Prof. Sam Wang, a biophysicist at Princeton, has revived his awesome "Meta-analysis of State Polls" that he started in 2004. If you remember how I was in the Fall of 2004, I used to check this site about 20 times a day.
In my opinion, the methodology is better than 538 since it does not include Nate's regression data, cool as it is. I think this is a major source of error for 538, as evidenced by Nate's big misses in the primaries (e.g., KY and SD). Also, Prof. Wang's model doesn't use a Monte Carlo simulation to estimate the probabilities, but actually calculates the exact numbers.
If you are an obsessive poll-watcher like me, I would highly recommend this resource.
Subscribe to:
Post Comments (Atom)
Posts
2 comments:
Too bad princeton doesn't matter.
Don't you wish your official title was Prof. Wang?
Post a Comment