This study shows the feasibility and added value of novel variations of the more common analyses in the field of behavior genetics, including genetic association analyses on factor analyzed traits and mediation analyses. New questions might be answered using such techniques, providing more insight into underlying common vulnerability patterns and etiological mechanisms. However, there were some limitations to the mediation analyses, including the lack of control for family relatedness and covariates in the Principal Components Analyses and the impossibility of calculating regression weights for the associations with CADM2. Also, we used Baron and Kenny's procedure to test for mediation only for outcomes that showed a significant relationship with CADM2. 47 Technically, mediation could arise in the absence of such a relationship. Bootstrapping is a more recently developed non‐parametric method that can increase power to detect mediation. However, this approach has not yet been implemented in the area of genetic association analysis. Future research might develop techniques to tackle these limitations. In conclusion, the mediation results in this study suggest that mediation testing may be feasible, but improved statistical tools applicable to behavioral genetics need to be developed.
