Co-evolution of Social Networks and Behavior in Social Dilemmas: Theoretical and Empirical Perspectives

Abstract

Social networks play an important role in explanations of outcomes of social dilemmas; situations in which goal-directed individual action can lead to a collectively suboptimal outcome. Among these dilemmas are the cooperation- and coordination problems that underlie many social phenomena. In most research on network effects in social dilemmas, networks are considered as static, exogenously determined structures. Yet, social network structures often result from individual choices of relations. What happens when actors can not only choose their behavior in social dilemmas, but can also purposively change the social network in which these interactions are embedded? This dissertation presents theoretical answers to this question, along with empirical studies in which implications of the theoretical models are tested. Four studies are presented. In two of these, theoretical models are developed for the co-evolution of networks and behavior in coordination problems and cooperation problems, respectively. Game-theoretic tools are use to derive formal results on characteristics of stable states of the co-evolution process, while computer simulation is used to predict how sable states depend on intitial conditions. In the remaining two chapters, predictions on coordination problems are empirically tested using a mix of methodological approaches. One chapter presents an experimental study of coordination in dynamic networks. The results show partial support for predictions on the group level, and stronger support for predictions on individual behavior. In the second chapter, the theoretical model for coordination in dynamic networks is applied to the co-evolution of drinking behavior and social networks among adolescents. The results show that the model can contribute to the explanation of differences in drinking behavior between groups of adolescents. The dissertation concludes with recommendations for future research. These include sketches for experimental and non-experimental empircal studies of cooperation in dynamic networks, designed to test predictions of the theoretical model developed in this dissertation.