There is a perception that polling in the 2016 presidential race melted down. But while that happened in key states, overall, most poll results weren’t far off the mark. The final RealClearPolitics polling average showed Hillary Clinton winning by 3.3 percent. Of course, Clinton didn’t win the presidency but did win the overall vote count by 2.1 percent. Clinton ran up her vote totals in heavily populated California, essentially doubling the number of votes Donald Trump received in a state she was never in jeopardy of losing.
A bigger problem emerged from state level polling in critical swing states in 2016. Non-college educated Americans were under-sampled in many critical swing state polls and very late deciders broke for Trump, which was missed by a dearth of individual state polls in the final days. Polling problems aren’t new in presidential elections and that’s the overarching theme in Campbell’s new book “Lost in a Gallup: Polling Failure in Presidential Elections.”
“In North Carolina, none of the three final polls showed Trump was ahead there; he won the state by almost four percentage points,” writes Campbell.
One of Campbell’s conclusions is that more “high quality polls at the state level are needed, conducted close to the election.” The bigger indictment though was the analysis of the polls which proved to be a media echo chamber far removed from America’s heartland.
Campbell’s book is a tour de force on the history of presidential polling. There is an entire chapter on the infamous upset of President Harry Truman over Republican nominee Thomas Dewey in the 1948 campaign. Most readers have seen the iconic photo of Truman holding up a “Dewey defeats Truman” news headline. Pollsters predicted a big win for Dewey and not unlike Trump, Truman railed against what he called the “sleeping polls.”
When it comes to electoral politics, most remember the chicanery and drama of the 2000 Florida recount between George W. Bush and Al Gore, but when it comes to presidential polling the 2004 election between Bush and John Kerry may deserve even more attention. After all, the exit polls predicted a massive win for Kerry on Election Day. One of Kerry’s aides, Bob Shrum, even referred to Kerry as “Mr. President” before the polls even closed. Kerry only worked on his victory speech on election night, paying no mind to a possible concession speech.
Campbell notes that many in the media believed and ran with those exit polls because they wanted to believe them. “On average, exit-poll results at the precinct level overstated Kerry’s advantage over Bush by 6.5 percentage points – the greatest such discrepancy in any presidential election for which comparable data were available,” writes Campbell. While it’s hard to pinpoint such a wide discrepancy, one of the reasons is Kerry supporters were more apt to participate in exit polls, in part because exit poll workers tended to skew younger and female, making them more likely to approach people who were similar to their own thinking and worldview.
Campbell points out in the 2000 election, data was already available that revealed “that Republicans were less likely than Democrats to participate in exit polls, perhaps reflecting suspicions about the media outlets that sponsored the surveys.”
Media bias and interpretation of polls are one of the biggest flaws of political polling. Perhaps there is no greater example of this than the 2016 presidential race. After all, a public opinion poll the following year revealed that 61 percent of Americans declared “they had little to no trust in public opinion polls.” This after incessant declarations by media members that Trump would never be president and several forecasting models predicting Clinton’s likelihood at victory over 98 or even 99 percent.
One of the most problematic trends in the media’s obsession with polls is its inability to understand or explain them. Campbell does a good job of highlighting journalists who have tried to blow the whistle on groupthink and ideological tunnel vision, a problem that appears to be getting worse and not better. Campbell cites Michael Kelly, editor-in-chief of the National Journal, and his words in the Washington Post in 1999:
Reporters like to picture themselves as independent thinkers. In truth, with the exception of 13-year-old girls, there is no social subspecies more slavish to fashion, more terrified of originality and more devoted to group-think.
Some journalists tried to make amends for the errors such as Liz Spayd, a former public editor for the New York Times. She criticized the Times for covering the campaign in a way that essentially championed “a juggernaut of blue state invincibility” for Clinton. Other prominent journalists vowed to listen more to citizens residing in places sometimes derisively dubbed “flyover country.”
In a political environment where the media has doubled down on their protracted war with Trump and many of his supporters, a surprise upset for Trump again might be beyond salvageable when it comes to media credibility. Undoubtedly, polling will be further tarnished and the old adage that “the only poll that matters is on Election Day” will perhaps be more accurate than ever.