A study using classroom-level data finds students are more likely to apply to college programs when classmates apply, suggesting information-driven peer contagion rather than admission success effects.
Researchers Ana M. Gazmuri and Elena Prager have taken a close look at a question that sits at the intersection of education and behavioral economics: how much do students’ college-application decisions respond to what their peers do?
In their study, the authors examine whether simply being exposed to classmates who apply to particular college programs changes an individual student’s likelihood of applying. The central challenge in identifying “peer effects” is separating true social influence from shared circumstances—such as a guidance counselor’s advice, a local economic shock, or a school-wide initiative—that could move many students in the same direction at once. To reduce these concerns, Gazmuri and Prager use a comparison that leverages neighbors who attend different high schools, a design intended to limit the role of common school-level shocks while still capturing meaningful peer exposure.
The analysis draws on detailed, classroom-level administrative data, allowing the researchers to observe application patterns at a fine level of granularity. The results point to a clear pattern: students are more likely to apply to a program when they are exposed to peers who have applied to that program. In other words, applications appear to be socially contagious, at least in the sense that one student’s decision to submit an application is associated with others becoming more likely to do the same.
Notably, the study finds that the influence seems tied to the act of applying—not the outcome. Exposure to peers who are admitted does not generate an additional boost beyond the effect of being exposed to peers who applied. The authors interpret this as consistent with an information-based channel: seeing peers apply may reveal that a program exists, clarify perceived prerequisites, demystify the process, or update beliefs about the feasibility or desirability of applying. If admissions outcomes carried extra influence, one might expect a distinct effect from seeing peers succeed; the absence of that extra effect points away from a purely “role model success” mechanism and toward information diffusion.
Beyond documenting these patterns, the paper also tackles a technical problem: modeling how students choose an “application portfolio,” or bundle of programs to apply to. In general, optimizing such a portfolio can be computationally intractable (the authors note that the problem is typically NP-hard). They make progress by focusing on a setting where admission probabilities are observable and can be treated as conditionally independent, which helps render the portfolio-choice problem workable for counterfactual exercises.
Those counterfactuals suggest the peer-driven shift in application behavior could be large in policy-relevant terms. The estimated effect is strong enough, in their simulations, to eliminate both urban–rural gaps and gender gaps in applications to selective college programs. While these are model-based exercises rather than direct policy trials, they indicate that interventions affecting what students learn from peers—or the information that peers transmit—could meaningfully reshape who applies to competitive programs.
The authors acknowledge a range of colleagues and seminar participants for feedback and note multiple sources of research funding, including support connected to the French National Research Agency’s Investments for the Future program, an ERC/EU grant (agreement No. 101162723 — EDDYNCHOICE), and the Washington Center for Equitable Growth. They also credit research assistance from Diego Fica and support for secure data work from staff at McMaster University’s Secure Empirical Analysis Lab (SEAL). The paper emphasizes that the conclusions are those of the authors and do not necessarily represent the views of OUAC, Ontario universities, funders, or the National Bureau of Economic Research.
Why it matters
College application behavior is a key gateway to economic opportunity, especially for selective programs that can shape long-run earnings and mobility. If information spreads through peer networks—rather than admissions success per se—then relatively low-cost steps that replicate that information (clearer guidance, targeted outreach, structured peer mentoring) may narrow persistent gaps in who applies. The next question is whether policy can reliably trigger these peer-like information effects at scale without widening inequalities in access to well-connected networks.