elected (phase 4). For this last phase voters are crucial: they decide who will be elected. Consequently, it is important to know how voters perceive candidates.
  Voters form impressions of candidates by inferring character traits (e.g. McGraw, 2003); such traits are cues for future political behaviour (Blais, 2011; McGraw, 2003). In a seminal study by Kinder et al. (1980), respondents had to describe an ideal president and based on the traits they mentioned, two clusters were compiled: competence and trust traits. Kinder (1983) specified the two ‘dimensions’ into four sub-dimensions: competence, leadership, integrity and empathy. The competence (sub)dimension entails traits such as technical skills to lead; leadership is about being an ‘heroic’ leader; integrity concerns being ethical or honest; and finally, the empathy subdimension is about whether a politician is understanding and compassionate (Kinder, 1983). Forty years later and based on a comprehensive meta-analysis of politically relevant character traits, Bittner concluded that these traits could again be ordered into two categories: integrity/character (integrity and empathy) and competence (competence and leadership). Traits belonging to the integrity/character category are for example being warm, honest and fair; intelligence, effectiveness and respectability belong to the competence category (Bittner, 2011).
  To infer a candidate’s character traits, citizens use heuristics. One commonly used set of heuristics are stereotypes (e.g. Lau & Redlawsk, 2006). According to Lippmann, who introduced the concept of a stereotype in relation to human groups (Blum, 2004), a stereotype is a ‘picture’ that comes to mind when thinking about a particular social group (Lippmann, 1922). Classifying an individual into a particular group facilitates drawing inferences about this individual and his or her intentions (Fiske et al., 2002; Kahn, 1996; Lippmann, 1922); stereotypes are cognitive schemas for processing information and simplifying reality (Hilton & Von Hippel, 1996; Kahn, 1996). Stereotyping is common when one needs to draw conclusions based on imperfect information, which during elections is often the case (Kahn, 1996; Lau & Redlawsk, 2001; Sanbonmatsu, 2002).
  “The most common political application found in research on (…) stereotype in person perception concerns response to male and female candidates and political leaders” (Sapiro, 2003, pp. 616-617). Together with physical appearance and race – and in an American context: partisanship – gender is the main category at the basis of stereotypical inferences (McGraw, 2003). When voters see a candidate, their first impression likely is related to the candidate’s sex or gender (Kahn, 1996) and this may activate a gender stereotype (Dolan 2014). Consequently, voters infer that an individual has certain character traits based on gender and connected gender roles, that is beliefs and ideas about personality traits of men and women and qualities and behaviour based on their socially identified gender (Eagly, 1987). These beliefs entail two main types of expectations or norms (Eagly & Karau, 2002): descriptive and injunctive norms. Descriptive norms are expectations about what men and women actually do; injunctive norms are expectations about what a group ought to do. Descriptive expectations are similar to stereotypes, but a gender role contains both descriptive and injunctive expectations.
  The fact that traditionally men have been in charge, inside and outside politics (Glick & Rudman, 2010), has established a conviction about appropriate behaviour connected to gender roles. Men are supposed to be leaders; therefore, men are considered competent and independent, that is agentic. Women are perceived as warm, expressive and supportive, that is communal. When specific individual behaviour does not match these social roles, this results in role incongruity. One of the consequences of role incongruity is that people are evaluated negatively. For example, a career-oriented woman is likely to be perceived as task competent – but not warm. This is contrary to the common idea that women are warm – but less competent (Fiske, 1998; Fiske et al., 2002; Okimoto & Brescoll, 2010).
  This ‘mismatch’ between observed and expected behaviour may occur when a politician is female, because men are supposed to be in charge (Eagly & Karau, 2002; Koenig et al., 2011). Political leaders are considered to have agentic traits and women do not fit: women are supposed to be communal and less agentic (Eagly & Karau, 2002). Because of such conflicting expectations – being a politician violates the female gender role – we expect:

  H1: Female politicians are evaluated less positively than male politicians.

  As argued, a ‘good’ politician should have different traits. Citizens ascribe both competence and integrity traits to their ideal politician. The integrity dimension entails traits such as being trustworthy and empathic, and the female gender role is not necessarily in violation with these traits; the female gender role implies being warm and supportive and this is similar to the integrity/character traits that are important for a politician. For the male gender role, on the other hand, it is expected that men do not, or to a lesser extent, embody these traits. Therefore: it is argued that female politicians will be evaluated more positively than male politicians on integrity traits:

  H2a: Female politicians are evaluated more positively than male politicians on integrity traits.

  On the other hand, female candidates fulfil a political function, which is out of line with being a ‘real’ woman according to conventional gender beliefs. Consequently, they may be evaluated relatively low, as role incongruity theory suggests (Eagly & Karau, 2002). Since the expectations concerning the evaluations of female politicians on the character dimension are ambiguous, we test the following competing hypothesis:

  H2b: Female politicians are evaluated more negatively than male politicians on integrity traits.

  Regarding the evaluation of the competence traits, the expectations are more straightforward. The competence traits fit the agentic dimension of the male gender role: being confident and competent. Women do not, or to a lesser extent, have these traits according to their gender role. Consequently, female politicians are expected to be perceived as less agentic/competent.

  H2c: Female politicians are evaluated less positively than male politicians on competency traits.

  In her overview of the literature on gender stereotypes in a political context, Sapiro (2003, p. 617) concludes that “[t]hese studies vary the stimulus, circumstances, or questions, but their conclusions converge: the public uses common gender stereotypes to fill in information about candidates, especially in low-information elections”. For example, female candidates are perceived as relatively compassionate and honest and men as stronger leaders and better crisis managers (Huddy & Terkildsen, 1993; Kahn, 1996). Moreover, gender stereotypes impact people’s willingness to support female candidates (Dolan, 2010). Consequently, the underlying gender stereotypes are relevant to explain the underrepresentation of women in politics; these gender stereotypes may influence why a voter rejects or prefers a female candidate.
  Stereotypes could impact candidate evaluations and indirectly also impact party or candidate choice (Dolan, 2014). Obviously, party choice is structured by numerous factors next to gender, making it difficult to isolate and estimate the separate effect of gender stereotypes (e.g. Kahn, 1996). However, it has been demonstrated that party cues do not rule out gender effects (e.g. Sanbonmatsu & Dolan, 2009). Fox and Smith explored the electoral effect of gender and observed that male candidates were consistently favoured over female candidates. They even found that when a man ran against a woman, voters liked the man better than when that same man ran against another male candidate (Fox & Smith, 1998)! Sanbonmatsu showed that voters have a gender baseline with respect to voting, that is a predisposition to support female over male candidates (or the other way around). She found that a preference for male candidates was explained by negative stereotypes about female traits and positive stereotypes about a man’s ability to handle ‘male’ policy issues. A similar preference for female candidates could be explained by positive stereotypes about women’s position and ability to deal with ‘female’ issues (Sanbonmatsu, 2002). Valdini (2013) showed that in countries with the opportunity to cast personal votes, gender bias, especially in conservative countries, has a negative effect for women. Other scholars show that women, to do equally well in elections as men, should perform better to achieve equal results (Crowder-Meyer et al., 2015; Fulton, 2012), indicating that female candidates must ‘compensate for being a woman’. Overall, gender bias thus matters in the electoral context.
  Despite various studies demonstrating that gender bias matters, there are mixed findings about how exactly stereotypes matter. Dolan (2014) hypothesised that gender stereotypes impact electoral choice either by directly influencing this choice or because gender stereotypes influence candidate evaluations and these evaluations impact on electoral choice, but she did not find the expected effects. Teele et al. (2018) also find no ‘outright discrimination’ against female politicians; in fact, the respondents in their study show a preference for female candidates.
  All in all, there are mixed and puzzling findings concerning gender stereotypes and their electoral effects (Dolan, 2010; Fox & Smith, 1998; Hayes, 2011; Huddy & Terkildsen, 1993; Kahn, 1996; Sanbonmatsu, 2002; Sapiro, 1981/1982). In such a situation another empirical test is in order; to start with, we need to test what the direct effect is of having a gender stereotype. If voters have gender stereotypes, does this influence candidate evaluation?¹x Stereotypes can have both negative and positive dimensions. We focus on the positive dimension, since this dimension entails the agentic and communal traits (Schneider & Bos, 2014) which are essential in role incongruity theory. Negative stereotypes are for example being emotionally weak (for women) and greedy (for men) (Schneider & Bos, 2014), but these traits are not typical agentic and communal traits. We expect the following:

