Panel Paper: Gender, Peer Advising, and College Success

Monday, June 13, 2016 : 12:10 PM
Clement House, 3rd Floor, Room 02 (London School of Economics)

*Names in bold indicate Presenter

Seth Gershenson and Jimmy Ellis, American University
Large, persistent socio-demographic gaps in college attendance, persistence, and graduation rates are troubling, as educational attainment, particularly postsecondary education, improves a variety of socioeconomic outcomes and facilitates upward socioeconomic mobility. The gender gap in college completion, which has reversed over the past 25 years and now significantly favors females, is particularly interesting since males and females live in the same households, attend the same K-12 schools, and have access to similar post-secondary resources. Compared to the K-12 educational context, however, we know relatively little about the interventions that improve postsecondary educational outcomes once students are enrolled. The proposed paper investigates the potential for one low cost, scalable intervention to improve postsecondary student outcomes, particularly those of male students: a voluntary peer advising program.

A midsized private university offered first-year undergraduates with 1-on-1, in person academic advising during their first semester on campus. It was hoped that using peer connections to offer advice, provide information about on-campus resources, and bridge the student developmental gap from high school to college would benefit students. Peer advisors are third and fourth year students who receive extensive training on working with students and handling sensitive academic issues, must maintain a GPA of 3.0 or higher, and are paid hourly to work 8 to 10 hours per week. Each advisor is assigned to a subset of students (i.e., the advisor’s potential advisees) via an email sent to each subset of advisees that provides their advisor’s name, contact information, and brief overview of the program, which is free to students.

The advisor-student assignments are made quasi-randomly, as each advisor is assigned to a specific quartile of the distribution of alphabetically sorted students’ last names; thus students can choose whether to meet with their advisor, but not which advisor they meet with. Conditional on a student’s sex and place in the distribution of alphabetically sorted last names, then, whether students are assigned a same-sex advisor is as good as random. We leverage this quasi-random assignment mechanism to identify causal estimates of (i) the impact of being assigned a same-sex advisor on participation in the peer advising program and (ii) the impact of participation in the peer advising program on postsecondary outcomes such as retention and GPA.

Preliminary estimates of (i) find that males assigned a male advisor are about 40% more likely to meet with their advisor than females and about 60% more likely to participate than males who were assigned a female advisor. Similarly, reduced form estimates find that being assigned a same-sex advisor has significant, positive effects on both retention and GPA. Since there is no mechanism through which the gender of your randomly assigned peer advisor could affect outcomes other than through increased use of peer advising, these reduced form estimates are highly suggestive of a causal effect of peer advising on student outcomes. We will formally estimate (ii) using “same sex advisor” as an instrumental variable for participation in the advising program, thereby accounting for endogenous participation in the program.