Drudge is currently featuring a Rasmussen Reports poll with the headline "Poll: All tied up". But there's no particular reason to give that much credence to the Rasmussen poll, which shows the candidates tied at 46% (including leaners). The filtered poll averages of Wisconsin's Charles Franklin at Pollster.com and UNC's Jim Stimson still show Obama with a 3-4 point lead, though both have him trending downward (which is consistent with research suggesting that anti-incumbent sentiment is higher six months before Election Day than it is in November).
At the macro level, the most recent forecasts from Ray Fair and Douglas Hibbs (IE only), who both have well-known models of presidential election outcomes, predict Obama will get 52% of the two-party vote, while a less credible model from a random economic forecasting firm reportedly predicts an Obama win by more than 10 points. Meanwhile, Intrade currently projects the likelihood of an Obama win at 65%, which I think is a reasonable number.
I don't understand why you like to cite Intrade. Based on this election, it seems like Intrade is completely reactive and no more predictive of the outcome than the gut instincts of the political punditry.
Is there any evidence that the political prediction markets are better than traditional polls?
Posted by: Jinchi | July 15, 2008 at 09:11 AM