Many Western polities display intense distrust and hostility across party lines. This affective polarization can prompt animosity toward partisan opponents as neighbors, coworkers, or family members (Iyengar et al. Reference Iyengar, Lelkes, Levendusky, Malhotra and Westwood2019), economic discrimination against partisan opponents (McConnell et al. Reference McConnell, Margalit, Malhotra and Levendusky2018), and willingness to violate democratic norms in pursuit of political objectives (Kalmoe and Mason Reference Kalmoe and Mason2018). Indeed, the January 6 U.S. Capitol insurrection, the attempted storming of the German Bundestag in August 2020, and the murder of British legislator Jo Cox during the 2016 Brexit campaign chillingly illustrate the violent consequences that may ensue from heightened political hostility. In response, scholars analyze the causes of affective polarization including policy disputes, economic conditions, levels of corruption, and electoral systems (e.g., Gidron, Adams, and Horne Reference Gidron, Adams and Horne2020; Reiljan Reference Reiljan2020; Wagner Reference Wagner2021).
No research to date asks whether who serves in office affects mass-level interparty hostility. Drawing on scholarship on both women elected representatives’ behavior and citizens’ beliefs about women politicians, we posit that partisans more warmly evaluate out-parties with higher proportions of women members of parliament (MPs). To test our claim, we analyze an original dataset on women’s presence in the parliamentary delegations of 125 political parties in 20 Western democracies between 1996 and 2017 combined with Comparative Study of Electoral Systems survey data on partisans’ affective party ratings. We show that women’s presence in parties’ parliamentary delegations is associated with lower levels of partisan hostility and that both men and women partisans react positively to out-party women MPs. Our findings thus suggest that increasing women’s parliamentary presence could mitigate cross-party hostility.
AFFECTIVE POLARIZATION AND WOMEN’S REPRESENTATION
Scholars often emphasize the negative consequences of affective polarization, including its role in democratic backsliding (Orhan Reference Orhan2022; though see Broockman, Kalla, and Westwood Reference Broockman, Kalla and Westwood2020). There is less consensus about its causes. One strand of research links mass-level partisan resentment to elite-level policy polarization (Lelkes Reference Lelkes2021; Orr and Huber Reference Orr and Huber2020). Others see affective polarization as rooted in emotional attachments to social identities that are “sorted” along partisan lines (Harteveld Reference Harteveld2021; Mason Reference Mason2018) or emphasize structural features like economic conditions (Stewart, McCarty, and Bryson Reference Stewart, McCarty and Bryson2020).
Though still in its early stages (see Wagner Reference Wagner2021 for a discussion), comparative research highlights the role of parties and electoral systems. This work documents the intense hostility between mainstream and radical-right parties (Harteveld, Mendoza, and Rooduijn Reference Harteveld, Mendoza and Rooduijn2021; Helbling and Junkunz Reference Helbling and Jungkunz2020; Reiljan and Ryan Reference Reiljan and Ryan2021). Democratic dissatisfaction and affective polarization are also more pronounced in majoritarian systems, which solidify “us-versus-them” political dynamics, compared with proportional systems, which incentivize elite cooperation in the form of coalition governments (Anderson and Guillory Reference Anderson and Guillroy1997; Gidron, Adams, and Horne Reference Gidron, Adams and Horne2019; McCoy and Somer Reference McCoy and Somer2019). Multiparty governments, for example, alleviate tensions between cogoverning parties (Bassan-Nygate and Weiss Reference Bassan-Nygate and Weiss2021).
