​Trust the Polls

As Donald Trump’s campaign spirals downward, with undecided voters and his party elites abandoning him with equal haste, the only person who doesn’t know he’s toast may well be Donald Trump himself.

That, of course, is hyperbole. His die-hard supporters are just as much in denial. In fact, they, not his ever-shrinking inner circle of former pollster Kellyanne Conway and erstwhile Breitbart editor Steve Bannon, are the ones enabling this delusion, roaring their approval as he dismisses the tales of his campaign’s demise as exaggerations by a scurrilous media. How could he possibly be losing Pennsylvania, where 5.7 million people voted in 2012, when he’s packing 8,000-seat venues in Wilkes-Barre with people who can recite his iconic phrases—now dismayingly including “locker room talk”—verbatim? (I’ll be the first to predict a Bob Dylan landslide in 2020, should he choose to accept it!)

And just check out the polls! No, the real polls, not the “crooked” ones that show him far behind Hillary Clinton in swing states and losing his grasp on Republican necessities like Arizona and even Alaska. If you want to really know who won Sunday’s debate, look at the (nonrandom and nonscientific) polls on the websites of Drudge, Time, and Breitbart (which he admitted was, I kid you not, “slightly conservative”).

The great irony is that Trump was once a great believer in actual polling. Trump taught data journalists (and optimists worldwide) a stunning, valuable lesson with his victory in the Republican nomination. Trump tweeted polls gleefully throughout the primary process; he took the lead in early July 2015 and never looked back. But pundits, often criticized as obsessed with the horse race, maintained a gradually waning confidence that he was a Herman Cain-style flash in the pan, even as his lead lasted...and lasted.

Nate Silver, a founding father of data journalism and frequent defender of the poll, wrote an uncharacteristically gut-feeling-based piece last August that defined the genre and possibly shaped the Trump-bearish conventional wisdom. He estimated Trump’s odds of winning the nomination at just two percent.

In defense of Silver, the two percent figure, which he called “absurd specificity,” may actually have been correct: Things with two percent likelihood happen all the time (about one in 50 times!), and Trump’s rise required the spectacular failure of the Republican establishment to unite behind an alternative. I would even argue that something as minor as the “Rubiobot” incident, which happened when Marco Rubio was coalescing establishment support following a strong finish in Iowa, could have given Trump the election, and the betting markets agree.

Silver also wrote a fantastic mea culpa, “How I Acted Like a Pundit and Screwed Up on Donald Trump.” The takeaway is that data journalists didn’t fail to predict Trump because the data was wrong; they failed because they ignored the data. Public opinion polls don’t have a perfect track record in primaries, far from it. But they’re among the only information we have. When confident Trump supporters and panicking liberals cite Trump’s unexpected primary victories as a reason not to count Trump out just yet, they’re missing the point: People underestimated Trump in spite of the polls, not because of them. In fact, Trump actually slightly underperformed his polls, perhaps because his voters typically had voted less in previous elections than other candidates’.

Now that they show Trump trailing by such a margin that a comeback would be unprecedented, the campaign that made tweeting polls cool is offering no shortage of increasingly dubious reasons that the polls don’t matter. Most are easily refuted with a simple question: What about, you know, the very recent past?

The polls are missing Trump’s supposed deluge of new voters? They didn’t miss them in the primaries. They oversample Democrats? Take it up with President Romney. The pollsters are just “crooked” and set on destroying Trump’s candidacy? They missed a big opportunity from January to June!

An international analogue, Brexit, is giving the left sleepless nights, but Donald turns out to be Mr. Brexit in more ways than one. In the final month before the referendum, the polling average showed Leave winning, but consensus failed to catch up.

But my favorite alternative take on the polls belonged to Kellyanne Conway—again, she is a former pollster—who argued that Hillary Clinton’s dismal numbers, just 46 percent in swing states, reveal a troubling lack of excitement for the first female nominee. (Clinton, I imagine, would frustratedly agree.) Anderson Cooper, who was interviewing Conway, observed that 46 percent is better than 37 percent. “One needs not be a pollster to know that,” Conway retorted. Well, exactly.

Trevor J. Levin, ‘19, a Crimson Arts executive, is (probably) a Social Studies concentrator living in Mather House. His column appears on alternate Thursdays. Follow him on Twitter @trevorjlevin.