“…As shown in TableA5, there is no discernible change in the results.15 As in most fixed effect models, our estimates are still subject to time-variant confounders.16 A number of prior studies investigate the importance of female introductory instructor for female students. For instance, whileBettinger and Long (2005) use an instrumental variable approach and find mixed results for different majors,Carrell et al (2010) exploit a natural experiment and find strong support for positive effects on high-achieving female students taking STEM courses with female instructors in introductory classes.17 The sharpened q-values are 0.011, 0.027, 0.027, 0.027 for 4-year retention, 1-year retention, 2-year retention, cumulative GPA, respectively.In other words, they remain statistically significant at the 5% level.18 For the efficacy of such an intervention, seeAlesina et al (2018).19 The literature on gender spillovers hypothesizes that the positive effects of gender pairing in the supervisor-supervisee relationship on supervisee career outcomes stem from direct channels such as mentoring of supervisees by supervisors as well as indirect channels such as inspiration and role-model effects. As discussed in Sections 1 and 2, the direct (mentoring) and indirect (inspiration and role-model) channels are also the primary sources of the positive effects of gender pairing in the adviser-student relationship on student outcomes.…”