Carlos Barria / Reuters

On Tuesday, a divided America returned a divided verdict on the tumultuous first two years of Donald Trump’s presidency. Rather than delivering a “blue wave” or a “red wall,” the election produced a much more divergent result than usual in a midterm.

Democrats made sweeping gains in the House, ousting Republicans in urban and suburban seats across every region of the country to convincingly retake the majority for the first time since 2010. They also made a dramatic recovery in the Rust Belt states that tipped the presidency to Trump in 2016: Democrats won both the governorship and Senate races in Pennsylvania, Wisconsin, and Michigan, the three bricks in the party’s “blue wall” that Trump dislodged to capture the White House.

But Republicans expanded their Senate majority across a belt of older, whiter heartland states, and crushed liberal hopes by denying victory to the three young Sun Belt Democrats who had captured the party’s imagination more than any other candidates in this election cycle. The Republicans Ron DeSantis and Ted Cruz respectively defeated the African American gubernatorial nominee Andrew Gillum in Florida and Beto O’Rourke, the Texas Senate hopeful who energized party activists across not only the state but the country. In Georgia, Stacey Abrams, seeking to become the nation’s first female African American governor, trailed Republican Brian Kemp, but refused to concede while waiting to see whether the count of remaining ballots would force a runoff in December.

The combined results reconfirmed the deep lines of division etched in Trump’s narrow 2016 victory over Hillary Clinton. The evening amounted to a simultaneous repudiation and reaffirmation of Trump from two very different Americas, and underscored the fundamental demographic, cultural, and economic changes reshaping America and its politics.

The results dramatized both the benefits and the costs of the electoral bargain Trump is imposing on his party. Behind his racially infused nationalism, the GOP is trading white-collar voters for blue-collar voters; suburban for rural; and younger for older. Those trends advantaged them in a Senate map centered mostly on white heartland states, and they also showed continued potency in Sun Belt battlegrounds such as North Carolina and Georgia, where overwhelming margins among working-class white voters allowed Republicans to overcome erosion in other categories.

In Senate races, Trump successfully mobilized his coalition to help Republicans oust Democratic incumbents in North Dakota, Indiana, and Missouri, all states that he carried in 2016. Montana remained too close to call on Wednesday morning, though many analysts believed Democrat Jon Tester had a clearer path to victory. Trump’s coattails in the Republican victories were apparent: Exit polls conducted in Indiana and Missouri showed Trump’s approval rating at 50 percent or above in each, with the GOP candidates winning 86 to 88 percent of the voters who approved of him.

But for the first time in Trump’s national political career, the electoral costs of his approach also came due. House Republicans were swept away in urban and suburban districts, from New York, New Jersey, Philadelphia, and Virginia in the East, to Chicago, Minneapolis, Detroit, Des Moines, and Kansas City in the Midwest, to Denver, Tucson, Orange County, and possibly even Salt Lake City in the West. Republicans fell even in suburb-heavy southern districts around Houston, Dallas, Oklahoma City, Charleston, and possibly Atlanta that traditionally have leaned much more reliably red than similar areas in other regions.

All of these suburban seats were in places where voters are doing best in the buoyant economy, but widespread discomfort with Trump’s style and values ignited a huge backlash among college-educated white voters—primarily women, but also an unusually large number of men. The exit polls put Trump’s approval rating among college-educated white voters at only about 40 percent. Burdened by that verdict, Republican House members were swept away in fast-growing, economically dynamic metro areas.

In the midwestern states that were key to Trump’s victory in 2016, the Democrats rebounded, while also facing reminders of the obstacles they may face in reclaiming those states from Trump in 2020. Democratic Senators in Wisconsin, Ohio, Michigan, and Pennsylvania—all of whom were considered at risk after Trump carried their states two years ago—won comfortable victories. The party also reelected a governor in Pennsylvania, beat Scott Walker in Wisconsin, and took over for the outgoing GOP governor in Michigan. Still, Republicans comfortably held the Ohio governorship, and narrowly prevailed in Iowa.

In the Sun Belt, Democrats were frustrated by a string of relatively narrow losses. But the strong showing of Abrams in Georgia, Kyrsten Sinema in Arizona (whose race had not been called as of Wednesday morning), and, above all, O’Rourke in Texas could signal a revival of their competitiveness in those states. O’Rourke actually won more votes in Texas than Hillary Clinton did in the 2016 presidential race, an incredible performance. At the same time, the dispiriting loss for Gillum in Florida underscored the difficulty Democrats face winning the state in a high-turnout election—an ominous sign for 2020.

