The Other McCain

"One should either write ruthlessly what one believes to be the truth, or else shut up." — Arthur Koestler

Fact Check: Were 2020 Polls ‘Rigged’? Is There Even Any Need to Ask Anymore?

Posted on | July 23, 2021 | Comments Off on Fact Check: Were 2020 Polls ‘Rigged’? Is There Even Any Need to Ask Anymore?

In Goldfinger, the eponymous villain tells James Bond: “Mr Bond, they have a saying in Chicago: ‘Once is happenstance. Twice is coincidence. The third time it’s enemy action’.” This is a fair appraisal of the alleged “mistakes” in 2020 presidential polls, as John Nolte explains:

Most pollsters, including all the media pollsters, blew 2016 and really blew 2020, and now they’re blaming former President Donald Trump for their own failure.
Something called the American Association for Public Opinion Research (AAPOR) released a report this week that admits 2020 national polling was the worst in 40 years and state polling was the worst in the history of state polling.
The far-left Politico reports the AAPOR is blaming – you guessed it – Trump [emphasis added]:

The most plausible — yet still unproven — theory is that the voters the polls are reaching are fundamentally different from those they are not. And Trump’s rantings about the polls being “fake” or rigged only exacerbate that problem.
“If the voters most supportive of Trump were least likely to participate in polls then the polling error may be explained as follows: Self-identified Republicans who choose to respond to polls are more likely to support Democrats and those who choose not to respond to polls are more likely to support Republicans,” the report reads. “Even if the correct percentage of self-identified Republicans were polled, differences in the Republicans who did and did not respond could produce the observed polling error.”

Oh, okay, guys… Then please explain to me why some pollsters got it pretty close to correct in 2016 and 2020?
Nowhere in Politico’s write-up of this pile-of-shit autopsy does anyone acknowledge the handful of pollsters who came out of 2020 looking pretty good. Among them, are Rasmussen, Trafalgar, and IBD/TIPP.
Obviously, someone knows how to poll even with all those mean Trumptards out there looking to screw up the polling game.
Good grief, this is not difficult.
We all know the problem, we all know the answer to the problem, and still, the pompously named AAPOR is pretending no one knows how to poll in TrumpLand, and nothing will ever improve.
Honestly, how hard would it be to mimic the methodology of pollsters who did it right in 2020? Not hard at all. But nowhere in the Politico report does anyone suggest that pretty gosh-darned simple idea (the far-left Politico sure doesn’t) because these serial liars aren’t interested in getting it right.

Exactly so. The problem with most 2020 polls is plainly evident — the “weighting” of their samples (especially in terms of self-identified Republican/Democrat/Independent voters) was wrong, and this “weighting” problem was so wrong that it could not have been accidental. The fact that a handful of pollsters were able to “weight” their samples accurately, so as to predict the election outcome within a few points, raises the question of why other pollsters did not do the same, given that all of the public polling companies publish their R/D/I numbers.

This was enemy action, as Goldfinger would say — polls as partisan propaganda for Democrats, part of a deliberate disinformation campaign intended to sow discouragement among Republicans. Among the very worst polls were those associated with prestigious major news organizations. For example, CNN’s final national poll, released Oct. 28, the Wednesday before Election Day, had Biden leading by 12 points — TWELVE POINTS! — which was nearly eight points higher than the 4.5 actual margin. You cannot be that wrong by accident. Here, from page 28 of CNN’s final poll report, is the key description of their methodology:

A total of 1,005 adults were interviewed by telephone nationwide by live interviewers calling both landline and cell phones. Interviews were conducted in English and Spanish. Among the entire sample, 33% described themselves as Democrats, 28% described themselves as Republicans, and 39% described themselves as independents or members of another party.
All respondents were asked questions concerning basic demographics, and the entire sample was weighted to reflect national Census figures for gender, race, age, education, region of country, and telephone usage.

So, from a raw sample that was D+5, the consequence of CNN’s “weighting” process was a 12-point advantage for Biden, and even with the benefit of hindsight, they cannot admit where the problem lies.

