The initial LR-related evaluations based on the Social Information Processing Model predicted that in a heavy-drinking society people with a low LR to alcohol and higher alcohol intake are likely to develop a bias in their social information processing that encourages them to accept heavy drinking as an expected behavior, which then enhances the probability they choose friends with a similar bias [22, 23]. Similar to elements that may contribute to other models (e.g., Deviance Prone Models), the selection of the peer group is also consistent with Peer Cluster Theory [24], and the subsequent reinforcement of heavy drinking behaviors relates to social developmental and social learning models [17, 24–28]. In our models, sex has been used as a covariate, with girls expected to have less severe alcohol problems, and, reflecting their size and alcohol metabolism, a lower number of drinks for effects [1, 29]. Age was also covaried reflecting higher risks for heavier drinking and problems with increasing age [30]. These elements of the hypothesized model are shown in Figure 1.