  H3: Respondents who hold gender stereotypes evaluate female politicians less positively on character traits than respondents who do not have such a stereotype.

• 4 Case Selection, Data and Method

  4.1 The Dutch Case

  We study gender stereotypes and their impact in the Netherlands, an advanced Western liberal democracy. Still, over a century after the introduction of female suffrage (1917/1919), the number of female MPs has never equalled the number of male MPs. After the 2021 elections for the Second Chamber, 59 members (39%) of all 150 MPs are women. In the Senate and at the regional and local level the picture is similar, with respectively 39%, 34% and 33% female politicians (Joop, 2019; NOS, 2018). Theoretically, the underrepresentation of women in Dutch politics could be due to a lack of supply. It could be that there are simply too few women opting for a political career and on candidate lists, because, for example, a lack of political ambition (e.g. Fox & Lawless, 2014). However, this explanation seems not applicable. In the 2021 elections, for example, 439 female politicians were on the candidate lists of all the parties then participating (Inter-Parliamentary Union, 2021b). The Second Chamber has 150 seats, so theoretically, voters could have filled the Dutch parliament with female MPs twice.²x For a more elaborate argument on the use of raw numbers instead of proportionality tests to study the supply side, see Ashe & Stewart (2012, p. 691).
  The puzzle of female underrepresentation is even more intriguing if we consider key features of the electoral system: the Dutch system should be advantageous for female politicians (Leijenaar, 2004). There is consensus on the positive effects of a proportional electoral system on female representation (e.g. Diaz, 2005; Paxton, 1997; Rule, 1987); the Dutch system is qualified as extremely proportional (e.g. Lijphart, 1999). Another factor is district magnitude. In multimember districts female candidates do not threaten or ‘push away’ male candidates from the list (Matland, 2005; Matland & Brown, 1992); the Netherlands with its nationwide 150-member district should be advantageous for female candidates. Finally, the fact that voters may cast a preference vote on an open list is beneficial (Golder et al., 2017), although Valdini (2013) argues that this electoral option may be disadvantageous for female candidates in a conservative cultural context. As regards the latter, the Netherlands is according to the Hofstede index a feminine society (Hofstede & Soeters, 2002) wherein “emotional gender roles overlap” (Hofstede et al., 2005, p. 140); this cultural feature should benefit female representation. Also, there has been a steady increase in the number of preference votes with almost 30% of Dutch voters casting a preference vote in recent national elections; many of these votes are cast for female candidates (Andeweg & Van Holsteyn, 2011; Van Holsteyn & Andeweg, 2012; Nagtzaam, 2019). This clearly signals that even though the Netherlands may still be a party-centred system, individual candidates are increasingly relevant as an element of the electoral calculus.
  Based on relevant institutional and cultural features of the Netherlands, one would expect equal representation. As already mentioned, this is obviously not the case – so how come that a country that is so likely to show equal representation, has never achieved this? Studying such least-likely or atypical cases deviating from theoretical expectations (Levy, 2008) may offer comparative insights as well. If women are underrepresented in countries such as the Netherlands, where we would not expect women underrepresentation, it might be even harder to understand and fight underrepresentation in countries where we would expect it.

  4.2 Data and Method

  We conducted a survey experiment with a post-test only design. The questionnaire consisted of question modules on (1) general political attitudes; (2) the evaluation of a political candidate; (3) the perception of societal attitudes with respect to character traits of men and women; and (4) women in Dutch politics.³x A copy of the questionnaire (in Dutch) can be obtained on request from the corresponding author. Key were vignette questions referring to a fictitious candidate for Second Chamber elections. We manipulated the gender of the candidate; following the logic of the experimental design (cf. Auspurg & Hinz, 2015) respondents were randomly assigned to a single scenario of a candidate running for office being:

  A1: male politician A who was presented with a short text only

  A2: female politician A who was presented with a short text only, identical to the text for A1, except for his/her name⁴x Since particular names may elicit (negative) responses (Newman et al., 2018), we have chosen names that are ‘common’ and easy to alter in an equivalent male or female version (Karel and Carolien). We also made sure that the names did not match the names of well-known politicians. The surnames are also common, belonging to the 50 most used surnames in the Netherlands (Nederlandse Familienamenbank, 2007). At the time of the experiment, there were no national politicians with these surnames. (see Appendix for the vignette text)