As glaring as this “enemy action” factor is in the worst 2020 national polls — and CNN’s poll was complete garbage — it was even more obvious in certain so-called “battleground” states that on Election Day proved to be solid for Trump. Consider the case of Iowa, which has been trending “red” for the past decade or more. If you were following the Real Clear Politics average, you might have thought in October that Joe Biden was on the verge of winning Iowa, because beginning in late September, a series of polls showed him either leading or tied with Trump in the Hawkeye State. I remember checking the RCP Electoral College map, which showed Iowa as a “toss-up,” watching Biden (supposedly) pull into a lead there, and thinking: “Damn, if Trump can’t even win Iowa . . .” The specter of a 1980 Jimmy Carter-style wipeout loomed as a shadow over the final month of the campaign. And why? Bad polls.

Not just off by a few points either, but astonishingly wrong — so wrong that only “enemy action” can explain how wrong they were. Consider the New York Times/Sienna College poll of Oct. 18-20, which had Biden leading Trump in Iowa by three points, 46%-43%. Just ask yourself, how likely was it, in the third week of October 2020, that 11% of Iowa voters were still undecided. Do you see my point here? Such a high number of undecided voters less than two weeks before an election ought to raise doubts, even if the numbers otherwise don’t seem strange.

But the numbers in this NYT/Sienna poll were very strange indeed. In 2016, Trump beat Hillary by nearly 10 points — 51%-42% — in Iowa, and so when NYT/Sienna claimed Biden was ahead by 3 points, what they were predicting was a 12-point swing in Iowa. Was that even possible?

Of course not. On Election Day, Trump won 53%-45% in Iowa — meaning that the NYT/Sienna poll missed it by 11 points. “Enemy action.”

Or consider the historical “bellwether” state of Ohio. I clearly recall the unreal experience of being on the ground there in October:

Brilliant yellow maple trees surround my mother-in-law’s home in rural Ohio, and the roads in her part of the state are lined not only with the colorful foliage of late October but also with signs expressing support for President Trump’s election. Anyone can look at the results from 2016 and see that Trump racked up majorities of 2 to 1 or more in this part of Ohio — 72 percent in Morrow County, 71 percent in Crawford County, 67 percent in Knox County, 66 percent in Richland County, 64 percent in Marion County. In fact, Trump won 80 of Ohio’s 88 counties four years ago, defeating Hillary Clinton by an eight-point margin statewide.
If you believe the polls, Joe Biden is neck-and-neck with Trump in Ohio.

And if you believed those polls, you were a damned fool. Trump beat Biden in Ohio 53%-45%, matching the 8-point margin by which he defeated Hillary there four years earlier. So why did so many people believe the race was “neck-and-neck” in the Buckeye State in October? Bad polls — and again, as in Iowa, NYT/Sienna shows up among the culprits, with an early October poll that had Biden leading 45%-44% in Ohio, i.e., nine points off the final. But the real stinkeroo among Ohio polls — arguably the worst single poll of the entire campaign — was Quinnipiac University’s Oct. 1-5 poll, which had Biden ahead in Ohio by five points, 50%-45%, missing the actual result by 13 points.

Anyone looking at the RCP Average of Ohio polls in mid-October would have believed that Trump was on the ropes there. Of the nine Ohio polls conducted between Sept. 17 and Oct. 27, Biden led in six of them, Trump led in two, and one showed a tie, so that in late September, Biden led by 3.3% in the Ohio RCP Average, and as late as Oct. 30, it was still a virtual tie. Only in the final few days did the Ohio RCP average shift slightly in Trump’s favor — by a single point, when in fact he won by 8 points and, everyone ought to be able to admit in retrospect, it was never really possible for Biden to win Ohio, nor was it likely he could even make it very close. The polls were wrong all along — “enemy action.”

When Trump said the polls were “rigged,” he was simply stating a fact.



Comments are closed.