Homophily in opinion networks affects collective risk perception in heterogeneous populations

Understanding how to accurately inform the population about the risks of disasters is key for the well-functioning ofour societies. Risk communication is also an essential feature of the disaster cycle, fundamentally contributing topreparedness but also to management. Here, we use an agent-based model (ABM) to investigate such an importantproblem. Specifically, we study the emergent behavior of a population of individuals who revise their opinion on therisk of a certain event, based on information received from an institution, processed through individual sensitivity,and discussed with peers. Such a complex process may include several biases, e.g., due to heterogeneous riskperception across the population, and homophily, i.e., tendency of individuals to interact with like-minded people.Our ABM, which encapsulates these crucial features, allows us to perform a campaign of numerical simulationstowards gaining mathematically-grounded insights into their impact on the emergent behavior of the population and,ultimately, on how accurately institutional information is received and processed by a population. Such insightscan be useful to design empirical studies to test them and, in case of empirical support, to use them to designrecommendations for policy decision makers.