The Role of University Policies and Practices in Student Behavior and Achievement

This dissertation examines the roles that three university-level policies or practices play in student behavior and decision making. The first chapter examines the practice of providing letter grades to students near the middle of the semester. Many institutions now give first-year students midterm grade reports to allow students to alter their behavior before the official end of course grades are entered onto their transcript. I use a regression discontinuity design to provide the first causal evidence on how students respond to information regarding their midterm grades. To do so, I merge institutional administrative data with information I collected on detailed intra-course grades for approximately six thousand economics and physics students. Consistent with the related literature, I find that lower course grades in economics and physics appear to discourage women and cause them to perform worse in the remainder of the semester. There is no similar effect among men. The second chapter studies the influence of roommate assignment on student academic performance. Many applied microeconomists have used randomly assigned college roommates to study peer effects. To date, almost all of these studies have examined students in contexts that are not very representative of the average U.S. college student, despite an emerging consensus from the literature that peer effects are highly context-specific. In this chapter, I study roommates at a large, public, urban university that is arguably more representative. Using a linear-in-means model, I find no evidence or peer effects at relatively precise levels, which is consistent with much of the extant research on the topic. Using non-linear modeling of peer effects, I find no strong evidence that either a student’s or roommate’s position in the pre-college academic ability distribution matter for the magnitude of the peer effect – a departure from many of the findings in past work. The final chapter examines the effects of academic probation. Specifically, we use a regression discontinuity design to study how academic probation affects outcomes and course-taking behaviors at a large public university in the US. Consistent with past work, students placed on probation improve their GPA in the subsequent semester. We document that part of this GPA improvement is attributable to strategic course-taking, and there is significant heterogeneity in these behaviors across race. Non-minority students placed on probation attempt fewer credits, easier courses, and are more likely to withdraw from courses in the following term. In contrast, underrepresented minorities exhibit few of these behaviors, consistent with past work that suggests black and Hispanic students are less likely to possess helpful institutional knowledge and use available support systems such as academic counseling.