The literature on women's descriptive representation has looked at the debate on open and closed lists as a choice between electoral systems. This article instead focuses on whether voters or the parties are biased against female candidates. Using data from six Estonian elections, the article finds that voters are not consistently biased against female candidates and open lists do not necessarily decrease women's representation. However, unknown and non-incumbent female candidates fare significantly worse than similar men. The analysis also shows that parties do not place women in electable positions on closed lists, and closed lists do not improve women's representation.Keywords: candidate gender; voter preference; political parties; candidate selection; electoral systems; Estonia
IntroductionIn most countries, women are far less likely to hold elected office than men, and there has been virtually no change in the gender composition of parliaments in some countries since the 1970s. This is the case despite advances made in women's education, employment rates, and substantial changes in public attitudes about female leadership. What or who is still standing in the way of electing women to office? This article looks at the debate on the effect of closed and open lists on women's representation, an issue that ultimately comes down to the question of whether political parties or voters disadvantage female candidates. While previous research in this area has tested theories about list types at the country level, this article reassesses the accuracy of this theory at the individual level, that is, at the level of the actual causal mechanism.I use data from six parliamentary elections in Estonia between 1992 and 2011. The Estonian case is insightful for many reasons. First, during the 20-year span, the electoral system has largely remained unchanged and simultaneously combines both closed and open lists. The uniqueness of the system allows us to test the effect of list types on women's election while holding everything pertaining to each election constant. Second, the availability of detailed party and candidate data permits the testing of the theories at the level of the mechanism rather than observing a relationship at the country level. Third, previous research on individual-level data has focused on candidate-centred systems -mainly single member districts (SMD), but also single transferable vote (STV) -and analysed only a single point in time. The Estonian data provide the opportunity to test the effect of gender on votes, election chances and list position in a party-centred proportional system over a 20-year period.