How do electoral rules affect women's legislative representation? Nearly every national legislature in the world underrepresents women compared to their proportion in the voting population, but the extent to which they are underrepresented (or that men are overrepresented) varies remarkably (Krook Reference Krook2009; Schwindt-Bayer and Squire Reference Schwindt-Bayer and Squire2014). Explanations for the variation in women's representation draw on the cultural and socioeconomic environment in a country as well as on its electoral rules and the characteristics of its party system. Recent studies suggest that the effect of electoral rules on women's representation depends on the cultural or historical context in which elections occur (Holli and Wass Reference Holli and Wass2010; Valdini Reference Valdini2013). Such complexity makes it challenging to effectively discern the role of these different factors. In this paper we focus specifically on the role played by the openness of electoral rules and examine whether women candidates win more votes under some electoral rules than others.
Existing research suggests that electoral rules matter for the electoral success of women candidates, but evidence for how it matters is mixed. A general finding is that proportional representation electoral systems are associated with greater legislative representation than majoritarian systems.Footnote 1 Studies have begun to examine distinctions among different types of proportional representation systems to determine whether closed party lists, where the parties decide how many female candidates are put forward as well as their positions on the party list, tend to result in greater female representation than open lists, which allow for preferential voting (Luhiste Reference Luhiste2015; Thames and Williams Reference Thames and Williams2010).
As we detail below, prior research has produced mixed findings. The mixed results may be due in part to a reliance on cross-national data in which individual vote preferences, candidate behavior, and electoral system features depend on the national context. Although cross-national studies are an obvious way to try to capture the effect of different electoral rules on women's electoral success, it is difficult to be confident that all of the other factors that might determine a voter’s propensity to vote for female candidates are taken into account.Footnote 2 The approach we take here complements the cross-national studies and allows us to isolate the effect of the openness of the electoral rules from other factors that might influence the proportion of votes cast for women.
In this paper, we draw on an experiment conducted during the 2014 European Parliament (EP) elections to evaluate the effects of different proportional representation electoral systems. Participants in the experiment were given the opportunity to vote for real EP candidates in three different electoral systems: closed list, open list, and open list with panachage and cumulation. The candidates on the party lists were randomly chosen from the actual members sitting in the EP, and each participant faced a different ballot, though it was the same for each participant under the three voting rules. Thus, our protocol has two distinct features that will play a key role in our empirical strategy:
-
1. The experimental protocol randomizes the ballots faced by each subject and, in particular, the gender composition of the different party lists. This allows us to directly assess the electoral performance of the different lists, randomly varying the gender composition of the lists.
-
2. The experimental protocol allows each voter to vote under three different systems (with the same ballot), which allows us to vary the influence that the voter has over choosing candidates while holding the voter and context constant.
This experiment enables us to examine several different research questions: Do voters show a general reluctance to vote for female candidates, and, in particular, are parties hurt when they have more women on their lists? Do female candidates attract more votes under more open electoral rules? Do all voters react the same way to the openness of electoral rules, or do women and men react differently? Finally, how does ideology affect support for female candidates in each system? We use the experimental results to address each of these questions, enabling us to ascertain more precisely how electoral systems affect support for women candidates. First, we find that parties with a higher proportion of women on their lists are not penalized. Second, we find that voters are more likely to vote for women under more open electoral rules. This pattern holds for both male and female voters, though the propensity to vote for women is more pronounced for the latter. Interestingly, the effect of ideology seems to operate mainly through the gender composition of the different parties rather than any direct effect of ideology on a voter's propensity to vote for female candidates.
ELECTORAL INSTITUTIONS AND VOTING FOR WOMEN
Scholars point to a wide array of contextual and institutional factors that could explain the discrepancy between the proportion of women in the electorate and the proportion of women elected to office, addressing the many stages in the process that results in women's representation (Kittilson and Schwindt-Bayer Reference Kittilson and Schwindt-Bayer2012). What determines how women participate in politics by voting, learning about politics, becoming a candidate, becoming a member of a party's elite, or serving in office if successful at the polls? Assuming that women put themselves forward as candidates (Fox and Lawless 2005) and that parties support them by placing them on the ballot (Cheng and Tavits Reference Cheng and Tavits2011), a crucial question for understanding women's representation in legislatures becomes, under what conditions will voters vote for them? Our focus in this article relates to this critical question—that is, the effect of electoral rules on a voter’s propensity to vote for women.
