College dropout rates are quite high, and at least partially offset the recent trend of increases in college enrollments in the U.S. First-generation students, those attending college for the first time in their families, experience particularly high dropout rates. This is troubling because college degrees facilitate upward socioeconomic mobility, yet these students who may benefit from college the most fail to obtain these rewards.
First-generation students are especially vulnerable to dropout during their first year of college, as they often struggle to acclimate to the college setting and lack access to a de facto college counselor at home (i.e., a parent with postsecondary experience). Accordingly, many postsecondary institutions have instituted various forms of advising and mentoring programs, with the aim of increasing student persistence in higher education, with a focus on the most vulnerable students.
However, evidence on the efficacy of such programs is mixed. One likely reason for the mixed results is that many early studies failed to account for selection into advising and support services. For example, if struggling students are more likely to utilize support services, naïve analyses will document a spurious negative correlation between take-up of support services and student outcomes. Another potential explanation of the mixed results that applies even to rigorous experimental studies is the relatively low take-up rates of voluntary support services, particularly among male students. That men are less engaged with campus support services is consistent with broader trends in gender gaps in educational attainment. Another understudied question, then, is what policy levers are available to university administrators and policy makers that might increase student engagement with advising and mentoring programs. We address both questions in a recent working paper that was partially supported by the W.E. Upjohn Institute.
The setting for our study was American University, a selective, midsized, private university in Washington, DC [Disclaimer: we work there]. The university sought to address these dropout concerns by launching an innovative peer-advising program about five years ago. The paid, trained peer advisors are high-achieving, third- and fourth-year undergraduate students at the university who were selected through a competitive application process. Specifically, peer advisors are tasked with offering socio-emotional support and information about the array of specialized, formal support services on campus. The program aims to serve as a gateway to the formal support services on campus by providing a friendly, low-stakes, informal venue for first-year students to voice concerns, ask for help, and become better acquainted with the campus and university. The idea is that students who are uncomfortable discussing their concerns and insecurities with faculty members or professional “adult” advisors might be more comfortable discussing such matters with peers who are closer in age and have successfully navigated social and academic life on campus.
We evaluate the program’s effects on students’ persistence at the university, as well as the malleable factors that promote student engagement with the program. Key to our analysis is the fact that students were conditionally randomly assigned to peer advisors, which created a natural experiment in which some students were randomly matched to a same-sex peer advisor, and others were not.
Motivated by recent research indicating that student-teacher demographic match increases student attendance and office-hours visits, we expect that males assigned a male peer advisor will be more likely to use the peer-advising services than males assigned a female peer advisor. Indeed, this is exactly what we find: males assigned a male peer advisor were 16.5 percentage points (38 percent) more likely to engage with the peer advisor than males assigned a female peer advisor. This is a large effect, particularly given the light touch of the intervention, which simply replaced female signatures with male signatures in emailed invitations to meet with a peer advisor. Importantly for college administrators, providing same-sex (or choice in) peer advisors is a low-cost, scalable, uncontroversial policy that can dramatically increase male students’ engagement with campus support services. This is important, since male students tend to be less academically engaged than female students on college campuses.
Of course, this finding does not speak to the efficacy of the peer-advising program. To address this question, we use data on students’ academic performance (GPA) and whether students returned to the university for their second year of postsecondary education. We find no evidence that meeting with the assigned peer advisor affected academic performance, which is perhaps unsurprising given the program’s focus on non-academic outcomes such as acclimating to campus. However, we do find large (11 percent), statistically significant effects of being assigned a male peer advisor on male students’ likelihood of persisting into the second year. It is hard to imagine a channel through which the sex of the randomly assigned name attached to an email invitation to the peer-advising program would affect persistence other than by causing students to use the peer-advising program, so this result provides strong evidence of a causal effect of the peer-advising program on student retention. The implied effect of actually meeting with a peer advisor is even larger, since not all students assigned a same-sex peer advisor did, in fact, meet with the advisor.
These results provide causal evidence of (i) a malleable policy lever that can increase male students’ engagement with university support services and (ii) that engagement with the peer-advising program can increase retention rates. These findings suggest that thoughtfully designed peer-advising programs are a low-cost, potentially useful way to increase student retention in postsecondary settings. Though the program was not targeted exclusively to first-generation students, such students shared in the gains, and offering the program to all first-year students may have reduced the potential stigma of meeting with the peer advisors. Another question is why being assigned a same-sex peer advisor did not affect female students’ take-up of the program. One possibility is that the majority of students at AU are female and the sex-match of the assigned peer advisor was made more salient by males’ minority status. Moving forward, it would be useful to similarly evaluate these types of programs in other educational contexts.
 Though the program was motivated by the needs of first-generation students, the program was offered and made available to all first-year students in the College of Arts and Sciences.
 Interestingly, there is no such same-sex effect for female students.
 We will investigate graduation rates in future work, once such data are available for all cohorts. We are also currently investigating whether engagement with the peer-advising program caused students to change major.
 An important caveat here is that since we identify the causal effect of the peer-advising program using random variation caused by the sex of randomly assigned peer advisors, our causal claim about the program’s efficacy only extends to students who were motivated to meet the peer advisor because it was a same-sex peer advisor, but would not have done so otherwise (i.e., the compliers). Hence the LATE (Local Average Treatment Effect) pun in the title of our paper.