Monday, November 08, 2004
What on Earth is Going On in Florida?
by Tom Bozzo
The odd result of the election is that George W. Bush carried, often by large margins, Florida counties with very large Democratic advantages in the partisan voter registrations.
I did enough poking to say that the important figures appear to check out. Florida's (partisan) voter registration data by county are here. The county-level returns are, among many other places, here.
I can think of some potential explanatory mechanisms -- stale party ID on old registrations, incompetent reporting of the registration data by the State of Florida, more evangelical-leaning Democrats, for instance -- but on the whole, the idea that Democrats largely stuck by Kerry (89 to 11, according to the exit polls; sorry, Ann Althouse) everywhere but Florida takes my credulity and drops it down a wormhole where it emerges in the Gamma Quadrant in no recognizable form.
Interestingly, the issue is with counties using optically-scanned ballots, which is generally considered the most reliable available voting technology. Though as a result, the optical scan jurisdictions would be under much less scrutiny than those using touch-screen voting. At least it should be straightforward, in principle, to validate the vote totals from the original ballots; I fervently hope someone with appropriate legal resources does so (hello, Florida Democrats)!
I've leaned towards acceptance of the results on the grounds that Bush won fair and square -- or at least as fair and square as you get with a system that admits the fear-mognering, gay-baiting BC'04 campaign -- and that we get another crack at his minions in two years. I'd go critical if there's been a massive fraud.
Update: A Pharyngula commentator makes a moderately persuasive case that the issue is screwy party IDs. In a couple of the counties of interest, presidential and congressional preferences have been fairly strongly disconnected in the past few general elections. So I've toned down the title of the post accordingly.
PZ Myers of Pharyngula (best Father's Day posts, ever), notes some highly irregular county-level voting patterns in the Florida returns. (He apparently got it from Brian Leiter.) I'd seen this kicking around before, but hadn't done any due diligence on the numbers and thus assigned it a low credibility rating.
The odd result of the election is that George W. Bush carried, often by large margins, Florida counties with very large Democratic advantages in the partisan voter registrations.
I did enough poking to say that the important figures appear to check out. Florida's (partisan) voter registration data by county are here. The county-level returns are, among many other places, here.
I can think of some potential explanatory mechanisms -- stale party ID on old registrations, incompetent reporting of the registration data by the State of Florida, more evangelical-leaning Democrats, for instance -- but on the whole, the idea that Democrats largely stuck by Kerry (89 to 11, according to the exit polls; sorry, Ann Althouse) everywhere but Florida takes my credulity and drops it down a wormhole where it emerges in the Gamma Quadrant in no recognizable form.
Interestingly, the issue is with counties using optically-scanned ballots, which is generally considered the most reliable available voting technology. Though as a result, the optical scan jurisdictions would be under much less scrutiny than those using touch-screen voting. At least it should be straightforward, in principle, to validate the vote totals from the original ballots; I fervently hope someone with appropriate legal resources does so (hello, Florida Democrats)!
I've leaned towards acceptance of the results on the grounds that Bush won fair and square -- or at least as fair and square as you get with a system that admits the fear-mognering, gay-baiting BC'04 campaign -- and that we get another crack at his minions in two years. I'd go critical if there's been a massive fraud.
Update: A Pharyngula commentator makes a moderately persuasive case that the issue is screwy party IDs. In a couple of the counties of interest, presidential and congressional preferences have been fairly strongly disconnected in the past few general elections. So I've toned down the title of the post accordingly.
