“When women run, women win – as often as men do”. In many elections the percentage of women elected tends to mirror the percentage of women running at the same elections, and therefore the electoral phase is perceived as gender-neutral. However, even though the first empirical observation is true, this does not automatically lead to the causal conclusion. In this paper, we claim that it is not sufficient to look at the net results, since there might be patterns under the surface that are overlooked. It might be the case that the “nominators”, i.e., the political parties and the “electors”, i.e., actually do have gender-biased candidate preferences, but since they go in opposite directions they are not detected. Our hypothesis is that the nominating political parties discriminate negatively against female candidates when they rank order the candidates, but that this effect is counterbalanced by the voters since they tend to favour female candidates when they cast their preferential votes. We study a case where elections are held in a multi-party setting, where each party-list include several candidates, where these candidates are rank-ordered by the parties, and where the voters cast preferential votes for candidates. In Denmark local elections are held this way, and at the same time the electoral statistics are available split by gender. We have built a data set which includes all electoral results from the 98 municipalities in existence in Denmark. A total of 903 lists were running at the elections in 2009 and we have data for all of the 9049 candidates running – including their gender, their ranking by the party and the number of preferential votes they received at the election. This unique data set enable us to isolate the effects of the parties nomination order and the voters preferential votes, respectively.