These diverging results—big Democratic gains in the House and respectable advances in governorships offset by a strong Republican performance in the Senate—were rooted in the intense polarization around Trump. Exit polls measuring the national vote in House races showed that Democrats posted big advantages among the groups most antagonistic to the president: young voters (they carried about two-thirds of those younger than 30 and three-fifths of those between 30 and 40), African Americans (about nine in 10), and Latinos (about two-thirds). Republicans in turn won about three-fourths of white evangelical voters and ran up a double-digit lead over Democrats among rural voters.

But the defining trend of the night—as throughout the Trump presidency—was the substantial gap between white voters with and without a four-year college education. That gap helps explain both the Democratic suburban gains in the House and the strong GOP performance in the Senate.

In the first decades after World War II, white voters without a college education consistently voted more Democratic than white voters holding a four-year college degree or more. That’s flipped in recent years, with Democrats at both the congressional and presidential level consistently winning a higher share of white voters with a college degree than those without one.

That inversion has intensified under Donald Trump. In both the 2010 and 2014 midterm elections under Obama, House Democrats won only about one-third of non-college whites and about two-fifths of whites with a college degree. In 2010, Democrats ran six points better among college whites than non-college whites; in 2014, the gap was seven points. But in 2016, with Trump on the ballot, the gap roughly doubled to 13 percentage points, as House Republicans improved further with non-college whites and lost ground among college-educated whites.

On Tuesday, the gap between the two groups expanded further. Democrats carried only 37 percent of white voters without a college education (compared with 61 percent for Republicans). But Democrats won a 53 percent majority of college-educated white voters (compared with 45 percent for Republicans). Tuesday’s Democratic performance among white voters without a college degree improved just slightly from their weak showings in the 2010 and 2014 midterms, when they carried only about one-third of them each time. But their showing with college-educated whites on Tuesday represented a big improvement from those two previous midterms, when they carried about two-fifths of them in each election, according to exit polls. This week, Democrats not only carried 59 percent of college-educated white women, an unprecedented number, but reached 47 percent among college-educated white men; they hadn’t reached even 40 percent among those men nationally in any House election since 2008.

That surge in white-collar support sparked the gains in suburban seats that produced the Democrats’ first House majority since 2010. But the persistence of Trump’s appeal with blue-collar voters, amplified by his frenetic and fear-infused final campaign swing, frustrated Democratic hopes in several districts with substantial rural populations that they had earlier hoped to capture, including seats in Kansas and Kentucky. Trump also helped Republican Ron DeSantis win the Florida governor’s race by following the same model the president employed to carry the state in 2016: big margins in small places, which allowed them to overcome commanding Democratic advantages in the urban centers. Rick Scott, whose Florida Senate race appears to be heading for a recount, benefited from this dynamic as well.

On each side there was some regional variation in this pattern. The Rust Belt Democratic Senators who won reelection in Ohio, Wisconsin, Pennsylvania, and Michigan each pushed past 40 percent among white voters without a college education. That was hardly a commanding performance, but it represented at least some recovery from Hillary Clinton’s anemic showing in the same states in 2016—and it is sure to encourage Democrats who want to focus on recapturing those states as the most straightforward path back to the White House in 2020.

Conversely, the Democratic performance among college-educated whites in the South—who tend toward more conservative positions than their counterparts elsewhere, particularly on social issues—continued to lag. O’Rourke did capture just over two in five college-educated whites, which was a notable improvement over earlier Democrats in Texas (who have often struggled to win more than 30 percent of those voters), but it wasn’t enough to overcome Cruz’s distinct advantage among non-college whites, who gave him about three-fourths of their votes, according to the exit poll. Abrams, even more strikingly, lost over four-fifths of whites without a college degree, while attracting just a little over one-third of those with one. That was also better than Georgia Democrats had done in the past, but—pending the final ballot counting—not enough to win. The key to Gillum’s loss, a big letdown for Democrats, may have been his inability to win more than about one-third of college-educated white men (even as he won nearly three-fifths of white women with a college degree).

Looming over all of this was the intensely divisive figure of Trump. As noted above, his approval rating stood at 50 percent or more in almost all of the states where Republicans notched important victories. But nationally, just 45 percent of voters approved of Trump’s performance, while 54 percent disapproved, according to the exit polls. And while 88 percent of those who approved of Trump said they backed Republican House candidates, fully 90 percent of those who disapproved said they voted Democratic. The correlation between attitudes toward the president and the vote in congressional races has been growing in recent years. But among both Trump’s supporters and his detractors, the connection between attitudes about the president’s performance and the House vote on Tuesday night was the highest recorded in exit polls since at least 1982.

Those numbers quantify the outsized shadow Trump is casting on American politics. Even as suburban voters in major metro areas from coast to coast registered an emphatic statement of discontent with his direction and performance, the small-town, exurban, rural, and blue-collar areas that powered his victory reaffirmed their commitment to his cause. Tuesday’s results only widened the persistent gulf between those coalitions and set up even more intense conflict between them moving forward.

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