Jumping on the bandwagon and off the Titanic: An experimental study of turnout in two-tier voting

We experimentally study voter turnout in two-tier elections when the electorate consists of multiple groups, such as states. Votes are aggregated within the groups by the winner-take-all rule or the proportional rule, and the group-level decisions are combined to determine the winner. We observe that, compared with the theoretical prediction, turnout is significantly lower in the minority camp (the Titanic effect) and significantly higher in the majority camp (the behavioral bandwagon effect), and these effects are stronger under the proportional rule than under the winner-take-all rule. As a result, the distribution of voter welfare becomes more unequal than theoretically predicted, and this welfare effect is stronger under the proportional rule than under the winner-take-all rule.