Polarization and Partisan Bias

Polarization, which has been growing over the last few decades, refers to the divergence between Democrats and Republicans. Polarization has been found to have a number of negative consequences: it contributes to legislative gridlock, decreases trust in government and institutions, and increases income inequality. Polarized partisan politics also contribute to election of ideologically extreme candidates to office, which gives rise to a concern that these candidates will not adequately represent a large segment of the population. Examination of polarization is worthwhile because it may prove useful in mitigating these negative effects. This dissertation primarily deals with psychological bases of polarization and partisan bias. First, I examine whether people with certain psychological traits approach partisanship in a more emotional, biased manner. Second, I examine whether the effects of partisanship vary by the context (namely political vs. nonpolitical). Finally, I have a methodological study on conjoint experimental design in political science. It examines how different types of conjoint design affect subjects’ preferences. More specifically, In the first chapter I examine whether some people are influenced by their partisanship to a greater extent than others. In the increasingly polarized political climate in the U.S. partisans tend to ascribe positive characteristics to co-partisans and negative to opposing partisans. Group attachment partisan literature has long argued that strength of one’s partisanship moderates this partisan bias. But are there other predictors that explain people’s tendency to exhibit partisan bias? In this chapter I examine whether people’s tendency to feel and understand emotions--Need for Affect (NFA)—is a predictor of partisan bias. I find consistent evidence that judgments of partisans are affected by both, in-group and out-group partisan bias. However, different subsets of the population have a different tendency to exhibit the two forms of bias. Out-group partisan bias is mostly driven by individuals who score high on the NFA scale. Moreover, there is no evidence that those who score high on the NFA scale exhibit in-group partisan bias. This finding is particularly significant for it shows that contrary to social identity theory, in-group bias is not an automatic consequence of identification with a group. In the second chapter I examine whether the influence of partisan bias on people’s evaluations varies by setting. Partisanship has long been known to be an influential force in shaping people's evaluations of political actors. More recent research has demonstrated that the influence of partisanship extends to people's decisions and evaluations in non-political settings. However, it remains unknown how the influence of partisanship on people's behavioral judgments compares between the political and non-political settings. In this chapter I experimentally manipulate the setting in which respondents evaluate candidates for jobs, scholarships, internships and awards. I find that respondents’ tendency to exhibit partisan bias (both in-group and out-group) is the same in both settings. I interpret these findings through the lens of conflict extension theory. Namely, conflict between Democrats and Republicans has spread from few policy issues to many, and eventually into non-political settings. As this conflict spread, it did not abate in intensity. The current partisan polarization among rank-and-file members is similar in intensity in the political and non-political setting. In the third chapter I examine the use of conjoint experiments in political science. In recent years they have been increasingly used in the discipline. However, political scientists have predominantly relied on the most commonly used form of conjoint: forced-choice conjoint (also referred to as choice-based conjoint). This reliance on one type of conjoint leaves the question open of how different conjoint designs affect responses, and whether some conjoint designs are more appropriate for some studies than for others. In this chapter I examine these research questions. I find that forced-choice conjoint results in statistically significant social desirability reporting for racial preferences, while other forms of conjoint do not. This finding is particularly significant as it suggests that the most commonly used form of conjoint is not appropriate for the study of sensitive topics. I also find that forced-choice conjoint results in higher measures of partisan bias than other forms of conjoint.Doctor of Philosoph