2025/07/18 by Piotr Bentkowski, Bentkowski, Piotr, Tomasz Gubiec +1
Mathematics · Physics and Astronomy · #COVID-19 epidemiological studies #Complex Network Analysis Techniques #FOS: Physical sciences #Opinion Dynamics and Social Influence #Physics and Society (physics.soc-ph)
paper · pdf · doi:10.48550/arxiv.2507.13848
openalex publication_date 2025/07/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We investigate how homophily in adherence to anti-epidemic measures affects the final size of epidemics in social networks. Using a modified SIR model, we divide agents into two behavioral groups-compliant and non-compliant-and introduce transmission probabilities that depend asymmetrically on the behavior of both the infected and susceptible individuals. We simulate epidemic dynamics on two types of synthetic networks with tunable inter-group connection probability: stochastic block models (SBM) and networks with triadic closure (TC) that better capture local clustering. Our main result reveals a counterintuitive effect: under conditions where compliant infected agents significantly reduce transmission, increasing the separation between groups may lead to a higher fraction of infections in the compliant population. This paradoxical outcome emerges only in networks with clustering (TC), not in SBM, suggesting that local network structure plays a crucial role. These findings highlight that increasing group separation does not always confer protection, especially when behavioral traits amplify within-group transmission.