Political pollsters: Apologize to Trump, Biden and the American people for getting it (really) wrong
Let’s say I’m a political poll executive and I’ve just switched careers: I’ve landed a plum job at The New York Times — but contingent on a writing test. I’m asked to proofread the first line of the Gettysburg Address, find some errors planted by the testing team, and correct them. And after hours of careful work, this is my final revision:
For score and 18 years ago our fathers brought froth on this incontinence, a nude nation, conceived in Jibbery, and dedicated to the prostitution that all men are created eagles.
After two straight elections of pollsters screwing up, I’m absolutely sure that’s what they’d come up with. A pot shot? Of course. But satire being what it is, I don’t care. Think I’m a moron? Go ahead, call me a moron. Many do. But if you can hang in there, you’re in for some well-reasoned arguments that I’ll bet you can agree with.
Across the polling universe, a bunch of black holes
This time around, the pollsters got it wrong. Not just wrong: utterly, horribly, inexcusably wrong. Again. When the stakes for American politics were at their highest. They showed up for the Super Bowl of Polling with ping pong paddles and waffle irons instead of shoulder pads and helmets.
Here I salute the Connecticut News Project and reporter Ana Radelat, an experienced journalist alert enough to fetch this Tweet from Bill Burton, a political consultant who served as deputy White House press secretary in the Obama administration:
“There was a failure across the political universe when it came to polling. There was not a serious person in the country who predicted these results in the House, the Senate or even the president outperforming his 2016 tally by millions of votes.”
The end result reminds me of a doctor who gets the dosage on your medication wrong and makes you sick as opposed to well. And then he can’t explain it — when he was at the root of it. So you’re sick. And yeah, I’m sick of this shit, too.
Election news flash: liberals, conservatives agree!
Finding common ground between conservatives and liberals in this election is like trying to lasso a four-headed unicorn with a snake. With two hands tied behind yer back, pardners. But I think all of us, red, blue and purple, can agree on this:
In mis-monitoring the sentiment of the American people, political pollsters influenced it to the detriment of our democratic process.
If you support Trump, you have every right to raise the point that Biden’s supposedly unflinching 10-point led many conservatives to believe “Why bother voting? This is a done deal.” Ironically, this also worked the other way — and those on the Left know the pain only too well. In 2016, inaccurate polls encouraged a lot of Hillary Clinton voters to stay home and watch “Seinfeld” reruns. They just knew they’d wake up bright eyed and Bush-y-tailed to America’s female president.
Look: Trump is wrong as usual when he calls the polls a fraud because he has no material proof. And he’s an idiot in this regard: a petulant, whiny child who excels at making things up. As usual. But that said: Donald Trump has a right to be angry — as do all of us — when the pollsters fail this badly. The polling organizations didn’t lie. But they did, by their own definition, make an error because they exceed that sacred “margin of error.” And in a sense, they borderline professed to know things they didn’t really know. For two straight elections, the vote counts and results have borne this out.
And no American likes being let down. Maybe the final results will or won’t. But polling organizations definitely did.
The Seven Habits of highly ineffective pollsters
What happened? I think I know, at least in part. But you’ll read none of this elsewhere because I’m not a navel-scratching pundit with seven doctorates. Rather, I’m a realist who’s been around the block many times: the proud son of a high school dropout mother who had more horse sense that a barn full of jargon-spewing consultants.
Channeling my mother, the late Geneffa “Genevieve” Carlozo, this is what I believe we’re up against. WARNING/CUIDADO/ACHTUNG! COMMON SENSE AHEAD!
- Total job security. When was the last time you heard of a pollster getting fired? They’re like weathermen.
- Outdated methodology and procedures — but universally accepted. There’s security in numbers when all the pollsters drink the same Kool-Aid.
- Not held accountable by any agencies or confluence of watchdog groups. (Let’s credit, though, the American Association of Public Opinion Research.)
- Basic misunderstanding of human nature. My baseline assumption here is that liberals talk way too much about their personal opinions to polling personnel, while conservative people one-on-one are more apt to keep things private. (That’s a wild generalization, sure. But can you imagine it the other way around?)
- Occupational arrogance. (See: Howell Raines of The New York Times as a snooty apologist for disgraced fabricator Jayson Blair.)
- Unwillingness to confront mistakes and make changes. After four years, they didn’t get it any better — and in fact got it worse.
