(Via Blue Mass Group) The National Election Data Archive has released a primer as to why they think the results in Ohio for the 2004 Presidential election were hokey, based on a comparison of exit poll results and actual precinct results.
I wonder if they’re overrelying on an assumption of general unbiasedness of exit poll results, or the assumption that the discrepancy between exit poll and precinct result is normally distributed…but it’s interesting to look at nonetheless.