Most polls show Joe Biden as the front-runner, but his lead swells when voters are asked about which 2020 contender has the best shot of beating Trump.Carlos Osorio / AP

In the 2020 Democratic primary, electability is like the end of The Sopranos: Everybody talks about it, but nobody agrees on what it means.

Third Way, a center-left think tank, offers important insights on the question of electability in an extensive new study of Democratic primary voters it is releasing this morning. The results, provided exclusively to The Atlantic, signal that primary voters may be judging electability on different grounds than most political insiders believe, and that the verdict on which candidates are most electable may be more malleable than many expect.

“So much of the Beltway chatter about electability is about race, gender, and ideology,” says Lanae Erickson, Third Way’s senior vice president for social policy and politics. “That was not what people talked about.”

But the research also indicates that however voters assess electability, it looms as an overriding factor in the decision making for most Democratic voters. That suggests former Vice President Joe Biden’s consistent lead in polls on the question of which Democrat is most likely to beat Donald Trump could be a sturdier cushion under his candidacy than his opponents may think.

“If those numbers don’t change,” Erickson told me, “it is going to be very, very hard for someone to overtake [Biden], because this is the question on Democrats’ minds: How do we beat [Trump]? I don’t think those numbers are set in stone at all, but that’s the argument [other candidates] need to make.”

To better understand Democratic voters’ views on electability, Third Way worked with Avalanche Strategy, a research firm based in Washington, D.C. Avalanche surveyed 1,600 likely Democratic primary voters using a technique that mixed typical poll questions with open-ended queries that allowed voters to express their priorities and perceptions in greater detail. The firm then used artificial intelligence to draw out patterns from the open-ended responses. The process essentially sought to combine the quantitative sweep of a traditional poll with the qualitative depth of a focus group.

The first major takeaway from the research isn’t surprising: Democratic primary voters feel an overwhelming imperative to beat Trump. Ninety-seven percent of them called defeating him either “extremely” or “very” important. But only about half of the primary voters the group surveyed called it “extremely” or “very” likely that Democrats would in fact do so. “The tension between Democrats’ urgency to beat Trump and their uncertainty that it will happen is why electability is driving the primary process,” writes Ryan Pougiales, a senior political analyst at Third Way, in a memo releasing the study. More than three-fifths of Democrats said they preferred a nominee with the best chance of beating Trump, even if they didn’t agree with him or her on most issues; only about one-fifth prioritized a candidate they agreed with most of the time.

“But how Democrats conceive of electability—specifically as it relates to the adoptable characteristics that they think make a candidate electable—is more complicated,” Pougiales argues.

Maybe the most striking pattern in how Democratic voters assessed electability was what they did not prominently mention. In open-ended questions, the race and gender of a candidate did not surface frequently as an important factor: In other words, at least in how people responded, electability was not just code for “old white man.” The research did not provide exact figures on how often voters raised those issues, but Erickson said, “People did not say race and gender—they didn’t volunteer that. Are they really thinking that in some way? Maybe. But they certainly didn’t volunteer that.”

Another consideration that surprisingly few primary voters raised in weighing electability was a candidate’s ideology. The share of respondents who thought that an aggressively liberal path improved the Democrats’ odds against Trump was slightly larger than the share who favored centrist positioning. But these ideology-focused voters amounted to only 6 percent of the respondents. These “are all tiny slices of the primary electorate,” Pougiales writes. “The bottom line is that Democratic primary voters generally don’t believe that ideology will be the key to beating Trump in 2020.”

Not many more voters said that they believe the issues a candidate stresses will be crucial to beating Trump. Those who did consider issues paramount cited health care followed by both climate change and immigration as the concerns they thought the nominee most needs to stress. But only about one in five cited a strong emphasis on any issue as central to victory.

If these Democratic voters didn’t measure electability primarily through ideology, issues, or identity, how did they measure it? By far, more voters—about two-thirds in all—picked a candidate’s personal qualities rather than any other choice. According to the analysis, the most common response was honesty and integrity, followed by compassion, strength, and then competence. Gender mattered here: Women were more likely to stress honesty and compassion, while men gravitated toward strength.

