How shocking!
For fun, I compared the 538 predictions with the current totals at CNN (which as of now are not 100% complete, so there’s some error here). I normalized the 538 predictions using the margin of error for each state, and compared all results to a normal distribution curve.
The differences fall almost exactly on the normal distribution; importantly, there doesn’t seem to be much bias in either direction. As you’d expect, two states fall outside of the 95% margin of error: Hawaii and West Virginia, but both are by small amounts.
So, pretty good job by my amateur eye.
Do not even think of kidnapping Nate to run your own private operations; he will outsmart you if you try.
What used to be on the site?
I agree with some of the posters who say that further in-depth analysis would be appropriate, but from the initial look of things, this does look like support for Nate Silver’s model.
And a big win for him personally. He achieved a lot of prominence in this election cycle, much of it due to attacks on him from conservatives, whose own predictions proved to be very wrong.
Silver is a huge winner. He is now one of the leading election/poll analysts in the country, and will no doubt be a featured commentator on all sorts of other media. (The only problem he has in this regard is that his expertise is very heavily data and analysis driven, and it’s harder for him to just show up on television shooting off his mouth at random than it is for the more conventional babblers.)
As I was reading this I thought “cool!” - Nate Silver is a Straight Dope fan. Then you burst the bubble.
On the other hand, it seems like SDMB is something he’d really enjoy. Does anyone know if he follows us? It’d be interesting to hear his views about the comments made here.
Fiction, apparently.
Sam Wang did better in the Senate races (though he picked Florida for Romney), at least if Tester holds on to his lead (as of an hour ago) in Montana.
One of the good things about this electoral cycle is that it sounds like we’ve now achieved SABRmetrification of poll analysis. I don’t know if the networks will get much better at reporting polls in context, but that kind of context will get more attention and it’ll be easier to find.
Did God smite CK with boils on this face or something?
Could i have simply taken the last say, ten, polls on every state and come up with pretty much the same results as Silver?
And he’s just not very good at television, although not as painfully bad as he was initially.
Yeah, pretty much. That’s what RCP does and their “no toss-up” map was exactly right as well (pending FL). However if Florida does go to Obama by razor-thin margins then Silver would have been marginally better there (RCP had Romney by 1.5 in their last average).
I have no doubt Nate (and others) will do some analysis over the next few weeks measuring the average errors of the various methods.
I dunno. He’s still giving Obama only a 90.9% chance.
I reported that poster for what I think is thread-shitting and trollish behavior. We’ll see what the Modmins think…
And of course we’ll have to listen to a bunch of old fogies droning on about pollsters not living in their mother’s basement.
But even the *way *he’s bad at television fits right into our preconceived notion of what a wonk should be.
He’s got a socially awkward laugh and a nervous smile. He’s a pocket protector away from being a self-parody. It’s perfect.
Off-topic: unskewedpolls.com got it wrong. Their last post - exit polling (done on the web) hadRomney with the vote well in hand.
Has anyone posted the Nate Silver parody twitter feed. It’s awesome stuff.
A sample:
Political commentary is just is off season gig. This is why you don’t hear much from him around the holidays.