Building on this scholarship, we argue that electoral rules do matter, but who is elected to office also influences out-party hostility. We posit that women’s descriptive representation in parties’ parliamentary delegations can potentially defuse affective polarization for at least two reasons. First, women may employ more consensual and participatory leadership styles. For example, studies of the United Kingdom (Childs Reference Childs2004; Sones, Moran, and Lovenduski Reference Sones, Moran and Lovenduski2005) and New Zealand (Grey Reference Grey2002) find that women representatives are less adversarial than men are. Women’s legislative speech in Austria (Haselmayer, Dingler, and Jenny Reference Haselmayer, Dingler and Jenny2021) and the UK (Hargrave and Langengen Reference Hargrave and Langengen2020) is also less negative. Work from the United States (Holman and Mahoney Reference Holman and Mahoney2018; Kanthak and Krause Reference Kanthak and Krause2012) and abroad (Barnes Reference Barnes2016) shows that women representatives engage in more collaboration and cosponsorship. This cooperative behavior likely reflects gendered socialization processes or women’s strategic efforts to overcome marginalization within political institutions. Regardless of their motivation, if women employ more cooperative, consensual leadership styles, citizens may feel more warmly toward rival parties with more women MPs.
Second, independently of whether women representatives behave differently from men, women’s descriptive representation affects both citizens’ and journalists’ political perceptions. US-based studies show that respondents hold gender-trait stereotypes, seeing women politicians as more caring and compassionate (Bauer Reference Bauer2019), more likely to compromise and build legislative consensus (Bauer, Yong, and Krupnikov Reference Bauer, Yong and Krupnikov2017), and having better interpersonal skills (Cassese and Holman Reference Cassese and Holman2017; Holman and Mahoney Reference Holman and Mahoney2018). Citizens in Norway (Matland Reference Matland1994), Belgium (Devroe and Wauters Reference Devroe and Wauters2018), the UK (Johns and Shepard Reference Johns and Shephard2007), and Israel (Ben-Shitrit, Elad-Strenger, and Hirsch-Hoefler Reference Ben-Shitrit, Elad-Strenger and Hirsch-Hoefler2021) likewise apply gender stereotypes to politicians. These stereotypes also influence media coverage of both candidates and parties. Analyzing newspapers from Australia, Canada, and the US, Kittilson and Fridkin (Reference Kittilson and Fridkin2008) find that women candidates are disproportionately linked to “feminized” issues and traits—including honesty, compassion, and noncompetitiveness. In European Parliament elections, the media connects parties with more women MPs to compassion issues independently of the issue content of parties’ platforms (Greene and Lühiste Reference Greene and Lühiste2018).
Perhaps unsurprisingly in light of these studies, related work shows that citizens prefer institutions with more women (Clayton, O’Brien, and Piscopo Reference Clayton, Piscopo and O’Brien2019) and report greater trust and satisfaction in institutions with higher levels of women’s representation (Badas and Stauffer Reference Badas and Stauffer2022; Ben-Shitrit, Elad-Strenger, and Hirsch-Hoefler Reference Ben-Shitrit, Elad-Strenger and Hirsch-Hoefler2021; Verge, Wiesehomeier, and Espírito-Santo Reference Verge, Wiesehomeier and Espírito-Santo2020). Political parties have even sought to capitalize on these beliefs. Weeks et al. (Reference Weeks, Meguid, Kittilson and Coffé2022), for example, argue that radical-right parties strategically increase their proportion of women MPs in order to defuse their extremist image and expand their support beyond their base. We thus posit the following:
The Women MPs Affective Bonus Hypothesis: All else being equal, partisans display warmer affect toward out-parties with higher proportions of women MPs.
DATA AND ANALYSIS
To test our hypothesis, we combine an original dataset on women’s descriptive representation at the party level with survey data from the Comparative Study of Electoral Systems (CSES) for 20 Western publics and 81 election-years between 1996 and 2017. Section S1 in the appendix lists the countries, elections, and parties in our dataset. The CSES surveys include a 0–10 feeling thermometer asking respondents to rate the parties in their country, where zero denotes maximum dislike and 10 denotes maximum liking.Footnote 1 The feeling thermometer is the most common measure of out-party dislike in affective polarization research (Iyengar et al. Reference Iyengar, Lelkes, Levendusky, Malhotra and Westwood2019), and it correlates with other affective measures (Druckman and Levendusky Reference Druckman and Levendusky2019). The survey also includes a question about party identification, which we use to classify party supporters.Footnote 2
Our dependent variable, [party i’s supporters’ evaluations of out-party j (t)], is the mean thermometer score that party i’s partisans assigned to out-party j in the CSES election survey administered in the year t. We analyze party dyads because studies find that out-party evaluations respond to ideological distances between the parties and also to their governing relationships. For example, Leftist party supporters evaluate left-wing out-parties more warmly than right-wing out-parties and governing parties’ supporters award a large “affective bonus” to cogoverning out-parties independent of ideological distance (Horne, Adams, and Gidron Reference Horne, Adams and GidronForthcoming).
