In arguing against a hand recount in Miami-Dade County, assorted attorneys, henchman, and surrogates for George W. Bush have been trying to make the case that the 10,000 uncounted ballots there are simply par for the course. A lot of people go to vote on Election Day and choose not to cast a ballot for president, they say. One of Bush's new lawyers, Irvin Terrell, elaborated on this point at a press conference yesterday:
In fact, those 10,000 nonvotes are about 1.6 percent of the votes cast in that county. If you look at other states and other counties--other states outside Florida and other counties in Florida, you find that in 34, at least 34 counties in Florida, there are higher percentages of nonvotes, and, in fact, in the states of Idaho, Wyoming, and Illinois, there are substantially higher percentages of nonvotes.
This was Terrell's first public statement as a Bush lawyer, and as far as I can tell, it contained not a single accurate figure, fair comparison, or valid inference. In Miami-Dade County there were approximately 10,000 "undervotes," meaning votes on which the tabulating machine detected no choice for president. That's 1.6 percent of the total. But undervotes are just one type of what Terrell calls "nonvotes." The other type is "overvotes," where a voter punched out two or more holes. There are about 18,000 of these, which when you add it to the 10,000 means that 4.4 percent of the votes cast in Miami were "nonvotes," not 1.6 percent.
There are not, as Terrell claims, 34 Florida counties with higher rates of nonvotes than Miami-Dade, only 26. It's not clear why this figure should matter in any case since the failure to count valid votes was a problem around the state of Florida, thanks to the widespread use of punch-card voting systems. If anything, the large number of counties with substantial quantities of nonvotes substantiates the argument that Bush should have agreed to a statewide hand recount. But if you're going to throw around numbers, you should use real ones, and Terrell's are wrong.
Terrell also spoke fiction in comparing the Miami nonvote tally to figures from other states. Only Idaho (5 percent), not Wyoming (3.6 percent) or Illinois (3.8 percent), had a higher rate of "nonvotes" than Dade County (4.4 percent), according to the Bush campaign's own numbers. These are all very high rates, mostly attributable to the same problem that occurred in much of Florida: punch-card systems that produce dangling and dimpled chad and then can't read them. These undervotes weren't counted by hand in other states because they clearly didn't matter. The margin of victory for either candidate in Illinois, Wyoming, and Idaho was larger than the total number of nonvotes in the presidential race. Had Bush lost Illinois by 1,000 or so votes, you can bet he'd be screaming for a hand recount.
Appearing on Larry King Live last night, Dick Cheney attempted to make the same point about the normalcy of nonvotes and ended up telling an even bigger whopper.
They've focused in on the 10,000 votes in Miami-Dade County that supposedly are unmarked. But there are some 34 counties in Florida that have a larger percentage of unmarked ballots for president than those in Dade County. If you go across this country, you'll find a large number of ballots cast by voters who go in, don't want to decide between the two candidates.
By failing to use Terrell's legally precise terminology, Cheney made an even more wildly inaccurate claim. While there are 26 counties with a higher "nonvote" rate than Miami-Dade, there are as few as seven that have a larger percentage of undervotes, or what Cheney calls "unmarked ballots," according to a Democratic source who has crunched the numbers. Again, it wouldn't really help his case if he were right, but he happens to be wrong. But the more significant deception is Cheney's broader assertion that the 10,000 undervotes in Miami-Dade merely reflect voters failing to make a choice among the candidates. This clearly isn't so. The optical scanning system used elsewhere in Florida produces approximately one-fourth the number of undervotes that punch-card systems do (0.4 percent vs. 1.5 percent). The problem isn't that Miami voters--or those in Broward County, Palm Beach County, or elsewhere--are four times less likely to express a choice for president. It's that the machine used to count their votes is much worse at detecting their preferences.
I wouldn't be too hard on Cheney, who appears to have been just retailing the day's inaccurate talking points without understanding much about the details. But there's no such excuse for James Baker, who said at yesterday's press conference:
It is wrong, simply wrong, and I would submit not fair to say, as our opponents do over and over, that these votes have never been counted. They've been counted just like all of the other nonvotes, not only in other counties in Florida, but across the United States of America have been counted. They've been counted, and they've been recounted by machines. They have not been manually counted, but neither have all these other votes that were thrown out in other counties in Florida and cross the United States.
Baker is the one who is wrong, simply wrong. You can say that the 10,000 votes in Miami should not be counted. But you can't say they have already been counted, except in the superliteral sense that we know how many of them exist. As I've noted before and as just about everyone who has been paying attention now understands, punch-card systems usually can't read ballots on which the chad hasn't detached. That's why, as election law in places like Florida and Texas recognizes, you have to examine the undervotes by hand in a close contest. Tom Lyons, a Florida newspaper columnist, aptly compares the votes rejected by a tabulating machine to dollar bills spit out by a change machine. It doesn't mean that the currency is counterfeit. It just means the machine can't read it.