[Sorry I haven’t blogged for a couple of days. Been busy.]
I do not know Nate Silver. I have never spoken to Nate Silver. I do not know how old Nate Silver is or anything about Nate Silver’s existence. I am not even sure Nate Silver is his real name. That sounds like a stage name, one that a numbers geek would pick because he secretly dreams of being a cowboy.
And you have to admit it’s a good name. If I had a name like that, instead of one that rhymes with Hole Schlockenbock, I’d have made the big time by now. But wait, this isn’t about me (I keep forgetting).
The point is, Nate Silver is under attack because his fivethirtyeight blog is forecasting a likely Obama victory.
Ezra Klein explains exactly what is going on, and does so quite persuasively, defending his fellow wonk. Ezra makes an immaculate free-market argument here, pointing out that Silver’s influence is based on his excellent track record and that if someone came along with a better formula for predicting election outcomes that person would supplant Silver as the go-to numbers geek. If you don’t like Silver, then build a better mousetrap.
And Ezra also makes the most important point: Silver is not saying Obama is definitely going to win. Silver is currently saying that in some scenarios Obama will win and in some scenarios Romney will win and the first batch of scenarios outnumber the second batch of scenarios by about 3 to 1.
When we look at Silver’s forecast of a 75 percent chance — oops, it’s now at 77 percent! — of an Obama victory, we intuitively see that as a large lead. But that’s not what the number really means. Silver projects a fairly narrow win, with Obama getting just 299 electoral votes, far fewer than in 2008. The 77 percent figure means merely that there’s a roughly three out of four chance that Obama will win 270 electoral votes or more. You win with 270 electoral votes the same way you win with 538. I bet the Obama folks in Chicago don’t look at that 77 percent figure and consider it a comfortable lead.
As you know, I’ve been talking to actual voters in Ohio, Florida, Virginia and Wisconsin, and have one more swing-state story to produce in the next day or so, this one about Ohio again. I’ve talked to some Obama ‘08 voters who will vote for Romney, but it’s almost impossible to find a McCain ‘08 voter who will vote for Obama this time. Based purely on that data point, I’d guess that Obama would lose narrowly in some of these swing states — except there are other data points to consider, such as the efficiency of Obama’s ground game, which has been years in the development.
As noted here recently, I am skeptical that any poll, or model, can fully capture the thinking and behavior of a highly diverse group of voters in a polyglot nation in which the demographics change, the early voting laws change, the weather changes, and so on. In a super-close election like this I do not think we can tell in advance what is going to happen.
So I agree with Silver’s ultimate point: The winner will be either Obama or Romney.
I’d put the odds on that at 100 percent.