The effects of Greek affiliation on academic performance

We use a difference-in-differences approach to estimate the effects of Greek affiliation on academic performance. There are strong negative effects in some periods but smaller effects in others: fraternity affiliation hurts performance by 0.32 standard deviations in the Freshman Spring; sorority affiliation hurts performance by 0.22 standard deviations in Spring semesters after Freshman year. We estimate both ceteris-paribus effects and non-ceteris-paribus effects which allow Greek affiliation to influence course choice behavior. We account for censoring of grades and show ignoring censoring leads to attenuation bias. We also document heterogeneity in treatment effects by student preparation and organization social status.