UC Berkeley

This dissertation consists of three essays on political polarization in the United States which, together, justify and build a new worldview model of polarization. I develop this theory to better address the issue of whether polarization is asymmetric at the mass level, a question about which quantitative work has thus far had relatively little to say, despite clear evidence that Republican elites are far more polarized than their Democratic counterparts.

The first essay takes the open ground of asymmetry as a starting point for a theory contest: some work on affective polarization defines polarization solely in terms of out-party animus, while other work sees both in- and out-party affect as constitutive. Given asymmetries at the elite and organizational levels, we expect Republicans to be more polarized than Democrats, but these two theories of affective polarization have clearly different expectations for what the asymmetry looks like. However, this apparently neat theory contest fails. Using data from the American National Election Studies, I look at how affective polarization developed from 1978-2016 and I find that Democrats are more polarized, but that that difference is driven by in-party affect (while out-party animus isn't asymmetric). These findings are anomalous for both theories, and I argue that they should disrupt the basic assumption of a straightforward link between affect, identity, and polarization.

The second essay brings together sociological work on cultural schemas, recent advances in the study of attitudinal polarization, and core insights on the role of affective polarization from political science, to propose a new theory which overcomes this disruption: the worldview model of polarization. By defining polarization in terms of persistent patterned relationships between partisan affect, perceptions of threat, and feelings of fear, I am able to approach asymmetry as a question not only of degree, but of kind. Using data from the 1978-2016 ANES, I not only clearly demonstrate that Republicans are indeed significantly more polarized at the mass level, but also suggest a mechanism for this asymmetry, namely, that Republicans are caught in a radicalizing feedback loop of fear of the opposition and dissatisfaction with their own party’s response to that perceived threat.

The final essay elaborates on this model by turning to the public opinion literature on belief systems. Developing the insight that polarization consists in the persistent patterned relationships between emotions and judgments, not either alone, it uses belief network analysis to compare the structure of these relationships over time. Comparing 1992 and 2016, it clearly shows the strengthening integration of affective and attitudinal survey items and overall stronger centralization of the networks, indicating both that political belief systems were becoming more emotionally grounded and that grounding was consistently oriented against a small number of issues (i.e. partisanship and ideology). This not only provides another strain of support for the basic insights of the worldview model, the inclusion of affective items in a quantitative assessment of belief systems is a major contribution to the study of public opinion more generally.