Journal Article
Do disadvantaged students benefit from attending classes with more skilled colleagues?

Using two rich administrative data sets and a rule of admission at one top university in Brazil that splits students into two classes, we apply a regression discontinuity design to study the effect of class allocation on academic performance and labor market outcomes. The last student of the first class will have higher-ability peers but a lower ordinal rank than the first student of the second class. 

These effects usually play in different directions. The main results suggest that the academic outcomes of affirmative action students in technology and health sciences majors are negatively impacted by being the last students in the first class. However, this negative effect does not translate into the labor market outcomes.