First of all, let me say that unlike almost every conservative on Earth right now, I don’t think Nate Silver is a hack.
I think he’s actually a really smart numbers geek. He writes the blog FiveThirtyEight, which is now hosted by the New York Times (though I’ve been reading his stuff long before it was part of the Times.) The reason he is controversial to conservatives right now and why liberals think he is a saint, is because he is predicting that Obama has a 78.4% chance of winning reelection. To a lot of people, that sounds completely insane.
Two quick things before we delve into the accuracy of that claim:
1) I would describe Nate Silver as an admitted liberal, with a soft spot for markets. He is strangely libertarian on some things, which I like, even though I don’t usually agree with him on policy.
2) He was one of the first people that I remember who predicted Republicans had a good chance at taking the House back after the ’08 election. This is during the time when most liberals (and some conservatives) were saying the Republican Party was about to become a regional party with no hope of ever winning another election. He also was recently yelled at by holier than thou ‘scientist’ Michael Mann about global warming. So, he can’t be all bad.
So, what about Nate Silver’s model as it stands right now? In my opinion, I think he is significantly overstating the chances of an Obama win, with a few caveats.
For example, his own model says that if he has predicted 0.8% of voters’ choices incorrectly, Romney would win the popular vote. If that were to happen (again, by his own model) Romney would almost certainly win the electoral college as well. (A Romney win in the popular vote and a loss in the electoral college has only a 5.1% chance of happening, according to Silver.)
Look, if you get 0.8% of voters wrong and your prediction falls apart—you probably aren’t 80% sure of it.
I don’t think Silver is intentionally making it look like Obama is a sure thing because he’s liberal. I just think he’s a tad too cocky on this one. That’s not the worst thing in the world. Wall Street stat geeks were too sure of themselves with the algorithms that led to the financial collapse.
Global warming scientists are too sure of themselves with their models of the future.
Human beings do such things.
One of the features of Silver’s model is that when the race remains static, and the election gets closer, whoever is ahead becomes more of a sure thing. That’s why his model seems to absurdly show Romney’s chances to be only slightly better than they were before the first debate.
Basically, to him, a 2 point lead that’s confirmed by numerous polls is incredibly convincing. That’s about what’s happening in Ohio, and if Romney loses Ohio, it’s going to be pretty difficult for him to win. I’d say Romney’s chances probably are about one in five if he loses Ohio, so it’s not completely ridiculous if you really trust the polls.
Many of the polls however, just look sketchy. They show samples that are more optimistic for Democrats than the electorate was in 2008. If more Democrats come out to vote than did in 2008, than yeah—Mitt Romney is losing. But, does anyone actually believe that’s reality? 2008 was a historic election for Democrats. Barack Obama is simply not going to repeat that enthusiasm again. It is not happening.
On the other side, while a 78% win seems like a sure thing– let’s put it in football terms. Essentially, Silver is saying the Boston Romney’s have the football, down by a field goal to the Chicago Obama’s with 2 minutes left in the game. First and ten from their own 31. Romney could get a field goal to send it to overtime, or score a TD and grab a win. Or they could go three and out and lose. I can’t say that sounds THAT far off from what is happening in the election, yet, an NFL team in that situation has only a 22% chance of winning.
To me, the data says Romney’s chances are about twice as good as Silver says. That still puts him as a slight underdog. Beyond that, I’m depending on divine providence, hanging chads, or the Koch Brothers hacking electronic voting machines with the help of Grover Norquist or something.
The bottom line is that if Obama wins, everyone is going to think Silver is a genius. If Romney wins, his credibility will be destroyed. Neither is fair, but both are painfully unavoidable.