As many might remember in July of this year we released the results of our presidential electoral model to predict the results of the 2012 election. As opposed to many predictive models out there (Nate Silver's 538 blog being the best known) that rely on publicly available data up to the day of the election, our model is based mainly on past results and demographic changes in the country. We do not use any polling or factor in who the candidates are.
The model not only performed exceptionally well in predicting the national support Obama would receive (only 1/10 of a point off), but we also predicted the state that would put Obama over the 270 threshold (Colorado). Additionally, when looking at the state by state results, the model was only off by an average of 2.3 percent across all 50 states. However, in the 10 swing states it was only off by .8 percent. 538 was off by .7 percent in the same 10 states. Additionally we were closer to predicting the results in 5 of the 10 states, while 538 was closer on 4 and we tied on one. Read more