Diagnostic symptoms of these disorders were assigned using DSM-III-R criteria by a team of at least two advanced clinical graduate students during a case conference. Symptoms were entered into a database and computer algorithms (based on DSM-III-R criteria) were employed to produce study diagnoses. For adolescents, a symptom of a disorder was counted toward the diagnosis if either the child or the parent informant endorsed it at threshold level as is typical in a best-estimate strategy (Bird, Gould, & Staghezza, 1992). These procedures produced diagnostic reliability coefficients (kappa) of 1.00 for alcohol use disorders, .84 for ADHD, .75 for CD, and .71 for ODD. To facilitate the analytical plan, symptom counts (sum of the number of symptoms met at threshold level) were tabulated for alcohol dependence, ADHD, and a combined variable for CD/ODD (sum of symptoms of both disorders). This latter variable reflects the fact that ODD was not controlled for when assessing CD. More importantly, latent variable research by Fergusson et al. (1994), using a community sample, and Hewitt et al. (1997), using a twin sample, have provided support
