A man holds a photo of Republican presidential candidate Donald Trump before a rally  March 1 in Louisville, Ky. (John Bazemore/AP)

In the 2016 presidential race, we hear a lot about anger. Voters are angry about the economy, about race, about the “establishment.” But knowing what voters are angry about doesn’t necessarily tell us why they are angry. There may be logical reasons, but there may also be a more basic, primal force.

My recent research focuses on one such force: the enduring power of group identity. In particular, the growing “sorting” of the American electorate along partisan, religious  and racial lines has created the conditions for the anger and intolerance that have been so obvious this year.

Belonging to a social group is intimately connected to our individual sense of esteem. Without social identities, we are alone in the world, and our losses make us losers. With social identities, we have groups whose victories can dampen and even eclipse our own personal failures, and possibly buttress our self-esteem in situations when we need it. These identities can provoke prejudice against outsiders who we must compete with.

Of course, we typically identify with more than one group. We are members of a political party, race, a religion, a class, a gender, a town and any other group that you could possibly call “we.”

Particularly important, then, is whether these identities “go together” or are in tension with each other. When these identities do go together — when they are “sorted” — most of the members of one group are (or are believed to be) also members of another group.

This is sorting is what has been happening among Americans. Partisan identities have become increasingly aligned with religious and racial identities. Republicans tend toward Christian and white identities, and Democrats tend toward non-religious and non-white identities. With these highly aligned identities, people tend to be more sensitive to threats from outsiders, reacting with higher levels of anger than those with cross-cutting identities.

In research forthcoming in Public Opinion Quarterly, I find two pieces of evidence that corroborate this pattern.

First, the electorate as a whole has grown angrier about their electoral opponents, and more proud of their own candidates, since the 1980s. Americans are increasingly taking sides and becoming emotionally engaged during election season.

Second, I conducted an experiment in which I threatened the status of either a party or a system of beliefs. The threats were artificial blog posts in which partisan bloggers from the opposing side predicted the defeat of a party (Republicans or Democrats) or a set of liberal or conservative policies.

For example, in the blog post that threatened a Democrat’s partisan identity, the blog post said “We’re going to defeat the hardcore socialist Obama, we are raising more money than Democrats, our Congressional candidates are in safer seats, and Democrats have obviously lost Americans’ trust.” The equivalent blog post for Republicans said: “Obama will easily win re-election against whatever lunatic the Republicans run, we are raising more money than Republicans, our Congressional candidates are in safer seats, and Republicans have obviously lost Americans’ trust.”

In both cases, those whose racial, religious, and partisan identities were well-sorted were more likely to react angrily to both types of threat, compared to those whose identities were not sorted. Strong partisans were emotionally engaged by the party-based messages, while strong ideologues were emotionally engaged by the issue-based messages. But only the alignment between partisan, religious, and racial identities predicted emotional reactions to all of the messages.

In this election, recent polls suggest that Republicans with well-sorted identities are more likely to support Donald Trump. His candidacy has been notable for its reliance on anger and “winning” — a message well-tailored to activating emotions and social identities.

Taken together, all of these findings suggest that we shouldn’t expect Americans to be any less angry any time soon. The poorly sorted, cross-cutting identities that can act as an emotional dampener are increasingly rare. There are simply fewer people in the “unflappable” portion of the population.

We are left with an increasingly volatile electorate.

Lilliana Mason is an assistant professor in the Department of Government and Politics at the University of Maryland.