While the ALSPAC dataset included several variables important to the evaluation of an LR-based model of drinking behaviors, some key influences were not measured. The first involves the expectations of the effects of alcohol (EXPECT), a domain shown in prior studies to be an additional mediator of the relationship of LR to ALCOUT in older teens [10]. Unfortunately, most standardized measures of the EXPECT are more time consuming than could be accommodated during the ALSPAC evaluations. The second absent domain of potential interest is the family history of AUDs. While these data are available for the biological mothers of the ALSPAC offspring, family history had not been systematically evaluated in the fathers, making it difficult to establish a reliable family domain to incorporate into the model. Finally, the ALSPAC protocol did not include a measure of using alcohol to cope with stress (COPE), a domain that performed well in the model in older populations [12].
