Attrition effects in the current study were examined using the software program Mplus (20) by regressing substance use counts on whether or not individuals had returned for later assessments (coded dichotomously as 0/1, with 1 indicating attrition from later evaluations) to examine differences between those who did and did not complete the measure at a follow-up time-point. Performing this evaluation as a regression in Mplus allowed us to control for the non-independent nature of twin data. There was no significant difference in 11-year-old substance use counts between participants who did and did not complete the 14-year-old assessment (rmales=0.15, z=1.67, p=0.09; rfemales=−0.01, z=−0.09, p=0.93) or between those 11-year-olds who did and did not complete the 17-year-old assessment (rmales=0.11, z=1.36, p=0.17; rfemales=−0.04, z=−1.00, p=0.32). There was a significant difference on 14-year-old counts in those who did and did not complete the 17-year-old follow-up (rmales=0.40, z=2.46, p=0.01; rfemales=0.41, z=2.035, p=0.04). However, the effect size, as indicated by Cohen’s d, was small (dmales=0.29, dfemales=0.26), with those who did not complete the 17-year-old assessment endorsing only slightly more substances at age 14 than those who
