President Trump is joined by supporters at a campaign rally for Senator Cindy Hyde-Smith of Mississippi in Tupelo.Kevin Lamarque / Reuters

Most Americans do not live in a totalizing bubble. They regularly encounter people of different races, ideologies, and religions. For the most part, they view these interactions as positive, or at least neutral.

Yet according to a new study by the Public Religion Research Institute (PRRI) and The Atlantic, a significant minority of Americans do not live this way. They seldom or never meet people of another race. They dislike interacting with people who don’t share their political beliefs. And when they imagine the life they want for their children, they prize sameness, not difference. Education and geography seemed to make a big difference in how people think about these issues, and in some cases, so did age.

One of the many questions the Trump era has raised is whether Americans actually want a pluralistic society, where people are free to be themselves and still live side by side with others who aren’t like them. U.S. political discourse is filled with nasty rhetoric that rejects the value of diversity outright. Yet, theoretically, pluralism is good for democracy: In a political era when the vast majority of Americans believe the country is divided over issues of race, politics, and religion, relationships across lines of difference could foster empathy and civility. These survey results suggest that Americans are deeply ambivalent about the role of diversity in their families, friendships, and civic communities. Some people, it seems, prefer to stay in their bubble.

In terms of both geography and culture, America is largely sorted by political identity. In a representative, random survey of slightly more than 1,000 people taken in December, PRRI and The Atlantic found that just under a quarter of Americans say they seldom or never interact with people who don’t share their partisan affiliation. Black and Hispanic people were more likely than whites to describe their lives this way, although education made a big difference among whites: 27 percent of non-college-educated whites said they seldom or never encounter people from a different political party, compared with just 6 percent of college-educated whites.

Even those Americans who regularly encounter political diversity don’t necessarily choose it, however. Democrats, independents, and Republicans seem to mingle most in spaces where people don’t have much of an option about being there. According to the survey, roughly three-quarters of Americans’ interactions with people from another political party happen at work. Other spheres of life are significantly more politically divided: Less than half of respondents said they encounter political differences among their friends. Only 39 percent said they see political diversity within their families, and vanishingly few people said they encounter ideological diversity at religious services or community meetings. Traditionally, researchers have seen these spaces as places where people can build strong relationships and practice the habits of democracy. The PRRI/Atlantic findings add to growing evidence that these institutions are becoming weaker—or, at the very least, more segregated by identity. “If you’re thinking from a participatory democracy model, you would hope to see these numbers much higher,” said Robert P. Jones, the CEO of PRRI.

Even Americans who are exposed to people from a different political party might not want to get too close. Almost one in five of the survey respondents said their interactions with people of a different political party are negative. This may be a reflection of deepening partisanship in America: Party affiliation influences not just how people vote, but cultural decisions such as what to buy or watch on television, said Lilliana Mason, an assistant professor at the University of Maryland. “As these other social identities have moved into alignment with partisanship, we’re seeing more animosity across partisan lines—not necessarily because we’re disagreeing about things, but because we believe the [person from the] other party is an outsider, socially and culturally, from us,” she said. “It also becomes really easy to dehumanize people who we don’t have identities in common with.” In recent decades, social scientists have seen increased use of the language of dehumanization, Mason said: people calling their political opponents monsters, animals, or demons, for example.

The survey from PRRI and The Atlantic only measured encounters with diversity—not the depth of those interactions or relationships. When asked how they would feel about their child marrying someone from the opposite political party, 45 percent of Democrats said they would be unhappy, compared with 35 percent of Republicans. This is a sharp increase from how Americans responded to similar surveys a half century ago, according to research by the Stanford professor Shanto Iyengar. While people who seldom or never interact with people of a different race, religion, or political party may live the most sharply segregated lives, a far greater number of Americans may have only cursory interactions with people unlike themselves. “Depth really matters,” said Jones. “A close friend or family member is different than somebody you brush shoulders with every day but never have an in-depth conversation with … What matters is whether that relationship is close enough that someone might feel safe enough to challenge a view.”

America is also divided along lines of religion and race. Roughly one out of five survey respondents reported that they seldom or never encounter people who don’t share their religion, and a similar proportion said the same for race. Certain subgroups were more cloistered than others: 21 percent of Republicans said they seldom or never interact with people who don’t share their race, versus 13 percent of Democrats. Similarly, more than a quarter of white evangelicals said they rarely encounter people of a different race, slightly more than any other major religious group included in the survey. Thirty percent of people over 65 said they seldom or never encounter someone of a different race, compared with 20 percent or less of people under 65.

Geography, along with education among white people, seemed to be an important factor in determining how much diversity Americans encounter. People living in rural areas were significantly less likely than those in cities to encounter racial, religious, or political difference. And among white people, education level made a huge difference: Those without a college degree were more than twice as likely as their college-educated peers to say they rarely encounter people of a different race, and more than four times as likely to say they seldom or never encounter people from a different religion or political party.

In general, the proportion of Americans who seem to live in fully homogeneous communities is small: In terms of identities such as race, religion, and partisan affiliation, only one-fifth to one-quarter of people usually said they seldom or never encounter people unlike themselves. But Americans also believe they are extremely divided along lines of identity: 77 percent said the country is divided over religion. Eighty-three percent said it’s divided over race and ethnicity. And fully 91 percent of respondents said the United States is divided by politics. Many respondents pointed to political parties and the media as two major causes of all this discord, with stark differences along partisan lines: 85 percent of Republicans said the media is pulling the country apart, versus 54 percent of Democrats.

Since the country’s founding, Americans have had to navigate conflicting impulses toward tolerance and a desire to build communities with thick, often homogeneous cultures. Some forms of this are indisputably ugly, such as racial segregation; others may be neutral or immensely enriching, such as tight-knit religious communities. Americans today are sharply divided over the value of multiculturalism: In the survey, 54 percent of Democrats said they prefer the United States to be made up of people from a wide variety of religions, compared with 12 percent of Republicans. By contrast, 40 percent of Republicans said they’d prefer a nation mostly made up of Christians, compared with 14 percent of Democrats.

Perhaps more than any other, this was the fracture line that animated the 2016 election. Even the iconography, from the Trump campaign’s “Make America Great Again” trucker hats to the Clinton campaign’s forward-pointing “H” and “Stronger Together” slogan reflected this divide, said Jones. “As certain groups reach a critical mass, I think it throws Americans as a whole back into a conversation about affirming these principles [of pluralism] or not,” he said. “If you think culture war today, it’s less about gay marriage and abortion than it is about American identity.”

Americans aren’t fully in control of the amount of diversity to which they’re exposed. Some isolation is a matter of geography and class: People living in rural Vermont, the whitest state in America, may not have many opportunities to meet people of another race, for example. Even self-segregation may not be malicious: “It’s hard to spend time with people who are not like you,” Mason said. People may not want to argue about deeply held political beliefs or explain their religious dietary needs to strangers. “At the most basic level, the place that prejudice comes from is not an evil place,” she said. “It’s just that it’s easier to spend time with people who are the same.”

And yet, the choices Americans make every day—about where to live or go to church or send a kid to school, about whose book club to join or whom to invite over for dinner—influence the way they see the world, and especially how they see politics. When people largely surround themselves with sameness, they may find themselves left shouting across perceived divides, unable to see their reflection in anyone who stands on the other side.

This project is supported by grants from the Joyce, Kresge, and McKnight Foundations.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.