Singling out the effect of electoral rules on voting behavior poses a challenge, because electoral rules can affect women's representation not only by having an impact on the behavior of voters, but also by having an impact on candidates and party elites. The most common argument about electoral rules and women's representation is that more proportional electoral rules lead to higher rates of women elected to office than is the case with majoritarian electoral rules (Kittilson and Schwindt-Bayer Reference Kittilson and Schwindt-Bayer2010). Some of the explanations for the difference in outcomes focus on the incentives of party leaders to include more women on the ballots in PR systems compared to majoritarian ones. There are at least a couple of explanations for this. One is that majoritarian, single-member districts are particularly likely to favor incumbents, and incumbents are mostly men (Fréchette, Maniquet, and Morelli Reference Fréchette, Maniquet and Morelli2008, Murray Reference Murray2012). Thus a party in a majoritarian system that wishes to protect its incumbents will be less likely to promote female candidates, at least in districts controlled by the party. In districts with larger magnitudes, parties have greater flexibility to nominate a more diverse set of candidates.
The second reason why majoritarian electoral rules might be associated with less female representation than proportional ones is because of actions taken by party elites due to their beliefs that voters tend to prefer to vote for men. As Roberts, Seawright, and Cyr (Reference Roberts, Seawright and Cyr2013, 1558) suggest, “Voters may be hesitant to choose women in head-to-head contests with men. This may lead parties to select fewer female candidates in plurality systems and fewer of the ones chosen to be elected.” Note that this explanation about the reluctance of party elites to select women as candidates does not require that voters actually be hesitant to vote for women, only that party elites believe this to be the case (Brooks Reference Brooks2013).
This anticipated reluctance to vote for women is at the heart of arguments that women can be more easily elected under closed list than under open list proportional representation (Matland Reference Matland, Ballington and Karam2005). Assuming that party elites can be convinced to place women in electable positions on the party list (whether by quota laws, placement rules, or otherwise), the advantage of a closed list system (where voters cannot affect which candidates on the list get elected) compared to an open list one (where voters can affect which candidates on the list get elected) is that voters cannot use their vote to elect the men on the list over the women (Thames and Williams Reference Thames and Williams2010). However, it is important to consider the question of whether (or under what circumstances) parties will place women in favorable positions on the lists. Schmidt (Reference Schmidt2009) suggests that a placement mandate, where women are placed in electable positions on party lists, is the institutional feature that matters most for the election of women, though placement mandates are only relevant in closed list systems. The natural tendency of a party whose leadership doubts voters’ enthusiasm for women candidates would be to deprioritize female candidates, at least in the absence of a quota requirement. Thus, closed list electoral systems would be better for female candidates only if an antifemale candidate bias exists among voters and parties are ready to place women in favorable positions on their lists in spite of the bias. If, on the contrary, voters are more favorable to women candidates than the party leadership is, open list systems would be better for female candidates.
This raises an empirical question. Is there evidence that voters are hesitant to vote for women? Many studies examining gender-based voting in majoritarian systems have argued that female candidates do as well as male candidates, suggesting that voters are as willing to vote for women as for men (Fulton Reference Fulton2014; Lawless and Pearson Reference Lawless and Pearson2008; Welch and Studlar Reference Welch and Studlar1988).Footnote 3 However, Schwindt-Bayer, Malecki, and Crisp (Reference Schwindt-Bayer, Malecki and Crisp2010) examine the electoral success of women in three countries that use the single-transferable vote type of proportional representation and find that the results vary by national context due to political culture. Holding the effect of electoral rules constant, and controlling for factors such as candidate strength, party characteristics, and district competitiveness, they find that being a female candidate is an advantage in Australian (Senate) elections, a disadvantage in Ireland, and that it is inconsequential in Malta. More generally, Thames and Williams (Reference Thames and Williams2010) examine the impact of electoral rules that allow voters to vote for candidates rather than parties across a wide set of countries and find that countries that use “party-centered” systems, such as closed list PR, encourage greater women's representation than countries that use “candidate-centered” systems, such as open list PR.Footnote 4 This suggests that a bias against female candidates may exist, but a recent study of the effects of electoral rules on the representation of women in the European Parliament finds that the level of openness of the electoral rules has no effect and argues that cleavage structures and party systems at the national level are the key determinants of women's electoral success (Fortin-Rittberger and Rittberger Reference Fortin-Rittberger and Rittberger2014). With respect to studies linking larger district magnitudes to the election of more women, Schmidt (Reference Schmidt2009, 193) cautions that larger districts tend to be in urban areas that are relatively affluent and are thus likely to be districts where cultural attitudes might favor voting for female candidates. In sum, existing empirical work on the electoral institutions that encourage voters to vote for women is mixed.
