Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Using data for students undertaking a series of real-world academic examinations with high future payoffs, we examine whether the differences in these evaluations' competitive nature generate a performance gender gap. In the univariate setting we find that women's performance is first-order stochastically dominated by that of men when the competition is higher, whereas the reverse holds true in the less competitive or noncompetitive tests. These results are confirmed in the multivariate setting. Our findings, from a real-world setting with important payoffs at stake, are in line with the evidence from experimental research that finds that females tend to perform worse in more competitive contexts.