We analyze out-party evaluations at the party-dyad level—that is, each partisan constituency i’s mean evaluation of each out-party j in the election year t—to account for these factors. Note that each party pair i, j in each election survey enters our dataset twice, as we analyze the mean thermometer rating that party i’s partisans assign to party j and the mean rating j’s partisans assign to party i. We analyze all dyads of parties i, j with at least four MPs each in the year before the current election survey, as smaller parties are arguably less consequential for affective polarization and also pose measurement problems because few survey respondents identify with these parties. In the appendix we show that our substantive conclusions also hold in analyses that include these smaller parties (see Section S2) and when using a stacked individual-level dataset (see Section S11) where each observation is an individual’s evaluation of a given out-party such that each individual enters the data as many times as they evaluate a party.
Our primary independent variable, [out-party j’s proportion of women MPs (t – 1)], is the out-party j’s parliamentary gender composition lagged one year prior to the current CSES survey, scaled from zero (all of j’s MPs were men) to one (all were women). For example, for the 2013 German parliamentary election, data on the gender composition of political parties is taken from 2012. Figure 1 displays the distribution of the proportions of women MPs across all of the party parliamentary delegations in our study, segmented into deciles (0.0 to 0.1, 0.1 to 0.2, etc.). Between 1996 and 2017, women MPs were significantly underrepresented in Western parties’ parliamentary delegations. The mean proportion of women representatives was only 0.29 (the standard deviation was 0.16), and roughly one in three party delegations featured fewer than 20% women. Fewer than 10% were majority women.
We expect the coefficient on the [out-party j’s proportion of women MPs (t – 1)] variable to be positive, denoting that partisans evaluate out-parties with higher proportions of women MPs more warmly. We control for governing coalition arrangements and for the ideological distance between the in-party i and the out-party j. The [i, j are coalition partners (t)] dummy variable equals 1 if parties i, j were governing coalition partners at the time t of the election survey, and we also include the dummy variable [i, j are opposition partners (t)] because research shows that opposition party supporters grant an affective bonus to co-opposition parties (Horne, Adams, and Gidron Reference Horne, Adams and GidronForthcoming). The variable [elite right–left distance i, j (t)] denotes the absolute right–left distance between parties i and j in the current election, based on the Comparative Manifesto Project (CMP) coding of the Left–Right tones of the parties’ election manifestos. We standardize this variable so that the coefficient predicts the variation in out-party evaluations associated with one-standard-deviation changes in the independent variable values.Footnote 3
Because error terms plausibly correlate within elections, we use ordinary least squares with robust standard errors clustered by election. Our models include country-year fixed effects to capture unmeasured factors associated with specific countries and periods such as economic conditions, electoral laws, media systems, and so on. Thus, our parameter estimates reflect within-country and within-election differences in partisans’ ratings of different out-parties. Section S10 in the appendix displays models with alternative fixed effects specifications. We show that our results hold in models with country—rather than country-year—fixed effects, which leverage variation across elections within the same country as well as within-year variation (Section S9). The results likewise hold when including individual fixed effects in models that use individual-level data (Section S11).
