My first chapter introduces a new measure of affect that reduces reductionism while providing researchers with an easy-to-use numerical output. The literature shows that partisanship drives negative emotional evaluations of out-partisans, but existing measures, like thermometers, candidate evaluations, and social-distance measures, discount the sentiment attached to individuals' negative attitudes. Our new measure captures the motivation behind partisans' attitudes by asking respondents to provide one-word descriptions of voters in their party and the opposing party, and to code the sentiment behind their word choice. This produces both qualitative and quantitative measures of respondents' affect. Our self-coded open-ended measure has strong face validity, correlates strongly with existing affect measures, and reveals a theoretically relevant dimension of affective polarization. This measure advances our understanding of partisan affect by allowing scholars a window into respondents' state of mind, and can easily be applied to other groups of interest.My second chapter uses a conjoint experiment to argue that partisans have genuine taste-based preferences against social engagement with out-party members and are not simply engaging in statistical discrimination. Research shows that partisanship can inform individuals' decisions in areas outside of politics, such as roommate choice, spousal selection, and economic behavior. However, few studies can systematically determine whether these decisions are based on partisanship or if they use partisanship to infer other characteristics relevant to a given social choice. To determine the extent to which partisanship informs decisions, we use a conjoint analysis to isolate the impact of partisanship on non-political considerations across three types of social decisions: selecting a spouse to marry, a neighborhood to live in, and a business to frequent. We find that partisanship influences all three social decisions, even while controlling for other salient considerations. Additionally, we find that the degree to which partisanship matters is similar to and, in some cases, exceeds other relevant considerations. Overall, our study shows that when individuals make key decisions that affect the trajectory of their life, partisanship is a fundamental consideration. My third chapter examines how affect interacts with the largest, fastest-growing, and most underrepresented religious group in the United States: the nonreligious. Religion is declining in the United States, as more Americans report low religiosity, have less attachment to religion, and an increasing number identify as nonreligious. However, in Congress, the story is different. Although a quarter of the public identifies as nonreligious, only one member of Congress does. This chapter uses a conjoint candidate choice experiment to examine how religious voters' bias against nonreligious candidates reduces support for them in electoral settings. It demonstrates that bias against the nonreligious affects electoral decisions and is causally linked to the exclusion of the nonreligious from government. Furthermore, it shows that nonreligious voters only exhibit ingroup support for candidates who explicitly identify as Atheists, not agnostics or candidates that merely lack a religious identity. My work supports the centrality of affect and identity in politics. Individuals' emotional evaluations and group identities drive their partisan evaluations, non-political decisions, and candidate choices. Individuals view out-partisans as bad people with poor character, weak intellect, and little value. They will make sacrifices to avoid social decisions that expose them to outgroup members and vote against candidates that do not share their identities. Rather than being rational actors with ideological preferences, this research suggests that political behavior is in line with predictions from Social Identity Theory (Henri Tajfel and Turner 1979). It highlights the need for further study on how to reduce identity-based engagement in politics.