Search

Introduction: Previous studies have found associations between alcohol use and having heavy-drinking social networks. This association is thought to be caused by (1) social influence, where ones social network influences his or her drinking, and (2) social selection, where an individual forms relationships with individuals who drink at similar levels. These processes are reciprocal and, when acting simultaneously, create a feedback loop with non-linear dynamics. Computer simulations have allowed complex systems to be modeled in many scientific fields; however, this method has not yet been used to study the dynamic associations between social networks and alcohol use. Method: The present study used computer simulations to model changes in drinking and social networks. Stochastic actor-based social networks were simulated using RSiena. Social ties between actors and drinking statuses evolved over time according to stochastic Markov processes. Social influence and social selection were each manipulated at three levels (none, medium, and high). Results: Correlations between actor drinking statuses and the percentage of heavy drinkers with whom actors shared social ties were approximately zero when neither social selection nor social influence were present, were positive but small when either social influence or social selection were medium or high, and were significantly higher when social selection and social influence were both present. Two individual-level manipulations, reducing the target actor's heavy drinking and reducing the target actor's susceptibility to social influence, reduced heavy drinking over time for the individual targeted for intervention. Reducing target actors' heavy drinking also reduced the heavy drinking of other actors not targeted for intervention. Discussion: Simulations of social networks offer a novel method for modeling dynamic associations between drinking and social relationships. These methods may be used to replicate findings from real-world populations and can help generate novel hypotheses involving nonlinear processes that can inform real-world prevention and treatment efforts.

Previous research has shown that drinking levels are influenced by social networks, yet causal mechanisms that explain this relationship are poorly understood. The present study assessed alcohol-related attentional bias as a hypothesized mechanism connecting social support with drinking. Using a 10-minute writing assignment, 84 participants were randomly assigned to focus on a network member that was either a heavy or light drinker. Modified Stroop tests assessed alcohol-related attentional bias before and after the assignment by measuring response latencies for naming font colors for alcohol and neutral words presented on a computer screen. Drinking quantity, alcohol-related problems, and acceptance were assessed using self-report questionnaires. Analyses were conducted to test the impact of the writing task on alcohol-related attentional bias, the mediating relationship of attentional bias on social support and drinking, and the moderating role of acceptance on attentional bias and drinking. Results indicate that, relative to neutral words, response latencies for alcohol words were not significantly affected by the writing task condition in the expected direction. Contrary to the expected results, low-intensity drinkers had significantly longer response latencies to alcohol words than neutral words at baseline, and high-intensity drinkers had significantly longer response latencies to alcohol words during the post-writing-task Stroop test, collapsing across experimental conditions. Attentional bias was not found to mediate social support and participant drinking, and acceptance did not moderate the relationship between attentional bias and drinking. Further probing of the Stroop test suggested that the test may have poor reliability that may have contributed to the failure to support the study hypotheses.