Explorer les réseaux à l'échelle de la triade: l'apport des modèles statistiques ERGM

In contrast to conventional statistical analysis, statistical models for social networks must account for dependencies between observations. To address this issue, a specific class of models has been developed, Exponential Random Graph Models (ERGM). The basis for these models is the fundamental idea of triads, a longstanding concept within sociology, as seen in Simmel's work. This article aims to present for the first time in French ERGM's theoretical foundations, and reviews the usefulness of the triadic approach for exploring social processes. Using a case study of a business law firm, the ERGM estimation process is presented in detail, followed by a review of recent research using ERGM.