  The vignette was followed by a randomised block of statements on competency and integrity traits, asking respondents to rate on a 7-point scale from fully agree to fully disagree whether the candidate was intelligent, knowledgeable, lazy, inspiring, effective, commanding respect, compassionate, connected, dishonest and decent.
  The third block contained questions on gender stereotypes based on the Stereotype Content Model (e.g. Fiske et al., 2002). To control for social desirability effects the questions did not ask for the respondents’ personal opinion, but for the perception of the general attitude of or public opinion in society (see for a similar approach e.g. Fiske et al., 2002; Kerevel & Atkeson, 2015; Schneider & Bos, 2014). The assumption is that these “perceptions” are (predominantly or at least partly) the result of the psychological process of projection, that is “seeing one’s own traits in other people” (Baumeister et al., 1998, p. 1091). Respondents were asked to rate men and women on a 5-point scale⁵x The exact wording (translated from Dutch by the authors) is: ‘The following statements are about what society feels about men and women. It is not about your personal opinion, but how you think Dutch society feels. How confident/competent/warm/sincere are men/women perceived?’ Every trait was mentioned separately for men and women. There were 8 statements in total. on four items representing the dimensions of gender stereotypes competence and warmth: self-confidence, competence, warmth and sincerity. We qualified respondents with extreme perceptions of societal stereotypes as citizens having these stereotypes themselves, that is treating perceptions as projections and consequently as indications of underlying personal attitudes. Respondents that are in the highest quartile of agentic perceptions of men, that is a combination of the self-confidence and competence scale, and respondents in the lowest quartile of communal perceptions of men, that is a combination of the warmth and sincerity scale, are classified as having strong men stereotypes. The typical women stereotype is its mirror image: the highest quartile on the communal dimension and the lowest quartile in the agentic dimension.⁶x Whether respondents have gender stereotypes was measured after the stimulus was presented. We have chosen to do so, since we considered it a risk to ask respondents about gender stereotypes first and subsequently ask them to evaluate a candidate. Therefore, we chose to do it in a reverse order, taking the risk that our stimulus has influenced the measure of gender stereotypes. However, we compared the outcomes of the Stereotype Content Model between experimental groups and no substantive differences were found. Therefore, we are confident that there is no post-treatment bias. A table with these checks is presented in the appendix (Table A1).
  A risk to our approach is that respondents who indicate that there are stereotypes in society do not hold these stereotypes themselves: they could simply observe the stereotypes in society. To check whether this may have been the case, we attempted to validate this approach. In our survey we included statements about women and politics. When respondents whom we classified as having gender stereotypes in fact do not have such stereotypes, they arguably are more ‘progressive’ in responding to these statements. However, there is no difference regarding the statement whether women are less interested in politics than men: 40% of the respondents with stereotypes about women agreed, 36% of the respondents with stereotypes about men agreed and 37% of the respondents without stereotypes agreed. There are also no differences with regard to the statement that women do not try hard enough to be elected. On the other hand, there is some difference about the statement that women have less opportunities to get into politics than men: 67% of the respondents with stereotypes about men agreed, 62% of the respondents about women agreed and 49% of the respondents without stereotypes agreed to this statement. So overall the findings of our validation may be mixed but do not reveal a pattern that convincingly contradicts our assumption. In conclusion, our approach leaves room for improvement but is not unprecedented nor obviously inadequate. Consequently, we use these projection data to test hypothesis 3.
  The data were collected via the online survey facility EenVandaag Opiniepanel (EVO).⁷x The second author was involved in the development of the EVO and has been a scientific adviser to this panel. This EVO consists of respondents who have signed up for participation. During the data collection (13-18 May 2016) about 50,000 people were registered; 19,384 respondents participated in our study. Most respondents are men (70%), while men are a minority of the voter population (49%). Regarding age, 4% of the respondents is between 18 and 35 years of age; 46% is between 35 and 65 years of age and the plurality of the respondents (50%) is 65 or older; older voters are overrepresented, since 22% of the Dutch voter population is 65 or older. With regard to education, 55% of our population is highly educated, while 25% of the voters are.⁸x Data about the voter population are from 2012 and are retrieved from Statistics Netherlands.
  The non-probability sampling of the panel means that the sample is not truly representative of the population and that external validity or generalisation may be problematic. However, such estimation of population parameters is not our main goal: our study primarily addresses the impact of different political candidates in a between-subjects experimental design, in which not the random selection but the random assignment of respondents to experimental and control groups is key. Moreover, it can be argued that “[w]e should not exaggerate the self-selection problem”: the seriousness of the problem depends on scholarly ambition, but also in various respects the EVO data are similar to actual, real outcomes. Moreover, the large size of the panel guarantees variation on relevant variables⁹x We have run our analysis on a weighted dataset, but this did not substantially change our results: female politicians were still evaluated more positively than male politicians. (Van Holsteyn, 2015; van Elsas et al., 2014). All in all, we concur with the American Association for Public Opinion Research (AAPOR) recommendation that there ‘are times when a nonprobability online panel is an appropriate choice’, in particular if researchers do not primarily aim for estimates of population values (Baker et al., 2010, p. 714). This article contains descriptive data, but its focus is on the causal impact of gender on the perception and evaluation of candidates at the mass level and individual characteristics that interrelate with the impact of gender. This scholarly and analytical aim, with internal validity trumping external validity, is perfectly feasible based on the EVO panel data.

• 5 Results – The Perception and Evaluation of Female (and Male) Politicians

  We expect male candidates to be evaluated more positively than female candidates. To test this expectation, a t-test is conducted, resulting in a consistent pattern: our female candidate is slightly more positively evaluated than her male counterpart (Table 1).¹⁰x The vignette used in the analysis is about a politician with an expertise in economics. However, the specific policy area might be gendered: economics can be seen as a ‘male’ policy area (Kerevel & Atkeson, 2015; Huddy & Terkildsen, 1993). To check for the impact of policy area, we included another vignette about a politician with an expertise in health care in the questionnaire and analyses. The results for both policy areas are similar. All differences are statistically significant at conventional levels, but the differences are small from a substantive perspective. Still, the data show a consistent pattern in the evaluation of candidates for political office: female candidates are considered to be better fit for office. Consequently, we reject our hypothesis that female politicians are evaluated less positively compared to male politicians.

  Table 1 Mean scores traits evaluations and t-test results¹¹x The politician presented in the vignette is married and has children. Some research suggests that being married and/or having children can be advantageous for a politician’s evaluation (e.g., Bell & Kaufman, 2015; Campbell & Cowley, 2018). In another vignette we did not include any information about parental or marital status, and in this vignette the evaluation pattern persisted: female politicians were rated relatively high (see the appendix for results).^,12x See note 9.

  	Gender
  	Male politician (n=9,516)		Female politician (n=9,628)
  Intelligence	4.70 (1.50)	<	4.93*** (1.46)
  Knows a lot	3.95 (1.56)	<	4.07*** (1.55)
  Hardworking	5.15 (1.68)	<	5.50*** (1.59)
  Inspiring	3.35 (1.81)	<	3.46*** (1.82)
  Effective	3.79 (1.52)	=	3.83 (1.54)
  Commands respect	3.83 (1.66)	<	4.04*** (1.66)
  Compassionate	4.34 (1.60)	<	4.59*** (1.56)
  Connected	4.19 (1.63)	<	4.40*** (1.57)
  Honest	5.12 (1.67)	<	5.39*** (1.63)
  Decent	4.89 (1.45)	<	5.12*** (1.41)

  Note: * p < 0.10, ** p < 0.05, *** p < 0.001, standard deviation in brackets
  Results for vignette 1

  Overall, there is also no pattern for the male candidate to be rated higher on competence traits¹³x In the result section, we discuss the two categories of character and competence traits. However, a principal component analysis shows that the ten traits do not load on the two categories (cf. Bittner, 2011). Instead, the eight positive phrased traits form one factor and two negatively phrased traits the second; see the appendix (Table A2) with factor loadings. Because of this result we did not create two variables of character and competence traits but proceed the analyses with all separate traits. compared to the female candidate. This finding contradicts our hypothesis that female politicians will be evaluated less positively than male politicians on precisely these traits. If anything, the opposite pattern is empirically supported: regarding politically relevant character traits, women are slightly more positively evaluated than their male counterparts.
  Although the differences between the evaluation of the female and male candidate are contrary to what was expected, we need to further analyse the role of gender stereotypes. It may be that Dutch citizens simply do not have such stereotypes and that the absence of active gender stereotypes explains the relative positive evaluation of female candidates.
  To gauge the existence of gender stereotypes, questions from the Stereotype Content Model were used. Our data show that stereotypes are present in Dutch society, at least according to our respondents: women are rated as being warmer and more sincere and less competent and self-confident compared to men (see Table 2). These differences between men and women are statistically significant (p < 0.001). If we combine the two characteristics for the agentic dimension, men score on average 7.8 and women 6.7; a similar score for the communal facet results in a mean of 5.7 for men and 7.3 for women (see Table 2).

