How Did the Polls Get It So Wrong? Here's What We Know

Trump supporters were either missed or they mislead pollsters, probably both – a pattern that mirrors the ambivalent yet ultimately loyal Likud supporters who consistently prevail in Israeli elections.

A trader reacts as he looks at financial data on computer screens on the trading floor at ETX Capital, a broker of contracts-for-difference in London, U.K., on Wednesday, Nov. 9, 2016.
Bloomberg

BOCA RATON - Every major poll showed Donald Trump trailing Hillary Clinton in the days before the election – just as they had done about a month earlier. Newspapers and magazines were being laid out with headlines like “Madame President.” Our worst-case scenario was one in which the billionaire who divides the world into winners and losers would refuse to lose – and would not recognize the results of the election. He would deem the system rigged and ask for recounts.

But virtually no one – except for the mysterious polls that Trump occasionally touted but refused to source – showed the Republican candidate beating Clinton in a landslide.

Read More: Iran could be the big winner of the Donald Trump presidency (Zvi Bar'el) || Trump's Win, the Greatest Victory for anti-Semitism in America Since 1941 (Bradley Burston) || I Still Love America. But, After Trump's Victory, I Don’t Trust It (Peter Beinart) ||  Where does Donald Trump stand on Israel? || Follow LIVE UPDATES here

Some polls showed Trump trailing Clinton with a gap of only one percentage point in key swing states like Florida but Clinton, according to most of these polls, would still triumph in the Electoral College. She had emerged as the winner in almost every major poll, racking up solid blocks of blue up and down the East and West coasts as well as nabbing more moderate Midwestern states that have gone Democratic before – the so-called blue wall. On Monday, the New York Times predicted that Clinton had an 85 percent chance of winning the election, and Nate Silver’s FiveThirtyEight put it at 70 percent.

How did everyone get it so very wrong?

As the actual results are unpacked, here is what we know. There was a much higher proportion of Trump supporters in reality than showed up in polling samples. They were either missed, or they were misleading by telling pollsters that they were undecided. Probably both – a pattern that mirrors the ambivalent yet ultimately loyal Likud supporters who consistently prevail in Israeli elections.

Yes, there was a historic surge of Latino voters in key swing states like Florida, many of them voting for the first time, choosing Clinton out of fear of Trump, his immigrant-baiting rhetoric and threats to build a wall. But African-American voters did not come out in as great numbers as they did in 2012 and 2008 to support Barack Obama. Millennials, who helped carry Obama last time around and were so inspired by Bernie Sanders six months ago, also did not show up in the numbers that could have ensured a Clinton win.

A majority of white men came out in support of Trump, and that includes college-educated men, not just the working class. But the pantsuit power crowd – women voting for Hillary and against Trump – did not roar as loudly and broadly as expected. Forty-two percent of women voted for the Trump, which means millions of women were undeterred by reports of the Republican candidate’s predatory behavior and the charges of serial misogyny that has dogged his campaign.

Still, we could have seen the writing on the wall all along. Were we missing it, or were the messages really that cryptic?

“When you ask people their opinion in a phone call, they could be telling you the truth and they could be making it up. Some people do lie,” says Monica Escaleres, an economics professor at Florida Atlantic University and the head of BEPI, the Business and Economic Polling Initiative. BEPI’s poll released Oct. 26 showed Clinton taking 46 percent of the vote to Trump’s 43 percent, with 6 percent undecided. Escaleres notes that among these 6 percent were people who were already inclined to vote for Trump, but were still on the fence – or just reluctant to admit that truth.  

The fact that today, many people only use cellphones could also mean that polls are less accurate than they would have been, say, two decades ago. BEPI uses mixed media collection – calling land lines and conducting online polling. Random and automatic calling to cell phones to collect data is far more expensive in terms of the manual labor required, and thus not used as often.

It’s possible to point to other factors that may have siphoned off votes from Clinton that has seemed all but assured one month ago. Of course, the endless Wikileaks dumps and Democratic Party hacks were not helpful. On October 28, FBI Director James Comey told Congress in a letter that he was examining a newly discovered trove of Clinton emails found on the laptop of disgraced Congressman Anthony Weiner, the estranged husband of Huma Abedin, Clinton’s closest personal aid.

This Sunday, two days before the election, Comey announced that in fact, he was not reopening the investigation against Clinton he closed back in July. The damage is hard to gauge, but it seems that many voters have had their minds made up as early as September. They were angry, they wanted change, they wanted to “take their country back” as if someone has walked off with it. They either were inspired by Trump’s maverick persona, or they simply did not trust Clinton’s promises.