To better understand electoral outcomes, it is important to focus on the voters themselves—is there evidence of a propensity to vote for women under some circumstances? In other words, is there a demand for women representatives, and, if so, from whom? Sanbonmatsu (Reference Sanbonmatsu2002) argues that voters have “baseline gender preferences” that affect their vote choice. She finds that citizens in the United States have an “affinity” toward candidates of their own gender. And, although she cautions that voter gender is only part of the explanation for this affinity, she also discusses a wide range of studies (again, in the American context) that show gender effects in voting (see, e.g., Plutzer and Zipp Reference Plutzer and Zipp1996). Outside of the United States, patterns of same-gender voting have been found in Finland (Giger et al. Reference Giger, Maria Holli, Lefkofridi and Wass2014; Holli and Wass Reference Holli and Wass2010). Thus, from this literature, one might suspect that same-gender affinity voting would be most obvious in open list electoral systems. If both men and women prefer candidates of their own gender, then, assuming equal turnout, the effect of open list electoral systems for electing women should depend on the relative proportions of men and women in the electorate, and on the relative strength of same-gender bias among male and female voters.
A common theme in the studies examining electoral outcomes and/or voter preferences is that the effects of electoral rules on the electoral success of women depend on characteristics of parties, political culture, candidate characteristics, voter partisanship, and other contextual factors. However, the cross-national analyses typically used to examine this issue all confront the challenge of adequately isolating the effects of electoral rules on voting behavior from these other factors. With observational data, one has to worry about what other differences between voters in countries A and B, aside from the electoral rules, are systematically affecting the propensity to vote for female candidates. Are parties more likely to act as “gatekeepers,” keeping women away from viable positions on party lists, in some countries than in others (Cheng and Tavits Reference Cheng and Tavits2011, Luhiste Reference Luhiste2015)? Is the legislature more powerful in one country than another (Schwindt-Bayer and Squire Reference Schwindt-Bayer and Squire2014)? While not denying the importance of contextual factors, we think that it is worth reexamining the question of the effects of electoral rules on a voter’s propensity to vote for women candidates using a different approach. If we could have the same voters vote according to different electoral rules, we would be able to compare the effect of the electoral rules themselves, holding constant other contextual factors that might be relevant to a particular voter's situation. Using the experimental approach described below allows us to do just that. Because we can hold voter preferences constant across three different sets of voting rules, we are able to more easily evaluate the extent to which female candidates are more or less disadvantaged by the electoral system itself.
We use our experimental data to evaluate some of the claims and expectations put forward in existing research. In particular, we explore four research questions.
Research Question 1
Does having more women on a party list pose a disadvantage to a party's electoral performance?
This question draws upon the general expectation, discussed above, that women candidates are an impediment to parties at the voting booth. Existing evidence suggests that this is a false expectation, but such research has been conducted within single countries or cross-nationally. We wish to replicate these important findings with our original experimental design as a starting point for our examination of electoral system effects. The experimental protocol randomized the ballots faced by each subject and, in particular, the gender composition of the different party lists. The data therefore allow us to measure the impact of the gender composition of a party list on its electoral performance, thus providing a clear test of the existence of a general bias against lists with many female candidates.
Note that if such a general bias were to exist, one could expect a moderating effect of the electoral system on the relationship between the gender composition of the list and its electoral performance. In a recent study, Blumenau et al. (Reference Blumenau, Eggers, Hangartner and Hixn.d.) show that mainstream parties are more successful under open list PR than under closed list PR because voters who have an extreme preference over a certain “niche” issue do not have to vote for a niche party to express it. Under open list PR, they can express this position by giving a positive preference vote for a similarly extreme individual candidate on a mainstream party list. If one were to apply the same line of reasoning to preferences for women candidates, one would expect the effect of the gender composition of a party list to be larger under closed list PR.
Research Question 2
Does the openness of electoral rules affect the propensity to vote for female candidates? In other words, how do voters respond to an open list?
