Mark it down on your calendars: Yesterday — Monday, Oct. 8, 2012 — may go down in the annals of history as the day political polling died.
It was the most ridiculous polling day among many preposterous polling days in the course of this long campaign.
The ludicrous comedy began early in the morning. Gallup reported that, when it combined the results of its three nightly surveys after Wednesday’s debate, the race between Mitt Romney and President Obama had become a tie, 47-47.
Around the same time came news from the bipartisan Battleground poll that though its top line had the race 49-48 for Obama, among “extremely likely” voters, it was 52-46 for Romney. And 85 percent of its survey was done before Wednesday’s debate.
Then, around 9:30 a.m. came the news from Rasmussen’s three-day poll: A 48-48 tie.
Sounds consistent, right? It wasn’t. The day before, in Rasmussen’s three-day poll, Romney had been leading 49 to 47. A two-point shift in Obama’s favor in a single day in a tracking poll is a significant move.
Even more sobering was Rasmussen’s tracking poll of swing states, which showed Obama in the lead, 49-47. Its state polls (all from a single day of surveying on Sunday) were also good news for Obama.
All of this suggested, in the words of the sober election analyst for the liberal New Republic, Nate Cohn, that “Rasmussen [had thrown] cold water on Romney’s post-debate bounce.”
Then came the big news at 1 p.m., when Gallup revealed the results of its daily tracking poll: Obama leading, 50-45!
Republicans, high as a kite since Wednesday night, instantly reached for the Lithium to control the downward swing of their manic depression.
“Hey, wait,” I hear you thinking. “Didn’t Podhoretz just say Gallup had released a tracking poll early in the morning showing a 47-47 tie?”
Yes, you did, just a few paragraphs up.
But see, Gallup’s 1 p.m. tracking poll with the five-point Obama lead is a seven-day tracking poll, whereas the early-morning poll that had it tied was a three-day tracking poll.
And that three-day poll was only for Thursday, Friday, and Saturday. The 1 p.m seven-day tracking poll included Sunday. So Sunday must have been very good for Obama.
Wait: All of a sudden, there came a tweet from — of all sources — the extremely anti-Republican Markos Moulitsas of the Daily Kos Web site that its weekly survey “has Romney in the lead. Bulk of calls were made Thu & Fri, after R’s post-debate bounce.”
The big boom erupted at 4 p.m.: The Pew poll — thought even by liberals to be the most likely to skew Democratic — featured an astonishing result: Romney up by four, 49-45.
Even more astonishing, it was a 12-point swing in his favor from the last Pew poll, taken before the debate — and included an 18-point swing in one week among female voters toward Romney.
So, in one day, a single pollster released two seemingly contradictory surveys. Another pollster showed what appears to be a swing back to Obama after a swing toward Romney.
And two other polls were deeply encouraging to Romney. One showed him winning with those most likely to vote by a six-point margin even before the debate. The other? Winning by four after a 12-point shift in his favor in a single week.
You know what you call data like these?
Garbage.
The disparity in these numbers and their trends are so broad that even the cautionary method of adding them all together and averaging them out — best done by the Real Clear Politics “poll of polls” — makes little sense.
After an election season lasting 14 months, which saw Herman Cain and Michele Bachmann and Newt Gingrich serially seduced into fantasizing they might win a GOP nomination they never ever would have, we go into the final month with no reason to trust that these surveys are capturing anything other than a close race.
Pollsters themselves, when challenged on their stats, say they’re just presenting a snapshot of public opinion. Fine, but these snapshots are wildly distorted.
The key hidden fact is that fewer than one in 10 respond to those who try to poll them.
People who screen their calls, hang up on people they don’t know or end the survey because they don’t have time to take it make up more than 90 percent of those phoned by pollsters.
Then there are issues with cellphone users and those who communicate pretty much solely by texts and e-mail, and the like.
All we can be sure of, in the words of the peerless Internet humorist Iowahawk, “political poll results accurately reflect the opinions of the weirdo 9 percent who agree to participate in political polls.”
What yesterday proved is that all bets are off. We’re judging the state of this contest with junk data, and we need to stop. Until pollsters can figure out how to avoid all these crazy mood swings and white noise, they should be put on political and pundit probation.