A study of outcome reporting bias using gender differences in risk attitudes

There is a strong consensus in the experimental literature according to which women are more risk averse than men. However, new evidence reveals that only a tiny fraction of the replications of the Holt and Laury (2002) risk elicitation task displays gender differences. This striking distance between the consensus and the data gathered with this elicitation method provides a clean test of the presence of an \textit{outcome reporting bias} in the risk and gender literature. \noindent Exploiting a large dataset of replications of Holt and Laury (2002) we find no evidence that the likelihood of reporting about gender differences is affected by obtaining results in line or against the consensus. Two variables significantly increase the probability of describing results along a gender dimension: the share of women among the authors and the fact that the study focuses directly on risk preferences. Both variables, however, are orthogonal to the results being in line with the consensus or not, confirming the absence of any outcome reporting bias.