  Table 2 Mean scores gender Stereotype Content Model

  	Gender
  	Men		Women
  Confidence	4.00	>	3.34
  Competent	3.81	>	3.41
  Agentic	7.8	>	6.7
  Warm	2.80	<	3.81
  Sincere	2.95	<	3.47
  Communal	5.7	<	7.3

  Assuming an underlying active psychological projection process, we qualified respondents with extreme perceptions of societal stereotypes as citizens having gender stereotypes. We classified respondents as having ‘extreme’ perceptions about men when they were in the highest quartile of agentic perceptions about men and respondents in the lowest quartile of communal perceptions about men. The stereotype about women mirrors this approach, resulting (in our sample) in 1,454 respondents (8%) identified with a male and 3,232 (18%) with a woman stereotype. This is not an exclusive male or female phenomenon: 40% of the respondents with this perception were female and 60% male. Most respondents with a woman stereotype were male (63%), completed higher education (64%) and were between 55 and 75 years old (67%). There is minor overlap (3%) between the groups of respondents who think that both stereotypes about both men and women are present.
  The third hypothesis stated that gender stereotypes impact evaluations: respondents with gender stereotypes evaluate female candidates relatively negative. A distinction is made between character and competence traits. With respect to the latter, women are expected to be evaluated even more negatively, since these traits are not in line with the traditional woman’s gender role. To assess whether gender stereotypes about men and women affect the evaluation of female and male politicians, a multiple regression analysis was conducted. Table 3 contains the results for five models with the dependent variables being the five different ‘competence’ traits. In this test we controlled for the ‘usual suspects’ political knowledge, age, left-right self-placement – not partisanship or party identification, since this ‘American’ concept has been proven not to travel well to the Netherlands; (see Thomassen, 1976; Thomassen & Rosema, 2009) – and the respondent’s gender. The main independent variables of interest are whether people hold stereotypes and whether they employ these stereotypes to evaluate female and male politicians. These are interaction effects.

  Table 3 Multiple regression analysis explaining evaluation on five ‘competence’ traits¹⁴x See the appendix for the results of the analysis on the data of the second vignette in Table A3.

  	Intelligence	Knowledgeable	Hard working	Inspiring	Effective
  (Intercept)	3.26***	2.77***	4.47***	1.70***	2.67***
  	(0.19)	(0.20)	(0.21)	(0.23)	(0.20)
  Vignette type (1=female)	0.20***	0.05	0.29***	0.04	−0.02
  	(0.03)	(0.03)	(0.03)	(0.04)	(0.03)
  Gender respondent (1=female)	0.12**	0.08*	0.16***	0.00	0.15***
  	(0.04)	(0.04)	(0.04)	(0.04)	(0.04)
  Stereotype about men (1=present)	0.08	−0.09	0.02	−0.15*	−0.14*
  	(0.06)	(0.07)	(0.07)	(0.07)	(0.07)
  Stereotype about women (1=present)	0.04	−0.02	0.00	−0.07	−0.01
  	(0.04)	(0.05)	(0.05)	(0.05)	(0.05)
  Age	0.01***	0.01***	0.01***	0.02***	0.01***
  	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
  Left-right self-placement	−0.01	0.01	−0.03***	0.01	−0.00
  	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)
  Political knowledge	0.13***	0.08***	0.08***	0.06**	0.11***
  	(0.02)	(0.02)	(0.02)	(0.02)	(0.02)
  Education level 2 (ref = 1)	0.63***	0.46*	0.18	0.62**	0.46*
  	(0.17)	(0.18)	(0.19)	(0.20)	(0.18)
  Education level 3 (ref = 1)	0.67***	0.38*	0.36	0.49*	0.46*
  	(0.17)	(0.18)	(0.19)	(0.21)	(0.18)
  Education level 4 (ref = 1)	0.59***	0.37*	0.19	0.54**	0.39*
  	(0.17)	(0.18)	(0.19)	(0.20)	(0.18)
  Education level 5 (ref = 1)	0.52**	0.08	0.25	0.19	0.24
  	(0.17)	(0.18)	(0.19)	(0.20)	(0.18)
  Vignette female × respondent female	0.06	0.18**	0.08	0.21***	0.20***
  	(0.05)	(0.06)	(0.06)	(0.06)	(0.06)
  Vignette female × stereotype about men	0.11	0.11	0.14	0.04	−0.02
  	(0.09)	(0.10)	(0.10)	(0.10)	(0.10)
  Vignette female × stereotype about women	0.05	0.09	0.12	0.15*	0.05
  	(0.06)	(0.07)	(0.07)	(0.07)	(0.07)
  R2	0.02	0.03	0.02	0.03	0.02
  Adj. R2	0.02	0.03	0.02	0.03	0.02
  Num. obs.	15302	14691	14790	16139	14030

  ***p < 0.001; **p < 0.01; *p < 0.05.

  Based on the regression analysis we again conclude that female politicians are more positively evaluated; see the positive effects of ‘Vignette type’. The control variables age and left-right self-placement have no effect, while political knowledge has a positive effect: respondents claiming to have more political knowledge are more positive in their evaluation of politicians.
  Regarding the stereotypes the results are intriguing. When a respondent has a stereotype about women and evaluates a female candidate, this respondent will do so in a relatively positive way; the coefficients for this interaction effect are positive. To facilitate interpretation, the effects for the trait being knowledgeable are presented in Figure 1, showing that respondents with and without a stereotype about women evaluate a male politician in a similar manner. When they are confronted with a female politician, both become more positive. However, respondents with a stereotype about women become even more positive than respondents with no such stereotype. Although the difference is small and except for the trait ‘inspiring’ statistically insignificant, this finding is unexpected: one would expect that respondents holding a stereotype about women, that is thinking that women ought to be communal instead of agentic, would not rate female politicians positively on more agentic competence traits.
  We also controlled for the respondent’s gender, since this is an important variable in explaining how people evaluate politicians (e.g. Sanbonmatsu, 2002). We included gender as a separate variable and as an interaction variable in which the respondent’s gender interacts with the gender of the politician. Our findings show that women in general tend to be slightly more positive than men. The interaction effect shows that women are even more positive when they evaluate a female politician.

  Predicted probabilities for the character trait knowledgeable based on the interaction effect of having a stereotype of women and evaluating the male or female politician.
  

  When we consider traits that are assumed to be related to the character dimension, we again see some conflicting findings (Table 4). The female candidate is a significant positive predictor for a higher evaluation; this concurs with earlier findings, although for the character traits the effects are larger. With respect to the stereotypes about men and women and the interaction with the politician’s gender, the findings are mixed. Having a stereotype of men does not seem to affect the evaluation of female candidates in an unambiguous way; the direction of the effect differs, and its size is small. There is no clear effect of having a stereotype about women and evaluating a female politician. However, for two traits the positive effect is statistically significant. These regression analyses thus warrant the conclusion that stereotypes about men and women are not influencing the evaluation of female and male candidates in a major and consistent way. Moreover, having gender stereotypes (according to our operationalisation, based on the assumption of projection) does not result in giving female candidates a lower evaluation. We reject our third hypothesis.

  Table 4 Multiple regression analysis explaining evaluation on five ‘character’ traits¹⁵x See the appendix for the results of the analysis on the data of the second vignette in Table A4.

