As part of his general effort to undermine confidence in any sort of information that suggests he is not the most brilliantly successful leader — nay, human being — in recorded history, President Trump famously disparages public opinion research that tells him things he doesn’t want to hear. Polls (with the exception of those from Rasmussen Reports, which systematically produces pro-Republican, pro-Trump findings) are for him a species of Fake News, and the product of the Fake Media that often sponsors them. He is willing, in fact, to elevate all sorts of phony-baloney pseudo-research above professional polling, or simply misrepresent findings, as Philip Bump noted earlier this week:
Tuesday morning dawned, and President Trump decided, for some reason, to re-litigate the 2016 presidential debates.
“As most people are aware, according to the Polls, I won EVERY debate, including the three with Crooked Hillary Clinton,” Trump said on Twitter …
Trump insists, here in the year of our Lord 2019, that he won all three debates “according to the Polls.” According to the Polls, Trump lost all three debates, by varying margins. According to CNN, Clinton won the three debates, according to 62 percent, 57 percent and 52 percent of those polled, respectively. NBC News polling had Clinton winning with 52 percent, 44 percent and 46 percent of support from respondents (polls that included about a fifth of respondents each time who said neither candidate won).
The “polls” to which Trump is referring are probably the same ones he cited after losing that first debate (which, remember, was in part a function of his faulty mic except also he won the debate). At the time, Trump pointed to useless online surveys like the one typically hosted at the Drudge Report or to little surveys posted on local news sites. These surveys are to polling what playing solitaire is to gambling in Vegas. If you try hard enough, you’ll probably be able to get the outcome you want.
When Trump can’t find the numbers he craves, from whatever shady source, he makes them up:
In an interview with Fox News’s Tucker Carlson that aired on Monday night, Trump also claimed that approval polls showed him excelling — or at least they would if polls were fair.
“A poll just came out today I’m at 54 or 55, and they do say you can add 10 to whatever poll I have, okay?” Trump told Carlson.
Actually, in the vast collection of approval rating polls compiled by RealClearPolitics — a site with strongly conservative ownership, by the way — Trump has never during his presidency achieved an approval rating as high as “54 or 55,” even from Rasmussen. And Trump’s “they do say you can add 10 to whatever poll I have” claim is a complete mystery. “They” in that assertion might as well be “I.”
All across the landscape of MAGA-land, of course, it’s considered axiomatic that adverse polls are “fake” or biased because they all missed the 2016 presidential results. Brad Parscale, Trump’s 2020 campaign manager, recently waxed adamant on the subject, declaring “[i]t was 100 percent wrong. Nobody got it right — not one public poll.”
Now technically speaking no poll — or prediction, or tarot-card prophecy — is going to get an election totally right, so it’s a matter of degree. Again using RCP averages, late 2016 general-election polls had Clinton leading Trump by 3.3 percent in the popular vote. She actually won by 2.1 percent. As Nate Silver pointed out in an exhaustive postmortem, the national polls did just fine, and the sparse state polling wasn’t all that far off either:
Trump outperformed his national polls by only 1 to 2 percentage points in losing the popular vote to Clinton, making them slightly closer to the mark than they were in 2012. Meanwhile, he beat his polls by only 2 to 3 percentage points in the average swing state. Certainly, there were individual pollsters that had some explaining to do, especially in Michigan, Wisconsin and Pennsylvania, where Trump beat his polls by a larger amount. But the result was not some sort of massive outlier; on the contrary, the polls were pretty much as accurate as they’d been, on average, since 1968.
Predictions of who would win in 2016 based on polls, history, various models, and coverage of the campaigns tended to be wrong for a lot of reasons, but the simplest was that Trump drew an improbable inside straight in losing the national popular vote by 2,868,000 votes and then winning every close state other than New Hampshire — notably winning Michigan, Pennsylvania, and Wisconsin by a combined 77,000 votes. Nobody predicted that because nobody could have predicted that. As veteran pollster and poll analyst Mark Blumenthal told me in a phone conversation this week: “Skepticism about the use of polls to forecast elections is warranted, particularly when it comes to state-by-state data that’s harder to come by. But that says little or nothing about the fundamental quality of polls.” And it’s no reason for calling them “fake” or “biased.”
