The unconditional hazard function in Figure 2 illustrates a gradual decrease in survival rates of alcohol onset during adolescence, with a steep decline from 14–18 years. Approximately 32% of offspring reported not having tried alcohol by the time of their most recent interview, and this is indicated in Figure 2 as a plateau after age 21. This graph, therefore, reflects the inclusion of younger subjects in the sample, a step incorporated into DTSA to allow for the use of covariate data from younger abstainers and drinkers to predict the pattern of drinking onset. However, the result is a hazard function that, reflecting the entire sample of subjects, never indicates directly that at older ages (e.g., 21, 22, etc.), all individuals at that age are drinkers. The latter requires observation of Figure 1 where the proportions of abstainers and drinkers at each age are listed. Of course, if the data are restricted to only the 100 offspring who reported a lifetime use of alcohol, the hazard function would demonstrate a steeper decline that extended to 0% abstainers in the early 20s.
