Donald Trump's enduring strength in his pursuit of the Republican presidential nomination has had a profound, and underappreciated, effect on the Democratic primary campaign. Trump has shown Democratic voters that their ability to estimate "electability" is bad — no one predicted a year ago that Trump would do this well for this long. If electability is so challenging to assess in topsy-turvy times, then it is hard to take seriously as a decision-making criterion.
Among the 34 percent of Democratic voters who told exit pollsters that "honesty and trustworthiness" are the most important qualities a candidate must have, Sanders trounced Clinton, 91 percent to 5. For the 12 percent of Democratic New Hampshire voters who said the thing they value most in a candidate is the ability to "win in November," the ratio was reversed: Clinton defeated Sanders 79 percent to 19.
"Electability" is one of the strangest criteria a voter considers when deciding whom to vote for. Other factors require first-order calculations: Do I agree with the candidate's views on abortion, or on how to respond to a North Korean satellite launch? If we disagree, how strongly do I feel about the issues on which we disagree?
Questions of character are more abstract: Do I trust that the candidate's stated beliefs line up with his or her actual beliefs? Do I think their nerves will hold up under the pressures of being president? Do I trust them to make decisions based on their best perception of the public interest, rather than to serve political expedience as I define it?
None of these questions are, as a rule, easy to answer. But electability adds a whole other layer of complexity: How do I think other voters will decide this set of difficult questions? Each individual voter is forced to become a pundit, speculating on the behavior of millions of other voters, deciding which poll is reliable and which is carrying water for one candidate or another.
Not many voters typically tell pollsters that electability is their single most important criterion. This number is consistently much smaller when there isn't an incumbent president running for reelection, for the obvious reason that electability is harder to judge before a definite opponent exists. In 2004, 20 percent of Democratic New Hampshire primary voters said "can beat Bush" was their top quality in a candidate, while in 2012, 35 percent of New Hampshire Republicans similarly ranked the ability to beat Obama highest. In 2008, only 5 percent of New Hampshire Republicans and 6 percent of Democrats put electability first.
Polls that ask which quality "matters most" will only take you so far — electability matters a great deal to many voters as a subsidiary concern. Most people don't want to "waste" their vote on a loser. In upturning the Republican campaign, Donald Trump has shown voters — both Republican and Democratic — that assessing who is "electable" is trickier than they thought it would be. This frees Democrats who were partial to Sanders, but thought his prospects of electability poor, to express their support for him.
This, in turn, sets off the familiar feedback loop: The more voters express a preference for a candidate, the more electable the candidate becomes. Good polling numbers drive trend stories in the media that drive good polls. "Momentum" is an apt analogy. This dynamic is not new; as Larry Bartels wrote in 1988:
At the beginning of 1976, Jimmy Carter was a relatively unknown one-term exgovernor or a medium-sized southern state...fewer than 5 percent of the Democratic party rank and file considered him their first choice for the party's nomination. Five months later, Carter was quite clearly about to become his party's nominee. ...The events that transformed the Carter of January into the Carter of June seemed to have little to do with politics in the traditional sense — ideologies, interests, or public policies. Carter proposed no innovative solutions to the major problems facing the country, nor did he mobilize any new and potent combination of powerful social groups...he parlayed early victories into media attention, resources, and popular support sufficient to produce later victories and eventual nomination.
Bartels continues: "Despite its recognized political importance, momentum has a certain ineffable quality about it. Experts claim to know it when they see it, but they are not very good at either defining or describing it."
"Momentum" is a word that has a very definite meaning in physics: the product of mass and velocity. You can tell how much momentum something has if you know how much it weighs and how fast it is going. Something with more momentum is more difficult to stop and tougher to turn.
But the analogy breaks when it comes to politics. No one can say, definitively, how much momentum any candidate has at any given time. Jimmy Carter did well in the New Hampshire primary in 1976, defeating Mo Udall, the second-place finisher, by almost 6 percentage points. This gave him a certain amount of momentum — enough, it turns out, to win the nomination, and then the presidency. Hillary Clinton narrowly won the 2008 New Hampshire primary; this did not give her much momentum.
