The hazard of initiating alcohol use was estimated using survival analysis (Cox proportional hazard models; Cox, 1972; Singer & Willett, 2003), using proc phreg in SAS. Survival analyses handle right censoring due to failure to consume a full drink because of attrition or study end. We handled ties in the dataset using ties=exact, which computes the exact conditional probability under the assumption that tied event times occur before censored times of the same value or before larger values.
