mutually influence each other and lead to a self-reinforcing process. Finally, the association could be non-causal and merely reflect confounding factors influencing both educational and substance use outcomes. For example, familial background may be related both to education and substance use and could create a spurious association between the two (Patrick et al., 2012; Melotti et al., 2011). Similarly, externalizing behavior problems in adolescence could confound the association (Breslau et al., 2009).
