The following article by Ruy Teixeira, author of The Optimistic Leftist and other works of political analysis, is cross-posted from The Atlantic:

As we move into the endgame of the 2022 election, the Democrats face a familiar problem. America’s historical party of the working class keeps losing working-class support. And not just among white voters. Not only has the emerging Democratic majority I once predicted failed to materialize, but many of the nonwhite voters who were supposed to deliver it are instead voting for Republicans.

This year, Democrats have chosen to run a campaign focused on three things: abortion rights, gun control, and safeguarding democracy—issues with strong appeal to socially liberal, college-educated voters. But these issues have much less appeal to working-class voters. They are instead focused on the economy, inflation, and crime, and they are skeptical of the Democratic Party’s performance in all three realms.

This inattentiveness to working-class concerns is not peculiar to the present election. The roots of the Democrats’ struggles go back at least as far as Hillary Clinton’s presidential campaign in 2016, and, as important, to the way in which many Democrats chose to interpret her defeat. Those mistakes, compounded over subsequent election cycles and amplified by vocal activists, now threaten to deliver another stinging disappointment for the Democratic Party. But until Democrats are prepared to grapple honestly with the sources of their electoral struggles, that streak is unlikely to end.

From 2012 to 2020, the Democrats not only saw their support among white working-class voters—those without college degrees—crater, they also saw their advantage among nonwhite working-class voters fall by 18 points. And between 2016 and 2020 alone, the Democratic advantage among Hispanic voters declined by 16 points, overwhelmingly driven by the defection of working-class voters. In contrast, Democrats’ advantage among white college-educated voters improved by 16 points from 2012 to 2020, an edge that delivered Joe Biden the White House

Polling points to a continuation of these trends in 2022. Democrats are losing voters without college degrees while running up the score among college-educated voters. In the latest national New York Times/Siena poll, Democrats have a 15-point deficit among working-class voters but a 14-point advantage among college-educated voters. (The American Enterprise Institute’s demographic-group tracker averages poll results and confirms this yawning gap in Democratic support.)

In part, this results from further deterioration of Democratic support among white working-class voters. But nonwhite working-class voters—especially Hispanic voters—may be following suit. Democrats carried Hispanic voters by 35 points in 2018 and 25 points in 2020. Available data and reporting strongly suggest that this further decline is being driven by working-class voters, the overwhelming majority of this demographic.

In a proximate sense, it’s not hard to see how this might be happening, given America’s economic situation and Democrats’ campaigning choices. But these struggles tie back to the 2016 presidential election. Hillary Clinton’s campaign made two fateful decisions that decisively undercut her ability to beat Donald Trump. During the primaries, facing a stiffer-than-expected challenge from Bernie Sanders, Clinton elected to counter his class-oriented populist economics by flanking him to the left on identity-politics issues. This built on the party’s attribution of Barack Obama’s reelection in 2012 to mobilizing the “rising American electorate,” which ignored his relatively strong performance among working-class voters in the Midwest. For Clinton, turning to identity politics was a way of making Sanders seem out of touch.

After Sanders unexpectedly came close to tying Clinton in the Iowa caucus, she went on the offensive, seeking to characterize Sanders’s class-oriented pitch as racist and sexist. As NBC News reported at the time:

“Not everything is about an economic theory, right?” Clinton said, kicking off a long, interactive riff with the crowd at a union hall this afternoon.

“If we broke up the big banks tomorrow—and I will if they deserve it, if they pose a systemic risk, I will—would that end racism?”

“No!” the audience yelled back.

Clinton continued to list scenarios, asking: ​“Would that end sexism? Would that end discrimination against the LGBT community? Would that make people feel more welcoming to immigrants overnight?”

She continued that line of attack until the moment she secured the nomination. And once that was accomplished and her campaign launched in earnest, she made her second fateful decision, choosing to concentrate on Trump’s character and all the ways he was out of step with the rising American electorate. Studies of her campaign-ad spending reveal that the overwhelming majority of these ads had nothing to say about policy or even policy orientation, instead attacking Trump’s character and his many divisive and offensive statements. Her campaign slogan, “Stronger together,” was an implicit rebuke of Trump on these grounds.

