Dynamics of Local Clustering in Temporal Social Networks

Abstract

While the local clustering coefficient is central to network analysis, it is typically treated as a static property. We introduce delta clustering to quantify changes in an individual's local clustering across adjacent time points in undirected networks. To address structural constraints, we develop a density-based decomposition expressing clustering as a weighted combination of subgraph-specific densities, clarifying how ego-network size and composition bound clustering change. We identify six triadic components distinguishing reconfiguration among persisting contacts from the integration and loss of ties. Using high-resolution university communication networks, we estimate fixed effects models accounting for network size interactions. Variation in delta clustering is driven primarily by density changes among persisting contacts, followed by the integration of new contacts, with magnitudes conditioned by ego-network size and baseline density. We further demonstrate the measure's explanatory power by linking daily clustering changes to individual behavioral outcomes. These results establish delta clustering as a size-aware, temporally grounded measure of local structural change in dynamic social networks.