The proposed path model provided a good fit to the data (see Figure 4) and study variables accounted for approximately 8% of the variance in past year marijuana use (included SNPs collectively accounted for 1.6% of the variance). The only genetic polymorphism with a direct effect was SLC6A2 rs36021; those with the AA genotype had higher rates of marijuana use (see Figure 4). Results were consistent with a mediating role of behavioral control in the association between GABRA6 rs3811995 and externalizing behavior. More specifically, the CC genotype was associated with lower rates of behavioral control. In turn, behavioral control negatively predicted externalizing behavior. This indirect effect was significant (0.54, 95% BBCI = 0.25 – 0.97) with 42.3% of the total effect of GABRA6 rs3811995 on externalizing behavior operating through behavioral control. Behavioral control also had a significant direct effect on marijuana use. The indirect effect of GABRA6 rs3811995 on marijuana use via behavioral control was significant (0.08, 95% BBCI = 0.01 – 0.18) with 62.0% of the total effect of GABRA6 rs3811995 on marijuana use operating through behavioral control.
