As he set about disembowelling Joe Biden’s Build Back Better bill last month, Joe Manchin paused to offer some cheerful advice for outraged progressives. “[A]ll they need to do,” said the West Virginia senator, “is elect more liberals.”

There was something slightly perverse about Manchin counseling the leftwingers whose policy agenda he was helping tear apart: he sounded like a burglar recommending a home security system as he made off with his loot. And yet his point is undeniable: if progressives hope to gain more leverage in American politics, they must win more elections.

Of course, this is just what they have been trying to do. In the wake of the 2016 Bernie Sanders campaign, an energetic cohort of progressives crashed into city halls, state legislatures, and Congress itself. In the House, “the Squad” may be the most visible avatar of this new leftwing politics, but the larger Progressive Caucus has grown from under 28% of all House Democrats in 2008 to over 43% today.

Still, as the fate of Biden’s bill shows, leftwing influence on national politics remains limited. That’s because progressive gains have not been not responsible for changing the larger balance of power. By and large, progressives have replaced sitting Democrats in deep-blue districts. They’ve failed to prove themselves in contested seats, and have yet to win a major statewide election. Without expanding the Democratic base, progressive advances reflect a larger trend away from the party’s working-class constituency and toward educated urban professionals.

American progressives aren’t alone facing this problem. Over the last half-century, center-left parties all around the world have suffered massive defections from their once-sturdy working-class electoral base. As labor unions have declined in size and power, and economies have shifted away from high-wage blue-collar jobs, millions of working-class voters have moved toward parties of the right. And though pundits obsess over the “white working class”, the phenomenon now clearly extends across racial lines, with non-white, non-college-educated voters breaking significantly toward Republicans since 2012.

But in order to win more elections, especially in swing districts, both Democrats and progressives must win more working-class votes. The numbers here are just overwhelming: in 2020, well over 60% of American voters did not have college degrees. In Congress, over four-fifths of House seats – and 96 of 100 Senate seats – are chosen by electorates where 60% of the voters lack college degrees. For progressives to accept an inevitable decline in working-class support is to accept their position as a permanent and punchless minority.

So how can progressives win back the working class? Here opinions divide sharply. Some liberal analysts argue that an increasingly professional-class Democratic party must restrain its own instincts to suit the preferences of less educated voters. These “popularists” urge liberal candidates to build their campaigns around the party’s safest, best-liked ideas, like lowering prescription drug prices, while tactically sidestepping less popular positions on immigration or police reform. In response, a range of progressive critics contend that such compromises with working-class opinion are either politically ineffectual – since candidate messaging doesn’t matter very much – or morally untenable, since, they say, it would mean replaying Bill Clinton’s rightwing racial pivot in the 1990s.

In conjunction with Jacobin magazine and the public opinion firm YouGov, our team at the Center for Working Class Politics decided to dig deeper into this question than an ordinary survey would allow. Rather than polling voters on isolated policies or beliefs, we designed an experiment to test how potentially Democratic working-class voters respond to electoral matchups. By asking voters to choose between hypothetical candidates, who presented a range of personal characteristics and campaign messages, we were able to develop a richer portrait of voter attitudes at the ballot box. And by presenting this survey to a representative group of 2,000 working-class voters in five swing states – a much larger sample of this demographic than appears in most polls – we were able to focus on these voters in much greater depth.

What we learned in the “Commonsense Solidarity” report may confound both sides of the ongoing debate. The strongest candidates in our sample made bread-and-butter issues their top priorities – jobs and the economy, rather than immigration and racial justice – and spoke about those priorities in universalist, rather than “woke”, identity-centered rhetoric. These differences were even more pronounced among the working-class voters that Democrats and progressives have struggled most to reach, including rural and small-town voters and voters in blue-collar jobs.

In this sense, our findings support the view that working-class voters are sensitive to candidate messaging, and that progressives who want to win their support should put economic issues at the center of their campaigns. But does that mean that Democrats must either tack hard to the center, or abandon their effort to win back a fundamentally “conservative” working class, as some analysts have argued? Not at all.

The voters in our sample preferred candidates who endorsed Medicare for All to those who supported an anodyne centrist alternative, “increase access to affordable healthcare”. And given a choice of political messages, they chose a populist, Bernie Sanders-style soundbite – pitting working-class Americans against wealthy elites – somewhat more often than a moderate, Biden-style message.

Nor does our study suggest that Democrats must “play it safe” by avoiding discussions of racism. Working-class respondents strongly backed candidates who promised to “end systematic racism” over those who offered a bland pronouncement of “equal rights for all”. They did not punish female or non-white candidates – in fact, black candidates performed significantly better than any other group in our sample, even among white voters.

Working-class voters will not punish candidates for advocating for civil rights. But when Democrats frame this struggle in a way that overshadows their commitment to delivering bread-and-butter goods, and when they adopt an activist-inspired, identity-based rhetoric, they are likely to lose working-class votes. Our survey turned up some very large gaps on this front. A populist candidate with a central focus on the economy earned 63% support, for example, while moderates and “woke” progressives with a focus on immigration or racial justice won under 50%.

Combining a populist message with a candidate from a working-class background, meanwhile, stretched these gaps even further. While a moderate military veteran – the kind of Democratic candidate often celebrated by party leaders and the press – received just 51% support, a progressive populist teacher earned over 65%. Strikingly, these preferences were shared not only by Democratic voters, but the critical swing demographic of working-class independents.

To be sure, a choice between hypothetical candidates is different from an actual campaign but unlike most other studies of this kind (using surveys or election data), our experimental approach allowed us to isolate the characteristics that either attract or repel working-class voters to a particular candidate.

Last week’s elections offered one demonstration of what happens when workers’ issues are ignored. In Virginia, Democratic ex-governor Terry McAuliffe was lured into a culture war with Republican Glenn Youngkin, with Toni Morrison’s novel Beloved claiming headlines in the campaign’s final weeks. As economic issues disappeared from view, McAuliffe – a paradigmatic “woke” moderate with no ability to reframe the debate – found himself swamped by working-class defections. Where Biden had won Virginia voters without college degrees by seven points, McAuliffe lost them by 20.

Joe Manchin’s arithmetic is unyielding. If progressives want to exert real power in American politics, they cannot be content to replace establishment liberals in deep-blue seats: they must also prove themselves as an alternative to the cautious centrism that swing-district Democrats prefer. The good news is that this doesn’t have to mean sacrificing bold economic policies or evocative populist rhetoric. But if progressives continue to insist that political messaging is inconsequential, or that it is impossible to adjust their program to the priorities of the working-class electorate, they risk condemning themselves to permanent irrelevance.

• Matthew Karp is an associate professor of history at Princeton University and a contributing editor at Jacobin. Dustin Guastella is director of operations for Teamsters Local 623 in Philadelphia