  	Commands respect	Compassionate	Connected	Fair	Decent
  (Intercept)	2.15***	2.86***	2.74***	4.17***	3.25***
  	(0.21)	(0.20)	(0.21)	(0.22)	(0.18)
  Vignette type (1=female)	0.17***	0.20***	0.16***	0.21***	0.23***
  	(0.03)	(0.03)	(0.03)	(0.03)	(0.03)
  Gender respondent (1=female)	0.11**	0.09*	0.10*	0.19***	0.16***
  	(0.04)	(0.04)	(0.04)	(0.04)	(0.04)
  Stereotype about men (1=present)	−0.16*	−0.04	0.01	0.07	0.03
  	(0.07)	(0.07)	(0.07)	(0.07)	(0.06)
  Stereotype about women (1=present)	−0.01	−0.02	−0.12**	0.01	0.11*
  	(0.05)	(0.05)	(0.05)	(0.05)	(0.04)
  Age	0.02***	0.01***	0.01***	0.01***	0.01***
  	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
  Left-right self-placement	0.01*	0.03***	0.04***	−0.01*	0.01*
  	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)
  Political knowledge	0.06**	0.10***	0.11***	0.01	0.07***
  	(0.02)	(0.02)	(0.02)	(0.02)	(0.02)
  Education level 2 (ref = 1)	0.44*	0.49**	0.53**	0.06	0.83***
  	(0.19)	(0.18)	(0.18)	(0.20)	(0.16)
  Education level 3 (ref = 1)	0.38*	0.48**	0.52**	0.24	0.82***
  	(0.19)	(0.18)	(0.19)	(0.20)	(0.17)
  Education level 4 (ref = 1)	0.31	0.38*	0.47**	0.01	0.79***
  	(0.19)	(0.18)	(0.18)	(0.20)	(0.16)
  Education level 5 (ref = 1)	0.08	0.22	0.26	0.14	0.76***
  	(0.19)	(0.18)	(0.18)	(0.20)	(0.16)
  Vignette female × respondent female	0.13*	0.11	0.12*	0.08	−0.02
  	(0.06)	(0.06)	(0.06)	(0.06)	(0.05)
  Vignette female × stereotype about men	0.09	0.02	−0.10	0.00	0.10
  	(0.10)	(0.09)	(0.10)	(0.11)	(0.09)
  Vignette female × stereotype about women	0.03	0.09	0.20**	0.19**	−0.01
  	(0.07)	(0.07)	(0.07)	(0.07)	(0.06)
  R2	0.04	0.03	0.03	0.02	0.02
  Adj. R2	0.04	0.03	0.03	0.02	0.02
  Num. obs.	15516	15500	15111	13979	14863

  ***p < 0.001; **p < 0.01; *p < 0.05.

  Finally, when we look at the interaction effect of gender, there is again a positive effect, except for the trait ‘being decent’. It thus seems that in general women are more positive in evaluating politicians, but they are even more positive when this evaluation refers to a female candidate. This concurs with earlier findings (Sanbonmatsu, 2002) and seems to suggest the existence of a gender baseline.
  Overall, our data strongly suggest that gender stereotypes do exist among Dutch citizens, but that these stereotypes do not influence the evaluation of political candidates, at least not to the disadvantage of female candidates – and this we did not expect to find. Based on role incongruity theory we expected that female politicians would be ‘punished’, because a political role and function would not fit with their conventional female role. However, according to our data and analyses, this in fact is not the case. Among Dutch citizens there appears to be no gender difference in the evaluation of political candidates. Women are considered to be as fit for office as men.

• 6 Conclusion and Discussion

  ‘Where is our female Prime Minister?’ (Waar blijft onze vrouwelijke premier?) was the heading of a newspaper article (NRC, 11 May 2019; also De Telegraaf, 7 May 2019) sketching the lasting puzzle that in a supposedly modern Western, highly emancipated, progressive country as the Netherlands the political elite is predominantly male. Women have been underrepresented in most legislatures and the Netherlands is no exception to this rule. The most recent national elections of 2021 resulted in only 39% female MPs.
  We tried to understand this underrepresentation by focussing on the attitudes of Dutch citizens. Do they evaluate male and female political candidates differently, and are female candidates evaluated less positively, maybe as a result of gender stereotypes? In response to these important questions, our study can be considered an optimistic contribution to the debate on women in politics: our findings strongly suggest that at the mass level there are no differences in the evaluation of female and male politicians! Any differences we did find were substantively small. Moreover, such differences were in favour of female candidates: if there is any difference, citizens are more positive towards female compared to male politicians. Also, we contributed to the existing literature by focussing on a European party-centred parliamentary democracy. Most relevant literature mainly focuses on the United States and, although valuable, these results not necessarily transfer well to the European context due to the different political institutions and electoral dynamics.
  Our results are intriguing, as they cannot be explained by the absence of gender stereotypes. Among Dutch citizens gender stereotypes likely exist. However, these stereotypes do not seem to influence the evaluation of politicians – and this is not to be expected based on the theory of social role incongruity, arguing that women entering the political arena are evaluated relatively negative because they do not ‘belong’ there. Our findings suggest that role incongruity, in the Dutch case, does not result in negative evaluations. And although our findings are rather surprising, they are not unique and may fit a more general but yet not commonly known pattern: a recent meta-analysis of Schwarz and Coppock (2019) suggests that female politicians may be slightly preferred with regard to vote choice above male politicians.
  We focused on gender stereotypes based on character traits, but our approach did not solve the puzzle of female underrepresentation in politics. This may be due to the existence of a different ‘stereotype’ for women in politics. Schneider and Bos (2014) found little overlap of stereotype content between women and female politicians, suggesting that female politicians may be a sub-stereotype of women. Brooks (2013) refers to this ‘leaders-not-ladies’ phenomenon: women entering the political arena are evaluated as leaders instead of women. Sanbonmatsu (2002) suggests that voters develop expectations about male or female candidates based on either gender stereotypes or experiences with previous candidates and officeholders: citizens differentiate and develop stereotypes for politicians based on their experiences with female and male candidates and officeholders. In future studies this intriguing phenomenon should be further explored.
  The relatively positive evaluation of women may also be explained by the women-are-wonderful effect, suggesting that women are overall more positively evaluated than men (Eagly & Mladinic, 1994). Another potential explanation for the ‘bias’ in favour of female politicians is that citizens are impressed by the female politician: assuming that women must have had a relatively hard time entering the political arena, citizens also assume that she is highly qualified. This idea that women in politics are of ‘superior’ quality (e.g., Crowder-Meyer et al., 2015; Fulton, 2012) must be theoretically developed and empirically explored in future studies. Yet another explanation refers to the mechanism proposed by Bauer (2020), who argues that the stereotype about women does not align with the masculine expectation of political leaders. Consequently, women are perceived as not so qualified for office, which creates a low qualification bar for female candidates. Voters simply expect less from female candidates: “a female candidate does not need much political experience for voters to consider her more qualified than a typical woman” (Bauer, 2020, p. 3). This may explain the positive evaluation in our experiment, suggesting that citizens compared the female candidate with a ‘typical woman’ instead of a politician; compared to a ‘typical woman’ the female politician is very qualified. Bauer further argues that when a female candidate is compared to a male candidate, the woman will be evaluated on the basis of the political role and in that case she will be evaluated less well than the male candidate. Importantly, this argument may suggest that although our results indicate a gender-neutral evaluation, we must be wary to conclude that gender does not play any role in the electoral calculus.
  Admittedly, the measurement of gender stereotypes can be considered a vulnerability of our study. To avoid social desirability effects, we asked whether respondents think that people in society hold gender stereotypes and it could be that respondents who indicated that there are stereotypes in society, simply state their perception of public opinion, and do not hold these stereotypes themselves, as we assumed based on projection. We empirically and satisfactory (but not fully!) validated our approach, but we acknowledge that this measure is imperfect and that an alternative measure is needed. Despite our suboptimal measure, the main conclusion that female politicians are more positively evaluated than men on different traits firmly stands and is intriguing in and of itself. If citizens do not make an electorally relevant difference between male and female candidates, how then to explain the political underrepresentation of women?
  As mentioned, in the Netherlands there are enough women on party lists to fill over two times the lower house of parliament with female politicians. Yet, overall there are fewer female than male candidates. So it could be that not citizens as electors, but gatekeepers of parties are influenced by gender stereotypes. We suggest that ultimately the party-political gatekeepers – predominantly men: party membership is a male affair (Heidar & Wauters, 2019) – are crucial in explaining the underrepresentation of women in party-centred democracies: they recruit candidates and order the candidate list, and by doing so ‘determine’ who will be represented in parliament. Not only can stereotypes play a role here, also the way candidate selection is (in)formally organised may impact on the candidate list and its ranking.
  Since the main hypotheses were rejected, this article essentially contains a ‘null finding’ – but the derogatory term should not suggest that our findings are irrelevant. “When null findings are not published, they cannot place anomalously large and statistically significant results into their proper context” (Esarey & Wu, 2016, p. 1). Presenting null results is relevant for various reasons (Franco et al., 2014). First, it may prevent future researchers spending time and resources in conducting studies that already are conducted. Second, if future similar studies about gender stereotypes and the effect on candidate evaluations do find statistically and substantively significant effects, the absence of null findings would suggest, erroneously, that these reported effects are more evident and stronger than they in fact may be (Franco et al., 2014). This would hinder the true explanation and real understanding of the problem of female underrepresentation in politics.
  Our study arguably adds to confusion instead of the contribution to the final solution of this puzzle. At the same time, we end on a positive note: we have empirically shown that female and male politicians or political candidates are evaluated in a similar and equal way at the mass level, at least in the Netherlands. This is a welcome finding from a normative democratic perspective – although we cannot deny that this result does not add a single woman to the female minority of about one-third of 150 MPs.

