An illustration of the relational event model to analyze group interaction processes

A fundamental assumption in the study of groups is that they are constituted by various interaction processes that are critical to survival, success, and failure. However, there are few methods available sophisticated enough to empirically analyze group interaction. To address this issue, we present an illustration of relational event modeling (REM). A relational event is a "discrete event generated by a social actor and directed toward 1 or more targets" (Butts, 2008, p. 159). Because REM provides a procedure to model relational event histories, it has the ability to figure out which patterns of group interaction are more or less common than others. For instance, do past patterns of interaction influence future interactions, (e.g., reciprocity), do individual attributes make it more likely that individuals will create interactions (e.g., homophily), and do specific contextual factors influence interaction patterns (e.g., complexity of a task)? The current paper provides an REM tutorial from a multiteam system experiment in which 2 teams navigated a terrain to coordinate their movement to arrive at a common destination point. We use REM to model the dominant patterns of interactions, which included the principle of inertia (i.e., past contacts tended to be future contacts) and trust (i.e., group members interacted with members they trusted more) in the current example. An online appendix that includes the example data set and source code is available as supplemental material in order to demonstrate the utility REM, which mainly lies in its ability to model rich, time-stamped trace data without severely simplifying it (e.g., aggregating interactions into a panel).