Rethinking Polarization: A Multilevel Approach to the Study of Mass Political Disagreement

Abstract

Social scientists have long looked to public opinion as a lens through which to understand political culture and social change. In recent decades, American scholars have been particularly interested in how opinions have become patterned by political affiliation, noting increasingly divergent patterns of beliefs between self-identified Republicans and Democrats. Despite the ubiquitous narrative of a “polarized public”, there remain several points of contention and potential issues in the literature on mass polarization and public opinion, more broadly. To better understand mass polarization–its origins and consequences–this dissertation examines mass beliefs at three levels: between parties, within parties, and cross-nationally. I argue that each level of analysis provides distinct insight into political-cultural change and the consequences of mass opinion for institutional politics and social policy development. Using several waves of data from the American National Elections Study (ANES), Chapter 2 examines trends in three measures of partisan disagreement: partisan polarization, partisan sorting, and partisan constraint across multiple issue domains (economic, moral, and civil rights). I then develop a new approach to measure polarization, which considers partisans as conglomerates of ideological subgroups. Across all measures, I find growing divergences between partisans, with each measure reaching its highest value in 2020. I also find that Democrats have shifted farther to the left and have become more ideologically homogeneous in recent years, compared to Republicans. Chapter 3 focuses on between- and within-party dynamics on a single salient issue: immigration policy. I employ computational text analysis techniques to analyze presidential stump speeches, and I find that Donald Trump emphasized immigration policy significantly more than Hillary Clinton. To understand the demand for these two divergent strategies, I model changes in the mean and variance of immigration policy attitudes over time using variance function regression. I find that Republicans had grown more conservative on average, as well as more homogeneous on immigration policy views, whereas Democrats had become more liberal on average, but also more heterogeneous. I argue that researchers should consider both measures of central tendency and dispersion in public opinion data to understand relative party strategy and how issues play out in elections. Chapter 4 chapter expands and updates analyses of “American Exceptionalism” in cross-national comparisons of welfare state policy attitudes. I measure the distinctiveness of social policy support among Americans compared to levels of support in other wealthy democracies over the last several decades. I find that the U.S. has remained an outlier in a cross-national perspective. But a more distinctive cross-national pattern arises when examining the level of political polarization across countries: in 2016, the U.S. has far and away the highest levels of polarization across all welfare attitudes. Further analyses reveal that U.S. distinctiveness in the aggregate has increasingly been driven by the relatively low levels of welfare state support among Republicans. The results point to the importance of considering within-country variation when constructing narratives about cross-national patterns.