References

• Andeweg, R. B. & van Holsteyn, J. (2011). Second Order Personalization; Preference voting in the Netherlands. Paper presented at the European Consortium for Political Research General Conference, Reykavik, August 25-27, 2011.

• Ashe, J. & Stewart K. (2012). Legislative Recruitment: Using Diagnostic Testing to Explain Underrepresentation. Party Politics, 18(5), 687-707.

• Auspurg, K. & Hinz, T. (2015). Factorial Survey Experiments. Sage.

• Baker, R., Blumberg, S.J., Brick, J.M., Coupter, M.P., Courtright, M., Dennis, J.M., Dillman, D., Frankel, M.R., Garland, P., Groves, R.M., Kennedy, C., Krosnick, J., Lavrakas, P.J., Lee, S., Link, M., Piekarski, L., Rao, K., Thomas, R.K. & Zahs, D. (2010). Research Synthesis: AAPOR Report on Online Panels. Public Opinion Quarterly, 74(4), 711-781. doi: 10.1177/08944393211006059.

• Bauer, N.M. (2020). Shifting Standards: How Voters Evaluate the Qualifications of Female and Male Candidates. The Journal of Politics, 82(1), 1–12. https://doi.org/10.1086/705817.

• Baumeister, R.F., Dale, K. & Sommer, K.L. (1998). Freudian Defense Mechanisms and Empirical Findings in Modern Social Psychology: Reaction Formation, Projection, Displacement, Undoing, Isolation, Sublimation, and Denial. Journal of Personality, 66(6), 1081-1124. doi: 10.1111/1467-6494.00043.

• Bell, M.A. & Kaufmann K.M. (2015). The Electoral Consequences of Marriage and Motherhood: How Gender Traits Influence Voter Evaluations of Female Candidates. Journal of Women, Politics & Policy, 36(1), 1-21. doi: 10.1080/1554477X.2015.985150.

• Bittner, A. (2011). Platform or Personality? The Role of Party Leaders in Elections. Oxford University Press.

• Blais, A. (2011). Political Leaders and Democratic Elections. In: K. Aarts, A. Blais and H. Schmitt (Eds.), Political Leaders and Democratic Elections (pp. 1-10). Oxford University Press.

• Blum, L. (2004). Stereotypes and Stereotyping: A Moral Analysis. Philosophical Papers, 33(3), 251-289.

• Brooks, D. (2013). He Runs, She Runs: Why Gender Stereotypes Do Not Harm Women Candidates. Princeton University Press.

• Campbell, R. & Cowley P. (2018). The Impact of Parental Status on the Visibility and Evaluations of Politicians. British Journal of Politics and International Relations, 20(3), 753-769. doi: 10.1177/1369148118775032.

• Celis, K. & Childs, S. (2008). Introduction: The Descriptive and Substantive Representation of Women: New Directions. Parliamentary Affairs, 61(3), 419-425. doi: 10.1093/pa/gsn006.

• Celis, K. & Childs, S. (2014). Gender, Conservatism and Political Representation. ECPR Press.

• Celis, K. & Meier, P. (2006). De macht van het geslacht. Gender, politiek en beleid in België. Acco.

• Crowder-Meyer, M., Gadarian, S.K. & Trounstine, J. (2015). Electoral Institutions, Gender Stereotypes, and Women’s Local Representation. Politics, Groups, and Identities, 3(2), 318-334. doi: 10.1080/21565503.2015.1031803.

• Dahl, M. & Nyrup, J. (2021). Confident and Cautious Candidates: Explaining Under-Representation of Women in Danish Municipal Politics. European Journal of Political Research, 60, 199-224. doi: 10.1111/1475-6765.12396.

• Devroe, R. & Wauters, B. (2018). Political Gender Stereotypes in a List-PR System with a High Share of Women MPs: Competent Men versus Leftist Women? Political Research Quarterly, 7(4), 788-800. doi: 10.1177/1065912918761009.

• Diaz, M.M. (2005). Representing Women? Female Legislators in West-European Parliaments. ECPR Press.

• Dolan, K. (2010) The Impact of Gender Stereotyped Evaluations on Support for Women Candidates. Political Behavior, 32(1), 69-88. doi: 10.1007/s11109-009-9090-4.

• Dolan, K. (2014). Gender Stereotypes, Candidate Evaluations, and Voting for Women Candidates: What Really Matters?. Political Research Quarterly, 67(1), 96-107. doi: 10.1177/1065912913487949.

• Eagly, A.H. (1987). Sex Differences in Social Behavior: A Social-Role Interpretation. Erlbaum.

• Eagly, A.H. & Karau, S.J. (2002). Theory of Prejudice Toward Female Leaders. Psychological Review, 109(3), 573-598. doi: 10.1037/0033-295X.109.3.573.

• Eagly, A.H. & Mladinic, A. (1994). Are People Prejudiced Against Women? Some Answers From Research on Attitudes, Gender Stereotypes, and Judgements of Competence. European Review of Social Psychology, 5(1), 1-35. doi: 10.1080/14792779543000002.

• Esarey, J. & Wu, A. (2016). Measuring the Effects of Publication Bias in Political Science. Research and Politics, 3(3), 1-9. doi: 10.1177/2053168016665856.

• Fiske, S.T. (1998). Stereotyping, Prejudice, and Discrimination. In: D. Gilbert, S.T. Fiske & G. Lindzey (Eds.), The Handbook of Social Psychology (pp. 357-411). McGraw-Hill.

• Fiske, S.T., Cuddy, A.J.C., Glick, P. & Xu, J. (2002). A Model of (Often Mixed) Stereotype Content: Competence and Warmth Respectively Follow From Perceived Status and Competition. Journal of Personality and Social Psychology, 82(6), 878-902. doi: 10.1177/2053168016665856.

• Fox, R.L. & Lawless J.L. (2014). Uncovering the Origins of the Gender Gap in Political Ambition. American Political Science Review, 108(3), 499-519. doi: 10.1017/S0003055414000227.