Trump’s ads, by contrast, talked a great deal about policy, albeit not in the careful and detailed way Democrats tend to prefer, but rather discussing in broad strokes issues including trade, immigration, and the betrayal of elites.

The Clinton campaign believed that her strategy was working right up to Election Day, despite signs of softening support in Rust Belt states (though the polls, as we now know, were still overestimating Clinton’s support). The campaign just could not believe that it was possible to lose to a candidate who was so clearly on the wrong side of history.

But lose it did. While carrying the popular vote by two percentage points, Clinton lost three states—Florida, Iowa, and Ohio—that Obama had carried twice, and also narrowly lost Michigan, Pennsylvania, and Wisconsin, three Rust Belt states in the “Blue Wall” the Democrats had carried in every presidential election since 1992. That made for a 306–232 Electoral College victory for Trump (before faithless electors were factored in), despite Clinton’s popular-vote lead.

And, although Democrats did gain six House seats and two Senate seats in 2016, they fell short of flipping either chamber, giving the Republican Party control of both Congress and the White House. The GOP also emerged with complete control of 25 state governments, compared with a mere six for the Democrats. The Democrats had assumed they could capitalize on Trump’s unpopularity to produce a wave election, consolidating power at all levels of government. Instead, it was the Republicans who did so.

Trump’s victory was attributable, above all, to the shift of white working-class voters, including many who had voted for Obama, into the Republican column. In the country as a whole, the Republican advantage among white working-class voters went up by six points to a staggering 31-point margin. White college-educated voters went in exactly the opposite direction, increasing the Democratic advantage among these voters by six points.

But white working-class voters are far more numerous than their college-educated counterparts, particularly in certain areas of the country, such as the Midwest. And it was here that the death blow to Democratic aspirations was struck. In Iowa and Ohio, where Clinton got blown out, white working-class voters moved, respectively, 22 and 15 margin points toward the GOP. And in Pennsylvania, Michigan, and Wisconsin, where Trump’s advantages were very narrow, Democratic support declined by eight, 11, and 13 points, respectively, among the white working class.

These are large shifts, and they were decisive. Simulations show that if Clinton’s white working-class support had matched Obama’s in 2012, she would have carried all these states easily. Indeed, if Clinton had simply managed to reduce her losses among these voters by a quarter, she would have been elected president.

As analysts sifted through the wreckage of Democratic performance, trying to understand where all the Trump votes had come from, some themes began to emerge. One was geographic. Across county-level studies, low levels of educational attainment among white voters were clearly a very robust predictor of shifts toward Trump. These studies also indicated that counties that swung in Trump’s direction tended to be dependent on low-skill jobs, to perform relatively poorly on a range of economic measures, and to have local economies particularly vulnerable to automation and offshoring. Finally, there was strong evidence that Trump-swinging counties tended to be literally sick, in the sense that their inhabitants had relatively poor physical health and high mortality due to alcoholism, drug abuse, and suicide.

The picture was more complicated when it came to the characteristics of individuals who voted for Trump, especially those who had previously voted for Obama. A number of views correlated with Trump voting, including some aspects of economic populism—opposition to cutting Social Security and Medicare, suspicion of free trade and trade agreements, taxing the rich—and traditional populist attitudes such as anti-elitism and mistrust of experts. But many Democrats paid the most attention to studies showing that “racial resentment” and “status threat” bore a strengthened relationship to Republican presidential voting in 2016.

A rigorous accounting of vote shifts toward Trump, however, shows that they were concentrated among white voters—particularly those without college degrees—with moderate views on race and immigration, and not among white voters with high levels of racial resentment. The political scientists Justin Grimmer and William Marble concluded that racial resentment simply could not explain the shifts that occurred in the 2016 election. In fact, Trump netted fewer votes from white voters with high levels of racial resentment than Mitt Romney did in 2012.

