Shared group experiences

Studies on the gendered formation of political preferences start from the premise that women experience gendered constraints and exclusions that differ from men’s opportunities and that give rise to shared political interests. According to Sapiro (Reference Sapiro1981), women’s interests originate due to the private division of labor between men and women in the household. Since women give birth to children and take on the main burden of childcare, their socio-economic position differs from that of men producing distinct interests and policy preferences. In industrialized societies, childbearing and family obligations have historically limited women’s entry into the paid economy while facilitating men’s acquisition of market-relevant skills (Wood and Eagly Reference Wood and Eagly2012). This difference gave men more resources and power compared to women and led to different roles in the family and society (Diekman and Schneider Reference Diekman and Schneider2010; Siegenthaler Reference Siegenthaler2013). Diamond and Hartsock (Reference Diamond and Hartsock1981) add that it is not only the private division of labor but also the gendered division of productive labor, namely that women stay at home while men work in the public sphere, which contributes to the diverging social and economic positions of women and men. This division gives rise to women-specific interests and different policy priorities of women and men MPs.Footnote ³ This argument is echoed by Phillips (Reference Phillips1994: 71f.), stating that ‘[w]omen occupy a distinct position within society: they are typically concentrated, for example, in lower paid jobs; and they carry the primary responsibility for the unpaid work of caring for others. There are particular needs, interests, and concerns that arise from women’s experience, and these will be inadequately addressed in a politics that is dominated by men’.

These gendered formations of political preferences and the associated group-specific experiences may enable women MPs to form more accurate perceptions of voters’ opinions on women’s issues. More than a century ago, John Stuart Mill (Reference Mill1861: 22) already claimed that ‘in the absence of its natural defenders, the interest of the omitted is always in danger of being overlooked; and when looked at, is seen with very different eyes from those of the persons whom it directly concerns’. In other words, women MPs can draw on their shared experiences to notice aspects of issues that others might miss and therefore are able to conceive what is in the best interest of their constituents (Franceschet, Lucas and Rayment Reference Franceschet, Lucas and Rayment2024).

Thus, as women MPs share gender-specific experiences with the women population, women MPs develop a sense of solidarity and linked fate (Dovi Reference Dovi2002) with women in the general population, which, in turn, positively impacts their motivation to be responsive toward their voters’ preferences regarding women’s issues (Celis Reference Celis2008; Sobolewska, McKee and Campbell Reference Sobolewska, McKee and Campbell2018). This is supported by previous research, showing that women MPs are generally more concerned with women’s issues than their men colleagues and that they more frequently engage with women-specific issues in their political work (e.g., Coffé and Reiser Reference Coffé and Reiser2018; Dovi Reference Dovi2007; Phillips Reference Phillips1995).

Central to our argument, this motivation and sensitivity encourage women legislators to invest in information gathering about women’s issues, as well as an intense engagement with diverse stakeholders. Given their heightened awareness of gendered policy effects, they are more likely to actively seek input from a broad range of sources, including women’s organizations, civil society actors, and direct constituency interactions. This systematic effort to gather and process information allows them to develop a more precise understanding of public attitudes toward women’s issues, including women as well as men voters’ perspectives on gender-related policies. During an interview conducted for this project, a German woman politician illustrated this argument: ‘I personally make this experience to be a woman with kids and to have a time-intensive job as a politician. Other persons with kids identify with this and talk frequently to me about this’.

In sum, direct engagement and shared gendered life experiences equip women MPs with knowledge that improves their assessment of voters’ policy preferences on women-specific matters. By contrast, many men MPs lack some of these experiences (e.g., childbirth, period-related issues) and are less often drawn into conversations about such issues. While men may sometimes encounter analogous situations (e.g., gender-related workplace discrimination in women-dominated occupations or sexual harassment), such experiences are rarer and less frequently politicized and therefore discussed with citizens. A recent study further supports this argument, revealing that both women and men politicians continue to perceive the representation of gender equality as primarily the responsibility of women MPs (Höhmann Reference Höhmann2025). Thus, women representatives should have a more accurate view of voters’ policy positions on women’s issues compared to their men colleagues.

