Not swinging together: Partisan defection in the age of political polarization.

Traditional studies of swing voters in American politics typically focus on general attributes of a monolithic group. More recent analyses of "partisan defection," address the partisan direction of swing voting but often limit analysis to a single election or a single direction of defection. I seek to expand upon this research by examining multidirectional partisan defection across successive presidential elections. Using ANES and CCES 2012 and 2016 presidential election data, I test individual correlates of self-reported major party vote switching for presidential-candidates in both directions. Results suggest one the strongest correlates of defection has been underappreciated in recent defection studies: self-identified ideology. More conservative past Democratic voters and more liberal past Republican voters are more likely to defect than voters whose ideology is more consistent with their past voting patterns. The two groups of defectors are also vastly different from one another with Whites more likely to defect to Republicans and African Americans more likely to defect to Democrats. Less progressive racial attitudes and financial hardship are also both associated with increased odds of defecting from Democrats but decreased odds of defecting from Republicans. Finally, results show intra-defector group similarity across both elections, suggesting consistent partisan defection trends, irrespective of candidates. Overall, these results call for a reexamination of voter defection to consider distinct partisan directionality, secular trends in defection, and further exploration of the role of self-identified ideology. [ABSTRACT FROM AUTHOR]