Clearly a much more complex explanation for Trump’s victory was—or should have been—in order. Trump’s supporters integrated hostility toward immigration, trade, and liberal elites with a sense of unfairness rooted in a conservative, race-neutral view of avenues to upward mobility. That is why voters in Trump-shifting counties, whose ways of life were being torn asunder by economic and social change, found his message so attractive.

Such understanding was nowhere to be found, however, in Democratic ranks. Instead, the party chose to see in Clinton’s defeat a validation of her message, that racism and xenophobia were the country’s defining forces.

Trump’s rhetoric and actions over his first two years in office provided plenty of evidence to support that interpretation. The growing cultural left linked this to its radical critique of American society as structurally racist, hostile to marginalized communities, and embedded in a rapacious capitalist system that will destroy the planet. In the left’s view, opposing Trump had to be joined to a struggle against all these aspects of oppression, and to social transformation. Otherwise, the oppression would remain even if Trump himself was removed.

This view spread through sympathetic cultural milieus where it already had a considerable presence—universities, media, the arts, nonprofits, advocacy groups, foundations, and the infrastructure of the Democratic Party itself—redefining what it meant to be progressive. In opposing Trump, who was himself so radical, it seemed only reasonable to be radical in return.

But that was not true outside these milieus, where many moderate-to-liberal voters simply wanted to foil the Republicans and get rid of Trump, whom they found profoundly distasteful. These were the voters who provided the shock troops for the #Resistance and powered the defeat of Republican candidates in 2018. These voters were not particularly interested in promoting a radical critique of American society and certainly didn’t see their organizing for Democratic candidates as having any higher basis of unity than wanting to beat Trump and the candidates who supported him.

The result, in 2018, was a very successful election for Democrats, who took full advantage of Trump’s unpopularity. In a historically high turnout for an off-year election, Democrats gained 41 House seats and carried the House popular vote by almost nine points. Those gains flipped the Republicans’ 241–194 House majority to a 235–199 Democratic one.

Democrats also gained seven governorships—including in the swing states of Michigan, Wisconsin, and Nevada—and more than 300 state legislative seats, which helped deliver control of half a dozen state legislative chambers. However, Republicans retained control of the U.S. Senate, actually gaining two seats. This happened because, ominously, they took out four Democratic incumbents in red-leaning states, outweighing the Democrats’ two flips of Republican seats in swing states.

Media analysis of the 2018 election results tended to emphasize suburban and college-educated voters’ shifts toward the Democrats relative to 2016, combined with these constituencies’ high turnout. These voters did indeed turn out at high levels, and there were strong shifts toward the Democrats among them. But detailed postelection analysis of the voter file and other data painted a much more complex—and interesting—picture.

To begin with, the shift toward the Democrats was actually larger in rural than in suburban areas and larger among rural white people than among suburban white people. There were also big pro-Democratic shifts among both white working-class and white college-educated voters—about seven points in each case.

Moreover, despite the election’s stellar turnout and a big surge in new voters, the big-data firm Catalistestimated that about 90 percent of the Democrats’ improved 2018 performance came from persuasion—from vote-switchers—not turnout.

The election also showed promising signs for the Democrats in the three Rust Belt states that delivered Trump’s 2016 victory: Michigan, Pennsylvania, and Wisconsin. Democratic victories in these states—and in Minnesota, which Trump lost by less than two points—were overwhelmingly driven by shifts among white voters away from Trump.

The Democrats who delivered the party’s gains in the House ran on health care, targeting the GOP’s attempts to eliminate popular Affordable Care Act reforms such as protecting people with preexisting conditions. Secondarily, they attacked the GOP Congress’s big tax bill, which primarily provided tax cuts to the affluent and had little noticeable benefit for ordinary voters. Criticism of Trump was generally implicit in those candidates’ stances rather than explicit.