The closest Team Trump has come to making anything like a substantive case about polling quality was in a dismissive mid-June remark from Brad Parscale to CBS News:
“The country is too complex now just to call a couple hundred people and ask them what they think. There are so many ways and different people that are going to show up to vote now,” Brad Parscale, Trump’s campaign chief, told CBS News in an interview taped Tuesday.
“And the reason why is it’s not 1962 anymore. It’s not a place where there’s only a few ways and decisions and every[one] lines up to vote like a good old American. Now there’s a lot of distractions from going to vote. The world’s changed.”
Now this may have just been an impromptu word salad motivated by the need to spin the Trump campaign’s decision to fire three pollsters who had allegedly leaked data showing Trump doing badly against Joe Biden in a large number of battleground states. Parscale was also highly motivated to wave away an especially deadly general-election poll from Quinnipiac showing Trump doing poorly in Florida, the state where he was about to hold his formal reelection launch rally. “None of these polls mean anything,” Parscale added. “It’s the biggest joke in politics. It’s the fakest thing. It’s the fakest thing.” Makes you wonder why his campaign kept two pollsters on the payroll while firing the leakers.
Giving Parscale the maximum benefit of the doubt, maybe he’s suggesting that polls miss the social media magic he supposedly deployed in 2016. He hinted at this in his remarks to CBS:
The way turnout now works and the abilities that we have to turn out voters, the polling can’t understand that, and that’s why it was so wrong in 2016.
Putting aside the fact — not the possibility, but the demonstrable fact — that the polling was not so wrong in 2016, is there something about turnout methods today that polling can’t capture? Has Parscale developed the dark arts of voter mobilization in a manner that eludes detection until the deed is done?
That’s extremely implausible. Yes, pollsters constantly struggle to estimate turnout; it’s the single hardest thing to do, though it’s harder in low-turnout midterms than in (relatively) high-turnout presidential elections. But if anything, they seem to be getting better at it, via a combination of objective data about established voting proclivities (via official voter files that campaigns have used for years, and that pollsters are increasingly using, too) and subjective data on voters’ enthusiasm for participation. Best we can tell, variable turnout patterns were not, in fact, crucial to Trump’s 2016 win, as the New York Times’ Nate Cohn noted after a good look at the final numbers:
[T]urnout improved Mr. Trump’s standing by a modest margin compared with pre-election expectations. If the turnout had gone exactly as we thought it would, the election would have been extremely close. But by this measure, Mrs. Clinton still would have lost both Florida and Pennsylvania — albeit very narrowly.
Could Parscale know something we don’t know about 2020 turnout? That’s not likely either. If there was any mistake about turnout patterns made by election analysts going into 2016, it was probably assuming they would closely resemble those of 2008 and 2012, in which African-American voting participation was elevated by Obama’s candidacy and presidency. Now nobody’s assuming black turnout will be higher in 2020 than in 2016, and all indications are that voter enthusiasm will be high among both party bases. If anything, the arrows are pointing in the direction of a Democratic turnout advantage for the simple reason that anti-Trump voters are not about to assume he’s toast, no matter what the polls and pundits are saying, given the 2016 shocker.
I asked Mark Blumenthal if he could think of any secret Parscale might have to justify his dismissal of polls as incapable of capturing the voters he could drive to the polls, and he carefully answered:
Sure, there is evidence that campaigns have become more sophisticated in turnout techniques, including those involving the social media platforms where Parscale made his reputation. But there’s no reason why polls that ask about enthusiasm or likelihood to vote would not detect rising turnout potential among voters activated by the newer methods.
All in all, it’s likely that Parscale is emulating the signature Trump campaign knack for pure BS in pretending polls that don’t show the president cruising to total victory in all states are a “joke.” Or worse yet, he’s scamming the master scam artist himself and justifying his exalted role in Trump 2020 by depicting himself as a turnout sorcerer whom mere mortals — including the president — cannot comprehend. Either way, he’s literally incredible.