Chapter 1 studies the change in women's college major choices in response to the dot-com crash. Although the dot-com crash had similar labor market effects for new graduates in engineering and computer science, it had different effects on who chose each major: women disproportionately left computer science, but not engineering. I investigate the mechanism behind the gender difference in reaction to the dot-com crash using administrative data on students from a four-year public university. At said university, the gender gap in grades (in favor of men) is larger in computer science than engineering. I estimate a structural model of major choice where students choose a major to maximize expected lifetime utility, conditional on grades, the labor market, and other factors. I find that if the distribution of grades had been the same in engineering and computer science, the gender difference in reaction to the dot-com crash would have been 33 to 42% smaller, suggesting that students reacted to the dot-com crash in accordance with their perceived comparative advantage. My results suggest that grades are an important component in retaining women in computer science degree programs. Universities hoping to encourage women to major in computer science should investigate the sources of gender gaps in STEM grades and work to help women improve their performance. Chapter 2 studies the change in women's college major choices induced by the introduction of male peers. Researchers have speculated gender differences in labor market decisions may originate in part from psycho-social factors such as gender norms and competition, many of which become more relevant to women when they are in more male environments. We leverage a unique setting that generated variation in women's exposure to male peers: colleges that transitioned from women-only to coeducation. At such colleges, we observe a steady decrease in the share of women majoring in STEM over the decade following the transition to coeducation. This corresponds to a 17% decrease in the share of women majoring in STEM for a 10 percentage point increase in the male share of the graduating class. Our results are driven primarily by peer rather than by role-model effects. Our results suggest that women's human capital investments are affected by the gender mix of their fellow students and have implications for gender gaps in the labor market. Chapter 3 studies long-run changes in men's and women's choices of college major over time, in particular whether a Schelling tipping pattern exists in the gender composition of college majors. I build a framework that can produce a tipping pattern in the gender composition of college majors. However, I find that no evidence of a tipping pattern in college major. By relaxing two assumptions in previous tipping models, I explain theoretically why tipping may not occur in this context. I test the modified framework and find that the lack of tipping is most likely explained by men facing only small utility costs of being in highly female majors. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com.bibliotheek.ehb.be/en-US/products/dissertations/individuals.shtml.]