These responses, in some ways, can seem like Democratic voters are projecting their own views on the electorate’s: They basically responded that the qualities that would make a candidate most electable are the same ones they would like to see in a president. But those projections are far from proven. Trump, after all, won in 2016 even as voters held grave doubts about his honesty and compassion and had mixed feelings about his competence.

The Trump precedent looms large over the final ingredient that Democratic primary voters picked as a key component of electability: campaign tactics. The research found that about three in 10 primary voters thought that specific tactical choices the nominee pursues before Election Day would be crucial to beating Trump.

This dimension, as clearly as any other, exposes the stark division among Democrats about how to respond to the norm-shattering president. The action that most of the Democratic voters surveyed thought would help beat Trump was taking the high road and uniting the country. But what ranked next was the exact opposite: fighting back and standing up for beliefs.

The same divergence was apparent when the survey asked Democrats head-on which traits in a nominee would cause them the most concern. “Too left” and “socialist” led the list, but combined for only about one-fifth of the primary voters. Slightly more voters said that they worried about an alternative set of concerns that included “too old,” “too middle,” “man,” and “white.” Older voters worried more about “too left”; younger voters expressed more concern about “too middle,” “man,” and “white.”

In some ways, this detailed research reinforces what strategists working in the Democratic race already know: Different wings of the party are operating on very different theories about what it will take to oust Trump. Progressives insist that winning in 2020 requires a vanguard liberal agenda to mobilize young and minority voters; moderates say that the goal must be balanced against the need to hold white suburbanites moving away from Trump and to recapture working-class white women.

The major new insight in this research is that as Democrats assess electability, both of these tracks may be less consequential than whether they think that a candidate demonstrates strength, integrity, and an ability to unite the party. “When I think about electability, I think about not writing Republican attack ads for them [with unpopular positions], but clearly a lot of Democratic primary voters are not doing the same formula that I am doing in my head,” Erickson said.

Against all of these conceptual measures of how Democrats assess electability, the 2020 Democrats face the reality that, so far at least, voters consistently give Biden a huge advantage on the question. In this week’s national ABC/Washington Post poll, for instance, 45 percent of Democrats and Democratic-leaning independents picked Biden as the candidate most likely to beat Trump, far more than those who selected Senators Bernie Sanders of Vermont (14 percent) or Elizabeth Warren of Massachusetts (12 percent). Biden’s lead was slightly narrower among young voters and liberals, groups that have been skeptical of him overall, but even with these voters he still leads both senators by at least a 2-to-1 ratio. With voters more favorable toward Biden, his electability advantage on that question expanded to huge proportions: Among voters older than 50, non-college-educated whites, and moderates, the majority of each group picked him as the most electable.

Other recent polls in competitive 2020 states have found a comparable gap. In a Quinnipiac University poll of Texas Democrats released yesterday, Biden led both Sanders and Warren by five to one on the question of electability; in a Quinnipiac Pennsylvania poll from May, two-thirds of Democratic voters 50 and older picked Biden as the most electable—no one else drew more than 3 percent of them.

To Erickson, the research points to a two-part explanation for that consistent advantage. Voters, she said, believe that Democrats must be united to defeat as formidable an adversary as Trump, and Biden—as a former vice president and senator with decades of experience—has the stature that they believe is necessary to coalesce the party. The logic, she said, is that “we know this person … He’s a known quantity who can unite. It’s the risk-averse way of thinking about things if you are worried that this [primary race] could get out of control.”

But because primary voters indicate that they are mostly measuring candidates’ electability on their personal qualities, Erickson believes that there’s still plenty of time for the other contenders to prove that they can also unite the party in the monumental struggle of taking on Trump. “This perception of electability is much more malleable than we think,” she said. “People are assessing them based on their qualities. If you can show you are the uniter, that you are the one who has these qualities, then you can become the electable candidate. It’s not, as much as some worry, a proxy for ‘white dude.’”

History supports that argument: Voters’ perceptions of who can win the general election often have been reshaped by watching who actually wins the primaries. If Biden, for instance, can’t beat Warren in the early states, or vice versa, more voters are likely to question whether the losing candidate can beat Trump.

The Third Way research challenges the dominant narrative in D.C. that when Democratic voters talk about electability, they are really expressing a reluctance to nominate a woman or person of color against Trump. In less than five months, voters in Iowa will start to answer which side of that argument is right.

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