Several longitudinal studies have been conducted with results suggesting both that early substance use predicts poor school achievement and that academic failure is predictive of substance use (Ellickson et al., 2001; Hayatbakhsh et al., 2011; King et al., 2006; Pitkänen et al., 2008). More informative than studies that simply test the association in one direction are longitudinal analyses which test the predictive associations simultaneously in both directions while adjusting for the developmental stability of academic and substance use behaviors (Beal and Crockett, 2010; Maggs and Schulenberg, 2005).Cross-lagged path analysis of longitudinal data allows this by modeling autoregressive and cross-lagged associations simultaneously, but such analyses are rare in the literature, and even rarer are studies involving more than two time points.
