Context matters: including global covariates in relational event models

Abstract

Relational event models have become popular in the network science literature, as a number of phenomena in various applied fields, such as sociology, ecology and finance, can be described via a network of entities interacting over time. A relational event model allows to describe the formation of instantaneous links over time and to identify its driving factors. Traditional inferential techniques, involving Cox’s partial likelihood, can estimate the effects of covariates that are node-specific, such as age or in-degree, or dyadic, such age difference of pairs of nodes or reciprocity. However, the partial likelihood cannot account for global covariates, i.e., factors that are constant for all pairs. Indeed, these covariates, being only time-dependent, drop out from the partial likelihood. Nevertheless, these factors, such as weather or time of the day, are often important in capturing and explaining the temporal nature of the studied events. In this paper, we address this challenge with the use of nested case-control sampling on a time-shifted version of the event process. This will result in a partial likelihood of a degenerate logistic generalized additive model from which we are able to recover effects of all kinds of covariates, including global ones