• Fox, R.L. & Smith E.R.A.N. (1998). The Role of Candidate Sex in Voter Decision-Making. Political Psychology, 19(2), 405-419. doi: 10.1111/0162-895X.00110.

• Franco, A., Malhotra N. & Simonovits G. (2014). Publication Bias in the Social Sciences: Unlocking the File Drawer. Science, 345(6203), 1502-1505. doi: 10.1126/science.1255484.

• Fulton, S.A. (2012). Running Backwards and in High Heels: The Gendered Quality Gap and Incumbent Electoral Success. Political Research Quarterly, 65(2), 303-314. doi: 10.1177/1065912911401419.

• Glick, P. & Rudman L.A. (2010). Sexism. In: J.F. Dovidio, M. Hewstone, P. Glick and V.M. Esses (Eds.), Prejudice, Stereotyping and Discrimination (pp. 328-344). Sage Publications Ltd.

• Golder, S.N., Stephenson, L.B., van der Straeten, K., Blais A., Bol, D., Harfst, P. & Laslier, J. (2017). Votes for Women: Electoral Systems and Support for Female Candidates. Politics & Gender, 13(1), 1-25. doi: 10.1017/S1743923X16000684.

• Hayes, D. (2011). When Gender and Party Collide: Stereotyping in Candidate Trait Attribution. Politics & Gender, 7(2), 133-165. doi: 10.1017/S1743923X11000055.

• Heidar, K. & Wauters, B. (2019). Do Parties Still Represent? An Analysis of the Representativeness of Political Parties in Western Democracies. Routledge.

• Hilton, J.L. & von Hippel, W. (1996). Stereotypes. Annual Review of Psychology, 47, 237-271. doi: 10.1146/annurev.psych.47.1.237.

• Hofstede, G., Hofstede, G.J. & Minkov, M. (2005). Cultures and Organizations. Software of the mind. Intercultural Cooperation and Its Importance for Survival. McGraw-Hill.

• Hofstede, G. & Soeters, J. (2002). Consensus Societies with Their Own Character: National Cultures in Japan and the Netherlands. Comparative Sociology, 1(1), 1-16. doi: 10.1163/156913202317346728.

• Huddy, L. & Terkildsen. N. (1993). Gender Stereotypes and the Perception of Male and Female Candidates. American Journal of Political Science, 37(1), 119-147. doi: 10.2307/2111526.

• Inter-Parliamentary Union. (2021a). Parline. Retrieved from https://data.ipu.org/node/121/data-on-women?chamber_id=13490 (last accessed 9 December 2021).

• Inter-Parliamentary Union. (2021b). Women in National Parliaments. Retrieved from http://archive.ipu.org/wmn-e/world.htm (last accessed 27 October 2021).

• Jennings, M.K. (2006). The Gender Gap in Attitudes and Beliefs about the Place of Women in American Political Life: A Longitudinal, Cross-Generational Analysis. Politics & Gender, 2(2), 193-219.

• Joop. (2019). Initiatief ‘Stem op een vrouw’ succesvol: aantal vrouwen in bestuur stijgt. Retrieved from https://joop.bnnvara.nl/nieuws/initiatief-stem-op-een-vrouw-succesvol-aantal-vrouwen-in-bestuur-stijgt (last accessed 4 June 2019).

• Kahn, K.F. (1996). The Political Consequences of Being a Woman. How Stereotypes Influence the Conduct and Consequences of Political Campaigns. Colombia University Press.

• Kerevel, Y.P. & Atkeson, L.R. (2015). Reducing Stereotypes of Female Political Leaders in Mexico. Political Research Quarterly, 68(4), 732-744. doi: 10.1177/1065912915607637.

• Kinder, D.R. (1983). Presidential Trait. ANES Pilot Study Report.

• Kinder, D.R., Peters, M.D., Abelson, R.P., & Fiske, S.T. (1980). Presidential Prototypes. Political Behavior, 2(4), 315-337. doi: 10.1007/BF00990172.

• Klammer, C. (2019, May 11). Waar blijft onze vrouwelijke premier. NRC. https://www.nrc.nl/nieuws/2019/05/10/zo-beweeg-je-vrouwen-de-politiek-in-te-gaan-a3959847.

• Koenig, A.M., Eagly, A.H., Mitchell, A.A. & Ristikari T. (2011). Are Leader Stereotypes Masculine? A Meta-Analysis of Three Research Paradigms. Psychological Bulleting, 137(4), 616-642. doi: 10.1037/a0023557.

• Lau, R.R. & Redlawsk D.P. (2001). Advantages and Disadvantages of Cognitive Heuristics in Political Decision Making. American Journal of Political Science, 45(4), 951-971. doi: 10.2307/2669334.

• Lau, R.R. & Redlawsk, D.P. (2006). How Voters Decide: Information Processing in Election Campaigns. Cambridge University Press.

• Levy, J.S. (2008). Case Studies: Types, Designs, and Logics of Inference. Conflict Management and Peace Science, 25(1), 1-18. doi: 10.1080/07388940701860318.

• Leijenaar, M. (2004). Political Empowerment of Women: The Netherlands and Other Countries. Brill.

• Lijphart, A. (1999). Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries. Yale University Press.

• Lippmann, W. (1922). Public Opinion. Harcourt.

• Mansbridge, J. (1999). Should Blacks Represent Blacks and Women Represent Women? A Contingent “Yes”. The Journal of Politics, 61(3), 628-657. doi: 10.2307/2647821.

• Matland, R.E. (1994). Putting Scandinavian Equality to the Test: An Experimental Evaluation of Gender Stereotyping of Political Candidates in a Sample of Norwegian Voters. British Journal of Political Science, 24(2), 273-292. doi: 10.1017/S0007123400009819.

• Matland, R.E. (2005). Enhancing Women’s Political Participation: Legislative Recruitment and Electoral Systems. In: J. Ballington and A. Karam (Eds.), Women in Parliament: Beyond Numbers (pp. 93-140). International IDEA.

• Matland, R.E. & Brown D.D. (1992). District Magnitude’s Effect on Female Representation in U.S. State Legislatures. Legislative Studies Quarterly, 17(4), 469-492. doi: 10.2307/439862.

• McGraw, K.M. (2003). Political Impressions. In: D. O. Sears, L. Huddy and R. Jervis (Eds.), Oxford Handbook of Political Psychology (pp. 394-432). Oxford University Press.

• Nagtzaam, M. (2019). Second-Order Electoral Personalization: Intra-Party Preference Voting in Belgium and the Netherlands. Universiteit Leiden (dissertatie).

• Norris, P. & Lovenduski, J. (1993). ‘If Only More Candidates Came Forward’: Supply-Side Explanations of Candidate Selection in Britain. British Journal of Political Science, 23(3), 373-408. doi: S0007123400006657.

• NOS. (2018). Veel meer vrouwen in gemeenteraad, vaak dankzij voorkeursstemmen. Retrieved from www.nos.nl/artikel/2224839-veel-meer-vrouwen-in-gemeenteraad-vaak-dankzij-voorkeursstemmen.html (last accessed 10 February 2019).

• Okimoto, T.G. & Brescoll, V.L. (2010). The Price of Power: Power Seeking and Backlash Against Female Politicians. Personality and Social Psychology Bulletin, 36(7), 923-936. doi: 10.1177/0146167210371949.

• Paxton, P. (1997). Women in National Legislatures: A Cross-National Analysis. Social Science Research, 26(4), 442-464. doi: 10.1006/ssre.1997.0603.

