Participants in both studies completed a SSAGA interview (Bucholz et al., 1994) which collected information including demographics, psychiatric history, and symptoms and onsets of substance use disorders, conduct disorder and antisocial personality disorder. Data from phase 2 of the COGA study, collected from 1995-2000, were used, as these included a full ASPD diagnostic section administered to everyone without structured skip-outs and also obtained ages of first and last occurrence for each behavior reported. Lifetime diagnoses used for the group assignments were generated from the responses to the interview questions, using scoring algorithms that were developed specifically for the SSAGA by experienced clinicians and researchers. Separate scoring algorithms were written for DSM-3R, DSM-IV and ICD-10 classification systems. The present analyses were based on DSM-IV symptoms and disorders. The SSAGA, when used as an instrument by trained interviewers and reviewed by clinical researchers, has high levels of diagnostic reliability, with kappa estimates for alcohol dependence of .84, for other substance dependence ranging from .70-,90, and for ASPD of .70 (Bucholz et al., 1994 ). Validity as assessed using the SCAN as the
