The influence of familial risk, parental alcohol or drug dependence and parental depression was evaluated in separate models due to missing data considerations. Specifically, all of the offspring could be classified by risk status of the pedigree from which they were drawn. Parental alcohol or drug dependence was not always possible to determine. Some of the probands did not know the whereabouts of the co-parent of the child. Analysis of all variables in one hierarchical model would have been more parsimonious but would suffer from problems associated with missing data for the parental diagnostic classification in some cases.
