I am SOO not looking forward to this holiday season's visiting.

Thanks, Burrell. The thing is, they don't post on my FB or even on their own stream (for the most part). It was the ticker on the right side (where it says so and so posted on so and so's wall, etc) where I clicked a few times in the past couple of weeks. That's where I could see the crap. It was depressing. And it was just also a message, in general, to that segment of the population who sincerely doesn't get why anyone would ever want to have their taxes go to for the good of someone else.

I don't mind at all that Ohio took the prize of granting the president another term, but I so wanted it to be Florida.

Actually, according to Nate Silver it was Colorado that provided the tipping point this time.

I'm all out of internet, and still waiting for TCG to get home.

Actually, according to Nate Silver it was Colorado that provided the tipping point this time.

WOOT!!!!

Though, the election coverage I was watching called Obama once the Ohio results were in. Either way...don't care.

It has been interesting hearing people's conversations post-election. I spent a few hours in a coffee shop today and I didn't overhear any political grumbling, just people happy that Obama won and that there are more liberals in Colorado than most people think.

I was on MSNBC and they called it with Ohio. Considering the abuse the state suffered with voter suppression and nasty ads, they deserved the prize. Clearly I believe we all won, even those that don't get it. I just have to take a Facebook holiday until they stop being Wrong or until I regain my natural ability to ignore.

Though, the election coverage I was watching called Obama once the Ohio results were in. Either way...don't care.

Ohio was the one that got called to put Obama over in terms of the coverage). When Nate looks at the tipping point state, he means when we rank them by winning margins. That is, Obama could've lost every state of his that was closer than CO (that's Virginia, Ohio and Florida), and he would still have won (with 272 votes); but if he lost CO too, then Romney would've won.

I kinda love the attention and adoration that has been bestowed upon Nate through this process. He certainly was my life raft in moments of doubt.

I want to have his little nerdling babies.

I kinda love the attention and adoration that has been bestowed upon Nate through this process. He certainly was my life raft in moments of doubt.

So true; so naturally this is the moment when I'm going to forward a criticism of him. Check out this page at the Princeton Election Consortium: "Feeding Karl Rove a Bug." [link] Sam Wang runs some statistical tests on how well Nate Silver's model did, and compares it to his own (and to making just random guesses). PEC's model is polls-only, and pretty much just takes the median of the most recent polls. Nate Silver's model includes many other factors, including national polls, pollsters' house effects, economic conditions and states' political leans.

Cliffs notes: In calling the Presidential outcomes, they both were very accurate, with PEC slightly better. In calling the ten closest Senate seats, PEC called ten out of ten and did even better than in the Presidential race; 538 missed two (gave Tester in Montana only a 1 in 3 chance; gave Heitkamp in North Dakota only a 1 in 12 chance), and actually didn't do very much better than random guessing.

This suggests that, when making forecasts, the features that Nate Silver tracks in addition to the polls look to have added more noise than signal. (The two Senate seats he missed were in Republican-leaning states, and he excluded a few polls, I think because of being associated with the Obama campaign rather than being stand-alone.) For making predictions, we may be better off following the polls, the whole polls and nothing but the polls.

Having said that, I think 538's more complex model has two key advantages over PEC. First, while it seems that the additional features wind up being fully reflected in the polls, when we're earlier in the campaign it may give some information about where things are likely to move. (Nate Silver gradually reduces the impact of economic variables to zero on election day; that might be a good idea for the other features too.) Second, and I think more importantly, Nate's full-time job is blogging for the New York Times. His richer model gives him an eye on many more events and their effect on the race, which leaves him (I think) better able to provide insightful, valuable commentary.

There may be a third advantage. On the eve of the election, Nate gave Romney a 9% chance of victory; Sam gave him a 0.2% chance. (They both predicted much the same most likely outcome, which was pretty much spot on.) I think Sam was closer to the truth; but I think Nate was closer to what could be readily accepted into mainstream discourse. As the quote says, a little inaccuracy sometimes saves a ton of explanation. (I'm queasy about this one. I'm a numbers person and prefer the best estimate. But it may give Nate an easier time reaching a wide audience.)

Bottom line though is that all the aggregators (we can include Linzer's Votamatic and Jackman's Pollster) did far better than the prognosticators; and that gap was much greater than the differences between them. And that was just awesome.

Edit: you may now commence discussion on just whether I actually comprehend what is meant by "Cliffs notes".