By Chris Gavaler
Gavaler is assistant professor of English at Washington and Lee.
I’ve heard it so many times from both Democrats and Republicans that it has the seeming credibility of encyclopedic knowledge: 2016 was an upset that no poll saw coming. Everyone knows that. It’s established fact.
Except it’s not true. Look at the actual numbers. According to the conglomerate polls at Real Clear Politics, the day before the election Clinton was up 3% with a margin of error of 3%. No, she didn’t therefore “technically” win. She actually won. By 2%. That’s the popular vote, what the polls were polling. They predicted 3%, she hit 2%, and with a 3% margin of error, that’s not getting it wrong. That’s a bullseye.
Okay, but, the argument goes, they still blew it because they weren’t looking at the Electoral College. They said Clinton would win there. Except they didn’t. Trump won Pennsylvania, Michigan, Wisconsin, and Arizona. For those to be “upsets,” the polls must have predicted Clinton. That was true only in Wisconsin. The rest were toss-ups because neither candidate was polling outside the margin of error. Going into the election, the polls said Clinton had only 203 safe electors, with 171 in the toss-up range.
Did any pollsters predict a Trump victory? Only one of the 11 I know of. But that’s no fault of the polling numbers. The numbers said that Clinton would win the popular vote and that the Electoral College was a toss-up. They were right.
What are the numbers saying about 2020? As with 2016, it will come down to toss-up states. Combining the four predictions so far (Wason Center, Inside Election, Crystal Ball, Cook Political Report), there are eight states in play: Arizona, Wisconsin, Florida, Pennsylvania, North Carolina, Michigan, New Hampshire, and Iowa. Those last three are outliers; only one of the four predictions include them as toss-ups. Three out of four predictions include the first three, leaving the middle two, Pennsylvania and North Carolina split.
According to Wason Center, it’s already over. Even if Trump takes all four of their toss-ups (Arizona, Florida, North Carolina, Iowa), he still loses in the Electoral College.
According to Inside Election, Trump needs all three of their toss-ups (Wisconsin, Florida, North Carolina), while the Democrat needs only one of them.
According to UVa’s Crystal Ball, both candidates need Pennsylvania plus either Wisconsin or Arizona, leaving New Hampshire irrelevant.
And according to Cook Political Report, though the Democrat has an 11-point lead, both candidates have more than one path through Arizona, Pennsylvania, Michigan, Wisconsin, and Florida to reach 270 in the Electoral College.
But which of these eight toss-up states are serious battlegrounds? Three of the four predictions give Iowa to Trump and New Hampshire and Michigan to the Democrat. Two of the predictions give North Carolina to Trump and Pennsylvania to the Democrat. One of the four gives both Florida and Arizona to Trump and Wisconsin to the Democrat. What does all that yield in the Electoral College?
It gives 259 electors for Trump, and 278 for his opponent. Even if he wins Florida, Arizona, and North Carolina again, he still loses.
Other numbers are worse. Morning Consult lists his approval and disapproval state-by-state, combing them for his net approval rating. In late September, he had a negative score in seven out of the eight toss-up states. The exception was Florida where his net approval was zero. Compare that to his inaugural numbers. From January 2016 to November 2019, Trump dropped from 19 to -4 in Arizona, 6 to -17 in Wisconsin, 22 to -2 in Florida, 10 to -9 in Pennsylvania, 17 to -3 in North Carolina, 7 to -13 in Michigan, 1 to -20 in New Hampshire, and 9 to -12 in Iowa.
Though few are calling Georgia a serious toss-up, he’s dropped from 18 to -3 there. Montana is 25 to -2. Utah is 27 to -3. Nebraska is 23 to -4. Even Texas has dropped from 21 to 2. These aren’t battleground states. These are pieces of his base. It’s not hard to see why the prediction market average at PredictWise.com have the Democrat leading by six points: 54% to 46%. People are literally betting against Trump.
What does this all say about what will actually happen in November 2020? Absolutely nothing. But if the current polls are as correct as their 2016 forebears, Trump is a one-term president.