The Operations Research / Management Science Blog

There are already several articles on this topic (see references), so I will just briefly talk about it.

Election day is only over a month away. Dr. Sheldon H. Jacobson, an OR professor at the University of Illinois at Urbana Champaign, and his students have created a mathematical model that dynamically forecasts the outcome of the 2008 US Presidential election.

Their website is at http://election08.cs.uiuc.edu/

According to the site, the motivation to create this model is that previous Presidential Elections suggested that popular vote is not a very good indictor for predicting the winner. The website aims to provide up-to-the-minute estimates for the probability that each candidate win the election. Currently, John McCain has a higher probability (0.60) of winning, while Barack Obama has a 0.38 chance of winning.

Bayesian estimators are used in the mathematical model to determine the probability of each candidate winning each of the states based on available state poll results. A dynamic programming algorithm then uses these state-by-state probabilities to figure out a probability distribution for the number of Electoral College votes that each candidate will win in the election.

For more information, see the articles listed below.

References
“Advanced math computes McCain ahead in early electoral college“, TG Daily, Sep 2008
“Math Model Shows McCain Ahead By As Many As 27 Electoral College Votes“, ScienceDaily, Sep 2008
“Predictions for the 2008 United States Presidential Election: A Bayesian Approach“, Election 08, Sep 2008