This argument also finds support in Jane Mansbridge’s (Reference Mansbridge1999, Reference Mansbridge2003) seminal work on political representation. For Mansbridge (Reference Mansbridge1999), women-specific concerns are uncrystallized, meaning that they are not clearly delineated and that they can change over time. This attribute implies that new topics affecting these concerns might surface unpredictably on the political agenda. In these situations, Mansbridge (Reference Mansbridge1999) argues that ‘descriptive representatives are, other things equal, more likely than non-descriptive representatives to act as their descriptive constituents would like them to act’ (Mansbridge Reference Mansbridge1999: 646). Women MPs are therefore in a better position to accurately assess their voters’ preferences and opinions on these issues.

In her conceptualization of different forms of representation, Mansbridge (Reference Mansbridge2003) refers to this type of behavior as ‘gyroscopic representation’ where MPs are expected to follow their ‘internal gyroscope’, which is derived from their own social backgrounds and experiences. Regarding the gendered assessment of policy priorities, this ‘gyroscope’ is a virtue that men MPs cannot possess, so they have a harder time knowing when men and women voters differ on certain issues. Therefore, men lack socialization as a mechanism for accurately perceiving voters’ preferences on these issues. These expectations are corroborated by a number of empirical studies showing that women legislators see themselves as representatives of the women electorate and have different priorities than men MPs (Childs Reference Childs2002; Coffé and Reiser Reference Coffé and Reiser2018; Reher Reference Reher2018; Schwindt-Bayer Reference Schwindt-Bayer2010). Taken together, we derive the following expectation from this literature:

Beyond potentially improving their overall assessment of public opinion on women’s issues, we expect that shared group experiences may bias women MPs’ estimates toward the policy preferences of their women voters. Since socialization postulates that women MPs can rely on their personal experiences to make judgments about public opinion, their assessment of the public preferences regarding women’s issues should be more similar and lean toward the preferences of women voters. Put differently, women MPs may place greater weight on the preferences of women voters when estimating their party’s overall position on women’s issues. As a result, their estimations may systematically reflect the perspectives of women voters within their party electorate:

It is important to note that not all women politicians will have a better understanding of women’s issues, just as not all men politicians will lack such understanding. Women’s interests are not homogeneous (Beckwith Reference Beckwith, Escobar-Lemmon and Taylor-Robinson2014), and their perception and understanding vary with individual experience, values, and ideology. We therefore do not assume that women MPs are perfectly accurate in assessing voters’ opinions; rather, we expect that, on average, they will be more accurate than their men colleagues.Footnote ⁴

Strategic behavior based on gendered role expectations

The shared-experience argument is unconditional: women MPs, by virtue of lived experience, are better able to assess party voters’ preferences on women’s issues. This section advances a conditional argument by building on the strategic considerations of women MPs. In a nutshell, we argue that women MPs must act strategically in a society that imposes gendered constraints and asymmetric expectations on them. As Murray (Reference Murray2018: 4) points out, ‘[women] are faced with expectations which are not raised about men’. For example, previous research has established that voters hold different expectations for women politicians (Sanbonmatsu Reference Sanbonmatsu2002) and that women MPs adapt their behavior – for instance, on social media – to cater to these gendered norms (Fridkin and Kenney Reference Fridkin and Kenney2014). We contend that these expectations extend to policy expertise: voters expect women MPs to be better informed and more knowledgeable about women’s issues than men MPs. In an interview conducted for this study, a woman politician outlined to us that she certainly gets asked more often what her opinion about a women’s quota is. Failing to meet these expectations can entail reputational and electoral costs. Such criticism can be especially damaging under conditions of electoral vulnerability, when even small losses of credibility or support can endanger re-election. Hence, we expect women MPs to be particularly invested in demonstrating that they fulfill these expectations, especially when their re-election is uncertain. We outline this argument in more detail below.

