Saturday, September 20, 2008

Electoral Projections at FiveThirtyEight.com

While there are a ton of polls and election projections on the presidential race, my favorite spot to take a look at who is winning and by how much is FiveThirtyEight.com.

I like the way they take all the polls, massage them and then run 10,000 Monte Carlo simulations to come up with their prediction of who is likely to win (more on the methodology here).

The image to the left is the result for 9/19. If I am interpreting correctly, Obama is 71.5% likely to win this election based on the polls and how well each pollster was able to predict previous elections. Or put another way, previous candidates who had Obama's the same lead in polls at this point in election cycle won 71.5% of the time.

What is interesting to me is that while the popular vote is so close (1.8% difference), the electoral vote is not nearly as close (56.3% to 43.7%), and McCain is only 28.6% likely to win. Part of the reason for Obama's high win percentage is that he is much more likely than McCain to win the electoral college and lose the popular vote.

If you look at the their breakdown of states by regions, there are really only two regions that are competitive: the Rust Belt and the Southwest (I am excluding the South Coast region with Florida and Virginia, because I am skeptical that Obama really has a shot there). I hate the fact that the Rust Belt states are battlegrounds yet again, as they appear to vote solely on which candidate is more likely to keep shitty jobs from going overseas. I am glad that the Southwest also has battleground states now, as these states are forward looking and want to discuss how the US needs to change to be successful in 2050, unlike the Rust Belt states that want to change the US to return to 1950.

The site is worth checking out if you are curious how the election is likely to turn out.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.