Does greater gender diversity in peer groups benefit students? I answer this question with a large field experiment involving 3,060 students randomly assigned to 765 study groups at the start of university. Students assigned to mixed-gender groups achieve significantly higher course performance in the first year and become less likely to drop out of their study program than those in homogeneous groups. Increasing group gender diversity also improves subjective well-being in the short and longer run. However, being assigned to gender-diverse groups causes students to adopt more traditional attitudes about family gender roles; this effect persists only for men. Mechanism analyses show that gender-diverse groups have more active and inclusive social dynamics, yet exhibit more gender-specific roles: While women tend to ask questions, men tend to explain and compete. These findings highlight the powerful yet double-edged effects of gender-diverse groups.

We study the impact of online instruction with a field experiment that randomly assigns 1,344 university students to different proportions of online and in-person lectures in multiple introductory courses. Increased online proportion leaves men’s exam performance unaffected but significantly lowers women’s performance, particularly in math-intensive courses. Online instruction also reduces women’s longer-run performance and increases their study dropout. Exploring the mechanisms, we find that women exposed to more online lectures report lower course satisfaction, greater difficulty in connecting with peers, and perceive instructors as less engaging. Our findings caution policymakers that shifting toward more online instruction may disproportionally harm women.

We identify and explore a gender concealment gap when individuals have the opportunity to hide information about their performance from others. In data from two universities that allowed students to replace letter grades with "credit" on their transcripts, we find that men are substantially more likely than women to conceal grades that will harm their GPAs. The gender concealment gap persists across student traits and course features and generates inequity: the option to conceal leads to GPA gains that are 50% larger for men than for women. University data and complementary experimental evidence suggests that women may conceal less because they expect others will make worse inferences about their concealed grades.    

We evaluate the effects of teaching negotiation skills to adolescent girls in economically vulnerable compounds in Lusaka, Zambia ten years later. Treated participants complete 0.27 more years of education. Consistent with greater relational empowerment, they also begin sexual activity later, marry later, have smaller age gaps with their husbands, are less likely to report a high likelihood of having HIV, and express less traditional gender attitudes. Social benefits and costs from HIV reduction and increased education suggest the intervention generated 7.8-16.2 dollars for every dollar spent and may have paid for itself in government revenues. 

We document a new way that discrimination operates: through sequential spillover effects. Employers in an incentivized resume rating experiment evaluate a sequence of hypothetical candidates with randomly assigned characteristics. Candidates are rated worse when following white men than when following women or minorities. Exploring the mechanisms, we find that spillover effects are inversely related to direct bias. When reviewing high-quality resumes or recruiting in STEM industries, employers directly favor white men and display no spillover effect. For low-quality resumes or non-STEM industries, we find no direct bias but a strong spillover effect. Results suggest that discrimination arises in subtle ways.

Do people around us influence our personality? We investigate this question through a field experiment where we randomly assign university students to study groups. We find personality spillovers along three dimensions: students become more conscientious when assigned to conscientious peers, more open-minded when assigned to open-minded peers, and more competitive when assigned to competitive peers. We find no effects for peers' extraversion, agreeableness, or neuroticism. Our findings are consistent with students' adopting peer traits that are predictive of academic achievement. Our paper provides novel evidence on spillovers in noncognitive skills and establishes that socialization with peers affects personality development.

This paper studies how access to public pensions affects old-age support and child investment in traditional societies. Guided by predictions from an overlapping generations model, we analyze the influences of a new pension program in rural China, using a difference-in-differences approach. We find that the program crowds out transfers from working-age adults, especially men, to their elderly parents. Interestingly, the impact on child investment significantly differs by child gender. While adult parents increase educational investment in sons, their investment in daughters appears to decrease. Our findings highlight the unintended consequences of public pensions on parental investment.