This question addresses the heart of the debate over female representation and electoral systems. The findings of observational research on open versus closed list proportional representation electoral systems are mixed. On the one hand, Shugart (Reference Shugart, Rule and Zimmerman1994) and Taagepera (Reference Taagepera, Rule and Zimmerman1994) argue that open lists and large district magnitude are best for female candidates; in such circumstances groups can coordinate to increase female candidates’ chances of being elected. On the other hand, Jones (Reference Jones1998) and Jones and Navia (Reference Jones and Navia1999) contend that closed lists with quotas can increase female representation most significantly. Further complicating matters, Schmidt (Reference Schmidt2009) finds that only placement mandates (in closed list systems) are relevant for the election of women and notes that the “recent tendency to view closed lists as more female-friendly has coincided with the increasing use of gender quotas and placement mandates” (192). We wish to examine the issue with our experimental data, as none of the other studies have isolated the behavior of voters in reaction to different ballots. We capitalize on the fact that each voter in the experiment voted under three different voting rules: closed list, open list, and open list with panachage and cumulation, facing the same ballot under the three rules. This means that we can test whether individuals adjusted their behavior in response to the different rules, while holding their preferences, the gender composition of the lists, and the context of the election, constant.
Research Question 3
Do voters show same-gender voting patterns?
Research on gender affinity in voting (women voting for women and men voting for men) suggests that it is more likely to occur when elections are candidate-centered (as opposed to party-centered) and when voters lack strong partisan identification (Holli and Wass Reference Holli and Wass2010, Plutzer and Zipp Reference Plutzer and Zipp1996). Potential explanations for gender affinity voting, which tend to focus on reasons why women would vote for women, include gender consciousness, a desire for descriptive representation, and an expectation of shared policy interests (Goodyear-Grant and Croskill Reference Goodyear-Grant and Croskill2011). To date, most analyses of same-gender voting examine countries with majoritarian electoral systems; in one of the few studies in a proportional representation context, Giger et al. (Reference Giger, Maria Holli, Lefkofridi and Wass2014) find evidence of a “gender gap” in gender affinity voting, whereby women are less likely to engage in same-gender voting than men, though they show that this gap varies according to contextual features such as district magnitude or the proportion of men and women on the party lists. While scholars have demonstrated substantial variation in the existence and strength of gender affinity voting, the literature suggests that we should not expect women to be less likely to vote for women, or men to be less likely to vote for men. At most, they might be equally likely to vote for a candidate of either gender. We wish to assess whether our voters exhibit gender affinity voting, as this is central to understanding whether women candidates are disadvantaged, or pose a disadvantage for their party, in the voting booth. We are also interested in studying whether same-gender voting patterns vary across electoral systems. We again use the fact that each voter in the experiment voted under three different voting rules, facing the same ballot, to address this hypothesis.
Research Question 4
Does ideology affect the propensity to vote for female candidates?
Our last research question draws upon existing literature that points to the relationship between left-wing parties and women's representation (Caul Reference Caul1999; Kittilson Reference Kittilson2006; Salmond Reference Salmond2006), the role of egalitarian attitudes about political leadership (Norris and Inglehart Reference Norris and Inglehart2001), and the importance of gender ideology (Paxton and Kunovich Reference Paxton and Kunovich2003). Extending this further, with the logic that those on the left of the ideological spectrum may be more ideologically predisposed to support gender equity, we examine whether left-leaning voters are more inclined to vote for women than right-leaning voters. If such a bias were to exist, it would provide some insight into the distribution of female representation around the world and across parties.
DATA AND METHODOLOGY
The data used in this paper are drawn from the EuroVotePlus project.Footnote 5 This project was run between May 4 and May 26, 2014, through an online platform, as part of the Making Electoral Democracy Work research project (Blais Reference Blais2010). The study was open to individuals from any country. The team in charge of the experiment advertised the website in national and local media during the three weeks preceding the EP elections. The website was available in 14 different European languages.
The purpose of the EuroVotePlus experiment was twofold. On the one hand, it sought to educate the public about different electoral systems used by some member states to elect their MEPs and to discuss some potential reforms of the EP electoral rules. The second objective of the project was to gather information about voting behavior under different electoral rules by presenting an experiment that would allow researchers to observe behavioral changes across candidate list configurations. Unlike using observational data in which voters from Country A may vote using a closed list system and voters from Country B may vote using an open list one, the researchers asked the same “voter” to use three different electoral systems.
Regarding the potential reforms, the website provided a discussion about having elections with Europe-wide party lists. Currently, each EU member state has a certain number of seats in EP elections and elects its own MEPs. Some have proposed to reform the EP electoral system by adding additional members, elected through Europe-wide party lists in a single pan-European constituency. However, which electoral system should be used for such a reform has not been established. Indeed, the diversity of electoral rules in use throughout Europe for the election of members of the European Parliament is remarkable.