We view MPs as rational actors driven by the intention of advancing their chances of being re-elected and to continue and advance their political careers (Müller and Strøm Reference Müller and Strøm1999). Different levels of electoral vulnerability thereby offer different incentives for women MPs to engage in activities enabling them to accurately assess the policy preferences of their voters regarding women’s issues (e.g., more frequent participation in discussions about women’s issues; reading and talking more about these issues with staff and voters). Although the shared gendered experiences of many women MPs might naturally bring them into a better position to accurately perceive the policy positions on women-specific topics, the difference in perceptual accuracy between women and men MPs might be more pronounced if an accurate knowledge of citizens’ preferences on women’s issues may affect women MPs’ prospects of being re-elected.

Public expectations that women MPs should be concerned about women’s issues can be understood through sociological institutionalism’s ‘logic of appropriateness’ and its feminist extension, the ‘gendered logic of appropriateness’ (Chappel and Waylen Reference Chappel and Waylen2013; Krook and Mackay Reference Krook and Mackay2011). Informal rules about acceptable gendered behavior produce distinct roles for men and women MPs, encouraging or constraining certain types of behaviors (Chappel and Waylen Reference Chappel and Waylen2013; Towns Reference Towns2003). Because women MPs share gender-specific experiences and perspectives with women in the population, the representation of women’s interests is typically seen as a function that should be fulfilled by women MPs (Bäck, Debus and Müller, Reference Bäck, Debus and Müller2014; Bergqvist, Bjarnegård and Zetterberg Reference Bergqvist, Bjarnegård and Zetterberg2018; Höhmann Reference Höhmann2025; Höhmann and Nugent Reference Höhmann and Nugent2022). Women MPs are expected to be more concerned with women’s interests than their men colleagues and to have a better sense of public preferences on issues such as the gender pay gap, protection against domestic violence, or the availability of breast cancer screening. This was also confirmed by one of our interviewed women politicians, who stressed that, in her opinion, voters expect women politicians to be competent in social affairs.

Due to the strong expectation that women MPs are responsible for the representation of women’s issues, they have to fear negative consequences if they do not fulfill this expectation and if they do not have a profound knowledge of women-specific topics. Media, feminist activists, or women’s organizations may criticize perceived shortcomings (Bergqvist, Bjarnegård and Zetterberg Reference Bergqvist, Bjarnegård and Zetterberg2018). When re-election is uncertain, women MPs will seek to avoid such criticism that could cost votes. Instead, they will respond vigorously to constituency demands under high electoral vulnerability, as a small vote gain or loss may be decisive for their re-election (Bernauer and Munzert Reference Bernauer and Munzert2014; Höhmann Reference Höhmann2020; Höhmann and Nugent Reference Höhmann and Nugent2022). We therefore expect that the accuracy of women MPs’ assessment of the public’s opinion on women’s issues increases if their electoral vulnerability is high. In these situations, women MPs are more motivated to gather information (e.g., through direct contact with citizens) about what their constituents want and – due to their ascribed role as the ‘women’s representative’ – women MPs should particularly be interested in learning about voters’ preferences regarding women-specific policy issues.Footnote ⁵

A psychological mechanism may reinforce this strategic behavior. Women MPs who face the prospect of losing an election are more likely to employ a risk-minimizing strategy by focusing on women’s policy issues. Recent work by Sheffer (Reference Sheffer2021) explores how women and men in office respond differently to being held politically accountable. Under heightened accountability, elected women adhere more strongly to status quo policies, while men become more likely to abandon them. This implicit risk-aversion might lead women MPs to focus on women’s policy issues.

