Partisan differences in health behaviors can impact respiratory disease dynamics.

The transmission of respiratory pathogens is fundamentally shaped by human behaviors such as interpersonal contacts, use of face masks, and vaccination. Political party affiliation has been shown to be associated with health-related behaviors. Yet, partisan heterogeneity in health-related behaviors is typically not included in infectious disease transmission models. Here, we leveraged uniquely detailed data from the Berkeley Interpersonal Contacts Study (BICS) on partisan differences in contact rates, mask usage, and vaccination patterns during the first year of the COVID-19 pandemic. We find substantial differences in health-related behaviors by political affiliation. Republicans reported a significantly greater number of average daily contacts, lower propensity of using masks and of getting vaccinated for COVID-19. These findings hold even after controlling for observable demographic and location-based differences across survey respondents. We adapt the classic Susceptible-Infected-Recovered (SIR) model to incorporate partisan-specific behaviors and varying levels of political homophily to simulate an outbreak of a hypothetical respiratory pathogen. We find that the observed behavior differences lead to simulated Republicans experiencing higher infection and mortality rates and earlier peaks compared to Democrats. Incorporating greater within-group mixing further amplified partisan differences in disease outcomes. Finally, we show that failure to incorporate partisan behavioral heterogeneity in disease models can lead to inaccurate predictions about the size and timing of outbreaks in a population.