Designing Preference Voting

Electoral systems in which voters can cast preference votes for individual candidates within a party list are increasingly popular. To the best of our knowledge, there is no research on whether and how the scale used to evaluate candidates can affect electoral behavior and results. In this paper, we analyze data from an original voting experiment leveraging real-life political preferences and embedded in a nationally representative online survey in Austria. We show that the scale used by voters to evaluate candidates makes differences. For example, the possibility to give up to two points advantages male candidates because male voters are more likely to give 'zero points' to female candidates. Yet this pattern does not exist in the system in which voters can give positive and negative points because male voters seem reluctant to actively withdraw points from female candidates. We thus encourage constitution makers to think carefully about the design of preference voting.