A goal of the online experiment was to see how respondents would react to the opportunity to take part in hypothetical elections with Europe-wide party lists. This project provides us with ideal data for our study of voting for women, as detailed above. Participants voted under three different voting rules that are currently used to elect (national) members of the European Parliament: closed list (example: France), open list (example: Latvia), and panachage (example: Luxembourg). Visitors to the website were invited to cast ballots for these hypothetical pan-European elections.
One challenge that faced the researchers when designing these pan-European elections was to set up elections that would mimic, from the point of view of the voter, what would be realistic elections of European delegates through pan-European party lists. The solution was to allow each participant to see seven lists, corresponding to the seven political groups that were actually registered at the European Parliament at the time of the real election. Each of these experimental lists was composed of 10 candidates. The candidates on each list were randomly selected from among the real MEPs registered in the corresponding group. Randomization was done independently among participants, meaning that any two participants saw different candidates. Each candidate was presented on the ballot under his or her group affiliation, with his or her first name, surname, nationality, and official picture. Further information about the candidates was made available by having the name of each candidate linked to the candidate's official Web page on the European Parliament's website. Note that because of the way these pan-European party lists were constructed, many candidates were unfamiliar to each respondent (because they were MEPs from other countries). Although gender was not explicitly indicated on the ballot, it could be easily inferred from the pictures of the candidates and often from first names (and, in some cases, last names).
Each participant voted under the three different voting procedures, in the following order: closed list, open list, and, finally, panachage. In the closed list system, participants were asked to indicate which party list they wanted to vote for. Voters had no direct control over which candidates within the lists were elected. In the open list system, participants had to choose one party list. Then, they were able to assign 0, 1, or 2 points to each candidate within this selected list.Footnote 6 There was no constraining maximum number of votes to be assigned (the maximum for each participant was two multiplied by the number of candidates on the list). In the panachage electoral system, each participant had a total of ten points to be distributed. Instead of being constrained to a single party's list of candidates, participants were able to distribute the points among any number of party lists. However, a maximum of two votes could be assigned to any single candidate. The ballots remained the same for a given participant under the three voting rules.
After completing the three pan-European votes, respondents were asked to complete a short questionnaire with standard demographic questions (including gender) and questions about politics in general and ideology. For our analysis, we use only observations where we have results from the vote in the three pan-European elections as well as answers to the question about gender. Our sample contains 1,810 participants (see summary statistics in Table A1 in the supplementary material). Participants are self-selected; they are not a representative sample of the general European population. Females are underrepresented (27% of the respondents), and there seems to be a general left-wing bias in our sample.Footnote 7 Young respondents are overrepresented: mean age in the sample is 36 years, and the proportion of respondents aged between 18 and 30 is 47% (compared to around 20% in the general European population). Respondents are also overeducated: 42% of the respondents in the sample are still studying, and among those who report having stopped studying, the mean age for finishing studying is 25 years old. As the sample is voluntary and nonrandom, it is also likely that the respondents are more politically interested and sophisticated than the general population. We asked the respondents how interested they were in politics in general, and on a 0 to 10 scale, the mean answer is 8.1 (1,791 observations). By comparison, a postelectoral survey conducted by the European Parliament (European Parliament 2014) reveals that more than 54% of Europeans declare not being interested in politics in general. Because our focus is on whether the rules alter choices, representativeness is not required to test our research questions effectively; each respondent voted three times, holding their own preferences, attitudes, and context constant. Further, even if we assume that our sample is more politically aware than the general population, the only likely implication is that our results represent effects among those most likely to be aware of the incentives provided by electoral systems. Therefore, while we do not expect the magnitude of our effects to be replicated in the general population, we do take them to be true effects among a sophisticated subset of the population.
Because of the randomization of the ballots, there was substantial variation in the percentage of female candidates seen by the respondents. The proportion of female candidates on the ballots is approximately normally distributed, with a mean of 32% and a standard deviation of 5% (the minimum value is 0, and the maximum value 90%). Figure A1 in the supplementary material reports the distribution of the proportion of female candidates seen by participants on the ballots (among the 70 candidates on the ballot).
Table 1 reports the statistics for the ballots seen by our respondents, by party list. Note that extreme parties (on the right and left) had, on average, fewer female candidates than the other parties. In particular, the list of the Europe of Freedom and Direct Democracy Party contained, on average, less than 6% female candidates.Footnote 8 The parties with the most female candidates were the Greens/European Free Alliance (48% female candidates on average), the Social Democrats (43%), and the Alliance of Liberals and Democrats (42%). The ideological aspect of female candidate variation occurred naturally because the lists were compiled from the roster of existing MEPs; parties with fewer women MEPs would naturally provide fewer potential female candidates for our lists.