This is a realistic picture of what happened in 2018—of the voters and the politics that delivered victory for the Democrats. But it did not become the dominant interpretation of the election in Democratic circles. Instead of highlighting their success in persuading many voters to switch sides, many Democrats looked at the high turnout of Democratic-leaning groups and credited victory to the mobilization energies unleashed by the rising left of the party.

This emerging conventional wisdom was greatly assisted by the media’s obsession with four new Democratic House representatives elected from deep-blue urban districts, two of whom got their party’s nomination by defeating liberal incumbents in primaries. None of these seats was a gain for the Democrats, but “The Squad,” as they came to be known—and especially the media-savvy Alexandria Ocasio-Cortez—captivated the press. Many in the party’s establishment concluded that it needed to respond to their youthful, radical energy by absorbing their cultural outlook and maximalist politics.

By the time the first Democratic presidential primary debates were held in late June 2019, leading candidates were seeking to outflank one another to the left. Many of these candidates endorsed a wide range of radical policy options: “Medicare for All” reforms that would eliminate private health insurance; a Green New Deal with an aggressive timeline for reducing reliance on fossil fuels; banningfracking; decriminalizing unauthorized migration over the Mexican border; providing health insurance to undocumented immigrants; allowing prisoners to vote; abolishing U.S. Immigration and Customs Enforcement; and promising reparations to the descendants of slaves.

While candidates for the Democratic nomination were floating such ideas, the latter half of Trump’s term unfolded on ever more polarizing terms, leading to Trump’s impeachment by the newly Democratic House on December 18, 2019. Shortly thereafter, the coronavirus pandemic hit the United States, leading to a massive nationwide shutdown followed by a sharp economic contraction and skyrocketing unemployment. In the midst of this, Trump continued roiling the waters with bizarre theories about the coronavirus and how to treat it, leading to a stark politicization of the public-health emergency.

In this situation, Joe Biden was able to take advantage of both his primary opponents’ radical ideas and the chaos of Trump’s governance by striking a moderate note, promising to pursue progressive but sensible policies, restore the “soul of America,” provide the help Americans needed to get through the crisis, and, of course and above all, beat Donald Trump. This was a congenial message to the Democratic primary electorate, starting with Black voters in South Carolina on February 29 and running through every demographic on Super Tuesday and beyond. It turned out that, despite the strenuous appeals of many candidates to the party’s rising left, most Democratic primary voters had more pragmatic and moderate views than the media-anointed advocates for a more radical party. Other candidates’ failure to understand this emptied the field for Biden, who cruised to the nomination after Super Tuesday.

While Biden was wrapping up the nomination, cultural radicalism continued to gather force on the left of the party and among progressive elites. In these quarters, being progressive meant not just opposing Trump and the GOP and supporting Democratic policy priorities, but also a deep commitment to the beliefs and practices of identity politics. They argued that racism, sexism, transphobia, and other forms of oppression were everywhere in America, including within seemingly liberal institutions and the most minute interactions of daily life.

Then the egregious murder of George Floyd in Minneapolis by a police officer sparked an unprecedented surge of protests. Corporations and the entire nonprofit and arts world competed to offer the most ringing affirmations of the need to fight systemic racism. The potential public-health problems posed by mass protests in the midst of a public-health emergency were waved away. The violence that sometimes attended these protests was defended as an unavoidable feature of a righteous uprising and of little importance compared with racist police violence. The Democratic data wunderkind David Shor was fired from the Civis data-analytics firm simply for tweeting out an academic study indicating that such violent protests typically help the right.

Even when “Defund the police” became a popular slogan in protest circles, Democrats did not want to hear a critical word about the Black Lives Matter protests. The slogan—adjacent to, and frequently embracing, police and prison abolition—made its way into the discourse with only mild pushback in Democratic circles. There it was in 35-foot bright-yellow block letters on 16th Street in Washington, D.C., (with a helpful equal sign equating it to Black Lives Matter signage on the same street). Democrats struggled to thoroughly dissociate themselves from “Defund the police” while remaining supportive of the BLM movement.

