Even from its first flurries, it’s already clear that a presidential race between Hillary Clinton and Donald Trump will radically accelerate the ongoing transformation in the identity of the two major political parties.

One of the key trends in modern American politics is what I’ve called the class inversion—the shift since the 1960s of working-class whites from the Democratic Party to the Republican, and the parallel movement of more white-collar whites from the GOP to the Democrats since the 1980s. A Clinton-Trump race that could prove more competitive than many expected threatens to finally uproot the last vestiges of the class-based political alignment that defined U.S. politics from Franklin Roosevelt through the 1960s.

This generation-long resorting has profoundly reshaped the balance of power both between and within the two parties. Combined with the growth in the minority population, the Democrats’ improving position among college-educated whites has allowed them to win the popular vote in five of the past six presidential elections, despite consistently large deficits among the non-college white voters who constituted the bedrock of their coalition from Franklin Roosevelt through Jimmy Carter. Conversely, the Democratic decline among blue-collar whites has been key to the recent Republican dominance in the House of Representatives.

The class inversion has also transformed each party’s primary electorate. The GOP primary has become more populist—as the growing base of blue-collar whites challenges the hegemony of the party’s traditional white-collar base. Simultaneously the Democratic primary electorate has grown more consistently liberal, as the growing populations of minorities, Millennials, and white-collar whites replace the working-class whites shifting toward the GOP.This year’s nominating primaries have already demonstrated the power of these trends. The growing strength of blue-collar whites inside the GOP keyed Trump’s march to the party nomination: He carried a commanding 47 percent of Republican voters without a college degree, compared to a much less imposing 35 percent of those with at least a four-year degree, according to a cumulative analysis of exit polls by the ABC pollster Gary Langer.

Meanwhile, although Bernie Sanders beat Clinton among whites without a college degree in almost every state outside of the South, she has marshaled enough support among minorities and college-educated whites (as well as older voters) to move within sight of the nomination. Clinton carried college-educated whites in 17 of the 26 states with exit polls; among whites without a degree the numbers were virtually reversed.

The class inversion has gained strength election after election almost regardless of the nominees. But Clinton and Trump—in their bookended strengths and weaknesses—are unusually well-positioned to intensify it. Trump has demonstrated a visceral appeal among many blue-collar voters, especially men, and raised deep alarms among many of those with advanced degrees, especially women. In both the voting during the primaries and in general-election polling, Clinton, in turn has shown extreme weakness among blue-collar men and generated the most support among college-educated white women. The national CBS / New York Times poll released last week captured a nearly perfect parallel: 68 percent of white women with a college degree said they viewed Trump unfavorably, while 67 percent of white men without a degree were unfavorable toward Clinton.

Trump faces big hurdles trying to dislodge the traditional Democratic advantage among non-white voters. So college-educated white men and non-college-educated white women are likely to become important swing groups.

In the history of modern polling dating back to 1952, no Democratic presidential candidate has ever carried most college-educated whites; even Lyndon Johnson fell slightly short during his 1964 landslide. (This analysis uses the American National Election Studies, a poll conducted immediately after the vote, for the elections from 1952 to 1976, and the exit polls conducted by a consortium of media organizations for the elections since.)

From 1952 through 1980, in fact, no Democratic nominee reached even 40 percent with college-educated whites, except Johnson. During that same period, no Democratic nominee failed to reach 40 percent of the vote with non-college whites, except George McGovern in 1972 and Jimmy Carter in 1980. Over these eight elections, every Democratic nominee except McGovern ran better, usually significantly better, among non-college-educated whites than among their college-educated peers. This was a world in which Democrats were the party of people who worked with their hands, and Republicans represented those who wore suits and worked behind desks.

But the period since 1984 has seen an accelerating reversal of that historic pattern. During his landslide defeat to Ronald Reagan in 1984, Walter Mondale ran slightly better among college-educated than non-college-educated whites. In the next three elections, Michael Dukakis and Bill Clinton ran almost exactly as well with both groups.

