101 of 102 if you include D.C., which you should since they have three electoral votes, the same as a bunch of states.
I think even Republican pollsters got DC right.
…Nope, they’re doubling down. 
He called Fla for Romney (albeit pale pink), which we know did not come to pass…
Some will double-down. But reality will sink in for others.
The interesting question, now as ever, is which faction will win. The Tea Party faction has a stranglehold on the House because they can easily primary any House member. In the rock-paper-scissors calculus, it’s probably the GOP media elite that needs to reshape the Tea Party for things to change.
To this end, Hannity has already embraced comprehensive immigration reform. If he could flip in 48 hours, we’re sure to see other changes. I suspect Republicans will settle on immigration reform as the magic bullet, chalking their woman problems up to Mourdock and Akin and not the Party.
On the other hand, the Tea Party faction can probably still win elections just fine in the mid-terms. I suspect we’ll see the same '08-'10 pattern again for '12-'14. Not as strong a wave, certainly, but still a much more favorable electorate for conservatives.
No, in the days before the election it moved blue too.
But that would make him inaccurate as a statistician, unless he also gave the states an average of 99% chance of going the way he predicted.
Of course, in the metagame Nate has more to lose by individual incorrect predictions than by inaccuracy over the long haul, so maybe he is being too conservative on purpose. Perhaps he knows that if he honestly calculates that a state has an 80% chance of going to the candidate, he will lose a lot more credibility [ETA: amongst those who look at him mostly as a pundit] on those honest 1-in-5 flukes than he would if he calculated it as merely 65% and it won. Ahhhh, metastatistics.
As Richard Parker points out, just before the election it shifted to pale blue. As far as I am aware, Florida is not called yet, but Obama is still ahead by a slim margin.
Haven’t you ever looked at the site? That’s actually what he does. He doesn’t “predict” which way a state will go, instead he gives the odds of a win by one side or the other. His maps are shaded according to the probability of a given result. For example, he had Florida at only a 50.3% chance of an Obama victory, Virginia at 79.4%, Iowa 84.3%, New York at 100%. He gave North Carolina a 74.4% chance of a Romney win. When I say he predicted the states correctly, what I meant was that all of them have gone according to the odds given by Silver, even in cases like Florida where Obama’s advantage was very slight.
Over the last few weeks of the campaign, he had Obama with anywhere between about 70% and 90% chance at victory nationally, based on the most recent polls in each state.
This election was a epistemological watershed, but given the modern conservative mentality we can predict many more of them. Will reality ever sink in?
Fordham University recently released a study evaluating the accuracy of various polling organizations. It was in keeping with past assessments: numbers one and two were PPP and DailyKos/SEIU/PPP. “Conservative comfort blanket Rasmussen was 24th out of 28.”
Because of their relationship with the DailyKos, conservatives distrusted PPP and I suspect the hard facts of the matter won’t change their mind. Rasmussen’s business model will require no tweaking.
Liberals are wired differently: they are wired for factual assessment. After the first Presidential debate, liberals looked at the polls and freaked. There was some disagreement: a few liberals like myself thought that Obama didn’t do that bad – though I believed that he had to improve his game. But on the whole we didn’t doubt the polls. And the Obama organization itself thought the first debate was bad: they were capable of assessing the reality on the ground.
Contrast with the Romney campaign, who decided to reweight their own polls, in the face of longstanding advice by polling professionals such as Gallup. If they assessed the evidence properly, they might have focussed better on GOTV efforts - which was unusually incompetent. Hey, I’m glad the Romney’s lost. I’m not so happy that the GOP nominated somebody temperamentally prone to policy errors.
One benefit of Silver’s statistical analysis being confirmed beyond reasonable doubt is that it erects a new banner, possibly higher than the existing ones, against such shenanigans – the credibility of any purported “election results” differing from the analysis outside a rather tight margin of error will be DOA.