UC Davis

Affective polarization - the growing dislike and distrust between supporters of opposing political parties - has become a growing concern in the United States and other Western democracies. While a number of potential explanations for this phenomena have been offered, such as elite ideological polarization or income inequality, in this dissertation I focus on the role of social identity and its relationship to polarization. I explore whether the growing inter-party hatred we witness is due less to what our opponents think, and more about who they represent. I build on social sorting theory to argue that individuals who perceive opposing parties as looking less like them will form more polarized views of the parties. In this dissertation, I offer three tests of this theory, and find several instances in which the alignment between an individual’s social identities — whether that be racial, gender, religious, age or other core identities — and those they attach to a political party, influences their feelings towards those parties.

The first paper develops a theory of identity alignment, explaining how and why we might expect individuals to form views of the parties that are based on a comparison between an individual’s view of themselves and their views of those who support each party. I then provide three experimental tests of this theory, two studies using undergraduate subjects from a large public university, and a third replication study using a national sample, and find tentative evidence that supports the idea that the groups we associate with a political party influence our affective evaluations of the party, perceived ideological placement of the party, and responsiveness to party cues.

The second paper studies the rise of social sorting in the United States and its relationship with affective polarization at the aggregate and individual level. I provide a new way to measure social sorting using the tools of machine learning, and validate this method by also applying it to the more well established case of ideological sorting. I then demonstrate that social sorting has risen at roughly the same rate as ideological sorting within the US, and that both phenomena ar2e independently associated with increasing levels of affective polarization. I show that individuals do indeed dislike their out-party because of what they believe, but they also form increasingly negative views because the two parties now represent wholly different social groups, comprised of few overlapping identities. This chapter provides strong evidence that the US parties now represent two groups who find it increasingly difficult to see themselves in the other side.

The final chapter provides a comparative perspective by studying the relationship between social sorting and one core social identity — gender. I explore whether the gender gap between two parties — the difference in the proportion of female supporters within each party — impacts the affective evaluations that partisans make of out-parties. I argue that the gender composition of one’s own party represents an important identity marker, an indication of the identity composition of the party, and a yard stick by which individuals judge other parties. I argue that the more a party differs in its gender composition from an individual’s own party, the less warmly the individual will feel towards the party. I test this, along with competing theories regarding the gender composition of party supporters, on a dataset containing evaluations of 159 parties across 25 countries between 1996-2020, and find that an increased gender gap between parties is indeed associated with increasingly polarized evaluations of the out-party. I also find tentative evidence that individuals, particularly women, prefer parties with more women supporters, but that this relationship is attenuated by the effects of the gender gap.

Taken together, these three papers present evidence that the increased partisan animosity that we see in many contemporary polities is driven, at least in part, by the social divisions between parties. As parties (particularly those in the United States) come to represent wholly different social groups who no longer share large numbers of cross-cutting identities, individuals find it increasingly difficult to find those among their political opponents who look like them, and so become increasingly negative, distrustful, and unresponsive to out-parties. Unlike other potential explanations for affective polarization such as elite ideological polarization, income inequality, or majoritarian electoral systems, this polarization driven by social sorting may be a much more difficult problem to solve. Without incentives for political parties to court supporters of social groups that are currently associated with their political opponents, it is hard to see how and why parties would take the necessary steps to decrease social-based polarization. Indeed, it is more likely that the results here are part of a vicious cycle — as groups become sorted with one party or another, this feeds into the stereotypes people have about the parties, forcing them to update their own identification and retrench into the party associated with their group, thus increasing levels of social sorting, and driving further polarization.