What We Actually Know About ‘Electability’

Welcome to Secret Identity, our regular column on identity and its role in politics and policy.

In the last edition of this column we looked at “electability” as a concept, and some of its more problematic undertones. This week, let’s explore the nuts and bolts of electability — what factors make someone more (or less) likely to win an election. I know everyone is thinking about this question in terms of who the Democrats should nominate to take on President Trump in 2020, but I want to explore both presidential races and down-ballot contests, since we have much more data on the latter.

First, two really important caveats. One, there have been only 58 presidential elections in U.S. history. That’s a fairly limited data set — and one of the main reasons why you should be skeptical when someone confidently suggests one potential presidential aspirant is more or less electable than another. Second, the way we think and talk about politics often overemphasizes candidates and campaign strategies and underplays what political scientists refer to as the fundamentals — factors like the state of the economy and which party controls the White House at the time of the election. One view of the 2008 election, for example, is that Barack Obama ran a brilliant campaign that overcame America’s racial divides. Another view is that once Obama won the Democratic nomination he was virtually a lock — the GOP was trying to win a third straight presidential term, which is generally hard to do, and there was a stock market crash a few weeks before the election.

With those caveats in mind, I’m going to start with a Gallup poll from 2015, because I think it’s telling in some important ways. Gallup asked people if they would vote for a presidential candidate from their own party if the person were “well-qualified” and were, say, Muslim, or Jewish or a socialist. (Gallup had 11 categories in all.) More than 90 percent of Americans said they would vote for a woman, a black candidate, a Jewish candidate or a Latino candidate, while fewer than 10 percent said they would not. By this measure, women and nonwhite candidates are pretty electable. (People could be lying to pollsters, but let’s take people at their word for now. And even so, 7 or 8 percent of the electorate isn’t nothing.)

In contrast, 50 percent of Americans said they would not back a socialist candidate. More than a third were opposed to an atheist or Muslim candidate.¹ Similarly, in a Pew Research Center 2016 survey, more than 40 percent of Americans said they would be less likely to vote for a candidate who is Muslim and more than half said they would be wary of someone who did not believe in God.²

So, if you want to have a conversation about whether an atheist or Muslim candidate for national office is “electable,” you at least have some data to support asking the question. (The better way to phrase that question, of course, is probably, “Are Americans too Islamophobic to elect _______.”) In addition, these are categories, not people. How does Americans’ resistance to a “socialist,” for example, relate to the once-and-maybe-future presidential candidate Bernie Sanders, who describes himself as a democratic socialist? We are not totally sure, but it’s worth remembering that many of his views are shared by other leading Democrats — and voters might consider him more of a Democrat than a socialist if he is the party’s 2020 nominee.

Electability concerns are still raised about plenty of candidates who don’t fall into those categories, particularly women, minority candidates and people with views more out of the ideological mainstream of U.S. politics. And just because Americans say they wouldn’t rule out voting for these candidates, that doesn’t mean those identities have no effect on how elections play out. So, let’s look at some of those other factors now, where the electability effects perhaps aren’t as explicit.

Women

This is a tricky question to answer. The best data we have on how gender affects elections is in down-ballot races. An extensive study of U.S. House races from 1982 to 2012 by the University of Georgia’s Jason Anastasopoulos found no “gender penalty” for women candidates. That was true in terms of raising money and in garnering general election votes.³

“Based on a systematic analysis of the 2010 and 2014 midterm House elections, we found not only that women win at equal rates, but also that the content of women’s and men’s campaigns looks the same, the volume and substance of the media coverage they receive is indistinguishable, and voters assess male and female candidates on a variety of issue competencies as equals,” said Jennifer Lawless, a professor at the University of Virginia, referring to research she did with Danny Hayes of George Washington University for the 2016 book “Women on the Run.”

“In short, it all comes down to party,” she added. “Whether there is a D or an R in front of a candidate’s name is far more important — to voters, donors, journalists, and their opponents — than the presence or absence of a Y chromosome in a candidate’s DNA.”

That’s Congress, though. Many observers have asked whether gender plays out differently in the context of a presidential election. Two scholars at the University of Texas at Dallas concluded that attitudes about gender were a much bigger factor in determining how Americans voted in 2016 compared to 2012 (when there were two male candidates), but on balance that helped Clinton because the number of voters with sexist views were outnumbered by those with less sexist views. Clinton did not do as well as Obama electorally, but that probably had more to do with the fundamentals of the race (the fact that she was running for the third term of her party was a disadvantage for example). Other scholarship also shows that attitudes about gender did correlate more closely with how people voted in 2016 than in previous elections — but so did racial attitudes. (So it’s likely that Trump brought gender and race to the forefront in 2016 as much as Clinton.)

But we should be cautious here. Clinton was the first woman to win a major party presidential nomination. If we don’t have a big sample of presidential elections in general, we have far less than even that for women running for president. So I don’t think we can dismiss the idea that Clinton’s gender played some role in the campaign. There were facets of the campaign that are difficult to explain without gender — such as the somewhat odd view of American voters throughout 2016 that Clinton and Trump were basically equally untrustworthy. From a scholarly perspective, however, Clinton followed the pattern — there is not a big penalty in terms of vote share for being a woman in U.S. elections.

