Elections, ballot counting, and the truth

I have a bunch of pages stuck open in my aggregator that I haven’t posted yet because I didn’t feel right about them. They’re all about alleged or actual errors, improprieties, or other issues with the voting in this presidential election. Today Scott Rosenberg at Salon posted what I feel is the more balanced perspective on this: if there was vote fraud, we‘ll report it, but the people who continue to insist that the election was stolen are beginning to sound like the folks who think Iraq had WMDs and caused 9/11. With that perspective, I can comment on these links and then move on—unless, of course, the vote stealing allegations are proven.

A lot of the articles start out with statistical analyses of the variances between exit polls and actual reported results, such as Blue Lemur’s Odds of Bush gaining by 4 percent in all exit polling states 1 in 50,000; Evoting/paper variance not found to be significant. This article sums up a lot of the threads going around as follows: It seems like Bush got an average gain of 4.15% between exit polling and actual vote tallies across the 16 states where exit polls were taken. That seems pretty high, and you can make a probability assessment that it’s pretty unlikely, but the article is careful to point out that the differences between exit polls and vote counts were higher in some paper ballot states than in e-voting states.

The authors of the paper want the raw exit poll data. This strikes me as scary, since that data has to be weighted against the actual population before it’s any good and if you’re going to go into the raw data and start weighting it yourself, you can make it tell pretty much any story. The only thing this approach buys is the ability to recreate the weightings that the polling organizations actually used, then second guess their methodology. Nice, but what I would really want is the actual vote counts.

Unfortunately, for every careful but ultimately futile article like that at Blue Lemur, you get a dozen roundups of anecdote and speculation, such as the one at bellacio.org: Too many voting “irregularities” to be coincidence. To which I reply, How many voting irregularities would constitute coincidence? And what is the chance of a voting irregularity in 2004, when we’ve all been sensitized by the 2000 election, compared with earlier days when no one would dream to ask the question? Don’t get me wrong, some of the errors, like the 4,000 extra Bush voters in Franklin County, Ohio, are pretty egregious. But some of the other observations, like the one at Commondreams.org about the correlation between voting for Bush and the minimum wage hike, seem pretty thin.

The frustrating thing about the obsession with the election being stolen is that the general tinfoilhatdom is obscuring some real issues, like the ease of hacking e-voting systems and optical scan computers. That’s where we need to put our time and energy, not re-fighting November 2 for four years.

Thankfully, Fury adds another dollop of balance by exploring the use of tin-foil hat as signifier for conspiracy theorist, including a full survey of current usage. Thank God for the academy.