Tuesday, September 25, 2012

Election Model: Ohio and Virginia

Brief update notes: 

  • I added a metric measuring President Obama's chances of exceeding his 2008 electoral vote total. They aren't good. He won Indiana in 2008 and his chances of repeating that feat are.. slim.
  • Arizona moved from safe Romney to likely Romney. This is the combined effect of a weak recent poll for Romney in AZ (+3 on 9/19) and the dropping completely out of an older stronger poll for him. (Ras +13 on 6/26)


Ohio and Virginia

The other day Nate Silver tweeted the following:

Today I decided to see just how OHVA the election would be if Mitt Romney lost both of those states.

I had the model keep track of how often Mr. Romney lost both Ohio and Virginia, how often he won at least one of those two states, and how he fared nationally in those two scenarios. I came up with the following:



It does appear that if Mitt Romney lose both OH and VA the election is effectively over, he manages to overcome that deficit less than 1% of the time. There is good news for Mitt, if he does manage to win one (or both) of those states.

60% of the times Mitt Romney wins at least one of  OH or VA, he also wins the election. The catch is, the model also gives him just a 16% chance to win one of those states..

It is important to remember the causality runs both ways here. swinging Ohio's 18, or Virginia's 13 electoral votes helps Mitt in the electoral college. Simulations where Mitt Romney wins one of OH or VA are, on average, simulations where he is doing better nationally and is likely contesting other key states like FL or MI.

If I did this same kind of analysis on say, F and MI, I would expect a similar kind of story. You can't make those into a cool pun though... MIFL?

No comments:

Post a Comment