What Is "Electability" and Who Is "Electable"?
Part 1: The term "electability" is not very useful and can be confusing.
Posted March 6, 2020
Questions about "electability" are on people’s minds these days as we move forward throughout the Democratic primaries. The notion of electability is a very tricky one, so let’s try our best to unpack it. There’s some psychology that might be useful here.
In the interest of transparency, I dislike the term "electability" for several reasons, and I don’t think it’s very useful for people in deciding how to cast votes strategically. In my opinion, this term is often applied in a very slippery, inconsistent, and biased way. There is a conflict of interest when politicians claim that they are more "electable" because the implication is that they deserve our votes more than their opponents do.
This mirrors other types of self-serving biases in day-to-day life. Psychology research shows that people often use mental gymnastics to demonstrate how they exemplify abstract qualities better than others do (like having a good sense of humor). Every politician probably believes they are more "electable" than their opponents, just like most individuals believe they’re higher than the average person in kindness or creativity or driving ability. It is statistically impossible for most people to be above average on any variable. So, you say you’re the more electable candidate? How nice for you! I’m sure that you have a completely objective point of view on this.
"Electability" is also often used to mask subtle prejudices by discouraging candidates who appear different from more "traditional" candidates. These are often women, people of color, homosexuals, religious minorities, and/or staunch progressives. The same can be said for comments that a candidate looks or sounds "presidential" (i.e. old, white, male, Christian).
I’m far from the only one to make this observation. When the primary season began last year, news correspondents began to caution that “electability has been a questionable indicator in the past” and “it’s heavily colored by individual prejudices.” In the 2008 primaries, plenty of people called Barack Obama "unelectable," but many observers saw this claim for what it likely really was: A coded message to dampen support for an African American candidate in a country with a long history of racism. Of course, Obama went on to win two terms as President by defeating his white Republican opponents, and both victories were by an unquestionable margin.
Another reason I don’t like "electability" is that whenever discussions about this come up, they always involve predictions about future events (e.g., the outcome of the November 2020 general election). But no one can see the future. Doing this type of forecasting involves a very high amount of uncertainty.
Would you cast your vote for candidate A or candidate B in eight months? You might respond, “Probably candidate A.” But you can’t know for sure. We haven’t seen their general election campaigns yet, and a lot of things about the world might change during this time.
It works the same way for other types of behaviors. Are you going to go on a diet in six months? Will you buy a car in seven months? Will you and your significant other still be dating in eight months? Probably? Perhaps? We can anticipate what is likely, but we’re not 100 percent definitely sure what’s going to happen until it happens.
That’s one of the reasons why Trump’s 2016 victory was so surprising. It caught a lot of pollsters off guard because much of the pre-election survey data had indicated that Clinton would likely win. To better understand the limitations of polling data, I suggest reading these comments from some analytics scientists who have worked on political campaigns.
As Matt Lackey points out, response bias is significantly higher now than in previous years. Only about 1 percent of people will agree to participate in a national survey, while the other 99 percent will decline to participate. Are those 1 percent of participants different from the general population? Likely yes, and this can skew the data: “For example, in 2016 the non-college white voters who took surveys were *not at all* representative of the full set of non-college white voters—the ones taking surveys supported Clinton way more.”
So what does this mean for electability? Well, if we’re trying to figure out whether a candidate like Bernie Sanders or Joe Biden is more "electable," we could look at survey polling data, such as the data from January/February 2020 reported by David Broockman and Joshua Kalla. Their analyses indicate that in order for Bernie Sanders to defeat Trump, he would have to turn out a lot more young voters than ever before. They suggest that this is unlikely to happen, and my own hunch is that they are probably correct. It’s worth noting that the primaries on Super Tuesday did not show an increased turnout for young people, which many Sanders supporters had hoped for. By contrast, Biden would not need as much support from young voters who have never voted before in order to defeat Trump. According to this logic, Biden would be more "electable."
Many of my social scientist colleagues are intrigued by these findings, but they also caution not to over-extrapolate. After all, these results are from survey data asking hypothetical questions about general election campaigns we haven’t seen yet, eight months prior to the election. A lot can happen in that time. Broockman and Kalla themselves say the same thing.
Some folks look at polling data and are concerned about Bernie Sanders’s electability. But another colleague suggested the very real possibility of Joe Biden or his family members coming under investigation right before the election by the Justice Department under Bill Barr, who has already confirmed that they are taking information from Ukraine seriously with respect to the former vice president. This could look really bad for Biden, even if he did absolutely nothing immoral or illegal. It’s perception that really matters, just as it did with the investigation into Hillary Clinton’s emails right before the 2016 election. And there’s no way for us to factor this into account with primary season polling data.
So who is more "electable" then, out of the remaining Democratic party candidates? We can’t really say for sure, which is why I dislike the term. It may not be useful for people who are trying to vote strategically because it leaves us with more questions than answers. Stay tuned and we’ll discuss this topic further in a second post.