Since then, every Democratic presidential nominee has run better with college-educated than working-class whites. From Al Gore in 2000 through Barack Obama in 2012, the share of the vote won by the past four Democratic nominees among college-educated whites has exceeded their performance among non-college-educated whites by four to seven percentage points.

That shift is rooted largely in the growing prominence of non-economic issues in national politics. On questions from abortion and gay marriage, to gun control, to whether immigrants benefit or burden American society, or Trump’s proposal to temporarily ban Muslim entry into the U.S., non-college-educated whites (especially men) consistently take more conservative positions than white-collar whites (especially women). Polls by the Public Religion Research Institute, for instance, have found that whites without a college education are significantly more likely than those with advanced degrees to believe the values of Islam are incompatible with American values. And while nearly three-fourths of college-educated whites say in Pew Research Center polling that the growing number of immigrants strengthens the U.S., nearly half of non-college-educated whites say they threaten traditional American values.

By contrast, these trends have not affected minority voters in the same way—or meaningfully dented the Democrats’ dominance among them. Through the 1990s, African Americans followed the pattern evident among whites in first decades after World War II, with Republicans running slightly better among those with college degrees than those without. But in every election since 2000, the Democratic nominee has won at least 90 percent of both college-educated and non-college-educated African Americans. Hispanics have not established a consistent pattern of class allegiance (perhaps partly because the sample size even in exit polls is relatively smaller). In 1988, 1996, 2000, and 2012, the GOP ran better among college-educated than non-college-educated Hispanics; in 1992 and 2008 it performed better among non-college-educated Hispanics, and in 2004, the groups split about equally. But Democrats have carried a majority of both groups in every election over that period.

Among whites, the Democratic gains with the college-educated haven’t yet been sufficient to allow any of the party’s nominees to capture a majority of them. But Democrats have remained more competitive among the college-educated than with their white working-class counterparts—drawing enough to help the party win its popular-vote majority in five of the six elections since 1992.

In early polling, the class inversion between Clinton and Trump is scaling unprecedented heights. In the CBS / New York Times  national poll released last week, Clinton narrowly led Trump among college-educated whites (drawing 47 percent of their vote) but trailed him by fully 20 points among whites without a degree (only 33 percent of whom supported her.) Similarly, the NBC/Wall Street Journal Poll released Sunday placed the two even among college-educated whites (with each drawing 44 percent) but gave Trump a 27-point advantage among those without degrees (only 31 percent of whom backed her.)  Even more dramatically, the ABC / Washington Post survey released Sunday also showed Clinton and Trump running almost exactly even among college-educated whites (with Clinton drawing 45 percent of them), but the Republican leading her by 40 points among those without degrees (only 25 percent of whom supported her). In each case, the gap between her strength among college-educated and non-college-educated whites would be much larger than the widest ever previously recorded for a Democratic nominee (Obama’s seven-point differential in 2008).

Several recent state polls show the same widening chasm. In a WBUR/MassINC New Hampshire poll released last week, Clinton trailed Trump by 14 percentage points among non-college whites but led him by 18 among those with degrees—enough to provide her a razor-thin overall lead. In a Vanderbilt University Tennessee poll also released last week, Trump led narrowly overall after crushing Clinton by 38 points among non-college whites, but only edging her by only four points among their college-educated counterparts (and trailing among minorities). The latest Quinnipiac University Pennsylvania poll showed a virtual dead heat overall, with Trump leading Clinton by 21 points among non-college whites there, but trailing her by five among whites with degrees.