“In terms of most of the academic and advocacy research we have, there is no voting or fundraising penalty in the raw data we have about who votes for women as nominees,” said Shauna Shames, a political scientist at Rutgers University and expert on the role of gender in politics.

“But that does not mean there is no gender penalty,” she added. “It just means, to my mind, that only the stronger female candidates run, and those can raise as much money and votes as the mediocre men. We should actually therefore see a penalty against men, in favor of the women who make it through the (tougher for women) gauntlet of becoming candidates, and we should see these women raising more money than men. But we do not. That the levels are equal, therefore, does not prove to my mind that there is no penalty for women — it suggests in fact that there is one, just one we don’t see.”

Race and ethnicity

African-Americans — Generally, there is evidence that black candidates increase turnout among black voters but do worse with white voters, perhaps because black candidates are perceived to be more liberal than white candidates with similar ideological stances.

Obama’s presidential campaigns seem to have conformed to this pattern. University of Massachusetts Amherst professor Brian Schaffner, for example, argued that Obama’s candidacy caused more voters to see the 2008 campaign through a racialized prism and that cost Obama support from people with more racist views. Schaffner estimated that Obama lost about 3 percentage points of the white vote, which comprised about 75 percent of the electorate in 2008. So that’s about 2 points overall. Economist and data scientist Seth Stephens-Davidowitz estimated that Obama lost about 4 percentage points in the national popular vote in 2008 and 2012 because of his race. Other experts, such as Nate Cohn, who’s now with The New York Times, have argued that that overestimates the racial backlash.

The flip side for Obama is that the black turnout rate was more than 65 percent in both of his runs, compared to about 60 percent in 2004 and 2016, when Democrats ran white presidential candidates, according to U.S. Census Bureau data. That matters. Clinton likely would have won Michigan and Wisconsin with 2012-level turnout among blacks.

Latinos — I found less research on Latino candidates. But there is some evidence they too face a penalty. A paper written last year by Neil Visalvanich of Durham University in the United Kingdom estimated that Latino Democratic candidates performed 6 percentage points worse with white voters than white Democratic candidates of similar ideology. The study, based on 2010 and 2012 congressional election results, found the penalty for Latino Democrats to be higher than that of black Democratic candidates (3 percentage points, according to Visalvanich).

There is evidence, based on mayoral and U.S. Senate elections, that Latinos are more likely to vote for Latino candidates. But direct appeals to Latinos, according to scholars, are likely to cause a backlash among some white voters. And Latino Democratic candidates, like black ones, are likely to be perceived as more liberal than white ones with similar ideologies.

Asians — We don’t have as much data here, But in his study, Visalvanich estimated that Asian Democratic candidates did better with white voters than even white candidates.

Ideology

Would it help the Democrats in 2020 if they had a “centrist” at the top of the ticket? All else being equal, it’s probably safe to conclude that candidates more removed from the mainstream of American political thought will do worse at the ballot box. There is some evidence, for example, that Republican presidential nominee Barry Goldwater (1964) and Democratic nominee George McGovern (1972) lost by larger margins than other factors would have predicted in their elections because of the ideological extremism of their voting records.

But ideology is somewhat complicated to measure, particularly for people who haven’t served in legislative bodies (like Montana Gov. Steve Bullock, a potential Democratic candidate in 2020) or in any political office at all (like Trump.) Trump’s Muslim plan was perhaps the most radical idea proposed by any recent presidential candidate, but voters had trouble pinning the candidate down on a left-right spectrum before the election. Trump, according to the Pew Research Center, won the plurality of 2016 voters who described their views as “mixed” and basically was even with Clinton among self-described independents.

Still, unlike race and gender, there is some solid grounding to claim that extreme candidates are less “electable.” Who gets labeled “extreme,” on the other hand, tends to be a less empirical exercise.

There are, of course, other characteristics that could make someone more or less electable. Would it help Democrats win the Midwest if they nominated someone from that region? What about the South? We didn’t get into age in this analysis, but in a 2014 Pew poll, 36 percent of people said they were less likely to vote for a candidate in his or her 70s. (Trump was 70 when he was elected in 2016.)

And that fact about Trump and age speaks to the point I started with: We don’t really know who is electable until the election. In terms of 2020, it’s hard to know how much it matters who Democrats nominate.

Maybe it will matter a lot –– because some of the fundamentals favor Trump (he is an incumbent and the economy is strong), so the Democrats may need a savvy candidate to win a race where Trump will be formidable. Or maybe the Democratic candidate won’t matter much at all — that the biggest fundamental of the election will be Trump’s very high disapproval rating, and Americans will vote for just about anyone to replace him.

If you have ideas for future Secret Identity columns, please reach out to me via email (perry.bacon@fivethirtyeight.com) or Twitter (@perrybaconjr.)