How do beliefs about skill affect risky decisions

•In this paper, we use a laboratory experiment to study the causal effect of beliefs about skill on risky choices.•The paper offers an innovative experimental test that is free of strategic confounds and based on revealed preference.•Low (high) skill subjects are more (less) willing to take risks on gambles where the probabilities depend on relative skill.•This suggests that the wrong people may engage in risky activities V such as entering competitive markets or career paths V while the right people may be crowded out.•Revealed beliefs are only moderately correlated with stated beliefs and so relying only on stated beliefs may be misleading.