There's going to be a blue Phantom on election day.
Nate Silver is The Man....
The accuracy of his November 2008 presidential election predictions—he correctly predicted the winner of 49 of the 50 states—won Silver further attention and commendation. The only state he missed was Indiana, which went for Barack Obama by 1%. He also correctly predicted the winner of all 35 Senate races that year.
Nate Silver - Wikipedia, the free encyclopedia
Yeah in 2008 he was very good. Then he signed on with the NYT and it's been all downhill ever since. He bases his calculations on weighting of the pollsters. Now this weighting is not based on historical accuracy, it's based on what he calls Pollster Introduced Error (or PIE). Unfortunately he refuses to release his criteria for what is PIE and what is just normal statistical error. He has been asked multiple times by multiple website and sources to divulge the information and he refuses. He also bases the weighting on whether a pollster belongs to his pet organization which is (hypocritically) an organization wherin pollsters divulge their entire statistical formulas. Those that belong to the organization he gives higher weight to because as he argues, he can check their methodology...real interesting since he refuses to release his own so people can check his.
The result is that polls like Marist and PPP which are historically WAY off in comparison to other polling agencies are now getting more weight in his calculations than organizations like Rasmussen and SUSA even though in presidential elections those two have been in the top three in terms of accuracy.
What Silver does is he recognizes that his calculations can create a bandwagon effect and, being the ridiculously biased liberal that he is, he has managed to weight his calculations so that polls with a liberal house lean get more weight then ones with a Republican lean. In doing so he can show that once you apply his weighting 2+2 does not equal four, it equals one. But when you ask to see his method for determining PIE which determines the weighting, his answer is "
no, just take my word for it, and who are you to be asking questions of the almighty me anyhow?"
Now...oh about two or three days before the election he will realize that his credibility is about to be publically exposed and SUDDENLY his model will start showing Romney winning the election. That's exactly what he did in 2010 and serious poll watchers laughed our asses off because we had been calling that for months and Silver finally came around and called what was clearly obvious months ago only right before the election.
Right now Silver is still trying to convince people to back Obama because "he's the winning team and everyone wants to back the winner". Provided the polling trends continue as they have been, that'll change.
BTW....I correctly predicted 49 of the 50 states in 2008 too as did most of the rest of the serious polling geeks. Big deal...that election was EASY to call.
