Multivariable regression adjustment was conducted using the generalized estimating equation to account for clustering of students within schools. Values in the cells are β-coefficients and represent the change in the outcome associated with the independent variable; P values are presented in parentheses next to the cell values. Only those outcomes that showed significant effects (P < .01) in multivariable analyses are shown in the table. Controlling variables included grade, gender, school, race, maternal education, and family income. Values in bold type are significant.