no event, and subsequent ages were treated as missing due to having experienced the event. Right-censoring occurred when a subject did not report alcohol onset by their most recent assessment. Such subjects would have “0” scores for all ages that included up to their age at most recent interview, with any subsequent ages being considered missing with a score of “99”. The inclusion of younger subjects (15–17 years old) who have lifetime abstinence provides valuable information about how the predictor variables relate to abstinence at these earlier ages, but their inclusion also affects the form of the hazard function in Figure 2.