Table 1. Fraction of women on the lists, by party
Note: Groups are ranked in the order presented on the EP official website.
See http://www.europarl.europa.eu/news/en/news-room/infographics (accessed November 24, 2015).
To give an idea of the ideological positions of these different parties, Table 2 gives, for each party, the average left-right ideology of respondents who voted for that party in the closed list system. To construct Table 2, we used the 1,755 respondents who answered the question about ideology that was asked in the short questionnaire. Ideology is measured on a 0–10 scale, with higher scores indicating a position that is more to the right (see the descriptive statistics for this variable in Table A1).
Table 2. Ideology of respondents voting for each list under the closed list system, by party
Note: Groups are ranked in the order presented on the EP official website.
See http://www.europarl.europa.eu/news/en/news-room/infographics (accessed November 24, 2015).
RESULTS
We begin by considering the broad question posed in our first research question: are voters biased against lists with many female candidates? As explained above, in the experiment the candidates on each party list were randomly drawn from the party's MEPs. Because of the way the ballots were built, the number of women on a party list is statistically independent of the number of women on the other lists and of the voter's characteristics. We can use this fact to assess how each party's electoral performance was affected by the number of women present on its list. As noted above, this possible bias against women candidates has been entertained as one reason parties may avoid putting female candidates on their lists. To perform this examination, we use the proportion of women on the ballot as the independent variable and overall votes for the party as the dependent variable.
More precisely, we measure the electoral performance of a list as follows. Under the closed list and open list systems, it is the proportion of voters voting for a list. Under the panachage system, we compute for each respondent the total number of votes given to candidates of a specific list, divided by the total number of votes given across all lists. As reported in Table A2 in the supplementary material, for each party under each system, the party's average performance depends on the number of women on its list. (The note below Table A2 further describes how the electoral performance of each party is computed.) Additionally, Figure A2 in the supplementary material offers a visual representation of electoral performance for each list, under each electoral system. From the graphs in Figure A2, it seems safe to conclude that, for each party, having a higher proportion of female candidates on a list does not present any clear disadvantage (at least in our sample of respondents). Across all of the graphs, there is no evidence of a clear decline in support for the party as the proportion of women on the list increases. In no case is the final bar, with the highest proportion of women, the lowest point in the series. If anything, having a higher proportion of women on the ballot may be a benefit, in particular in the case of the Greens/European Free Alliance, to a lesser extent for the Progressive Alliance of Socialists and Democrats, and, maybe more surprisingly, for the European Conservatives and Reformists, as they seem to attract more votes when more female candidates are present on their lists. We observe only small differences across voting rules.
In order to further assess the impact of the proportion of female candidates on the electoral performance of a list, we estimate the following model for each list K and each electoral system S:
$$Vote\; for\; ListK\; under\; SystemS_i = \beta _0 + \beta _1 Fraction\; Females\; ListK_i + \varepsilon _i $$
where:
-
• Vote for ListK under SystemS i is a dummy variable equal to 1 if respondent i votes for list K, when the electoral system S is either the closed list rule or the open list rule. When the electoral system S is the panachage system, Vote for ListK under SystemSi is the fraction of votes that individual i gives to list K—that is, the total number of votes given by the individual to candidates on this list, divided by the total number of votes given by the individual.
-
• Fraction Females ListK i is the fraction of female candidates on list K on the ballot seen by individual i.
We estimated 21 models in total (seven lists under three electoral systems), using a binary logit model for the closed list and open list electoral systems and an OLS model for the panachage system. Table 3 reports the coefficient for the Fraction Females ListK variable in each case, along with the associated p-value.
Table 3. Impact of the proportion of female candidates on the electoral performance of the list (proportion of votes): results of logit (closed and open systems) and OLS (panachage system) analyses (coefficient and p-value)
Note: Number of observations = 1810; *p < 0.1, **p < 0.05, ***p < 0.01.
With the exception of the Alliance of Liberals and Democrats, the sign of the coefficient on Fraction Females ListK is uniformly positive across all parties and electoral systems. This confirms that having more females on a list does not hurt the electoral prospects of a party, and this does not seem to depend on the voting system. Furthermore, the positive effect of having more females on the ballot is significant for the Progressive Alliance of Socialists and Democrats (except in the open list system) and for the Greens/European Free Alliance, as well as for the European Conservatives and Reformists in the open list system. Therefore, having women on a ballot may even be a benefit in some cases.
