Methods: Participants (n= 208) were Amazon Mechanical Turk (mTurk) workers screened for eligibility (e.g. ages 18-29 and at least one alcoholic beverage consumed in the past year) who also indicated they had a friend “for whom they have concerns about their substance use.” We utilized latent profile analysis to identify patterns based on four subscales (punish drinking, support sobriety, enable drinking, and withdraw from friend) of an adapted version of the Significant Other Behavior Questionnaire (SBQ) for young adults (Smith et al., 2013). To identify the ideal number of patterns in the sample, model fit indices were used (e.g., AIC, BIC, LMR-LRT, Entropy). Item probability plots of the patterns were also compared to identify the most parsimonious number of patterns fitting the data. After the selection of the patterns, posterior probability Wald chi-square tests were used to understand how patterns of peer response to a friend’s substance use related to race, gender, education, parental education, age of first onset of alcohol use, and alcohol-related problems.
Results: Although the fit indices suggested that the optimal pattern was beyond a six-class solution, the four-pattern model was selected as it reflected the most parsimonious number of patterns based on the item probability plots. Additionally, it revealed high classification accuracy (entropy = 0.85). The four-patterns were respondents who were low-all across all domains (3.3%), those with high supporting and enabling scores (6.8%), those who reported high punishing, supporting, and withdrawing (53.1%), and those high-all across all domains (36.8%). When comparing across the four-patterns, notably, more Whites, males, and parents with post-secondary education were clustered into the pattern of high supporting and enabling. This pattern also reflected higher levels of alcohol use. In contrast, the pattern with high punishing, supporting and withdrawing reflected the lowest levels of alcohol use.
Conclusion and Implications: A majority of concerned emerging adult peers do not endorse their close friend’s substance use, as evidenced by their self-reported use of non-enabling strategies when dealing with their friends. Still others report high levels of all behaviors, including enabling substance use. Prior theory suggests that enabling and confronting substance use or enabling and supporting change are mutually exclusive. However, these models were based on older cohabitating couples. When adapting couples therapy models for emerging adult peers, teaching skills to avoid confrontational (i.e., punishing) and enabling responses, while increasing social support, is important.