Information interventions in developing countries are usually judged by whether recipients learn what they are told, not by whether they believe it. This paper provides evidence that the two can diverge. In a field experiment at an Indian business school, 970 MBA students evaluated an instructor who had taught them for several weeks, after being randomly shown either the instructor’s name or the instructor’s credentials. Disclosure raised evaluations by 0.14 standard deviations on average. Male and female students retained the credentials similarly, but for male instructors only male students credited them - male students’ rating of these instructors’ achievements rose by 0.27 points on a five-point scale while female students’ did not, and the two groups’ evaluations diverged by 0.34 standard deviations. Stated beliefs about gendered access to academic careers explain little of the gap. Because evaluations feed into hiring and promotion, information meant to make judgments rest on merit can widen gender gaps in the records that decide who advances, and an evaluation that checked only whether the information was learned would miss it.