The data were analyzed via SEM in Mplus 5.2 (Muthén & Muthén, 2009) using the Weighted Least Squares Means- and Variance-Adjusted (WLSMV) estimator, due to the categorical nature of some of the selected indicators. Model fit was evaluated using the comparative fit index (CFI) and the root mean square error of approximation (RMSEA). School was specified as a clustering variable. Prior to analysis, adverse consequences, alcohol problems, early conduct problems, and early depressed mood were log transformed to help normalize their distributions. After transformation, skewness across all variables ranged from .03 to 3.5. Missing data procedures incorporated into Mplus resulted in a full analysis sample of 208 (control) cases.
