not detected by widely-used approaches to the estimation of spillovers. In a field experiment using university students …

generated by changes in the classroom gender composition and those generated by changes in the behavior of students. To control … for potentially confounding unobserved characteristics of schools and students that might be correlated with peer gender … results suggest that an increase in the proportion of girls leads to a significant improvement in students' cognitive outcomes …

intensified among high achievers and in schools with more interracial contact, but non-existent among students in predominantly …