Now that the election results (or, estimates) are in, I was curious to see how the election prediction model of Sheldon H. Jacobson along with co-authors Steven E. Rigdon, Edward C. Sewell and Christopher J. Rigdon has performed. OR bloggers Micheal Trick and Laura McLay wrote about this in their blogs earlier. If you are not familiar with it, here is a link to their website and here is a link to their paper that explains the model. From their paper:
It uses a Bayesian estimation approach that incorporates polling data, including the effect of third party candidates and undecided voters, as input to a dynamic programming algorithm … to build the probability distribution of the total number of Electoral College votes for each candidate.
Their results were last updated on Tuesday, Nov 4th, using the latest polling data. Of course, the model is as correct as the polling data that is given as an input to the model. However, take a look at this:
11:45pm AZ time: Obama has 338 and McCain has 159 electoral votes. The prediction model has 338 Safe Electoral Votes for Obama and 157 Safe Electoral Votes for McCain. They define “safe” when they predict the candidate has a 0.85 chance or better for winning. Montana, Missouri, Indiana and N. Carolina still not decided. In the prediction model, these are the states that are not “safe” along with N. Dakota.
12:15am AZ time: Indiana goes Blue… Hmm… Their model tends to Red… (the only time it does not match)
—>> Can’t wait to see the final results. . .