Taking both mechanisms together:

If our strategic logic holds, women MPs should develop a better sense of the public opinion on their entire electorate on women’s issues, because they seek to persuade both women and men voters. It follows that, unlike the shared-experience logic, we do not expect that electoral vulnerability amplifies any tendency to weight women voters’ preferences more heavily.

All of our hypotheses explicitly focus on MPs’ estimates of their party electorate’s opinion, rather than public opinion in the entire population, for two reasons. First, we aim to avoid any perceptual biases stemming from party ideology (c.f. Helfer, Giger and Breunig Reference Helfer, Giger and Breunig2024; Sevenans, Walgrave, Jansen et al. Reference Sevenans, Walgrave, Jansen, Soontjens, Bailer, Brack, Breunig, Helfer, Loewen, Pilet, Sheffer, Varone and Vliegenthart2023). Potential biases could occur if conservative MPs generally assume that citizens are more critical of policy proposals on women’s issues or, in turn, if MPs from progressive left-wing parties overestimate the general support for these policy proposals in the population. This bias should be less severe if we only use the assessment of the MP’s respective party electorate. In addition, asking politicians for estimates of their partisan electorate implicitly emphasizes each representative’s ideologically delimited target population.

Second, the responsible party model of representation implies that responsive politicians should primarily be concerned about knowing and acting on the preferences of their respective party voters. In the majority of countries with a multi-party system, politicians commonly express a stronger concern for their party’s supporters than for the overall public (see Dudzinska, Poyet, Costa et al. Reference Dudzinska, Poyet, Costa, Weßels, Deschouwer and Depauw2014). In such instances, there might be less necessity for precise assessments of public opinion, as long as politicians accurately gauge the preferences of their own party’s voters. A more accurate estimation of the position of the party voters is additionally supported by the more frequent encounters of MPs with their party supporters and members.Footnote ⁶

Dependent variable: Perceptual accuracy

The dependent variable of the analysis is MPs’ perceptual accuracy of the respective party voters’ opinion on women’s issues. Defining women’s issues is a contested task (Reingold and Swers Reference Reingold and Swers2011; Yildirim Reference Yildirim2022). We refrain from painting an essentialist image of women and accept that there is no fixed set of interests that are shared by all women across time and space. We nevertheless assume – as already outlined above – that women in society experience gendered constraints and exclusions that differ from men’s opportunities and give rise to women-specific interests. These gendered experiences are diverse, and not all women are faced with the same constraints or disadvantages. According to Beckwith (Reference Beckwith, Escobar-Lemmon and Taylor-Robinson2014: 20), only ‘within specific contexts similar shaping forces exist and have similar consequences for women’. Thus, women’s interests are not homogeneous or uniform, but are instead context-specific, changing over time and often only shared by a subset of women (Reingold and Swers Reference Reingold and Swers2011; Severs Reference Severs, Tanasescu and Dupont2019).

Consequently, we define women’s issues as those topics that explicitly have women, or some subset of women, ‘as their primary subject matter’ (Reingold Reference Reingold2000: 166f.). According to Carroll (Reference Carroll1994), women’s interests are those ‘where policy consequences are likely to have a more immediate and direct impact on significantly larger numbers of women than of men’ (Carroll Reference Carroll1994: 15). Building on Celis (Reference Celis2008), this definition identifies women’s interests as those topics that – for either biological or social reasons – disproportionately affect women more strongly than men, or those that address a social condition in which women are disadvantaged compared to men.

Based on this definition, we selected different policy statements to measure the perceptual accuracy of MPs regarding women’s issues. In Switzerland, MPs were asked to assess their voters’ opinions regarding the following issues:

1. 1. Hospitals need to have a baby hatch where parents can leave their infant anonymously.

2. 2. Sexual harassment at work needs to be punished more severely.

3. 3. Same-sex couples who have registered their partnership should be allowed to adopt children.Footnote ⁷

4. 4. Wedded people need to be assessed separately for taxation.

In Germany, we measured MPs’ perceptual accuracy on the following policy issues:

1. 1. If equally qualified women should be privileged in the labor market.

