Skip to content

Poll Axed

As pollsters crash, media credibility plummets

As America’s diehard corps of data journalists and omni-explainers wail over the smoldering ruin of their shiny, digital forecasting models, the time seems ripe for the rest of us to ask why the activity of polling is tethered to the news business in the first place. Yes, yes: public opinion surveys help to gauge general shifts and eddies in the American electorate’s mood, and to flesh out the public’s outlook on various issues and culture-war obsessions. And undeniably, mainstream news outlets position themselves as interpreters-of-first-resort when said mood and outlook pivot unexpectedly—or predictably, for that matter—within the terms set out by their research apparatus.

Now that the past two years’ worth of survey-taking, poll-aggregating, and cross-tab comparing have been exposed as a historic calamity, the race is on to tweak the collation of voter preferences, to iron out the grotesque kinks in presently existing poll analysis or otherwise explain away the core cognitive failures of our pundit class. We hear a good deal about the “shy Trump voters”—the retiring rightist souls who just couldn’t bring themselves to mention their make-America-great ballots in the polite company of a media pollster. (Funny thing, though: in victory, Trump supporters aren’t looking very shy at all.) More plausibly, there’s the suggestion that Trump’s campaign brought a big new contingent of first-time voters on board that stubbornly eluded the “likely voter” filters employed by professional polling operations.

These methodological disputes will rage on, and others will certainly catch fire, in the long season of post-election melancholy ahead. But the effort to fine-tune our plainly broken public-opinion measuring devices sidesteps the bigger question here: why is it, exactly, that such speculative prognosticating comes under the rubric of “news”? As a bevy of postmortems and think pieces ponder what’s wrong with the American polling establishment, shouldn’t we be questioning what public interest polling professes to serve in the first place?

The sensible course of action is to institute a Glass-Steagall style wall of separation between the American press and whatever remains of the polling industry.

As a practical matter, after all, polls are mainly prompts that national campaigns employ to set priorities and detect opportunities on the electoral map. (These live-or-die mandates may explain why the most reliable homestretch polling data appears to have come from within the Trump campaign.) When this model of political resource allocation is uncritically adopted by media campaign operations, it’s a recipe for all sorts of cognitive mischief. Polling can depress turnout when one side seems to be slipping toward ballot-box humiliation—or, for that matter, when a campaign appears poised for a blowout victory. Contingent results based on a slew of flawed premises—that ordinary citizens might openly bruit their innermost ideological preferences with a stranger; that representative voters continue to talk on landlines; that a respondent couldn’t voice an opinion on the spot, and then, Trump-like, reverse it shortly after ringing off a pollster’s call—thus get smuggled into a perverse scheme of historical inevitability. As independent media polling of the electorate took off in earnest in the 1970s, the Gallup Poll’s managing director David Moore offered this glum, prophetic appraisal of the overmeasured American polis: “media polls give us distorted readings of the electoral climate, manufacture a false public consensus on policy issues, and in the process undermine American democracy.” Summing up the woeful state of the media polling world last year, Jill Lepore noted that “turning the press into pollsters has made American political culture Trumpian: frantic, volatile, shortsighted, sales-driven, and anti-democratic.”

It’s also, not incidentally, a fundamental trespass against news values. Yes, weather announcers and sports reporters will indulge in speculative scenarios on their beats—but their readers are not in any position to determine the outcome of the weather or the World Series. (Assuming, of course, that neither Arnold Rothstein nor Simon Bar Sinister is the Platonic image of the news consumer in either profession.) The disastrous Heisenberg effect of simultaneously professing to measure, and instantaneously broadcasting, the public’s impressions of its own views to the public ad nauseam breeds the sort of inescapable paradox that’s fueled many a caper film and M.C. Escher painting. After all, when reported Republican confidence in the country’s economic direction skyrockets immediately after the election, what you’re measuring isn’t so much a cross-section of sober indicator-assessment on the part of conservative market watchers: It’s closer to a collective fuck you delivered in the face of a putatively out-of-touch elite press—via a medium conveniently furnished by the selfsame press in its doomed quest to perfectly register non-elite opinion. (This is all the more the case, of course, when the one safe prediction about the Trump economic team’s program is that it’s not about to generate any meaningful benefits to citizens beyond the golden inner circle of the 1 percent.)

When something as enormous, unrelenting and well funded as the media-polling complex produces a singularly worthless body of research on the scale of the 2016 national polls, the sensible course of action is to institute a Glass-Steagall style wall of separation between the American press and whatever remains of the polling industry. Such a plan would also yield untold collateral benefits. Without media-generated polls driving our news cycles, political punditry would be exposed as an even greater farce than it already is. Cable newscasters and pundit panels might actually have to discuss issues, policy, and ideas rather than incanting faux-scientific statistics in the hulking shadow of an interactive, multi-media election map. Voters could be left in peace, without having to bark out their provisional ballot leanings to bogus social scientists on the make. At the very least, we’d never have to take Chuck Todd seriously again.