In this paper the conventional wisdom about the electoral impact of party primaries is revisited. We have created an original dataset with data from 289 regional elections in Canada, Germany and Spain since the 1990s. Our argument is that the electoral impact of primaries depends on whether the two main parties, only one of them, or none of them select their top candidates by this new selection procedure for a given election. Our results show that selecting candidates by primaries does not have any effect no matter what the rival does; however, parties not using primaries to select their top candidates are punished when their main competitor does. We conclude that citizens do not reward parties for using primaries, but they do punish them if they do not use primaries and the rival party does.
