Goodness of fit in relational event models

Abstract A relational event process involves temporally ordered interactions between actors, where past network configurations may influence future ones. The relational event model (REM) can be used to identify the drivers of the underlying network dynamics. Despite the rapid development of REMs over the past 15 years, an ongoing area of research revolves around techniques for evaluating their goodness-of-fit, especially when they incorporate time-varying and random effects. Current methodologies often rely on comparing observed and simulated events using specific statistics, but this can be computationally intensive. We introduce a versatile framework for testing the goodness-of-fit of REMs using weighted martingale residuals. Our focus is on a Kolmogorov-Smirnov type test and its multivariate extensions designed to assess if covariates accurately model the dynamics. A simulation study is performed to assess the test’s power and coverage. Furthermore, we apply the method to a sociological study involving 57,791 emails sent by 159 employees of a Polish manufacturing company in 2010. The method is implemented using the R package .