Thursday, November 6, 2008

Princeton Election Consortium's Performance

Pretty damn good. A lot of people are focusing on how Nate Silver's FiveThirtyEight did in the predictions game, which is fine as he did quite well, but I wanted to take time to extend congratulations to the less well known Professor Wang for his outstanding efforts with a much simpler and more elegant model. (Electoral-vote.com also nailed it)

He predicted 364 Electoral votes (as I did using his analysis), but like everybody else, Indiana and Missouri were picked for the wrong sides. He predicted a 53% to 46% margin, where the actual result was 52.4% to 46.3%. He called for 257 Democratic seats in the House, and we (appear to have) won 256. The Senate is a giant mess right now, but he called for 58 and that appears to be the most likely outcome... though it really is hard to say what is going to happen in Alaska and Minnesota.

And finally:

The Electoral College and state win predictions were done using state polls only. The cell-phone adjustment was done using Pew Center data. The Senate and House predictions were done using Congressional polls only. The popular vote share prediction was done using national polls only. Turnout was estimated using the work of Curtis Gans and data from InTrade and the Census Bureau. In no case was any demographic or pollster-specific information used. Overall, the results show that a high degree of accuracy is possible without complex model-building.

Yet another victory for Occam's Razor.