- Polling organizations have never had to repent before and it’s not in their repertoire anyway. (See Nate Silver’s comment below.)
Which leads me to…
Here’s your apology script, pollsters: Repeat after me…
There’s no excuse for the miss this time around. None. Not in 2020 when a single IBM computer chip the size of your fingertip can hold an astounding 30 billion transistors. (I believe the polls said 30 transistors; I’ll fact check that later.)
All this to say that we live in an age of data analytics, artificial intelligence and big data. Our smartphones can crunch digital information with incredible speed and accuracy unimagined a generation ago. But as my work for Intel and interviews with thought leaders in technology taught me, the tech only works as good as the numbers fed into it: GIGO, or garbage in, garbage out.
Thus the polling organizations of America, having worked with garbage, fed us garbage. And they owe us all an apology. No excuses. No rationalizations. No explanations. No hollow promises. And definitely no pushback. They don’t buy that kind of false repentance crap in 12-step programs and we shouldn’t buy it here.
Just be grown ups and accept it. This is the script I suggest:
To the American people: We apologize for getting the presidential election projections wrong for a second straight cycle. If you can no longer trust us, we get it. We screwed up. We own it. And if we knew why, it wouldn’t have happened. Worse yet, we may have influenced the outcome in negative ways. We must take an unflinching look at ourselves, identify what we did wrong and correct it. No questions asked.
Except that it may already be too late.
Parting thoughts: A Silver lining
FiveThirtyEight Editor In Chief Nate Silver is a giant in polling, one of the most respected in his field. He explains things by way of hard statistics and analysis that other experts can’t. He’s frank in his assessment that pollsters do sometimes miss the mark. Plus, he’s passionate about baseball — and if you visit FiveThirtyEight you’ll see that it’s about a lot more than politics. (Check out this piece on how the Los Angeles Dodgers built a baseball juggernaut.)
As Silver attended the University of Chicago, I’m hoping if he ever comes through O’Hare again that we can grab a beer and talk about sports and politics and life. I respect him. I could learn a lot from him.
But I’d also engage a tough debate about Decision 2020. Because in his defense of polling in this election, he’s myopic: failing to realize that in this day and age of instantaneous information relay, polls don’t just report information. As I’ve said, they influence outcomes. It creates a feedback loop. Maybe that’s not the intention. But it sure as hell is a consequence.
Following Election Day, Silver got defensive. Fox reported that on his political podcast, he fired back at his critics: “If they’re coming after FiveThirtyEight, then the answer is, f–k you, we did a good job.”
Hmm. Sounds fairly Trumpian to me.
Because based on the numbers, the pollsters didn’t do a good job.
They were wrong.
That is fact.
And ten points off in the age of 30 billion transistors, machine learning and big data repositories that could digest every presidential poll in history down to the nano-hair? That’s a fucking joke.
Numbers don’t have opinions. They are either correct or incorrect. Margin of error is not the “kinda sorta margin of error that deserves another margin of error.” Maybe FiveThirtyEight and the whole lot did indeed do a “a good job” given what they had to work with. But clearly, they didn’t have enough to work with. Was it with bad methodology? And did it take into account the right variables, as any scientific experiment would? (See number 4 above.)
As for the profession as a whole: Saying polls aren’t scientific, but then touting the statistical survey minutia and advanced mathematical rigor that go into them, is trying to have it both ways.
You won’t see any headlines lauding the work of pollsters in the 2020 U.S. presidential election. But you’ll see many variations on the theme of “What went wrong?” And occasionally, you’ll get see to some honesty from a pollster that’s refreshing, even if sad, because you can sense his disbelief.
“The political polling profession is done. It is devastating for my industry.” — Republican pollster Frank Luntz
Should I take a poll on this to see how y’all feel? With my tea leaves and Ouija board and Magic 8-Ball, I’ll definitely get it right. Or righter. But not wronger. Trust me.
Lou Carlozo is Qwoted’s Editor In Chief. All opinions expressed have a zero percent margin of error. email@example.com
Biden? Trump? Mickey Mouse? Get away from the political mud bath and listen to Lou Carlozo’s “Bankadelic” as finance experts assess the impact of COVID-19, six months later.
February 14, 2023
Combating Disinformation in the Age of Artificial Intelligence
January 5, 2023
Qwoted Announces Top 100 List for Q4 2022
July 31, 2022