• Phillips, A. (1994). Democracy and Representation: Or Why Should It Matter Who Our Representatives Are? In: A. Phillips (Eds.), Feminism and Politics (pp. 224-240). Oxford University Press.

• Phillips, A. (1995). The Politics of Presence. Clarendon.

• Pitkin, H.F. (1967). The Concept of Representation. University of California Press Ltd.

• Rule, W. (1987). Electoral Systems, Contextual Factors and Woman’s Opportunity for Election to Parliament in Twenty-Three Democracies. The Western Political Quarterly, 40(3), 477-498. doi: 10.1177/106591298704000307.

• Sanbonmatsu, K. (2002). Gender Stereotypes and Vote Choice. American Journal of Political Science, 46(1), 20-34. doi: 10.2307/3088412.

• Sanbonmatsu, K. & Dolan, K. (2009). Do Gender Stereotypes Transcend Party?. Political Research Quarterly, 62(3), 485-494. doi: 10.1177/1065912908322416.

• Sapiro, V. (1981/1982). If U.S. Senator Baker Were a Woman: An Experimental Study of Candidate Images. Political Psychology, 3(1/2), 61-83. doi: 10.2307/3791285.

• Sapiro, V. (2003). Theorizing Gender in Political Psychology Research. In: D.O. Sears, L. Huddy & R. Jervis (Eds.), Oxford Handbook of Political Psychology (pp. 601-624). Oxford: Oxford University Press.

• Schneider, C.M. & Bos, A.L. (2014). Measuring Stereotypes of Female Politicians. Political Psychology, 35(2). 245-266. doi: 10.1111/pops.12040.

• Schwarz, S. & Coppock A. (2019). What Have We Learned About Gender From Candidate Choice Experiments? A Meta-Analysis of 42 Factorial Survey Experiments. Retrieved from https://alexandercoppock.com/papers/SC_gender.pdf.

• Taylor-Robinson, M.M., Yarkoney-Sorek, A. & Geva, N. (2016). Context and Attitudes about the Ability of Women to Govern: Findings from an Experimental Study in Israel and Costa Rica. Annual Meeting of the Midwest Political Science Association. Chicago.

• Teele, D.L., Kalla, J. & Rosenbluth, F. (2018). The Ties That Double Bind: Social Roles and Women’s Underrepresentation in Politics. American Political Science Review, 112(3), 525-541. doi: 10.1017/S0003055418000217.

• Thomassen, J.J.A. (1976). Party Identification as a Cross-national Concept: Its Meaning in the Netherlands. In: I. Budge, I. Crewe & D. Farlie (Eds.), Party Identification and Beyond (pp. 63-79). Wiley.

• Thomassen, J.J.A. & Rosema, M. (2009). Party Identification Revisited. In: J. Bartle & P. Belluci (Eds.), Political Parties and Partisanship. Social Identity and Individual Attitudes (pp. 43-59). Routledge.

• Valdini, M.E. (2013). Electoral Institutions and the Manifestation of Bias: The Effect of the Personal Vote on the Representation of Women. Politics & Gender, 9(1), 76-92. doi: 10.1017/S1743923X12000700.

• Van Es, M. (2019, May 7). Waar blijft de vrouwelijke premier? De Telegraaf. https://www.telegraaf.nl/nieuws/3550892/waar-blijft-de-vrouwelijke-premier.

• Van Elsas, E.J., Lubbe, R., van der Meer, T.M.G. & van der Brug, W. (2014). Vote Recall: A Panel Study on the Mechanisms that Explain Vote Recall Inconsistency. International Journal of Public Opinion Research, 26(1), 18-40. doi: 10.1093/ijpor/edt031.

• Van Holsteyn, J. & Andeweg, R. (2012). Tweede Orde Personalisering: Voorkeurstemmen in Nederland. Res Publica, 54(2), 163-191.

• Van Holsteyn, J. (2015). Angels gedoopt in honing: politieke tekeningen en hun betekenis. Res Publica, 57(3), 319-349.

Appendix

The Appendix is only available online and can be found here: https://www.elevenjournals.com/tijdschrift/PLC/2022/1/PLC-D-20-00007A.

Noten

• 1 Stereotypes can have both negative and positive dimensions. We focus on the positive dimension, since this dimension entails the agentic and communal traits (Schneider & Bos, 2014) which are essential in role incongruity theory. Negative stereotypes are for example being emotionally weak (for women) and greedy (for men) (Schneider & Bos, 2014), but these traits are not typical agentic and communal traits.

• 2 For a more elaborate argument on the use of raw numbers instead of proportionality tests to study the supply side, see Ashe & Stewart (2012, p. 691).

• 3 A copy of the questionnaire (in Dutch) can be obtained on request from the corresponding author.

• 4 Since particular names may elicit (negative) responses (Newman et al., 2018), we have chosen names that are ‘common’ and easy to alter in an equivalent male or female version (Karel and Carolien). We also made sure that the names did not match the names of well-known politicians. The surnames are also common, belonging to the 50 most used surnames in the Netherlands (Nederlandse Familienamenbank, 2007). At the time of the experiment, there were no national politicians with these surnames.

• 5 The exact wording (translated from Dutch by the authors) is: ‘The following statements are about what society feels about men and women. It is not about your personal opinion, but how you think Dutch society feels. How confident/competent/warm/sincere are men/women perceived?’ Every trait was mentioned separately for men and women. There were 8 statements in total.

• 6 Whether respondents have gender stereotypes was measured after the stimulus was presented. We have chosen to do so, since we considered it a risk to ask respondents about gender stereotypes first and subsequently ask them to evaluate a candidate. Therefore, we chose to do it in a reverse order, taking the risk that our stimulus has influenced the measure of gender stereotypes. However, we compared the outcomes of the Stereotype Content Model between experimental groups and no substantive differences were found. Therefore, we are confident that there is no post-treatment bias. A table with these checks is presented in the appendix (Table A1).

• 7 The second author was involved in the development of the EVO and has been a scientific adviser to this panel.

• 8 Data about the voter population are from 2012 and are retrieved from Statistics Netherlands.

• 9 We have run our analysis on a weighted dataset, but this did not substantially change our results: female politicians were still evaluated more positively than male politicians.

• 10 The vignette used in the analysis is about a politician with an expertise in economics. However, the specific policy area might be gendered: economics can be seen as a ‘male’ policy area (Kerevel & Atkeson, 2015; Huddy & Terkildsen, 1993). To check for the impact of policy area, we included another vignette about a politician with an expertise in health care in the questionnaire and analyses. The results for both policy areas are similar.

• 11 The politician presented in the vignette is married and has children. Some research suggests that being married and/or having children can be advantageous for a politician’s evaluation (e.g., Bell & Kaufman, 2015; Campbell & Cowley, 2018). In another vignette we did not include any information about parental or marital status, and in this vignette the evaluation pattern persisted: female politicians were rated relatively high (see the appendix for results).

• 12 See note 9.

• 13 In the result section, we discuss the two categories of character and competence traits. However, a principal component analysis shows that the ten traits do not load on the two categories (cf. Bittner, 2011). Instead, the eight positive phrased traits form one factor and two negatively phrased traits the second; see the appendix (Table A2) with factor loadings. Because of this result we did not create two variables of character and competence traits but proceed the analyses with all separate traits.

• 14 See the appendix for the results of the analysis on the data of the second vignette in Table A3.

• 15 See the appendix for the results of the analysis on the data of the second vignette in Table A4.