American politics feels like it has become blood sport. Democrats and Republicans report hating each other, and Trump and his supporters have vocally called for the jailing (and perhaps worse) of their opponents. A number of studies have shown that partisans are increasingly prejudiced against their political opponents, which manifests as, for instance, avoidance and vitriolic rhetoric, or taking material losses in order to avoid helping the other side (for a review, see: Shanto Iyengar et al., “The Origins and Consequences of Affective Polarization in the United States.” Annual Review of Political Science (Forthcoming)). Evidence for increasing hostility (or affective polarization as some call it) has stacked up over the past years. However, we have had far less evidence that explains how we got here. In Uncivil Agreement: How Politics Became our Identity, Lilliana Mason has offered the clearest and most convincing explanation for the current state of affairs.

Drawing on a rich bevy of literature from political science, psychology, and sociology, Mason blames the rise of affective polarization on changes in “social sorting,” i.e., the decline of cross-cutting identities. In the mid-twentieth century, the Democratic and Republican parties were big tents that held a wide array of social groups. The New Deal-era Democratic party, for instance, consisted of Southern Democrats and African Americans, farmers and labor unions, working-class voters and urban liberals. Today, the parties are far more internally homogeneous. The Republican Party is largely white, rural, Christian, and conservative, while the Democratic party contains more non-Christians, non-whites, and urban liberals. This process increases the likelihood that people will think of their partisanship as a social identity.

Decades of psychological research has shown that dividing ourselves into different social groups, leads to in-group favoritism and, often, out-group hostility. Past work has understood this social sorting as a sign of polarization. For instance, we have long known that rural and evangelical voters are more likely to vote for Republicans. Mason, however, sees the process of social sorting as the cause of (affective) polarization. This insight not only helps us see how we got to the fairly unpleasant status quo of American politics, but may also help us see how we may change things.

After discussing her theoretical framework, Mason shows that the parties are more social sorted than they have been in the past 50 or so years. She then demonstrates that partisans who are socially sorted are more likely to dislike the other side and report strong negative emotions towards them than partisans who are not, even among people who hold similar issue attitudes (hence the title of the book). The socially sorted partisan is also more likely to engage in politics.

Finally, Mason offers some potential solutions to the problem of affective polarization. Most of these come from the toolkit of social psychology, e.g., contact theory, affirmation, and priming a shared identity that cuts across partisanship. Some evidence in favor of the last suggestion already exists, but I’m skeptical that the effects are large enough and that any intervention would scale-up enough to make a substantive difference. If the sorting of identities caused the problem, then only the unsorting of identities can fix the problem.

Mason offers some hope that the Republican party may become less homogenous, or at least that a segment of the Republican party may break away due to norm violations by the current administration. Institutional reforms also offer a way forward. For instance, Lee Drutman suggests that proportional representation (which can be implented at the state level) would reduce polarization ( “This Voting Reform Solves 2 of America’s Biggest Political Problems.” Vox, 2018). Moving politics away from a winner-take-all system would surely reduce the likelihood that partisans see each other as members of warring factions. However, since major state-level electoral reforms and a social realignment of the political parties are both unlikely, things are likely to get worse before they get better as the United States teeters towards become a majority-minority country, exacerbating any threat felt by white Americans.

While Mason offers convincing evidence that increased social sorting is associated with increased affective polarization, the book falls short of providing definitive causal evidence. Her primary tool for causal inference is to create groups of sorted and unsorted by matching on a number of observable variables. Obviously, it would be impossible (or at least very difficult) to randomly assign citizens into sorted and unsorted conditions but nagging questions will remain without something cleaner than the current setup. For instance, perhaps exposure to elite rhetoric causes people to become both more polarized and more sorted (thus making the relationship spurious). Michele F. Margolis provides some evidence that this may be the case—exposure to right-wing rhetoric pushes some Republicans to increase their religiosity (“How Politics Affects Religion: Partisanship, Socialization, and Religiosity in America” The Journal of Politics 80(1), 2018), which also makes someone more sorted according to Mason’s measures, and other studies have shown that exposure to elite rhetoric also increases polarization.

In addition to stronger causal evidence, I would have liked to see more explanation or even tests of the mechanism linking identity-driven sorting to affective polarization. Why does the decline in social sorting lead to increased affective polarization? Is it because we are no longer in contact with people from the other side, which results in distorted stereotypes that shape our attitudes? Does it lead us into to echo chambers where we only hear from vitriolic elites attacking the other side? Does social sorting automatically lead to affective polarization, as per the minimal group experiments?

Finally, Mason acknowledges that the media may have played a role in affective polarization, but she also sees the media as a conduit of social sorting. The impact of partisan sources may also interact with fully formed identities, thereby exacerbating affective polarization. Social identifiers are also more open to identity-consistent information: they are less likely to process information critically when information comes from their group, and more likely to dismiss or counterargue information from the other side. If a Republican leader tells them, for instance, that the other side is “crooked,” they accept that information without much hesitation. Social sorting is likely only one variable in a chain of processes that explains affective polarization.

These unresolved questions indicate that more work is needed on social sorting and affective polarization. In this way, Mason’s book is an intellectual achievement: it opens the door (as well as sets the agenda) for future research. Her work has certainly made me think about affective polarization and American politics differently. Mason’s work is necessary reading for anyone wishing to understand the processes that explain the status quo, and is already having a large and well-deserved impact on academic and public conversation.