Our first research question related to the performance of parties/lists as a whole. Next, we turn to consider our remaining hypotheses, which address the relative performance of male and female candidates: does the individual propensity to vote for female candidates depend on the electoral system and the gender and ideology of the voter? In these analyses we focus directly on voting for women candidates, designating it as the dependent variable. Accordingly, for each system and each participant, we create a variable Votes for Women describing the respondent's support for female candidates. This variable was constructed differently for each electoral system, in such a way that the comparison across electoral systems is meaningful. In the closed list system, voters have no control over which candidates are elected, and our measure of support for female candidates is simply the proportion of females on the list that was chosen by the individual. In the open list system, participants were able to assign 0, 1, or 2 points to each candidate within the list they voted for. To create a measure of votes for women, we divide the total number of points given to female candidates by the total number of points distributed (within the chosen list). In the panachage electoral system, each participant had a total of 10 points to be distributed, possibly across lists. Our measure of support for women in this system is the total number of points given to female candidates divided by the total number of points distributed (across all lists), in the same way as for the open list electoral system. In Table 4 we report the mean value of votes for female candidates in the whole sample (in the column labeled “All”), in each type of electoral system. The last two columns in the table report the mean votes for women by male and female participants, respectively.
Table 4. Mean Votes for Women (proportion of votes), by electoral system
Note: Standard deviations in parentheses. The variable Votes for Women is constructed differently for each electoral system. In the closed list system, it is the proportion of females on the list that was chosen by the individual. In the open list system, it is the total number of points given by the individual to female candidates, divided by the total number of points distributed by the individual (within the chosen list). In the panachage electoral system, it is the total number of points given by the individual to female candidates, divided by the total number of points distributed by the individual (across all lists).
Let us first consider the results for the whole sample. The mean percentage of women candidates on the lists is 32%. Average support for women in the closed list system is 39%; under the open list system, 41%; and under panachage, 44%. These data provide evidence that more open electoral systems are associated with more votes for women, as clearly the panachage system has the most votes for women (44%). Furthermore, since the average support for women is always higher than the proportion of women candidates on the ballots, this provides further evidence that there is no systematic bias against female candidates. All the differences are significant at the 1% level. (See Table A3 in the supplementary material for reports of the associated statistical tests.)
Next we address our research question about same-gender voting. The last two columns of Table 4 demonstrate that, for both male and female voters, open electoral systems are associated with significantly more votes for women (see Table A3 for reports of the associated statistical tests). The data also demonstrate that women vote for women more than men do in the open and panachage systems, but there is no statistical difference in the closed list system.Footnote 9 This confirms the existing research that finds female affinity effects, while also providing evidence that women candidates are not disadvantaged among male voters. When given the opportunity to vote for more men under the more open electoral systems, male voters do not seem to engage in same-gender voting patterns.
Finally, we consider whether ideology moderates support for women candidates. Figure 1 shows how the propensity to vote for women depends on ideology, for each gender and each electoral system. The five ideology categories have been constructed based on the ideology variable: Extreme Left corresponds to ideology between 0 and 2 [number of males = 337 (26% of male respondents), number of females = 158 (33% of female respondents)]; Center Left to ideology equal to 3 or 4 [number of males = 472 (37%), number of females = 161 (34%)]; Center to ideology equal to 5 [number of males = 168 (13%), number of females = 158 (15%)]; Center Right to ideology equal to 6 or 7 [number of males = 209 (16%), number of females = 55 (12%)], and Extreme Right to ideology between 8 and 10 [number of males = 140 (11%), number of females = 39 (8%)].Footnote 10 The top graphs present the results for female respondents, and the bottom graphs for male respondents. The results are presented for the closed list system (left), the open list system (middle), and panachage (right).
Figure 1. Mean proportion of votes for women in the closed, open, and panachage electoral systems by voter ideology categories for female respondents (top) and male respondents (bottom).
What is immediately obvious in Figure 1 is that, regardless of gender and the electoral system, ideology is very important—the propensity to vote for women declines as one moves from the left- to the right-hand side of the ideological spectrum. The effect is especially strong for voters on the extreme right-hand side of the ideological spectrum. The graph also confirms that both male and female respondents tend to vote more for female candidates in more open systems, with a much larger effect for female voters. Interestingly, Figure 1 also shows that this last effect holds true regardless of the ideology of the respondent: voters tend to increase their support for women when one moves from a closed list system to a more open system, and they do so in proportions which do not vary significantly with ideology.
