The Washington PostDemocracy Dies in Darkness

Team Trump uses a hilarious mishmash of polls to boost the president

Meanwhile, the race isn’t tightening.

President Trump speaks at a campaign rally at HoverTech International in Allentown, Pa., on Monday. (Alex Brandon/AP)
Placeholder while article actions load

For President Trump, it was a rare moment of honesty.

“That’s when I like polls,” he said at a rally in North Carolina on Saturday. “You know, when I get them in my favor, I like them. Right? When they come in my favor, I repeat them all the time. I say they’re great. When they’re not in my favor? I don’t. I don’t discuss them.”

With no intended irony, he went on to disparage polls showing him trailing former vice president Joe Biden, particularly ones from Fox News that showed Biden with leads in several battleground states. The example of a poll result he liked and would share? A Twitter poll in which he was viewed as the winner of the last presidential debate by an 80-point margin.

This is a bit like bragging about your proficiency as a parent based on your kids giving you a “#1 Dad” mug. Father’s Day gifts are the real polls.

Again, though, Trump’s overall commentary was completely on the mark. He will elevate any number that seems even remotely positive and disparage anything that is the opposite, regardless of the quality of the bad poll or the crumminess of the good one.

His son-in-law and senior adviser Jared Kushner offered a variant of this argument in an interview on “Fox & Friends” on Monday.

“I do believe that polling with phones to people is an obsolete method, especially in the era of cancel culture,” Kushner said. “You’ve got a lot of, you know, snake-oil salesmen who are — kind of been in the business a long time, and they do this. They were all completely wrong last time, and they didn’t make any modifications going forward.”

An impressive amount of nonsense packed into three sentences.

His “era of cancel culture” line is a tacit reference to the idea that there are “shy Trump voters,” people who are unwilling to share their honest views of the race with pollsters. This assertion has been evaluated repeatedly since before the 2016 race without any finding of significant evidence for the idea. Kushner manages to make the idea goofier, by suggesting that these (vanishingly rare) individuals are worried about being “canceled”: that, in essence, they are afraid that telling a pollster they plan to vote for Trump will lead to that pollster telling the public about their specific preference and the public then culturally exiling the voter? Kushner’s not really saying this will happen; he’s just tying two buzzy Trumpworld concepts (“shy voters” and “cancel culture”) together.

Then there’s his claim that pollsters who have been in the business a long time are hustlers who got 2016 wrong. Maybe he believes that. But, despite some misses in state polls in 2016 (not egregious ones by the standards of polling, mind you), polling in 2016 was quite good. The race was close enough, particularly in the Upper Midwest and Pennsylvania, that the errant polls ended up suggesting that a Hillary Clinton win was significantly more likely than a Trump one. But a Trump win was also a possibility in late polling and forecasts. What’s more, pollsters have in fact refined their methodologies to reduce the likelihood that they miss in the same way that they did four years ago.

As of this writing, even if the polls are as wrong as they were in 2016 in each state, Biden will win the presidency with more than 300 electoral votes.

Reporters flying on Air Force One over the weekend were treated to some delicious polls, freshly picked off Twitter for their consumption.

Among those who haven’t voted yet, Trump leads by 19 points in Florida, 10 in Georgia and 17 in North Carolina. A dominant performance by the incumbent in three must-win states!

It seems quite unlikely that any reporters who have been in the business long enough to earn travel time on Air Force One found this convincing. There’s an obvious question introduced by the isolated data included in the tweet: How is Trump faring among those who have already voted?

The answer is: badly. Floridians who have already voted prefer Biden by 24 points. Georgians prefer Biden by 12, and North Carolinians prefer the former vice president by 25. Overall, therefore, the poll shows close races in all three states.

Remember, each of these is a state that Trump won in 2016 — by 1.2, 5.1 and 3.7 points, respectively. Trump’s campaign manager reportedly thinks that a Trump victory this year depends on holding all three of these states: states where he’s tied (Georgia) or down two points (Florida) or four points (North Carolina).

In other words, this isn’t really good news for Trump. Particularly, given that any campaign would prefer the support of a voter who has already voted than one who plans to. Voters who plan to vote don’t always vote. Part of Trump’s strategy this year has been to argue that votes already cast shouldn’t necessarily count (hence his jeremiad against mail-in ballots), but you get a major storm in one of these states on Election Day or a souring of enthusiasm for the president and things get dark for him quickly.

Again, Trump very much could still win, particularly with polls this close in an unusual political year. But it remains the case that, while the polls tightened a lot in 2016, they haven’t done so to the same degree this year.

Here’s our look at the last 28 days of the 2016 and 2020 campaigns, using polling averages from FiveThirtyEight. It has three columns: overall support for each candidate (using the same colors as above), margin between the two and undecided or third-party voters. Data from 2016 are indicated with dotted lines; final data are indicated with circles at far right.

Your browser cannot display this graph.

The important factor is the extent to which the polls have changed over the last week. Thanks in part to the letter from then-FBI Director James B. Comey that dropped late in the 2016 race, the shifts four years ago were much more significant than this year in most places.

Polling averages, Kushner will be happy to know, are a blend of both traditional pollsters and new additions to the field. These averages include obviously partisan polls of the sort Trump likes to cite. They are still subject to error, but they are at least scientific.

That they don’t say what Trump and Kushner want to hear isn’t their fault. That the president chooses not to talk about them doesn’t mean they don’t exist.