In this study, the authors use administrative data from North Carolina middle schools to estimate the impact of “delinquent” students (i.e., those with three or more suspensions in a school year) on their grade level peers (i.e., students with two or fewer suspensions). To accomplish this, they take advantage of the “wide-scale remixing” of peers that occurs when students transition from fifth grade to sixth grade to implement an “instrumental variable” approach that plausibly addresses the challenge of selection bias that is endemic in peer effects studies.
Overall, their results suggest that a 10 percent increase in exposure to suspension-worthy acts results in a .06 standard deviation decrease in math scores—or roughly half of the decrease one might expect if the average class size doubled from ten to twenty students. Somewhat surprisingly, they find little evidence that the magnitude of this effect differs depending on student or teacher characteristics.
Unfortunately, because the study relies on suspensions data to identify delinquency, the authors are unable to disentangle the effects of a student’s misbehavior from those of the school’s disciplinary response. And their results don’t tell us everything we‘d like to know about the sort of group dynamics that may be at work. For example, would the costs of delinquency be larger or smaller if delinquent students were more evenly distributed across schools and classrooms? (Unsurprisingly, the study finds that the current distribution of delinquent students is highly uneven.)
Overall, the study confirms what most teachers know instinctively: that disruption has costs for students. But that is the start of the conversation, not the end of it. As the authors put it, “We may have documented the effect of peer offenses, but it is unclear what the prescription would be to improve peer behavior. Indeed, it is not clear what the cost would be (or whether it is even feasible) to effectively deter misbehavior on a large scale.”
In a world of finite resources, such deterrence is likely to be inadequate; so it would be nice if educators and policymakers could weigh the likely benefits of transferring a dangerous or disruptive student to an alternative setting—inside or outside the school—against the likely costs of such a transfer for the student in question. This study brings us incrementally closer to that ideal. But we still have a long way to go.
SOURCE: Tom Ahn and Justin G. Trogdon, “Peer delinquency and student achievement in middle school,” Labor Economics (January 2017).
