Computational Analysis of Previously Observed Drinking Behaviors: Group Dynamics and Social Choice
Background/Purpose: Drinking behavior among young adults typically occurs in groups across both public and private settings. Settings in which alcohol is consumed differ in risk and protective factors. Understanding how individual motivations interact in a dynamic group network is critical to understanding the etiology of high risk drinking.
In the theoretical analysis of the dynamics of groups it is assumed that people’s behavior in groups is driven by the interactions between the members of the group and the environment. We have conducted several large federally funded field studies to examine drinking behavior in situ. Such studies are very expensive and increasingly difficult to fund. Computer simulations represent a potential approach to complement empirical research at a low cost. This study used the empirical findings from past field studies, to generate conceptually cogent computer simulations of group preference and drinking behavior.
Methods: Based on our prior field research, we used the formula B=f(P,E), where the behavior of a group (B) is a function (f) of the mutual interaction between its members with their personal preferences and characteristics (P), and the environment (E). To characterize this interaction among individuals and their environments, we developed mathematical models based on differential equations. Given the space in which each person’s preferences and/or qualities are represented and a network topology that defines the connectivity of the members of the group, the dynamics of the group are determined mainly by two components: (i) a component that models the attraction and repulsion between members of the group, in which the strength in the attractions and repulsions may remain fixed or change as part of the group dynamics, and (ii) a component that evaluates the position of each person according to the group norms and social settings.
We conducted our simulations in Matlab/Simulink. An iterative process was used to conceptually specify the model parameters and evaluate the ecological validity of the simulations. Once conceptually viable simulations were obtained (relative to empirical and conceptual considerations), we ran Monte Carlo simulations to examine dynamics across a range of events with varying values.
Results: Dynamic simulations for bar choice (selection of drinking settings) will be presented. Monte Carlo simulations will also be presented. We found that the simulations for both bar choice behave consistent with theoretical considerations and past research. Namely, drinking motivations and gender interact to influence selection of drinking locations.
Conclusions and Implications: Computer simulations of complex social behaviors have the potential to complement social research in economical ways. Social work research might benefit from future work in this area.