The best fitting model decomposed variance in cannabis involvement into additive genetic (74%) and individual-specific environmental factors across men and women (Table 5; Supplemental Table S2). As the association between cannabis involvement and ideation did not vary by duration, SI was modeled as a binary measure with additive (9%), non-additive (33–47%) and individual-specific environmental contributions (Table 5; Supplemental Table S2). To compute covariance, A and D were combined to estimate broad heritability (G=47%; Table 5) for SI as has been previously recommended (Maciejewski et al., 2014). The phenotypic correlation between cannabis involvement and SI was primarily attributable to shared genetic factors (rG=0.47) with overlapping individual-specific environmental factors also playing a role (rE=0.21).