The imposing gender gap in attitudes toward Trump and Clinton sharpen this picture. In the national CBS/NYT poll, Trump led Clinton by 27 percentage points among non-college-educated white men, while she led him by 17 points among college-educated white women, according to figures provided by CBS. The ABC / Washington Post survey recorded an even greater contrast: it gave Trump a staggering 62-point advantage among non-college-educated white men and Clinton a 24-point lead among college-educated white women. State surveys reinforce the pattern. In the Pennsylvania Quinnipiac survey, Trump led among non-college-educated white men by 43 points, but trailed by 23 among college-educated white women. In Quinnipiac’s latest Ohio survey, Clinton’s vote among college-educated white women was 20 points higher than her showing among blue-collar white men.

The big problem for Republicans in this long-term resorting of voters is that the white groups that favor them most are shrinking in the electorate, while the college-educated white women most open to Democrats have been growing.

Overall, the share of the total vote cast by whites without a college degree has declined in every election since 1980 (except between 1996 and 2000 when it remained unchanged). Cumulatively, non-college-educated whites have tumbled from 65 percent of the vote in 1980 to just 36 percent in 2012, according to exit polls. (Census figures show a higher overall vote share for non-college-educated whites, but track a similar pattern of decline.) Meanwhile, college-educated whites increased their vote share from around one-fourth in the 1980s to just over one-third in the 1990s and have essentially remained there since: Exit polls also showed them casting 36 percent of the vote in 2012, the first time they have matched the share of the vote cast by non-college-educated whites.

The remaining portion of the vote has been filled by non-white voters, who have surged from 10 percent during Reagan’s first victory to about 13 percent in Bill Clinton’s initial win to 28 percent in 2012. Looking forward to 2016, Ruy Teixeira, the co-director of the non-partisan States of Change Project, which studies demographic change in the electorate, projects that minorities will increase their vote share by another two percentage points, while non-college-educated whites will fall by slightly more than two points, and college-educated whites will increase marginally.

That growing minority presence allowed Obama to win reelection comfortably in 2012 even though Mitt Romney carried a higher share of white voters than Reagan did in his 1980 landslide. Obama also benefited from the shifting composition of the white electorate itself. In 1984, when Reagan won reelection, white men without a college degree constituted 28 percent of all voters; by 2012, that had fallen to just 17 percent. Conversely, college-educated white women grew from just 11 percent of the vote in 1984 to 19 percent last time, moving past blue-collar white men in the process. (Over that same period, college-educated white men have remained stable at about one-sixth of voters, while blue-collar white women, sometimes called “waitress moms” have slipped from about one-third of votes cast to 19 percent, leaving them equal to college-educated white women.) If the four white groups had represented the same vote share they did in 1984, and voted as they actually did in 2012, Obama’s deficit among whites would have been even larger—perhaps just large enough to tip the election to Romney.

Given these trends, it’s possible that college-educated white women will comprise the largest single block of white voters in 2016. That compounds the challenge that the growing minority vote share creates for Trump; in fact long-term trends suggest that between them, non-white voters and college-educated white women could cast either 49 percent or exactly half of the ballots this year.

Looking at whites by education and gender, those well-educated white women are consistently the best group for Democrats (who have carried them in four of the past six presidential elections and essentially tied among them in a fifth). But the matchup may be especially problematic for the GOP this time: The 55 percent Clinton drew in the CBS/NYT poll (or the 57 percent she attracted in the ABC / Washington Post survey) would represent the Democrats’ best showing among those women at least since 1980, and probably their best performance ever. By comparison, Obama won 52 percent of those women in 2008 and 46 percent in 2012.

In that CBS/NYT survey, Trump holds Clinton to less than 40 percent support among the other three white groups: college-educated white men and non-college-educated white men and women. The ABC / Washington Post survey gave Trump even bigger advantages with those groups.

Even so, given the resistance Trump faces among both minorities and well-educated white women, he may find it extremely difficult to reach a national majority just by winning the remaining whites by an even greater margin than Romney and other recent Republican nominees have done. To top Clinton, job one for Trump may be changing not so much the preferences of the electorate as its composition—by turning out enough working-class whites to reverse the decades of steady decline in their share of the vote.