Introduction to using relational hyperevent models to study criminal networks

Relational hyperevent data capture time-stamped/-ordered interactions among any number of nodes in a network. Such data are central in criminal network research, both in studies on social organisation of crime and on co-offending. Until recently, researchers had to coerce these data for standard network methods, risking loss of information and biased results. The relational hyperevent model (RHEM) solves this by modelling hyperevent data directly, preserving their temporal and structural detail.This article introduces RHEM to criminal network researchers. We show how to collect and prepare data for RHEM, outline the model’s logic in non-technical terms, compare it to other models and highlight the theoretical questions RHEM enables to answer. Two examples – a directed communication network and an undirected co-offending network – were used to illustrate the full workflow from data processing to model interpretation. Finally, we point the readers to RHEM applications in both criminal and non-criminal network studies for inspiration.