The Role of Noise Variance on Effort in Group Contests

We theoretically and experimentally study the effect of noise variance and prize valuation on effort in individual contests and in three different types of group contests: perfect-substitutes, best-shot, and weakest-link. For all contest types, we use the rank-order contest model, where effort determines performance together with random noise. The theoretical model for individual contests predicts that effort will increase with prize valuation but will decrease with noise variance. As expected, in the experiment, all subjects with high- and low-prize valuations decrease their efforts as noise variance rises. However, there is no effect of prize valuation on subjects’ efforts. For group contests, each group consists of two players — one with high-prize valuation, referred to as a strong player, and one with low-prize valuation, referred to as a weak player. The theoretical model predicts that both strong and weak players (with the exception of weak players in best-shot contests) will decrease their efforts with noise variance in all group contests. It also predicts that for both noise variance levels, strong players will exert higher effort than weak players in perfect-substitutes and best-shot contests, and that both players will exert similar efforts in weakest-link contests. Our experimental results show no effect of noise variance on either strong or weak subjects’ efforts in perfect-substitutes and weakest-link contests. They also show that in best-shot contests, strong and weak subjects decrease their efforts as noise variance increases. In both noise variance levels, strong subjects exert higher effort than weak subjects in best-shot contests but not in perfect-substitutes and weakest-link contests. Finally, we compare the efforts of subjects in individual and group contests; we find that subjects in all group contests exert as much effort as subjects in individual contests