“…We selected social infrastructure types by drawing on the domains proposed by Rhubart et al (2022): (a) free and publicly available places (e.g., religious organizations, civic organizations, and libraries); (b) low‐cost commercial places (e.g., coffee shops, bars, and fast‐food restaurants); (c) personal services (e.g., salons, barbershops, weight‐loss counseling, spas, and tattoo parlors); (d) creative, athletic and entertainment places (e.g., spectator sports, fitness and recreation, bowling alleys, museums, and performance arts); and (e) social services (e.g., organizations that provide social assistance, services for the elderly and persons with disabilities, other individual and family services, and community food services). Given that previous work has shown similar levels of social infrastructure availability across these domains (Rhubart et al 2022), and that many forms of social infrastructure are substitutable, we combined all domains into one social infrastructure measure. We aggregated census tract counts to the county level and used population‐based density (per 1,000 residents) to calculate availability of social infrastructure.…”