We estimated our model parameters on the 1,842 directed party dyads in our dataset. This represents every pair of parties with at least four MPs each in the 20 Western party systems we study, with most pairs observed at multiple years t across the 81 CSES election studies we analyze.Footnote 4 Table 1 reports the parameter estimates for our multivariate model (column 2), along with the estimate on a reduced-form model without controls (column 1). The estimates support our women MPs affective bonus hypothesis: the estimate on the [out-party j’s proportion of women MPs (t – 1)] variable is positive and statistically significant (p < 0.01) in both models. The estimate for the full model, +1.73 in column 2, denotes that moving from zero women MPs to all women MPs in a party’s parliamentary delegation improves predicted out-party evaluations by 1.73 units on the 0–10 thermometer scale, when controlling for coalition arrangements and left–right distance. And, moving from one standard deviation below to one standard deviation above the mean value of the [out-party j’s proportion of women MPs (t – 1)] variable—that is, from a proportion of 0.13 to 0.45 out-party women MPs—improves predicted out-party evaluations by 0.55 thermometer units, about one-third of the out-party dislike variable’s standard deviation (which is 1.54 units). All else being equal, partisans evaluate out-parties with higher proportions of women MPs more warmly.
Note: The dependent variable, [party i’s supporters’ evaluations of out-party j (t)], is the average thermometer rating on a 0–10 scale that party i’s partisans assigned to the out-party j in the CSES election survey administered at time t. The ordinary least squares regression models were estimated with standard errors clustered on elections. Section S1 in the appendix lists the countries, elections, and parties in our study. *p < 0.05, **p < 0.01 (two-tailed tests).
The estimates on our control variables confirm that ideology and governing coalition arrangements also drive out-party evaluations. The negative coefficient on the [elite right–left distance i, j (t)] variable (p < 0.01) indicates that out-party evaluations cool as the left–right distance between the parties increases. The positive coefficients on the [i, j are coalition partners (t)] and [i, j are opposition partners (t)] variables (p < 0.01) denote that partisans award an affective bonus to out-parties from the same side of the aisle compared with the baseline category of a party pair consisting of one governing and one opposition party. In particular, cogoverning parties’ partisans grant their coalition partners a large affective bonus of nearly one thermometer unit on the 0–10 scale, all else being equal.
EXTENSIONS AND ROBUSTNESS CHECKS
We conducted additional analyses to assess the robustness of our results. First, we examined whether our findings hold for both men and women survey respondents. Because existing literature yields competing expectations on this front (see Clayton, O’Brien, and Piscopo Reference Clayton, Piscopo and O’Brien2019; Klar Reference Klar2018; Ondercin and Lizotte Reference Ondercin and Lizotte2021; Stauffer Reference Stauffer2021), we reestimated our models separately on all self-identified women survey respondents and then on men. These estimates, reported in columns 3 and 4 of Table 1 above, show that both women and men partisans reward out-parties with higher proportions of women MPs.
Next, we estimated models controlling for the out-party’s family to assess whether partisans dislike some out-parties for reasons beyond policy disputes and coalition arrangements. These analyses continue to support our substantive conclusions (see Section S3 in the appendix). We also considered whether these effects are related to the country’s electoral system proportionality, finding that the women MPs affective bonus is not mediated by—and continues to hold when accounting for—proportionality (see Section S6). Finally, our results hold across both the earlier and later parts of the 1996–2017 time span of our study, despite the shift across this period to greater reliance on online campaigning, where partisan hostility can be mobilized by racist and sexist memes (see Section S4).
We also analyzed whether the women MPs affective bonus is related to the representation of women MPs in the in-party and whether partisans’ tendencies to reward out-parties for representing women in parliament diminishes as the out-party’s share of women passes parity. Sections S5 and S7 of the appendix report these analyses, which again support our conclusions. Finally, we assessed whether the women MPs affective bonus differed for out-parties with women leaders versus out-parties led by men. As reported in appendix Section S8, our analyses of out-parties led by men—which constitute nearly 75% of the cases in our study—strongly support our conclusions and imply an even larger affective bonus than the estimates reported in Table 1 above. Our estimates on out-parties led by women suggest a smaller (but still detectible) women MPs affective bonus effect.
DISCUSSION AND CONCLUSION
Affective polarization is associated with social distancing, economic discrimination against out-partisans, and even partisans’ willingness to condone violence against political opponents. Yet although scholars link affective polarization to systemic factors including elite-level policy disputes, economic conditions, and electoral laws, existing work has not considered a potentially more malleable predictor: the gender composition of political parties’ elected officials. In analyses across 20 Western democracies, we find that partisans more warmly evaluate out-parties with higher proportions of women MPs.