The best and brightest data scientists might try to pin down momentum by getting more and more granular information about the present state of each candidate's standing with the electorate. Get good enough information, the theory goes, and one can infer the momentum each candidate has, à la Laplace's demon. But as Jill Lepore pointed out in an excellent November 16 New Yorker essay, this is something of a fool's errand. She quotes sociologist Herbert Blumer (though not quite at this length):
The formation of public opinion occurs as a function of a society in operation. I state the matter in that way to stress that the formation of public opinion does not occur through an interaction of disparate individuals who share equally in the process. Instead the formation of public opinion reflects the functional composition and organization of society. The formation of public opinion occurs in large measure through the interaction of groups. I mean nothing esoteric by this last remark. I merely refer to the common occurrence of the leaders or officials of a functional group taking a stand on behalf of the group with reference to an issue and voicing explicitly or implicitly this stand on behalf of the group. Much of the interaction through which public opinion is formed is through the clash of these group views and positions.
No amount of polling by Hillary Clinton a year ago could have definitively predicted that Trump would do as well as he has for as long as he has. To the extent that Democratic primary voters can be said to have a "group view," it is fair to say that the "group view" — the consensus — a year ago was that a "democratic socialist" in general, and Bernie Sanders in particular, simply could not win a general election. By convincing the public in general, and the Democratic group in particular, that, this year, no one knows what might happen, Trump directly enabled Sanders's rise.
Blumer again: "In any realistic sense the diversified interaction which gives rise to public opinion is in large measure between functional groups and not merely between disparate individuals."
Just because there aren't many swing voters deciding between Trump and Sanders doesn't mean that the interaction between two functional groups — Republicans and Democrats — is insignificant. The number of voters who think highly of both men is likely vanishingly small. A Quinnipiac University poll taken just before the New Hampshire primary found that 83 percent of Democrats had an unfavorable view of Trump (with only 13 percent favorable), while 66 percent of Republicans view Sanders with disfavor and 13 percent view him favorably.
According to exit polls, Trump and Sanders did well in New Hampshire across income and ideological demographics. Trump got more than 30 percent of the vote in all income categories, while Sanders beat Hillary Clinton among all voters except those earning more than $200,000 a year, a category he only narrowly lost, 46 percent to 53. More surprisingly, Sanders beat Hillary Clinton not only with voters who identify as "very liberal" but also among "somewhat liberals" and "moderates." Similarly, Trump's support was roughly the same across "very conservative," "somewhat conservative," and "moderate" Republican primary voters.
The prospect of Trump makes Sanders seem electable. In that same Quinnipiac poll, he beats Trump 49 percent to 39. By contrast, a Sanders-Cruz election held on that day would be closer (46 percent to 42) and Sanders-Rubio a tie at 43 percent. Of course, a Rubio matchup is less likely today than before New Hampshire (as is a Cruz matchup). But if we take Rubio as a stand-in for a generic "normal" Republican candidate, the point holds.
These poll numbers will surely change. The point I want to make is not one focused on reading the tea leaves of polls too closely. As the past few months make clearer than ever, it makes sense to think of politics as a complex system. Trump's popularity stems not particularly from his ideology, which is not conspicuously more conservative than that of rivals like Ted Cruz, but because of his peculiar personal appeal. Absent Trump's sui generis candidacy, Sanders would not be doing nearly as well as he has been.
As this New Yorker cartoon puts it: "Who cares if he's not electable? Nobody's electable anymore." Which is to say, everybody is. More precisely: Donald Trump is seen as easy to defeat in a general election. Democrats begin to imagine they will not have to fight for the last swing voter in the suburbs of Cleveland and need not ponder how such a swing voter would see a candidate, but can instead vote for the candidate who shares their values and positions on the issues, both categories that Sanders won decisively in New Hampshire.