Counter-demonstrators and supporters of President Trump fight over a flag during a People 4 Trump rally in Berkeley, California, in 2017. Stephen Lam / Reuters

For the past 18 months, many political scientists have been seized by one question: Less-educated whites were President Trump’s most enthusiastic supporters. But why, exactly?

Was their vote some sort of cri de coeur about a changing economy that had left them behind? Or was the motivating sentiment something more complex and, frankly, something harder for policy makers to address?

After analyzing in-depth survey data from 2012 and 2016, the University of Pennsylvania political scientist Diana C. Mutz argues that it’s the latter. In a new article in the Proceedings of the National Academy of Sciences, she added her conclusion to the growing body of evidence that the 2016 election was not about economic hardship.

“Instead,” she writes, “it was about dominant groups that felt threatened by change and a candidate who took advantage of that trend.”

“For the first time since Europeans arrived in this country,” Mutz notes, “white Americans are being told that they will soon be a minority race.” When members of a historically dominant group feel threatened, she explains, they go through some interesting psychological twists and turns to make themselves feel okay again. First, they get nostalgic and try to protect the status quo however they can. They defend their own group (“all lives matter”), they start behaving in more traditional ways, and they start to feel more negatively toward other groups.

This could be why in one study, whites who were presented with evidence of racial progress experienced lower self-esteem afterward. In another study, reminding whites who were high in “ethnic identification” that nonwhite groups will soon outnumber them revved up their support for Trump, their desire for anti-immigrant policies, and their opposition to political correctness.

Mutz also found that “half of Americans view trade as something that benefits job availability in other countries at the expense of jobs for Americans.”

Granted, most people just voted for the same party in both 2012 and 2016. However, between the two years, people—especially Republicans—developed a much more negative view toward international trade. In 2012, the two parties seemed roughly similar on trade, but in 2016, Hillary Clinton’s views on trade and on “China as a threat” were much further away from the views of the average American than were Trump’s.

Mutz examined voters whose incomes declined, or didn’t increase much, or who lost their jobs, or who were concerned about expenses, or who thought they had been personally hurt by trade. None of those things motivated people to switch from voting for Obama in 2012 to supporting Trump in 2016. Indeed, manufacturing employment in the United States has actually increased somewhat since 2010. And as my colleague Adam Serwer has pointed out, “Clinton defeated Trump handily among Americans making less than $50,000 a year.”

Meanwhile, a few things did correlate with support for Trump: a voter’s desire for their group to be dominant, as well as how much they disagreed with Clinton’s views on trade and China. Trump supporters were also more likely than Clinton voters to feel that “the American way of life is threatened,” and that high-status groups, like men, Christians, and whites, are discriminated against.

This unfounded sense of persecution is far from rare, and it seems to be heightened during moments of societal change. As my colleague Emma Green has written, white evangelicals see more discrimination against Christians than Muslims in the United States, and 79 percent of white working-class voters who had anxieties about the “American way of life” chose Trump over Clinton. As I pointed out in the fall of 2016, several surveys showed many men supported Trump because they felt their status in society was threatened, and that Trump would restore it. Even the education gap in support for Trump disappears, according to one analysis, if you account for the fact that non-college-educated whites are simply more likely to affirm racist views than those with college degrees. (At the most extreme end, white supremacists also use victimhood to further their cause.)

These why-did-people-vote-for-Trump studies are clarifying, but also a little bit unsatisfying, from the point of view of a politician. They dispel the fiction—to use another 2016 meme—that the majority of Trump supporters are disenfranchised victims of capitalism’s cruelties. At the same time, deep-seated psychological resentment is harder for policy makers to address than an overly meager disability check. You can teach out-of-work coal miners to code, but you may not be able to convince them to embrace changing racial and gender norms. You can offer universal basic incomes, but that won’t ameliorate resentment of demographic changes.

In other words, it’s now pretty clear that many Trump supporters feel threatened, frustrated, and marginalized—not on an economic, but on an existential level. Now what?