What a Failure to Predict Brexit Means for the U.S.

Should the United Kingdom referendum serve as a warning not to underestimate Donald Trump?

Carlo Allegri / Reuters

The United Kingdom’s decision to leave the European Union has set off shockwaves across the globe—and rightfully so. Betting markets, pollsters, and pundits suggested the Brits would uphold the status quo. It didn’t take long before Americans started wondering what a failure to predict the Brexit could mean for the United States in a year where Donald Trump, the Republican Party’s presumptive presidential nominee, has outperformed expectations.

There are certainly parallels to be drawn. A similar demographic profile unites many of the people who voted for Britain to leave the EU and voters who have stood by Trump. “Brexit supporters mirror Trump voters,” Reuters reported, “in that they tend to be older, white, less affluent, and less likely to live in urban areas.” So should the referendum results serve as a warning to Americans not to underestimate the potential of a Trump presidency?

Yes, but there are caveats, according to Henry Olsen, a senior fellow at the Ethics and Public Policy Center who tracks conservative politics in the United States and Europe. “I don’t think that the Brexit means Trump is going to win. That would really be getting ahead of ourselves,” Olsen told me in an interview. “But I think it means that as a political matter, you have to take working class viewpoints seriously, and you have to be open to the possibility that traditional measures of polling are underestimating the political impact of those opinions.”

Polling showed close competition between support for the U.K. to “leave” or “remain” in the EU on the eve of the referendum. In early to mid-June, polls had indicated that voters favored leaving the EU, The Huffington Post polling model shows. But that dynamic reversed in the final days ahead of the referendum. Adding to a false sense that “remain” would prevail, additional surveys released on the day of the referendum showed a preference to stay in the EU edging out an inclination to exit.

One notable trend that emerged, however, was that online and telephone polling tended to yield different results. The results of online surveys ran neck-and-neck when respondents were asked if they supported leaving the EU in the months leading up to the referendum. Meanwhile, telephone surveys tended to favor the “remain” campaign, though there were exceptions. In an analysis of referendum polling, The Huffington Post reported: “the internet poll average estimated a 1.2-point lead for ‘leave,’ while live phone polls had ‘remain’ up by 2.6 percentage points,” concluding that “internet polls were clearly more indicative of the victory for ‘leave.’”

The U.S. presidential primary race also witnessed a split between online and live polling. Trump tended to perform better in online voter surveys relative to live interviews during the primary competition. As Andrew McGill wrote last December, one theory as to why that might be the case is that “a sizable percentage of poll respondents, though willing to punch a phone key to say they support Trump, are still too embarrassed to actually tell another human being.”

It’s not possible to say definitively that a similar dynamic was at work in Brexit polling, said John Curtice, a polling expert who tracks U.K. attitudes toward the EU and the EU referendum. “The trouble is that people can come up with theories as to why saying that a vote to leave would be socially unacceptable, while others might say that a vote to remain would be socially unacceptable,” Curtice said. “When you’re looking at a one-off event you don’t have the litmus test to pick up social desirability bias.”

Over the past few years, Olsen has been tracking how populist political parties fueled by working class voters have fared across Europe. He has observed that “working class populist parties, movements and candidates do better in polling the more anonymous it is,” and “quite often outperform the results predicted for them by pre-election polls.”

In Olsen’s view, “there seems to be a problem in pollsters worldwide inaccurately assessing the propensity of these voters to turn out and what they are going to do.” That might be because voters are less likely to accurately report their political convictions when talking to a person in a live interview out of fear they will be judged harshly, given that critics of conservative populism often criticize supporters of such political movements as xenophobic. It could also indicate that pollsters are underestimating the likelihood that working class voters will turnout to vote.

Anthony Wells, a director of political and social research at YouGov’s London office, said in an e-mail that while “there is academic evidence to support people answering more honestly when there is no interviewer … in the case of the referendum we don’t think it was the reason” for the difference between online and telephone polls. In an attempt to determine what accounted for the discrepancy, “we tried to pin it down using parallel phone and online surveys, including interviewing the same people online and by telephone,” he said, but “people gave the same answer.” Wells noted, however, that  this did not rule out the possibility entirely.

There are many reasons why it would be difficult to draw lessons from the Brexit vote in order to understand the potential for a Trump presidency in the United States. Apart from the obvious differences of culture, geography, and the actual event, there a variety of factors that could contribute to a disparity in online and live polling, such as the way questions are structured. There are also reasons to be careful in assigning significance to any kind of polling disparity. Reflecting on the primary season, Nate Cohn of The New York Times, wrote in May that despite Trump’s advantage in online polling and triumph in a crowded Republican field, the actual results still “weren’t as good for Mr. Trump as the balance of online surveys predicted they would be.”

If nothing else, the results of the Brexit referendum should be a reminder to American pollsters, journalists, politicians, and pundits not to gloss over polling results that don’t fit a preconceived worldview. “If there’s one thing that’s pretty clear it’s that the last people to take a working-class, populist, political movement seriously are people in newsrooms, trading rooms and political corridors,” Olsen said. “They will interpret things against that result coming to pass until the evidence comes up and pokes them in the eyes.” It’s one thing to find flaws in polling. It’s another to discount, intentionally or not, evidence that points to a conclusion that people in positions of power simply don’t want to believe is true.