Estimates of genetic and environmental influences on diversity of substances used were computed in Mx using a multivariate Cholesky model. The model was fit to a log-transformed diversity of substances used [ln(substances+1)] due to the positively-skewed nature of the raw data. The overall model fit the transformed data significantly better than those models in which age, sex, or both were set as equal (p<0.0001 for all, compared on a chi-squared distribution, see Table 2). As shown in Figure 2, population variance attributable to additive genetic factors increased with age and generally had a greater effect on males than on females across the ages. Conversely, both shared and unique environmental effects decreased across age and were more influential in females than in males.
