Recent scholarship has documented the rise of social sorting in the United States, the phenomenon whereby liberals identify as Democrats and conservatives identify as Republicans. This dissertation analyzes voters who cannot be explained by their social identities, voters whom I call political heretics. In order to better understand the motivations of these renegades, I marshal original ethnographic, observational, and experimental data to investigate one of the largest of these groups of defectors: white evangelicals who support the Democratic Party. I find that the social context in which they reside – the everyday milieu that animates their daily lives – shapes their partisan identification and conditions their political behavior. I display that the most powerful social context for these evangelicals is their local congregation. Beyond this, I explore and critique current measurement schemes of evangelicalism, offering recommendations to the discipline on better ways to understand and categorize one of the largest religious social groups in the United States. These findings have implications for American democracy as political heretics may reduce stereotypes around the parties, thus reducing Americans' negative views of the out-party.