as a predictor for escalated use, our second aim is to perform a genomewide search for GVs that give rise to individual differences in age at onset. To model age at onset as a function of genotype we will apply statistical methods based on survival analysis. This approach utilizes all available information on the age at onset among initiateds and takes into account the censored nature of the observations collected in those who did not initiate at the time they were last seen (i.e., they might initiate at a later time point). The approach is expected to show superior power relative to an analysis of the “ever-never” dichotomy or an analysis restricted to those who initiated (see e.g. Kiefer et al. 2013). To our knowledge, a genomewide survival analysis of age at onset of cannabis use has not yet been reported.
