The epistemic crisis of political polling
America has an epistemology problem, and it's wreaking havoc on how we think about presidential election polls and forecasts. We think we know things we do not; we reject knowledge legitimately available to us; and there's a serious risk in acting on what we wrongly think we know.
Epistemology is the study of knowledge: What do we know? How do we know it? What's the difference between knowledge, opinion, and conjecture? What's a legitimate source of knowledge, and what is not? Can we trust our senses, our reasoning, our memory, each other?
Most of us, most of the time, don't think about epistemology — and really, how could we? Whatever we believe, we have to live as if knowledge is attainable and we, individually and as a species, are fairly competent in attaining it. It isn't possible to function otherwise in ordinary life.
It is possible to function otherwise in politics, however, and that's increasingly what we're doing. We can't agree about what we know or how we know it. The border between knowledge and opinion is disputed territory. We try to have political debates at the usual level of analysis of established knowledge ("What should we do about this?") without realizing we haven't agreed on the knowledge itself ("What is this?"). But noticing the need to settle that lower-level question is no guarantee against frustration, either, when every potential source of knowledge is subject to doubt.
Evidence of this epistemic crisis is everywhere. It's part (though certainly not all) of the decades-long decline of trust in the news media. It's implicated in the gullibility and quarrelsomeness that has us spreading fabrications and fallacies on social media; so-called "satire" intended to fool rather than enlighten and bemuse; and political memes that too often consist of apocryphal quotations, cheap manipulations, and bad math.
The most quotable indicators of our epistemology problem come from the right. There's President Trump's incessant cries of "fake news," howled in rejection of every journalistic product he dislikes, from the most polemic opinion piece to simple reporting of statements he has made on camera. White House counselor Kellyanne Conway's "alternative facts" is at home in epistemological fog, as is Rudy Giuliani's "truth isn't truth." The instruction here is to doubt other people — particularly the media, but really any "elite" or "expert" offering unwanted claims — and our memories. This is an epistemology, to put it crudely, of living in the moment and ignoring the haters.
The left, meanwhile, will have us doubt our own senses and reasoning. This is an outgrowth of postmodern attention to identity, context, and perspective — the notion that what you can know is dependent on who you are, where you are, and what you can see from that vantage. That's not false, but as a complete epistemic framework it precludes communication rather than nuancing it. It says you can't see what I see, so you can't know what I know or challenge what I believe. "You must understand my experience, and you can't understand my experience," as Columbia University political scientist Mark Lilla has critically summarized.
Epistemology this confused can produce unwarranted certainty and uncertainty at once, and this muddled paradox is showing up in our electoral expectations.
Accurate national polling for presidential elections is less than 100 years old. Within living memory, most people simply could not know who would win. You'd know who was popular in your community, but national data wasn't available. Uncertainty was the warranted conclusion. Then, for several decades, we had relatively reliable polling and relatively high levels of public trust in the press. Feeling reasonably certain you knew who would win made sense.
In recent years, polling has become remarkably granular and accessible. Where once the average voter might catch a few simple statistics on the radio or in the paper, now we can parse detailed demographic breakdowns, check averages and trend lines, play around with various models and scenarios. But at the same time, Americans have come to distrust the media generally and polling specifically, so much so that a majority of Americans report (to pollsters, ironically) they often or almost always disbelieve poll results.
Some of that skepticism is prudent, as 2016's upset demonstrated well. But it's not only prudence, because rejection of what the polls say about who will win the presidency is almost never followed by agnosticism about the likely outcome.
Poll doubters don't want to return to the uncertainty of the pre-poll era. They want to retain the certainty polling provides without being beholden to any unwanted knowledge a more coherent epistemic framework might require them to accept. The claim is not, "We don't know what will happen" (which is eminently defensible). It's rather, "The polls are wrong, but I know the truth" (which is almost always a mistaken extrapolation of anecdotal observations).
This isn't merely a philosophical curio. Epistemic crises can have practical consequences. People who reject polling data in favor of personal impressions of the race could decide that an election outcome that doesn't match those impressions is illegitimate — "election fraud" or a "stolen election," depending on which way it goes. It's not clear what would persuade them otherwise, nor that none of them would respond with violence. The old aphorism says "knowledge is power," but in this case, perhaps, knowledge is peace.
Want more essential commentary and analysis like this delivered straight to your inbox? Sign up for The Week's "Today's best articles" newsletter here.