Monday, October 13, 2008

2008 Election Trends

A long time ago, I made some charts using data from the 1996, 2000, and 2004 elections to predict possible trends in how states vote. I did this by removing the overall swing from each state's results and looked out how votes in each individual state changed independent of national trends. I then extrapolated these numbers to find what the results might be in 2008. One of the things I found was that many of the states that were very close in the 2004 election were trending towards becoming more Democratic, but nearly all of the states that were trending towards Republican were already Republican. Based on these trends I found that if the final popular vote in 2008 was exactly the same as in 2004, the Democrats would gain 10 electoral votes, from Nevada and New Mexico. Not enough to win, but give the Democrats just a 1% swing, they would win Iowa, Ohio, and Colorado and thus the presidency. To win, Republicans would have to keep the national swing towards the Democrats to less than 1%.

I had forgotten about all this for several months, but I became interested again after North Carolina became a swing state. I didn't remember anything in my model saying that North Carolina could go Democrat even in the best scenario. I used the current numbers (as of Oct. 12) on Fivethirtyeight.com to find the predicted swing for the 2008 election, 7.9%. Putting this number into my model gives McCain just a 3% lead in North Carolina. While this is still a comfortable lead for McCain, it's much closer than Bush's 12.4% margin of victory in 2004. I found a lot of other random things in my model when comparing it to the current polls, like Obama is doing incredibly well in Alaska despite Sarah Palin, but I'll leave all that until after the election.

No comments: