Modeling Network Dynamics
Given that social networks are inherently dynamic phenomena, characterizing their structure, precursors, and consequences can be improved by methodologies that incorporate such dynamism. This chapter discusses several longitudinal network modeling approaches that seek to understand the process of network change, on one hand, and to predict future network states, on the other. These include the relational event model (REM), exponential random graph model (ERGM), and stochastic actor-oriented model (SAOM). These models focus on different temporal resolutions and differentiate instantaneous events from relations with longer durations, among other distinctions. The chapter identifies commonalities and unique features of each model, both conceptually and via an application to a longitudinal network dataset of dominance interactions within a herd of Eurasian red deer. Throughout, the chapter emphasizes each modeling framework’s assumptions, data requirements, and parameter and model interpretation.