November 16, 2004
More Election Denial
I came across this study, originally titled "Irrefutable Evidence: Ohio Election Rigged".
What he does is he takes a county, averages the votes that the Democrat and Republican got over the last four elections, declares it a "trend", and then compares how different the total was this year from that average. If it's over a couple percentage points, then he says that is irrefutable evidence of fraud.
I'm not going to go into the details here on how stupid that methodology is. I think I can safely assume that most of my regular readers can figure out that an average is not a trend, and that voting patterns don't follow those expectations.
But I decided to go ahead and explain to him why the study was flawed. It's all there up on his weblog - you can go read it and see the argument I lay out in the comments.
After his first couple of responses, it was pretty obvious that he wasn't going to be receptive to the possibility he was wrong. Under normal circumstances it would have been best to just roll my eyes and move along. But I wanted to see what would happen if I just doggedly kept trying to prove why it didn't make sense.
His reaction was pretty priceless - he then banned my IP address so I couldn't comment on his weblog anymore.
Another really great example of the kind of post-election denial that is among the left these days.
There are plenty of legitimate election issues to be upset about as Democrats. This study isn't one of them.
For an exploration of some of the worthwhile issues, please visit this wiki page of Ohio Irregularities. I've been the main maintainer of the site, and there is a lot to follow up on. One study that is begging to be made is the one that compares how many voters per voting machine there were in Democratic precincts versus Republican precinct. That's where you can get real evidence of disenfranchisement.
(Given that other folks on that site are upset at me for being too much of a skeptic of these election studies, that should prove that I'm not some sort of Republican plant. I guess they're too lazy to read my archives.)
Update: I actually made a math error in my last comment on that weblog, by looking at the variance of vote growth for each side, rather than the variance of the vote total compared to the "projection". However, the error doesn't change the fact that the study is silly. Here are the numbers for Mercer using the exact same math that Peter Smith uses:
Variance from '88 to '92:
Dem: 16.7%
Rep: -7.45%
Variance from '88 + '92 to '96:
Dem: 25.42%
Rep: -12.63%
Variance from '88 + '92 + '96, to '00:
Dem: -18.29%
Rep: 10.31%
So, you can see that Peter Smith's insistence that a variance of over 4% being fraudulent is completely silly. Unless you believe that Clinton committed gross fraud to beat Bob Dole.
By the way, Peter Smith also deleted my trackback that appeared on his blog, so that people wouldn't see the link over to here. :-)