Reply #21: where did I say Ohio wasn't dirty?

[OnTheOtherHand]

There isn't much point in trying to carry on a discussion if you're going to attribute someone else's (or maybe no one else's) views to me. I have a right and responsibility to evaluate the truth of arguments actually made; it doesn't mean that I am somehow disagreeing by implication with every other argument anyone else conceivably might have made. Do I really have to explain this?

A problem with Joe Knapp's plot is that most of the precincts are Kerry precincts to begin with, so the visual is misleading in that respect. Nonetheless, it is very well established that residual votes were higher in black precincts. And no one who studied past punch card results would be surprised. Exactly why is unclear. (Sorry, but I don't get to claim to know things that I don't.) As far as I know, the rather extensive review of punch card residual votes in Florida in 2000 didn't unearth evidence of actual tampering with the ballots, but it may have happened. It is also widely conjectured that machines in black precincts may be defective or improperly configured. It is not inconceivable that more black voters do have trouble voting on punch card machines.

As for the machines in Bexley 3A and 3C, I have no idea. Walter Mebane has forgotten more about vote counts from individual DREs than I will ever know. Those machines may have been crooked (or FUBARed) as hell, for all I know. I do know that statewide, the results from DRE counties aren't out of trend with the results from counties that used other voting methods. And Mebane and Herron found that overall, at precinct and war levels, the 2004 results were in trend with the 2002 results. That doesn't mean that all the vote counts were right, but it's a real finding that I wish some folks would at least try to grapple with.

"Also @ 1:07 AM CNN posted female votes in Ohio 53% Kerry 47% bush
male votes in Ohio 51% Kerry 49% bush."

Well, they posted that as early as 7:30 or so -- I saw it myself. They didn't get around to reweighting the exit poll results to vote counts until about 1:30 in the morning. That doesn't tell us whether the vote counts were right or wrong. However, anyone who was watching the Ohio vote count come in knows that it didn't take a sudden right turn around 1:30 in the morning. So there is no particular significance to when the exit poll results were reweighted.

"As a scientist you must believe in stats .... and for the thousands of "glitches*" nationwide
to all benefit bush then there must have been a causative agent behind them. Not random
chance."

Well, the problem is, that's a totally anecdotal sort of argument. What are the thousands of "glitches" that "all" benefited Bush? I'm not asserting anything about random chance, but given my professional norms, I see no point in making an argument that I can't actually defend. The basic problem with the vote switch reports is that there is no good evidence about prevalence. Even if vote switches favored Bush 100 to 1, one wants to know how many votes were affected. And analysis of election returns tends to indicate that not many were.