2. 2. There should be a right to full-time child care until the end of elementary school.

Based on the questions outlined above, we calculate two different dependent variables to test our theoretical expectations about the gendered perception of voters’ public opinion. Our first dependent variable is the perceptual accuracy score for each MP and each question with regard to the entire partisan electorate’s public opinion. Our measure of the dependent variable draws on extant work (Converse and Pierce Reference Converse and Pierce1986; Erikson, Luttbeg and Holloway Reference Erikson, Luttbeg and Holloway1975; Esaiasson and Holmberg Reference Esaiasson and Holmberg1996; Sigel and Friesema Reference Sigel and Friesema1965) and represents the inverted size of the percentage error in politicians’ estimations of the share of the electorate supporting the policy. It is calculated in two steps: First, we calculate the absolute distance between MPs’ estimation and the actual share of the party’s electorate that is in favor of this policy. Second, we take the complement (100% – X) of the absolute error to get an accuracy score (instead of an inaccuracy score). For example, a perceptual accuracy score of 90% means that a politician’s estimation of the partisan voter’s public opinion is 10 percentage points off. Thus, the higher the value of the perceptual accuracy score, the better the MPs’ assessment of their party voters’ opinion regarding women-specific issues. Thus, the first dependent variable reflects the general quality of representation. It measures how accurately MPs estimate the preferences of their party’s voters as a whole on women’s issues, and thus provides an overall indicator of representational competence and responsiveness.

The second dependent variable addresses a different theoretical mechanism. It captures the idea that women MPs’ socialization and shared group experiences may shape how they form perceptions of public opinion. If women MPs rely more strongly on their own experiences or identify more closely with women in the electorate, their assessments might systematically reflect the perspectives of women party voters rather than the entire party electorate. Our second dependent variable captures this idea as the perceptual accuracy score for women party voters, that is, we just measure the political preferences of eligible women party voters. We essentially apply the same approach as with our first dependent variable, but now only use the preferences of women in the citizen survey as our measure of public opinion. The accuracy of MPs’ assessments of public opinion is then calculated as the absolute distance between MPs’ estimation of the party voters’ opinion and the actual share of the women electorate that is in favor of this policy.Footnote ⁸ Compared to their men counterparts, we expect that women MPs’ assessment of their party electorate will be more in line with the opinion of the women party voters.

Employing both measures is crucial for disentangling the mechanisms underlying potential gender differences in perceptual accuracy. The first variable captures whether descriptive representation improves the overall accuracy of representation, while the second tests whether women MPs’ perceptions are selectively oriented toward women voters’ preferences, as proposed in Hypothesis 1b. The dual operationalization thus provides empirical leverage to determine whether women MPs’ advantage lies in general perceptual accuracy or in a gendered orientation toward women voters – thereby contributing to a more nuanced understanding of how descriptive representation translates into meaningful substantive representation.

Covariates

The key explanatory concepts are gender and electoral vulnerability. The first two hypotheses (H1a, H1b) focus on the different experiences of politicians between genders. We classify women MPs based on their self-identified gender on each parliament’s record. For our strategic expectation argument (H2), we measure the electoral vulnerability of MPs. Instead of using previous election results to calculate an objective estimate of the re-election probability of individual MPs, we employ MPs’ subjective perceptions of their electoral situation. During the interviews, MPs were asked to assess their individual chances of getting re-elected in the upcoming elections. The subjective re-election probability of MPs is measured on a scale from 0 to 4, where 0 indicates that MPs think that their re-election is very unlikely and 4 stands for a very likely re-election.Footnote ⁹ To get a measure of MPs’ electoral vulnerability, we reversed this scale so that a value of 0 indicates very low levels of electoral vulnerability and a value of 4 stands for very high levels of electoral vulnerability.