The above results on gender and ideology are quite compelling, but an additional way to test our expectations is to conduct multivariate regression analyses. To do so, we created a stacked dataset, where each respondent is entered into the dataset three times, once for each electoral system. This allows us to directly compare the effects of gender and ideology on support for female candidates across all systems. It also enables us to examine whether the effects observed in our graphical analyses continue to hold when we control for other variables as well. We present these analyses in the supplementary material. They confirm, and bring additional support to, the main claims made above. In brief, we find that the positive effect of electoral system openness on voting for women is not restricted to women voters, although the effect is greater for them. This suggests a same-gender voting effect for women as well as a propensity to support women overall. Our analysis of the role of ideology suggests that its effect is mostly driven by the different gender compositions of the party lists, rather than by voters’ own ideologies. Indeed, while we observe a large effect of ideology in the closed list system, there are no significant additional effects of ideology in the other systems (once one takes into account the closed list vote). This is consistent with the observations made about Figure 1 above.
CONCLUSION
The analyses above present us with a number of observations that allow us to answer our research questions. Regarding our first query, as to whether having more women on a party list disadvantages parties, our findings suggest it does not. Whatever the party, having a higher proportion of female candidates on its list does not hurt its electoral prospects (at least in our sample of respondents). If anything, it helps, such as in the case of the Progressive Alliance of Socialists and Democrats and Greens/European Free Alliance parties. We also observe that in more open systems, when voters are given more freedom to express their preferences about the candidates, the propensity to vote for women increases (see our second point below).
Our second research question asked whether the openness of electoral rules affects the propensity to vote for female candidates. Our results strongly suggest that it does: the openness of electoral rules does influence support for female candidates overall. The data provide clear evidence that open electoral systems are associated with more votes for women and that the panachage system brings about the most votes for women.
Third, we sought to investigate whether voters show same-gender voting patterns. The answer is yes and no, depending on the electoral system. In closed list systems there is no evidence of same-gender voting, in that male and female respondents vote for lists with similar gender characteristics. In more open electoral systems, and more specifically in the open list system with panachage and cumulation, there is evidence of a strong same-gender effect for women but not for men. We observe that both men and women alike are more supportive of female candidates in more open electoral systems, but this effect is much stronger for female voters.
Finally, our fourth research question addressed whether ideology affects the propensity to vote for female candidates. Again, the answer is yes and no. We do find that, whatever the voting rule, voters at both extremes of the political spectrum—and more particularly so those at the extreme right end—are less likely to vote for women candidates than voters with a moderate ideology. This is interesting given we anticipated that more left-wing voters might be more supportive of female candidates. We can think of two different mechanisms that might explain this finding. The first one relates to the very large differences in the proportions of female candidates across party lists, extreme parties having fewer female candidates. Given this heterogeneity in the composition of party lists, even if voters were gender-blind and voted solely based on their ideological proximity to parties, one would observe this strong negative correlation between ideological extremism and support for female candidates. A second mechanism, which could reinforce the first one, is that voters with extreme ideology, and in particular with right-wing ideology, do not like to vote for women candidates: when given a chance, they are less likely than moderate voters to vote for female candidates. It appears that, in our sample, only the first mechanism is at play. Indeed, we observe that when we compare voting under open rules to the closed list system, voters with extreme ideology increase their support for female candidates in proportions similar to those of other voters; they do not seem to exhibit any dislike of female candidates. If on average they vote for fewer female candidates, it is only because extreme parties give them less opportunity to do so.
To conclude, our results suggest that, at least in our sample, voters do not exhibit a systematic bias against female candidates. Parties are not hurt when they have more women on their lists, whatever the electoral system. Furthermore, the propensity to vote for women increases when voters are given more freedom to express their preferences about the candidates. This is true for both male and female voters, even if the effect is particularly strong for female voters. Interestingly, and contrary to some expectations, this is true independent of ideology. Our experimental results therefore suggest that concerns about voters being put off because of the presence of women on the ballot appear to be unfounded.
In trying to explain the small number of women MPs (and the low level of women in the political elite in general), two competing hypotheses that our study allows us to address are that (1) parties do not put females on their lists because of their lower chances of being elected, the latter being driven by voters’ bias against females; and (2) parties, historically composed of men, discriminate against female politicians. Our results do not support the logic behind the first hypothesis. Instead, our study suggests that it may be worthwhile to further examine, and shift the emphasis to, the second interpretation, according to which the culprit is party discrimination. This finding supports the arguments of others about the importance of ballot structure and placement mandates (Jones Reference Jones1998; Jones and Navia Reference Jones and Navia1999; Schmidt Reference Schmidt2009) and provides a more solid empirical foundation for having faith that voters are not inherently biased against female candidates.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/S1743923X16000684.
