President Trump’s theory — or, at least, the theory he presented to the audience at a campaign rally in Minnesota last week — is that Democrats are pushing for impeachment solely because they want him out of office, and they don’t think they can beat him in 2020.

It doesn’t take too much digging to identify two big flaws with that theory. The first is that 20 Senate Republicans are almost certainly not going to join with Democrats to agree to boot Trump from office. The second is that polls consistently show Trump trailing most of the possible 2020 Democrats in national and state polling, and often by wide margins.

For that, Trump has a response.

“Now the do-nothing Democrat con artists and scammers are getting desperate,” Trump said in Minneapolis. “Thirteen months, they got to move fast because they're not beating us at the polls and they know it despite the phony, despite the phony polls that you see all the time. They have phony polls, you know, polls are no different. Remember, I always used to talk about polls. I know polls very well. Polls are no different from crooked writers. They're crooked polls, crooked polls — no different.”

Trump did use to talk about polls a lot — back when they showed him winning the Republican primary by a wide margin. Here is a sample tweet:

Since he took office, though, the polls have been fairly grim. He’s resorted to repeatedly lifting up outlier polls from consistently Trump-friendly pollsters that show him hovering around 50 percent approval. On more than a dozen occasions, for example, Trump has highlighted a poll from Rasmussen Reports showing his approval at or over 50 percent.

Trump's line about polls being like reporting is accurate in one sense: If he doesn't like what it says, he dismisses it as fake — even when it's obviously not.

Polling is, of course, a statistical endeavor, in which pollsters collect information and weight it to accurately represent the population being polled. There’s subjectivity involved in determining what that weighting looks like, who’s polled, and what’s asked but it’s largely math — with a long track record of accuracy.

In recent days, Trump’s frustration about polling has been focused not on his approval rating or his chances in 2020 but on impeachment. When Fox News’s pollsters last week announced that 51 percent of respondents believed Trump should be both impeached and removed from office, Trump turned against his favorite cable news network.

“From the day I announced I was running for President, I have NEVER had a good Fox News Poll,” he tweeted on the day the poll came out. “Whoever their Pollster is, they suck.”

As is often the case, Trump was joined in his efforts to brush away bad news by allies and the conservative media ecosystem. In short order, Trump defenders started digging into the Fox poll, raising various specious questions about how it was conducted and who was targeted.

There emerged during the 2012 presidential race a concept called “unskewing,” in which polls that showed President Barack Obama leading Republican nominee Mitt Romney were adjusted by third parties to (in their estimation) better reflect the electorate. Those unskewed polls generally showed Romney winning, though, as history books can confirm, Romney did not win.

The New York Post decided to unskew the Fox News poll.

“Princeton, New Jersey, pollster Braun Research, which conducted the survey, noted 48% of its respondents were Democrats,” the tabloid’s Mary Kay Linge wrote on Friday. “But the actual breakdown of party affiliation is 31% Democrat, 29% Republican and 38% independent, according to Gallup.”

The “real” values? “A poll weighted for party affiliation would have concluded that 44.9% favored impeachment and 44.4% opposed it,” Linge wrote, pointing to “Post analysis.”

There are all sorts of misleading aspects of that “analysis,” including the use of tenths of a percent to make the results appear more statistically accurate. Doesn’t 44.9 percent look more precise than Fox News’s original 51 percent value? Well, it isn’t.

Among the many reasons? The assumption about what Gallup shows is itself wrong. Yes, Gallup’s most recent polling shows the party breakdown above. But most independents typically vote with one party or the other. Democrats and Democrat-leaning independents make up 49 percent of the population — about where the Fox News poll had them. There are a range of other problems with reweighting solely on party and without access to the original numbers, but we think we’ve made our point.

Trump, of course, loved the Post article. On Monday afternoon, he used it to try to debunk a New York Times article that referenced the Fox News result.

Shortly afterward, his 2020 campaign decided to try to fundraise on the idea that the polls were out to get the president. It highlighted a Drudge Report headline about the Fox poll.

“The FAKE POLLS are at it again,” an email from the campaign read. “The same people who GUARANTEED that Crooked Hillary would win in 2016 are once again saying that the American people have turned against me.”

(An amusing footnote to the email from the Trump campaign that was sent to my personal email? The line reading, “I’m much more interested in what Philip from New York thinks, than what Fake News Outlets think.” Why not both?)

I’m not sure that Fox News ever “guaranteed” Clinton would win in 2016, and, in fact, I seem to recall something quite different from that. But Fox News — like most pollsters — did predict Clinton had a national lead over Trump. Which, of course, she did, winning nearly 3 million more votes than Trump. (What did polling get wrong? Some state polls were off the mark on their models of who would turn out, giving Clinton an edge in states that she went on to lose, throwing off electoral vote calculations.)

Again, all of this is very much in keeping with Trump’s strategy from day one. Anything negative is contrived, fake and emanating from an untrustworthy source, even if that source has been celebrated as reliable by Trump in the past, and even if the information provided is ultimately proved to be accurate. Trump lives in a very narrow window of time, a bell curve of existence in which yesterday and tomorrow quickly vanish from view in favor of the now. As long as he keeps his supporters living in that same moment, the immediate worldview he offers them is sustainable. It’s an existence that relies on the assumption that everyone is biased and fickle, which is in itself probably revealing.

But, as the saying goes, the truth will out. And on Tuesday morning, it did, in the form of a new poll from Scott Rasmussen, formerly of Rasmussen Reports. According to Rasmussen, fully 50 percent of the country supports impeaching and removing Trump from office. That’s higher than the percentage of voters who approve of Trump’s job as president in the estimation of Rasmussen’s old firm.

No doubt the New York Post’s crack team of statisticians is already digging into the party breakdowns.


The article was corrected to indicate that the new poll came from Scott Rasmussen, not Rasmussen Reports.