Our findings are important for gender and politics scholars. There is increasing interest in the “symbolic effects” of women’s collective representation, including citizens’ feelings of efficacy and trust. We demonstrate that variation in women’s collective representation is also linked to out-party hostility, and we add to a small but growing literature on gender and affective polarization (Klar Reference Klar2018; Ondercin and Lizotte Reference Ondercin and Lizotte2021). Future work should examine the mechanisms driving our empirical findings, in particular whether the women MPs affective bonus stems primarily from substantive differences in women’s versus men’s leadership styles, from partisans’ gender stereotypes or preferences for descriptive representation, or from differences in how the media depicts women versus men MPs. Whereas some of these causal pathways require citizens to recognize the gender compositions of different parties’ parliamentary delegations, others do not. And, though existing work suggests that citizens are broadly aware of the gender composition of political institutions (Dolan Reference Dolan2010; Stauffer Reference Stauffer2021) and respond to the makeup of political parties (O’Brien Reference O’Brien2019), more research is needed to identify the individual-, party-, and system-level factors that predict knowledge of party gender composition.
Our results also extend the nascent comparative affective polarization literature. Most affective polarization research analyzes the U.S., and no study has asked whether who represents us influences out-party hostility. Future studies should examine whether the inclusion or exclusion of other marginalized groups is related to affective polarization. This work should also address the consequences of affective polarization for representatives, as high-profile women politicians are disproportionately targeted for violence (Håkansson Reference Håkansson2021) and uncivil messages on social media (Rheault, Rayment, and Musulan Reference Rheault, Rayment and Musulan2019). Though women MPs may provide an affective bonus to their parties, we must acknowledge the costs these legislators bear.
We likewise note that although women’s parliamentary representation has increased over the past two decades, there has been no corresponding diminution in affective polarization across Western publics (e.g., Boxell, Genzkow, and Shapiro. Reference Boxell, Genzkow and Shapiro2020; Gidron, Adams, and Horne Reference Gidron, Adams and Horne2020). Yet, this does not suggest that women’s representation is irrelevant to out-party hostility. Rather, the 1996–2017 period featured other developments that intensified cross-party hostility, including the rise of radical-right parties, the growing salience of cultural issues relating to multiculturalism and national identity, and economic shocks including a global recession (see Gidron, Adams, and Horne Reference Gidron, Adams and Horne2020). In the absence of women representatives, we may have observed even higher levels of affective polarization.
Finally, we acknowledge the mixed normative implications of our findings. On the one hand, our results are promising for practitioners who are seeking to ameliorate affective polarization, as they suggest that parties can “do well by doing good.” By nominating and electing more women MPs, parties can broaden their electoral appeal and defuse affective polarization while also providing better descriptive gender representation. On the other hand, some parties may use this effect strategically. For instance, populist radical-right parties—who are strongly disliked by mainstream partisans (Harteveld, Mendoza, and Rooduijn Reference Harteveld, Mendoza and Rooduijn2021; Helbling and Junkunz Reference Helbling and Jungkunz2020; Reiljan and Ryan Reference Reiljan and Ryan2021)—may use women’s representation to enhance their affective standing in the general public (see Weeks et al. Reference Weeks, Meguid, Kittilson and Coffé2022). Given that opposition to extremist parties—particularly those that promote illiberal, antidemocratic stances—is arguably justified, it is troubling if these parties can defuse this hostility simply by promoting women. These normative implications of the women MPs affective bonus, both negative and positive, suggest that this is an important area for future research.
To view supplementary material for this article, please visit http://doi.org/10.1017/S0003055422000491.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/AHQRVR.
Earlier drafts of this work were presented at the University of Liverpool and the 2021 American Political Science Association Conference. We would like to thank our discussants, fellow panelists, and audience members for their useful suggestions. We would also like to thank Leslie Schwindt-Bayer and Jonathan Homola for their excellent feedback on early versions of the paper and Rachel Bernhard for her advice on a later draft.
CONFLICT OF INTEREST
The authors declare no ethical issues or conflicts of interest in this research.
The authors affirm this research did not involve human subjects.