Multiplex Communities and the Emergence of International Conflict

Recent advances in community detection reveal new insights into multiplex, time-dependent, and heterogeneously structured networks. Less work, however, empirically investigates the relationship between membership in these communities and observable outcomes in social systems. We investigate this question in the domain of politics with an application to a crucial outcome of scholarly interest: the onset of violent conflict in the international system. We differentiate between stronger signals of observed relations and weaker signals of affinity and perform multilayer community detection to locate dense multiplex blocs of countries. Although it is often assumed that detected communities are associated with cooperative outcomes, we find evidence that membership in these communities can also be a significant predictor of conflict. The results point to the need to devote more empirical attention to the relationship between different tie types and observed network outcomes in community detection research.