Primary Submission Category: Applications in Health and Biology
Quantifying the Causal Impact of Political Polarization on COVID-19 Dynamics: A Longitudinal Proximal Causal Inference Approach
Authors: Jian Yang, Rebecca Smith,
Presenting Author: Jian Yang*
The politicization of public health has been a defining feature of the COVID-19 pandemic. However, estimating the causal effect of political affiliation on infection rates is challenged by unmeasured, time-varying confounders, specifically community-level behavioral compliance and risk attitudes. Standard regression models often fail to account for these dynamic latent factors, leading to biased estimates. This study proposes a longitudinal Proximal Causal Inference framework to identify the causal effect of county-level political leaning on weekly COVID-19 case growth rates from 2021 to 2023. We address the challenge of unobserved confounding by employing a Two-Stage Least Squares strategy with distinct proxy variables. Crucially, to overcome the 2022 discontinuity of commercial mobility datasets, we leverage the Bureau of Transportation Statistics‘ data as a continuous, outcome-inducing proxy. Conversely, we utilize the historical presidential election results as a static, treatment-inducing proxy. Our model incorporates a specific lag structure and time fixed effects to control for incubation periods and temporal dependencies. By formally separating static political treatments from dynamic behavioral mechanisms, this research provides a robust identification strategy for evaluating how polarization structurally impacted pandemic trajectories over an extended period.