Exploring the dynamic interplay between communication and co-offending using relational hyperevent data on an Italian mafia network

Abstract This study investigates the interdependence between criminal actors, relational hyperevents, and different types of events in an Italian mafia network. Relational hyperevents are time-ordered interactions involving any number of actors. We distinguish between communication (in-person meetings and phone calls) and co-offending (the commission of crime), arguing that communication reflects the social organisation of crime, whereas co-offending represents its outcome. We examine three multiplex relational mechanisms – triadic closure, interaction repetition, and interaction accumulation – alongside their uniplex counterparts. We also test the effects of age, leadership, group membership, and kinship. We analyse court records from the Italian mafia investigation Infinito, documenting the ‘Ndrangheta’s drug trafficking and territorial expansion in Lombardy between 2006 and 2010. The data comprise 346 actors and 1,131 events. We apply Relational Hyperevent Models (RHEM), using the eventnet software package to generate relational statistics and estimate Cox proportional hazards models of event occurrence. Descriptive analyses show that the network is held together primarily by higher-order in-person meetings, whereas phone calls are more diffuse and co-offending is highly fragmented, frequently occurring as single-actor crimes. Ignoring the relational hyperevent nature of the data would therefore result in substantial information loss. RHEM results indicate that communication often precedes co-offending, although significant effects are also observed in the opposite direction. Interaction repetition and accumulation consistently predict future events, whereas triadic closure exhibits only limited support. Age, leadership, kinship, and group membership also significantly shape event formation. Our findings bridge the literature on the social organisation of crime and co-offending by modelling communication and criminal acts within a unified relational framework. They further demonstrate the value of RHEM for criminal network research and suggest that commonly reported triadic closure effects may partly reflect artifacts of data aggregation rather than genuine behavioural tendencies.