WASHINGTON — The most important factor in the 2020 Democratic presidential primary is also the most unknowable: Electability.
Unlike primary voters in every other recent presidential election, Democrats are still so rattled by the 2016 election that they keep telling pollsters they'll vote for whichever candidate they think has the best chance of beating President Donald Trump, whether or not they like that contender most.
The problem is that no one, whether it's the voters or the professional prognosticators, can predict the future enough to figure out who it will take to beat the president some 500 days from now.
“Electability is in the eye of the beholder," said Patrick Murray, who runs the well-respected Monmouth University poll. "Voters are not very good at predicting electability...It really is ill-informed, but we know that it's not facts —but perception — that drive voter behavior."
At times, it can feel like Democratic primary voters are playing a giant game of Apples to Apples, where you win not by picking the cards you like the most, but by guessing which ones you think your fellow players will like — only in this case, the other players are imaginary swing voters from the future.
And many argue that perceptions of electability are driven largely by national media coverage, creating what analyst Nate Silver called a "feedback loop" and Washington Post media critic Margaret Sullivan dubbed a "self-perpetuating effect."
Some pundits and voters say one candidate — former Vice President Joe Biden — is the most electable, so they tell pollsters they want Biden, which produces media coverage that reinforces the idea that Biden is most electable, which then filters back down to voters concerned about "electability."
"In the old days, it wouldn't matter, but now the national narrative is seeping into voters' conscious in a way it hasn't before," Murray said. "So their answers when pollsters call is based on the national narrative, which then feeds back into the national narrative."
When the research firm Avalanche asked Democratic voters who they would make president if they had a "magic wand," it was often a different candidate than who they said they'd vote for. Many didn't think their "magic wand" candidate could win, so they threw their support behind Biden instead.
So is Biden actually the most electable candidate?
What does electability mean?
Both the reality and perception of electability have been the subject of rigorous research by academics and others.
On ideology, an influential analysis of 30 years of congressional election results published last year by Stanford University's Andrew Hall found that moderate candidates tend to do better, earning as much as a seven percentage point advantage over more extreme candidates.
Other researchers have challenged that finding, arguing that as the country has become more partisan and polarized in recent years, "this gap has disappeared," as Boise State University's Stephen Utych wrote in a paper published in January.
On gender, numerous studies have found that, when women are on the ballot, they tend to win as often as men. For instance, a recent study of congressional elections by Lefteris Anastasopoulos from the University of Georgia found "no evidence of a gender penalty." However, others have found that may be different when it comes to electing a president and sexism may have hurt Hillary Clinton at the ballot box.
The picture is a bit more complicated when it comes to race, with some studies suggesting black candidates drive up African American turnout while performing a bit worse with white voters, though it's not clear race is the cause.
The Reflective Democracy Campaign, which maintains what it says is the only comprehensive database of candidates and elected officials by race and gender, points to the massive surge of women and people of color elected in 2018 and concludes that the data shows that people of color win just as often as anyone else.
The research suggests the overrepresentation of white men in politics is less about voters' discrimination than barriers to entry that keep fewer women and people of color from running in the first place.
Aimee Allison, the founder of She the People, a network of women of color, argues that the conventional image of an electable candidate — typically a white, middle-aged man — is based on old data of who has won the past, not who can prevail today.
"We have to turn the conventional idea of electability on its head to understand what it means for 2020," she said. "It's almost like Joe Biden is campaigning to an electorate two decades late."
Biden's electability is based on his popularity with working-class white men in the Midwest, whose lack of support helped cost Clinton the 2016 election.
But Allison and others argue his represents only one path to victory for Democrat. They say Democrats would actually do better by focusing on their most loyal voters — by boosting turnout in Midwest cities like Detroit, where African-Americans voted in lower numbers for Clinton than they did for Barack Obama.
With a different electoral strategy for taking on Trump, the picture of the most electable candidate changes.
It's an argument Sen. Kamala Harris, D-Calif., made to the Detroit branch of the NAACP last month to promote her own electability.
"There has been a conversation by the pundits about electability and 'who can speak to the Midwest?' But when they say that, they usually put the Midwest in a simplistic box," Harris said, referring to white working-class voters. "It's shortsighted. It's wrong."
Most of the research on electability is based on congressional elections, for which there is vastly more data than presidential races.
And voters think about electing a commander-in-chief differently than they do about electing a member of Congress. For instance, some researchers have found evidence that sexism hurt Clinton's prospects in 2016 and could presumably dampen support for a female candidate in 2020.
"The presidency is a gendered institution, wherein power has been allocated to men and masculinity," was one conclusion from a report by the Center for American Women and Politics at Rutgers University and the Barbara Lee Family Foundation, which uses research to help women get elected.
Whatever the reality may be, it's voters' perceptions of electability that matter most.
To try to understand how voters think about it, NBC News conducted dozens of interviews with Democratic-leaning voters in Iowa in recent months about what type of nominee has the best shot to beat Trump.
The conversations in the first-in-the-nation caucus state revealed electability concerns about the age, gender and race of various candidates.
There was also widespread disagreement on what conclusions to draw from the 2016 election.
Some said Trump's victory taught the lesson that the most electable candidate is a man who can appeal to the Midwest and offer a return to normalcy. Others said the best pick would be someone who can match the president's boldness and aggression. Still others said the most electable candidate would be someone new, different and relatively young.
In other words, there was no overriding consensus — and plenty of conflicting views.
"Every caucus I have walked into, I've known who I wanted," said Sean Bagniewski, the Democratic Party chairman in Polk County, which includes Des Moines. "I have no idea who I want now."
"Winning begets winning"
Heading into next week's Democratic primary debate in Miami on Wednesday and Thursday, candidates have realized they need to argue that they're the most electable.
"We need to beat Donald Trump. So the number one issue for Democrats is electability, as it should be," underdog Sen. Michael Bennet, D-Colo., said on MSNBC on Thursday.
Every campaign has been focused on this, but Sen. Bernie Sanders' team has been particularly preoccupied, since his advisers are aware the conventional wisdom that says the Vermont senator is poorly positioned to take on the president, even though he often polls better against Trump than anyone but Biden.
The subject line of a recent Sanders fundraising email read: "Bernie beats Trump. Bernie beats Trump. Bernie beats Trump. Bernie beats Trump. Bernie beats Trump. Bernie beats Trump. Bernie beats Trump. Bernie beats Trump."
So how does a candidate change perceptions of electability?
"One of the realities of the perception, which is not really right, is that winning begets winning," said Jeff Weaver, Sanders' longtime close adviser. "You do well in primaries and caucuses, because then people begin to believe you are a viable general election candidate...whether that's true or not."