Competition increases the magnitude of dishonest reporting even when controlling for reward uncertainty

Does competition increase cheating? This question has been investigated by both psychologists and economists in the past and received conflicting answers. Notably, prior experimental work compared how people behaved under competitive and non-competitive tasks that were associated with different levels of uncertainty about the reward that people would receive. We aim to experimentally disentangle the effect of competition from the effects of uncertain rewards. We conducted an incentivized, pre-registered study featuring real-time interaction between participants (N = 765). We introduce an uncertain non-competitive incentive scheme along with the certain non-competitive scheme and the (uncertain) competitive scheme. We find that competition significantly increases the magnitude (but not the prevalence) of cheating relative to both non-competitive schemes, with the effect of competition being larger when the level of uncertainty is held constant across schemes.