For alcohol use behaviors, the regression coefficient for EXT-GR was non-significant at age 12 (β= −0.016 [−0.035; 0.003], p-value=0.107), increased significantly (p<0.001) to a peak at age 14 (β=0.127 [0.078; 0.176], p-value<0.001), and had lower point estimates at ages 17 (β=0.063 [−0.002; 0.128], p-value=0.058) and 22 (β=0.078 [−0.030; 0.185], p-value=0.155); the differences between age 14 and later ages were non-significant. In contrast, AUD-GR started at a low value at age 12 (β= −0.018 [−0.033; −0.003], p-value=0.018) and then rose nominally from age 12-17, and reached a peak value at age 22 (β=0.190 [0.109; 0.271], p-value<0.001). Only the point estimates for ages 12 and 14 were significantly lower than the point estimates at ages 17 and 22 (p-values<0.001). [Table 3. Alcohol Specific and General Externalizing Genetic Risk Scores Predicting the Twins’ Alcohol Use Behaviors across Development]
