As bad as last night’s election was for Democrats — and it was very, very bad — it was at least as bad for the professional poll and punditry class.
As vote totals accumulated Tuesday evening, with calls for swing states like Ohio and North Carolina, viewers were left with long, stunned, awkward silences from assembled anchors and talking heads. At one point on the liberal-leaning MSNBC Tuesday night, one panelist could be heard gasping, “Jesus!” after a particularly startling reversal of conventional wisdom.
On Wednesday, there was plenty of furious back-filling on the morning shows, with lots of oblique references to assertions made days, weeks or months ago suggesting the possibility of a Donald Trump victory. But the hard fact is that American journalists and their heretofore vaunted analytics compatriots were proven all but universally wrong in judging the appeal of Trump to the country’s voters.
One the current theories (from the same people) for why this happened is a fundamental lack of contact with and/or understanding of Trump’s base voters, as well as the high reluctance of the upper classes of traditional Republicans to admit publicly — to a real live person tallying numbers over the telephone — that they were indeed voting for Trump. The insinuation being that Trump’s reputation for racism, xenophobia, misogyny, yadda yadda precluded polite people from telegraphing their preference for him over Hillary Clinton.
The performance of algorithmic maestros like FiveThirtyEight’s Nate Silver leaves the value of their future analyses under a dense cloud of skepticism. Signing off last night, Silver wrote, “In an extremely narrow sense, I’m not that surprised by the outcome, since polling — to a greater extent than the conventional wisdom acknowledged — had shown a fairly competitive race with critical weaknesses for Clinton in the Electoral College. It’s possible, perhaps even likely, that Clinton will eventually win the popular vote as more votes come in from California. But in a broader sense? It’s the most shocking political development of my lifetime. We’re going to get some sleep, and then we’ll have much more to say over the next days and weeks about how Trump won and what it means for the country.”
Princeton’s Sam Wang, more accurate than Silver in several previous calls, drew the curtain early Wednesday morning writing, “Going into today’s election, many races appeared to be very close: 12 state-level Presidential races were within five percentage points. But the polls were off, massively. And so we face the likelihood of an electoral win by Donald Trump. … In addition to the enormous polling error, I did not correctly estimate the size of the correlated error – by a factor of five. As I wrote before, that five-fold difference accounted for the difference between the 99% probability here and the lower probabilities at other sites. We all estimated the Clinton win at being probable, but I was most extreme. It goes to show that even if the estimation problem is reduced to one parameter, it’s still essential to do a good job with that one parameter. Polls failed, and I amplified that failure.”
Both gentlemen get points for candor and humility, for what that’s worth.
Over in Wisconsin the (until last night) much-respected Marquette University Law School Poll wasn’t even close to accurate. In The Badger Herald, a University of Wisconsin student paper, there’s this: “Pre-election polling consistently had Clinton leading the country and the state, and the Democratic nominee didn’t make a single stop in Wisconsin after losing in the state primaries in April. The latest Marquette University Law School poll for example, had Clinton with a 46 percent to 40 percent lead over Trump. Barry Burden, a University of Wisconsin political science professor, said it remains unclear how the polls both here and nationally failed to predict a Trump victory. ‘There was almost never a poll with Trump tied or ahead, so something has gone fundamentally wrong here,’ Burden said. ‘I tend to think it’s a Trump-specific phenomena. He has just defied all the conventions and expectations we have in electoral politics.’ ”
Among the surpassingly few polling operations that correctly assessed the public mood was the USC Dornsife/LA Times Presidential Election Daybreak Poll. It was routinely dismissed by analysts as an unreliable outlier because of its use of “weighting” and because it is conducted online, not live person-to-live person on the phone. Untraditional though it is, it may have been the best method for this election because it protected the identity of voters either too embarrassed or ashamed to publicly declare for Trump.
As David Lauter explains in the Times, “Most of the summer and fall, the poll’s results have been about 6 percentage points more favorable to the Republican than the polling averages. As of Tuesday morning, the poll’s final forecast for the election showed Trump leading by a little over 3 points, 46.8% to 43.6%. … The biggest difference between the Daybreak poll and most other surveys involves what pollsters refer to as weighting, the process of adjusting a poll’s data to make sure it properly represents the diversity of the population. The Daybreak poll uses a weighting plan that is more complicated than most other surveys — perhaps too complex, critics said. As Ernie Tedeschi, a Washington-based economist and former Treasury Department official, has shown, if you take the Daybreak poll’s data — which USC made available to the public — and weight it more in line with the usual system pollsters use, you get results that largely match the polling averages.
“But as Sean Trende, the political analyst at Real Clear Politics, wrote of the Daybreak poll several weeks ago, ‘truth is not decided by committee.’ The fact that the Daybreak poll was weighted differently doesn’t mean that it was weighted incorrectly, it just means that it is different. Some of the worst failures of polling have come about because pollsters, whether deliberately or not, converged on a single view of an election, in what is often referred to as ‘herding.’ “
Woe to those who consume only the wisdom of the herd.