It may look as if pollsters got the presidential election horribly wrong, considering that Hillary Clinton had been favored over Donald Trump, who ended up winning. But they weren’t as wrong as it looks.
For one thing, Clinton seems assured of winning more of the total popular vote once all the returns are in. It’s a quirk of America’s electoral system that votes are awarded on a state-by-state basis. The founding fathers set up the Electoral College that way by design, to balance the power given to big states (like New York) against smaller states (like Wyoming). That played to Trump’s strength in rural areas.
Because of the Electoral College system, pollsters will have to de-emphasize popular-vote predictions in the future, according to David Rothschild, the economist at Microsoft Research who runs a political prediction market called PredictWise:
One thing for certain, future of polling will need to be radically different, because avg accuracy across country is worthless with E.C.
— David Rothschild (@DavMicRot) November 9, 2016
Of course, Rothschild and other analysts tried to account for the state-by-state effect. The fact is that most models didn’t capture this year’s demographic nuances. Even the Republican National Committee’s polling models showed Trump losing in the key states he ended up winning: Pennsylvania, Florida, Iowa and Wisconsin, Politico reported.
The enthusiasm gap was arguably a decisive factor: Although the complete figures aren’t in yet, nationwide turnout looks as if it will be in the range of 55.6 percent this year, compared with 57.5 percent in 2012 (and 62.3 percent in 2008).
Before the election, polls suggested that Clinton’s support from Latinos, African-Americans, millennials and women would more than make up for Trump’s support from white men.
On Election Day, however, exit polls showed that Latino support for Clinton was actually weaker than it was for Barack Obama four years earlier (65 percent vs. 71 percent, according to CNN). The same went for African-American support (88 vs. 93 percent) and the 18-to-29 age group (54 vs. 60 percent).
NBC’s exit-poll analysis showed Clinton with the edge among non-white women – attracting 94 percent of the vote among black women, for example. But white women favored Trump, 53 vs. 43 percent, and that segment of the population accounted for 37 percent of the electorate. That prompted some soul-searching among Clinton supporters.
Fellow white women, we need to talk about some things. pic.twitter.com/ea9OxzDrCc
— Andi Zeisler (@andizeisler) November 9, 2016
Trump’s strong populist appeal brought rural, lower-income voters to the polls in stronger numbers than expected, despite all the questions about his private behavior (remember “grab them by the…”?) and his finances (remember the billion-dollar tax write-off?).
This time, the enthusiasm was at its peak for white voters rather than the “Obama coalition.” Setbacks for voting access in key states like North Carolina didn’t help.
“Polling is a mix of estimating the voter population and the sentiment of the voter population,” Rothschild wrote in today’s poll postmortem. “I assume both were off: too few lower educated white voters and educated white women going more heavily for Trump.”
At least one poll built the enthusiasm factor into its analytical model: the USC Dornsife / Los Angeles Times Daybreak poll, which weighted its results based on how strong respondents felt about the support for their candidate on a zero-to-100 scale.
The poll consistently showed Trump with higher ratings than other polls showed, leading many to question the methodology. Today, the Los Angeles Times said its poll “saw what other surveys missed: a wave of Trump support.”
Others, including FiveThirtyEight polling guru Nate Silver, weren’t so sure:
Clinton's likely to win the popular vote by 1-2 points, so the LA Times will still wind up being among the least accurate national polls. https://t.co/iVVwf0s6OP
— Nate Silver (@NateSilver538) November 9, 2016
In any case, the enthusiasm factor is likely to raise lingering questions about FBI Director James Comey’s on-and-off statements pertaining to the long-running Clinton email investigation. When Comey told lawmakers (and the nation) that more emails were being looked at, it brought the issue back into the spotlight at a crucial time.
Even though Comey declared a little more than a week later that there were no grounds to revive the investigation, the tide had clearly turned. And that’s likely to fuel debate long after Trump’s inauguration:
We will always wonder — if Comey hadn't been so arrogant and jumped gun with "It's something…no it's nothing" what would have happened.
— Kurt Eichenwald (@kurteichenwald) November 9, 2016
The story is not Comey, or Johnson, or Stein. The story is that white conservatives did not reject this man, no matter what he did.
— Tom Junod (@TomJunod) November 9, 2016
Corrections for 12:25 p.m. PT Nov. 9: Pollsters aren’t the only folks who get things wrong. I’ve corrected a couple of typos, including a “Trump” when I meant “Clinton,” and a “2018” when I meant “2008.”