Given that we are interested in how electoral vulnerability affects the behavior of individual MPs, we intentionally refrained from applying an objective measure of re-election probabilities (e.g., André, Depauw and Martin Reference André, Depauw and Martin2015; Stoffel and Sieberer Reference Stoffel and Sieberer2018). A subjective judgment by an MP is more appropriate and conceptually closer in our case. Ultimately, it is not an objective number based on previous election results that affects the behavior of representatives, but the individual perception of MPs as to how likely or unlikely they consider their personal re-election. The advantage of a subjective measure of electoral vulnerability lies in its ability to capture more nuanced, as well as MP- and context-specific factors that may influence the electoral situation in a district. In particular, it takes into account the perceptions of the MPs themselves and thereby allows for a deeper understanding of the specific political and social dynamics within a given constituency or region. Perceptions capture local issues, shifts in public sentiment, or situations in nomination bodies such as local parties, that may not have been accounted for in objective metrics from a past period, but which are noticed by individual MPs and taken into account when assessing their individual re-election prospects at the moment of the interview. Subjective measures are also preferable to objective measures of vulnerability because MPs themselves estimate their joint re-election probability when running as both list and district candidates (as is the case in Germany).Footnote ¹⁰ While it is certainly true that MPs also use previous election results to inform themselves about their electoral situation and their chances of being re-elected in an upcoming election, we assume that these objective facts are updated by the individual perceptions of MPs about current political developments as well as MPs’ current political situations and their relationship with voters in their districts and their local party branch, which may influence their renomination chances. As a consequence, MPs’ perceptions and understandings of their electoral vulnerability may differ significantly from what objective measures of re-election probabilities may suggest. And while it is important to acknowledge that our measure is subjective by nature and that it can be influenced by biases, limited perspectives, or inaccurate perceptions, these individual considerations eventually determine MPs’ political behavior and their decision to inform themselves about public opinion regarding women’s issues.

We include several adjustment variables in the analysis. Previous research on the perceptual accuracy of MPs has identified three larger groups of variables that may affect how accurate politicians perceive public opinion: MPs’ political career, their role orientation and representational style, as well as MPs’ information sources that they use to inform themselves about what the public thinks (c.f. Walgrave, Jansen, Sevenans et al. Reference Walgrave, Jansen, Sevenans, Soontjens, Pilet, Brack, Varone, Helfer, Vliegenthart, van der Meer, Breunig, Bailer, Sheffer and Loewen2023).

Therefore, on the individual level, we account for age, seniority (both measured in years), and an MP’s membership in the women’s committee or the committee on social affairs and health. Legislators with more legislative experience, as well as members of committees that frequently deal with women’s issues, are better equipped to develop an accurate sense of what their voters prefer. Moreover, we account for the general style and the focus of MPs’ representational behavior. MPs were asked how important it is for them to represent a specific group in society (measured on a scale from 1 to 5, with higher values indicating less importance), to what extent they follow their own convictions while working in parliament (measured on a 10-point Likert scale with higher values indicating that MPs follow their own convictions), and how useful they find direct contact with citizens to inform themselves about what the public wants (measured on a scale from 1 to 8, with higher values indicating less importance).

Additionally, all models include a country dummy for Switzerland as well as a dummy variable controlling for MPs’ membership in a regional parliament (Kantonsparlament). Lastly, we control for party left-right placements based on RiLe scores from the Manifesto Project (MARPOR). In the sample, the leftmost political party (Swiss Greens) scores −17, and the rightmost (Swiss EDU) is at 73.Footnote ¹¹

Estimation

Since MPs had to assess public opinion on several issues, the unit of analysis is the MP policy statement dyad. We estimate two model specifications for each of the two dependent variables (perceptual accuracy for the whole party’s electorate, and perceptual accuracy for only women party voters’ opinion). The first model specification includes the covariates mentioned above. The second specification additionally includes a multiplicative interaction term between woman and the electoral vulnerability. These four model specifications test all facets of our expectations. After excluding MPs with missing information on one of the variables mentioned above, our analysis is based on 565 observations. The estimation is based on a linear regression with robust standard errors clustered by MP since different assessments from the same MP are not independent of each other.

