compared to resiliency as a mediator of family SUD risk and substance abuse (Martel et al., 2009). However, lower resiliency was associated with higher rates of externalizing behavior, which then predicted increased rates of substance use. This is consistent with prior work indicating that low resiliency predicts externalizing behavior (Hofer et al., 2010; Martel et al., 2009), and that externalizing behavior is a strong correlate of problematic substance use (Dodge et al., 2008; Trucco et al., 2014). Accordingly, at a personality level, the risk associated with specific SLC6A2 variants may be mechanistically related to early difficulties modulating distress. It is important to note that although we have been able to establish temporal precedence, a critical factor when testing mediation, causality cannot be inferred with non-experimental data (Kraemer, Stice, Kazdin, Offord, & Kupfer, 2001). A progression from early childhood risk to more problematic behaviors in the current study does not, of itself, prove that temperament and externalizing behavior are causal risk factors in the onset of substance use. However, our findings are consistent with cascade models.
