(Dis)Connected: Political Polarization, Social Connection, and Social Trust in the United States

Abstract

This thesis examines the relationship between political polarization, social connection patterns, and social trust in the United States through a multi-scale quantitative analysis. Using county-level data from the 2016 and 2020 presidential elections, social connection data from Meta, demographic information from the US Census, and data on trust from Gallup surveys, I analyze how these data influence five specific dimensions of polarization: spread, dispersion, divergence, group distinctness, and size parity. My findings reveal that within-county social connectedness is significantly associated with increased dispersion polarization, suggesting that denser social networks support more varied political attitudes. My analysis of cross-county social connections showed that stronger connections between counties slightly decrease overall polarization measures but also increase political similarity between connected counties. These relationships are strongly moderated by geographic and demographic factors, with distance, education, and racial diversity playing particularly important roles. Contrary to expectations, national data (though limited) show that some measures of polarization and all measures of institutional trust increased simultaneously between 2016 and 2020, challenging the hypothesis that polarization necessarily erodes trust. These findings highlight the importance of examining polarization as a multidimensional construct and of considering the multiple scales at which social connections operate to influence political environments.