Very interesting article in Rolling Stone about evidence of voter fraud in the 2004 Presidential election, especially in Ohio. The piece convincingly shows that at the very least the Secretary of State of Ohio, Kenneth Blackwell, went to great lengths to impede voters from casting their ballots and at the worst there was actual ballot tampering and vote falsification. For one thing Blackwell was simultaneously the Chief Elections Officer of Ohio (as part of his duties as Secretary of State) and co-chair of the Committee to Re-elect George Bush in Ohio, which doesn't seem to be a situation amenable to a fairly carried-out election.
The article discusses three distinct happenings in the 2004 election: the difference between exit polls and final vote tallies in several states, the measures taken in Ohio to deny the right to vote to many voters and the shadowy evidence of actual vote tampering in 12 rural Ohio counties. The first part is most interesting to me, because exit polls have generally been refined to the point where they predict final outcome in most elections with a margin of error of a percent or so. In several cases in 2004, however, the predictions and final results were wildly off. The author of the article tries to tie this into the subsequent allegations of GOP vote tampering, but I'm not entirely convinced. For one thing, some of the discrepant counties were swapped in the opposite direction (the final tallies were more heavily Kerry favored than the exit polls predicted). Even though these counties were the significant minority of the discrepancies, one would expect zero such cases if all the error were due to GOP interference. So I'm curious as a statisitician why the exit polls were so far off.
As for the other accusations, they're quite a bit more depressing. There's substantial evidence that Blackwell flew in the face of Federal and State election law to raise illegal barriers to voting. The author makes the case that many of these activities were also targeted to urban areas with mostly minorities (i.e. Democratic strongholds). This evidence is less convincing but certainly not unreasonable. The final discussion of out-and-out altering of ballots is even more worrisome, but not yet fully proven. Still, it is pretty depressing that we can't seem to manage a fair and free election for the most powerful office in the country.