These findings should be interpreted in the context of certain strengths and weaknesses. One strength is that the analyses are cross-reporter, from mother reports of early temperament to child reports of adolescent alcohol use; therefore, correlated rater bias is not an issue. However, a limitation of these findings is that all reports on early temperament derive from the mother. It is at least encouraging that the 2 temperament factors associated with the alcohol factors embody constructs that are stably indexed across multiple assessments and using different measures. Second, although ALSPAC is a large epidemiological cohort, with data at any one point available on >12,000 individuals, there are considerable missing data. Alcohol-related outcome data at age 15.5 were available on only 4,594 individuals. Analytic methods were used to account for missing data and make use of all available data. ALSPAC is particularly well suited to data imputation methods, which are most accurate when large amounts of auxiliary data are available for the prediction model, as is the case in ALSPAC. However, continued participation in ALSPAC is known to be correlated with
