The level of women’s parliamentary representation often increases after armed conflict, but do voters in postwar societies actually prefer female electoral candidates? We answer this question by analyzing a unique data set containing information on nearly 7,000 candidates running in three elections with preferential voting in postwar Croatia. Our analysis demonstrates that voters’ gender bias is conditional on the local electorate’s ideology and exposure to war violence, with voters of right-wing parties and voters in areas more affected by war violence being more biased against female candidates. These effects of ideology and exposure to war violence also exhibit a strong interactive relationship, suggesting that bias against women is strongest among right-wing voters in areas exposed to war violence and reversed among left-wing voters in areas exposed to war violence. Our findings highlight the need to better understand the relationship between gender, ideology, and violence in postconflict societies.

Previous studies have found similarities with presidential candidates or party leaders to be an important factor in explaining voting behaviour. However, with the exception of gender, few studies have structurally studied voter-candidate similarities in intra-party electoral competition. This study investigates the Belgian case and argues that voter-candidate similarities play a role in the decision-making process of citizens when casting preferential votes. Moreover, it investigates whether underrepresented groups, and especially women, are more guided by these voter-candidate similarities than overrepresented groups. To achieve this aim voter and candidate characteristics are modelled simultaneously. This enables an investigation of the decision-making process of voters while taking into account structural inequalities at the supply side. The results demonstrate that citizens are indeed more likely to cast preferential votes for candidates similar to themselves and that these effects are stronger for underrepresented groups. Hence, preferential voting could ultimately pave the way for better descriptive representation.

Around the turn of the century, political developments in Northern Ireland, Fiji and Papua New Guinea encouraged claims that preferential voting systems could steer polities in the direction of ‘moderate’ multi-ethnic government. Sixteen years later, we have a longer time period and larger volume of data to reassess these verdicts. This article investigates ballot transfer and party vote–seat share patterns in the seven deeply divided polities with some experience of preferential voting for legislative elections or direct presidential elections (Northern Ireland, Fiji, Papua New Guinea, Estonia, Sri Lanka, Bosnia-Herzegovina and Southern Rhodesia). We find little support for centripetalist claims that such systems encourage ‘moderate’ parties. We argue that where district magnitude is low, where voters are required to rank preferences and where ticket voting prevails, departures from vote–seat proportionality may favour ‘moderate’ parties, but such heavily engineered systems may simply advantage the larger parties or yield erratic outcomes.