Methods: Participants included 1,991 third- and fourth-grade students nested in 131 classrooms in 27 New York City public elementary schools. Data were collected during winter and spring 2016 via teacher surveys, child surveys, and observation of classroom interactions. School-level racial/ethnic diversity was calculated based on proportions of enrollment of Hispanic, Black, White, Asian, and Other students using Simpson’s index. Child outcomes included child-reported depression, anxiety, and aggression, and teacher-reported aggression, social competence, and academic skills. Three-level hierarchical linear models were used to estimate variation in child outcomes in spring based on school diversity, controlling for winter scores, and assess whether these associations were moderated by classroom interactional quality, controlling for child gender, age, classroom type (general vs. special education), and school random assignment to intervention or control group.
Results: No main effects of school-level diversity on child outcomes were found. A trend-level interaction effect was found for child-reported aggression, in that higher diversity was associated with higher aggression for children in lower-quality classrooms, but was associated with lower aggression for children in higher-quality classrooms. Additional analyses will examine classroom-level racial/ethnic diversity as a predictor of the same child outcomes, with classroom interactional quality again assessed as a potential moderator.
Conclusions: Implications for interventions targeting teacher training to promote high-quality classroom interactions, school integration policy, and within-school classroom assignment practices will be discussed.