Risk Aversion, Over-Confidence and Private Information as Determinants of Majority Thresholds.

We present and experimentally test a theoretical model of majority threshold determination as a function of voters' risk preferences. The experimental results confirm the theoretical prediction of a positive correlation between a voter's risk aversion and the corresponding preferred majority threshold. Furthermore, the experimental results show that a voter's optimal majority threshold negatively relates to the voter's confidence about how others will vote. Moreover, in a treatment in which individuals receive a private signal about others' voting behavior, the private signal tends to replace confidence.