The material on this page was produced using the model described in the my article “How the Electoral College Influences Campaigns and Policy: The Probability of Being Florida” published in the June 2008 issue of the American Economic Review.

Topics

• How the candidates should spend their time

• Have Obama and McCain
  visited states to maximize their win probabilities?

• Win probabilities

• Run your own simulations

How the candidates will spend their time

The graph to the right shows how the model recommends candidates to visit states, if they wish to maximize the probability of winning the election. The estimates are based on opinion polls available September 10. After the top four – Florida, Ohio Pennsylvania, Michigan – follows a set of states which were much less important in the 2004 election: Virginia, Colorado and California. To a lesser extent, this also holds for New Jersey, Nevada, Arizona, Indiana and New Mexico. The importance of the states in 2008 relative to 2004 (measured in mid September) is plotted in Graph 2. States above the 45-degree line are more important in 2008 than in 2004.

Graph 2. Expected visit shares in 2008 relative to 2004

Have Obama and McCain
visited states to maximize their win probabilities?

To see how well the model explains the 2008 candidates’ visits so far, I downloaded data on campaign visits from the New York Times web site. The correlations between the actual and equilibrium visits are 0.9 for Obama and 0.8 for McCain (who spent a lot of time in New York). Graph 3 shows the actual and predicted visits.

Graph3.
Expected visits and actual Obama + McCain visits August-September.

Forecasted votes, win probabilities, and campaign importance of states

Table 1 contains a set of statistics based on 100,000 simulated election outcomes, based on data from September 10. See here for a description of the simulation procedure and how to run your own simulations. The first column is the forecasted Democratic share of the two-party vote. The second is the probability that the Democrats will win the state. The third column is the probability that the Democrats get between 49 and 51 percent of the votes in the state, at the same time as the state decides the election outcome (in the sense that, ex post, moving the state from one candidate’s column to the other’s changes who wins the presidency). This is what happened in Florida in 2000, and (almost) in Ohio in 2004. The third column shows the probability of this happening in 2008 for the different states based on the information available September 10. Qs is essentially the same number (see paper), computed using an analytical formula from the equilibrium condition of presidential candidates maximizing their win probabilities. 

Table 1. State statistics