Strengths of the study include the use of a large, population-based sample with data on educational achievement and substance use on four time points from early adolescence to young adulthood. The cross-lagged modeling allowed us to test the directionality of the associations while adjusting for autoregressive effects by simultaneously including educational and substance use variables as both predictors and outcomes. Standardized and uniform teacher-reported grade point averages were available in our sample, improving the validity of the educational achievement measures in adolescence. The teacher-reported measures were highly predictive of student status and educational attainment later in life.