Usually, candidates attempt to move toward the center in preparation for a general-election campaign. But Biden did the reverse. He formed six “unity task forces” jointly coordinated by Biden and Sanders campaign figures, covering climate change, criminal-justice reform, the economy, education, health care, and immigration. The co-chairs included such lions of the left as Ocasio-Cortez and Pramila Jayapal, now the chair of the House Progressive Caucus, and the task forces themselves were well stocked with Sanders (and Elizabeth Warren) supporters. The task forces produced a blizzard of positions and language considerably to the left of the “moderate, normie” politics upon which Biden had built his successful campaign. And these positions and language found their way into the Democratic Party platform, were incorporated into Biden’s campaign promises and, importantly, determined how the Biden administration made staffing and policy decisions. Despite Biden and his team’s initial insistence that the strenuous leftism found on Twitter wasn’t real life, by the end of the campaign they seemed to be quite happy to act as though it was.

If any in the Biden campaign had qualms about the campaign’s evolution to the left, they might have taken comfort from two facts. One, Biden was ahead by eight points nationally in the middle of September, and by 10 points in the middle of October. In FiveThirtyEight’s final forecast, Biden was projected to carry not only the three Rust Belt states that had handed the 2016 election to Trump (Michigan, Pennsylvania, and Wisconsin) but also Arizona, Florida, Georgia, and North Carolina. Ohio was projected to be close—within a point—and Iowa and Texas within two points. The projected victory margin for Biden in Michigan and Wisconsin was eight points, and in Pennsylvania, it was five points. Democrats were also projected to take back the Senate, attaining a three- or four-seat margin in the body, and to pad their lead in the House.

The campaign’s second comforting aspect was that the obvious contrast it wanted to draw—between the disruptive and chaotic Trump presidency, unable to manage the pandemic and the economy, and the reassuring approach of a moderate, trusted Biden, who would restore America to normality in both realms—appeared to be defining the contest. The stage seemed to be set for a resounding rejection of the incumbent administration and perhaps another wave election for the Democrats.

This made Biden’s decision to absorb the left in his campaign look like a winner, or at least harmless. A big victory was in the offing. But to the Democrats’ great shock, this was not the election they got. Once again, the polls had overestimated Democratic strength across the board.

Biden carried the national popular vote by a little more than four points, better than Clinton in 2016 but far off the gaudy margin that the campaign and political pundits had anticipated. Biden did carry Michigan, Pennsylvania, and Wisconsin, but they were much closer than projected. He won Michigan by less than three points, Pennsylvania by less than two points, and Wisconsin by less than a point. The breakthrough states for Biden—Arizona and Georgia—were taken by razor-thin margins of three-tenths and two-tenths of a percentage point, respectively. And Biden got blown out by eight points in both Iowa and Ohio, which were supposed to be so close.

Democrats, to their astonishment, managed to lose 13 seats in the House, reducing their majority to a narrow 222–213. And they managed to gain the Senate only by improbably winning two runoff elections in Georgia, as Trump prospectively declared the elections “illegal and invalid” and subverted his own party’s electoral chances by suppressing turnout among his supporters. That gave the Democrats a 50–50 tie in the Senate with newly elected Vice President Kamala Harris as the tiebreaker in their favor.

Majorities don’t get much narrower than that. In addition, at the state level, Democrats fell further behind in both governorships and control of state legislative chambers. This was hardly the wave election that Democrats had anticipated.

But Democrats did attain control of the House, the Senate, and the White House—albeit by the thinnest of margins—allowing most of them to ignore, or downplay, the many troubling signs from the election, especially their deteriorating working-class support. That has led them down the path to their current situation. The aftermath of the 2022 election will likely give them another opportunity to reexamine their approach. Will they return to their historical roots? Or will their long goodbye to the working class continue?