Wharton Professor Weighs In On The Elections

Caesar S. writes "Recent research directed by Wharton School Professor J. Scott Armstrong takes political forecasting to the next level by using innovative techniques to combine forecasts from polls, enonometric models, betting markets and predictions by experts (Delphi surveys). Check out Polly's page to hear Polly the parrot objectively predict this year's presidential election. There's lots more interesting stuff on this site about how electronic markets and Delphi surveys can be used for forecasting. Definitely worth a read."

Comments

Re:Interesting site, but there's a fatal flaw

[coaxial]

Yeah, Election Projection is obviously wrong. They have Oregon going for Bush. There's no way those latted drinking flanel wearing hippies :) are going to vote for Bush.

I used to look at electoral-vote.com too, but I decided that it's predictions were to volatile, so I made my own. [url:http://www.cs.siu.edu/~jkoren/electoral_vote. html]

Mine averages the assorted polls roughly based on "trustworthiness". For instance: Gallup is weighted lower than Zogby, but not because Zogby polls Kerry consistently higher. It's because Zogby was the most accurate poll of 2000, and made a strong argument about what was wrong with Gallup's polling this year.

It's updated daily.