She the People, an organization that describes itself as a “national network connecting women of color to transform our democracy,” released a survey of political organizers, donors, and activists on potential 2020 Democratic candidates. The results showed Sen. Kamala Harris (D-Cal.) at the top, followed by Representative Beto O’Rourke (D-Tex.) and former Vice President Joe Biden. Senator Bernie Sanders (I-Vt.) came in a miserable eighth.
Clara Jeffery, editor in chief of Mother Jones magazine and longtime critic of Sanders, sent out a misleading tweet summarizing the poll. Because this landed directly at the intersection of online fury—Jeffery is a frequent target of Bernie supporters’ rage, and early polling seems to be a particular object of their interest—this turned into a multi-day firestorm on Twitter. For those lucky enough to have missed it, suffice to say that the entire thing was ridiculous, since the idea of a poll taken of a tiny self-selecting subset of Americans more than a year before any election and months before anybody has even declared their candidacy is completely ludicrous.
Polls in December a full year before the first primary elections are notoriously inaccurate. In late 2014, Bernie Sanders was pulling 4 percent, and our current president wasn’t even on the list. (Another way to gut check this: Do you really think Democrats would nominate Joe Biden, old handsy himself, the man best known for destroying Anita Biden and rolling out the carpet for serial sexual harasser Clarence Thomas, two years after the Year of the Woman 2.0? If early polls are to be believed, they will.) Besides the difficulty of predicting who will actually be on the ballot (Michael Bloomberg keeps threatening to inflict a candidacy on America, and I remain skeptical that Biden will run), it just doesn’t make sense to guess an outcome without knowing any of the themes. It’s like trying to predict the ending of one of Pynchon’s more ambitious novels on the basis of its title. There will be attack ads and support ads and surrogates and debates and, with any luck, actual platforms of substantive policy proposals that will, at least in theory, influence the voters. Taking a poll now, before any of that, is meaningless.
It’s time for leftists to begin treating horse race polls in general with the suspicion they so richly deserve, and to ignore them wherever possible. Now note that I am not talking here about public opinion polling, the sort that shows, for example, that 70 percent of Americans support single-payer health care. That’s useful and actionable information.
But polling whose sole purpose is to predict elections today is less than useless. The fact that one candidate receives a majority of support from a poll’s respondents means nothing in the real world. It is not a reason to vote for or against a candidate, unless your moral barometer is stuck in 10th grade and you base your vote on popularity. At best, these polls let you sound smart as you make predictions at the kind of extremely boring dinner parties where people make predictions about elections. At worst, they’re ruining democracy.
The modern system of horse race polling began more or less as a marketing ploy. George Gallup, who pioneered public polling, only began conducting prediction polls in order to test the validity of his other polls. The thinking went that if a poll accurately predicted the outcome of an election, it must also be showing an accurate picture of public opinion. He, himself, realized the limited use of the tool: “While such forecasts provide an interesting and legitimate activity, they probably serve no great social purpose.”
Even as they provide nothing more than “an interesting and legitimate activity,” polls are an enormous opportunity cost. Back when I wrote a weekly newsletter for Bay Area for Bernie, I devoted an entire section of it to the polls. That little section took hours to compile, hours that I could have, and should have, spent on other things. The media, too, tend to invest vast resources into conducting, reporting and analyzing polls. This, obviously, leaves less time for issues like racism, labor, and poverty. In all of 2016, climate change, which is on track to wipe out life as we know it within the next century, got a whopping 50 minutes of coverage on the major broadcast networks’ Sunday and evening shows. Every minute spent drooling over the polls represents a minute that the media didn’t spend on impending ecological collapse. You can walk and chew gum at the same time, but it’s best not to do it on air.
And it’s not just that polling takes up time and energy best spent elsewhere. They are actively harmful to the democratic process. By creating narratives of momentum, they become self-reinforcing. In late 2015, a few months before the first primaries, I listened regularly to the podcast produced by the number gurus at FiveThirtyEight. In one episode, the hosts read a question from a liberal in a state with open primaries who supported Bernie but wanted to see Trump defeated. The hosts’ advice was to vote for John Kasich since all the polls showed that Bernie wasn’t going to win anyways.
I have no clue whether the listener took the advice or how widespread the practice of ditching a preferred candidate in order to game the election is. (It’s difficult enough to get people to cast a ballot in a primary at all; doing it strategically must be a heavy lift indeed.) But there is a direct correlation to a candidate’s place in the polls and how much media attention he or she receives. (Trump was a major exception; he received media attention that far outstripped his numbers.) And the more media attention a candidate gets, the higher name recognition climbs. It becomes a vicious self-perpetuating cycle: A media organization takes a poll, reports the numbers and diverts its resources to covering the frontrunner; voters are exposed disproportionately to the frontrunner; a sense of inevitability settles in; the media takes another poll and finds that even more voters prefer the frontrunner. The observer effect is in full force when it comes to polling. By attempting to measure public opinion, pollsters actively shape it.
The vast machinery of polling has exactly two purposes: perpetuating itself and entertaining its consumers. Reading the polls is fun, and it can even give you something to talk about when conversation otherwise lags. It gives you a spurious air of wisdom and savvy. Throw the words “crosstab” and “regression” around, and you too may get someone to buy you a drink, if only to shut you up. And as long as you understand that polls contribute exactly the same value to the world, and extract exactly the same cost from it, as the New York Yankees, then you’ll approach them with the right attitude.
A brief note on general election matchups: I do find them useful. These are the polls taken during the primary that match hypothetical Republican nominees against hypothetical Democratic nominees. Around February 2020, they begin to be predictive, and they may be a good guide for who will be best the best candidate to nominate. The rest of it is all just silly. My plea to everyone who intends to participate in the Democratic primary: Enjoy the polls if you want, but don’t take them too seriously. We don’t even know if the US will still be a constitutional republic in recognizable form in 2020, much less who will win the Iowa caucuses. Just sit tight, read up on the candidates, make some buttons, knock on some doors and pray the world doesn’t end before you get to cast a vote.