Posted on | October 20, 2012 | 20 Comments
Politics is a lot like pro football — it’s always more interesting if you’ve got money riding on the outcome. I don’t know if Hugh Hewitt is much of a football fan, but if he’s a gambler and you’re an Obama supporter, Hugh’s talk of the “ongoing collapse of President Obama’s campaign” identifies Hewitt as an easy mark. (Make him give you a point-spread.)
On the other hand, if Hugh wanted to clean up by making sucker bets with Obamabots, Nate Silver looks like a gold-plated chump. Nate’s Incredible Genius Forecasting Model™ was giving better than 2-to-1 odds for Obama as of Friday, so that Hewitt could bet $100 against Silver’s $200, if either of them was willing to put his money where his mouth is.
Some friends and I have been having an e-mail conversation about Nate Silver, and one of my friends — with a graduate degree in political philosophy — name-checked Richard Rorty in the discussion. (Show-off!) This required a response from the Hayekian Public Intellectual™, and here’s what I wrote:
Rather than invoking Rorty, I think that the basic problem is Hayekian. Like economic “planners,” Silver succumbs to the illusion of expertise and is seduced by the apparent exactitude of data, so that the Forecasting Model becomes a sort of Five-Year Plan. Silver’s method fails to compensate for situations (not really that rare in politics) where unexpected events falsify the predictive value of polling. Remember that we are talking about the opinions of fallible human beings en masse, as opposed to the behavior of a select group of professional athletes, which is why Sabremetrics cannot be too much help in predicting how the polls will shift in response to events that could not have been anticipated:
- The multilayered fuck-up in Benghazi.
- An utterly lopsided victory by Romney in the first debate.
If you’ll go look at the RCP national average, you will notice that Obama’s lead was already starting to slip a few days before the first debate. I interpret that as the first glimmerings of a Benghazi backlash, and Romney’s Godzilla-stomps-Tokyo act in the first debate (which exceeded the expectations of even Mitt’s most enthusiastic supporters) accelerated that trend. We don’t really know how far that trend may go, but the fact that Ryan made a campaign stop in Pittsburgh yesterday is kind of a hint that Team Mitt thinks it might go pretty damned far. I’m willing to say, “I don’t know,” and concede the possibility that Obama might yet find a way to pull out a win.
Nate, however, has the burden of expertise obtained through his statistical fluency, and he can’t just throw out an anecdotal gut hunch: “Hey, this Benghazi mess might turn into a sucking chest wound for Obama if he’s not careful,” or: “Watch out for that Romney in debate — he’s a crafty weasel.” Or even: “Hey, y’know, maybe swing voters in Iowa are a bunch of closet racists who have just been waiting for a plausible excuse not to like Obama any more.”
Finally, I will once more repeat: Polls are a lagging indicator. When a game-changing moment happens, it may take a week or two for the polls to fully reflect the extent of the shift. And the 90% refusal factor means that the margin of error could either (a) hide the magnitude of the shift, or (b) exaggerate the shift.
It’s Hayekian, as I say, because there is no reliable way for the “expert” to discern which data inputs are most accurate and which are anomalous outliers.
My gut-hunch is that Silver’s 70% confidence in Obama’s re-election is about 180 degrees opposite of the true state of the race. Silver is a statistician who trusts his data, however, and he has no reason to trust my gut-hunch. So all we can do is wait and see. But if Nate wants to give me 2-to-1 odds, I’d happily bet a beer against his two beers.
P.S.: I still owe John Fund a beer because Christine O’Donnell lost, that no-good witch!
Just kidding about that “witch” thing, Christine. Nobody ever listens to my advice, so no hard feelings, OK?