Placebo analysis

We perform a placebo analysis to strengthen our basic findings.Footnote ¹³ Following Eggers, Tunon and Dafoe (Reference Eggers, Tunon and Dafoe2024), we run a placebo outcome test that replicates the core analysis with an alternative dependent variable that should be unaffected by the main independent variable if our research design is valid and if our theoretical expectations about the behavior of women MPs are correct.

If our expectation is correct that electorally vulnerable women MPs are particularly interested in learning about voters’ preferences regarding women-specific policy issues because they react to gendered role expectations, we should not find any significant effects for gender and electoral vulnerability in policy areas that are not explicitly women-specific. Thus, we use our original model specification but now estimate the interaction effect of woman and electoral vulnerability on MPs’ perceptual accuracy of the party voters’ opinion on non-women-related policy proposals. The placebo test covers a broad set of issues. For Swiss MPs, these included, for example, proposals on defense procurement (purchase of new fighter jets), immigration control, taxation, internet voting, and pension reform. For German MPs, the items comprised policy statements on EU integration, environmental regulation, redistribution, criminal justice, and asylum policy, among others.Footnote ¹⁴ By including such a diverse range of policy domains, the placebo analysis ensures that the gendered effects identified in the main models are not driven by a general tendency of women MPs to be more accurate or attentive across all policy areas, but are instead specific to women-related issues.

Except for the alternative dependent variable, we estimate exactly the same models as in the core analysis to ensure that potential flaws of the original research design (which might have produced spurious results) also operate in the placebo analysis. The dependent variable in Model 5 and Model 7 is the perceptual accuracy of the assessment of the overall public opinion among the party electorate. The dependent variable in Model 6 and Model 8 is perceptual accuracy with regard to women voters’ opinions.

The estimates in Table 2 for women MPs as well as the interaction effect of women MPs and their electoral vulnerability are statistically non-significant. The results of the placebo test thus indicate that the electoral context of women MPs is irrelevant for estimating public opinion on policies unrelated to women’s issues. The insignificant effects of electoral vulnerability in Table 2 additionally show that we are also not observing a more general pattern where all MPs become more motivated to inform themselves about their voters’ policy preferences if they are electorally vulnerable. The additional evidence implies that the electoral vulnerability of women MPs only induces them to focus on citizens’ preferences regarding women-specific policy issues to fulfill their ascribed role as the ‘women’s representative’ and to avoid negative publicity.

Table 2.

Placebo analysis. Estimation of the public opinion on non-women’s issues. Ordinary least squares regression. Perceptual accuracy of assessment of party voters’ opinion on issues that are not women-specific as the dependent variable. Model 5 and Model 7: Assessment of the public opinion of the entire party electorate. Model 6 and Model 8: Assessment of the public opinion of women in the party electorate. Standard errors (in parentheses) are clustered by MPs

Significance levels: *p < 0.05, **p < 0.01, ***p < 0.001.

Robustness tests

Besides the placebo test, we conduct several analyses to determine the robustness of our main findings. The detailed description and results of these additional analyses can be found in Appendices A4 and A5. In sum, the robustness tests show that our findings are robust toward the inclusion of policy statement fixed effects, that all substantive findings remain unchanged if we re-estimate our core model specification without the control variable for membership in the women’s committee (potential post-treatment bias), and that the results are not sensitive to the inclusion of specific policy statements in the estimation of the perceptual accuracy score. Lastly, we explored whether personality traits that are neither caused by shared group experiences nor by working harder toward understanding women’s issues impact perceptual accuracy. The idea is that women may have certain personality traits that make them better suited to understand women’s interests. However, the results of a mediation analysis in Appendix A5 show that all estimated effects are not statistically distinguishable from zero, meaning that personality plays little role in estimating public opinion.