Lifetime cannabis use was assessed with the question “Have you ever used (ever experimented even once with) marijuana or hashish?” During data collection for Sample 1, to reduce respondent burden, only 6 CUD symptoms, including two abuse symptoms1 and four dependence2 symptoms were queried. These criteria showed good sensitivity and specificity, and a high level of agreement compared to two other national surveys (Lynskey et al., 2002), indicating that this is a valid measure of CUD. While the full set of DSM-IV criteria were available for Sample 2 (i.e., 4 abuse and 6 dependence criteria, and withdrawal), we only selected the subset of 6 items that were consistently available across both samples. For cannabis involvement, participants were divided into one of four mutually exclusive groups: (0) no cannabis use (N=3,566); (1) cannabis use only (having used at least once, without endorsement of CUD symptoms; N=4,084); (2) endorsement of 1–2 CUD symptoms (N=939); and (3) endorsement of 3 or more CUD symptoms (N=994). No cannabis use served as the reference category. However, to satisfy the assumption of multivariate normality, groups 1 and 2 were combined to create a three-level measure for the twin analyses.
