"Bush, not Gore, bigger victim of voting disparities in Florida in 2000" by Steve Sailer for UPI; ballot, precinct, county, voting machine

Bush, not Gore, bigger victim of voting disparities

UPI, December 8, 2000

December 8, 2000 (UPI) -- Did unfair disparities in voting machines and ballot designs rob Al Gore of certain victory in Florida? Did his supporters have crucial ballots thrown out as uncountable because they tended more than Bush voters to live in counties with antiquated voting systems and confusing ballots, such as Palm Beach County's notorious butterfly ballot? Surprisingly, a statistical analysis implies that difficult voting conditions may have victimized George W. Bush's supporters more than Gore's backers. In reality, what may end up costing Gore the Presidency was that his followers' appear to have found it more difficult than their Bush-backing neighbors to correctly follow their county's voting procedures.

This conclusion stems from comparing a Nov. 19 UPI study of disqualified votes by county to a Dec. 3 Miami Herald study of bad ballots by precinct; The county analysis provides more insight into the effect of local rules on the two candidates' vote totals. The precinct analysis focuses more on how how effectively Republican and Democrat voters within a county responded to the rules.

Here is what the county level analysis found. Owing to differences in voting systems, ballot designs, and other factors, Florida's counties diverged radically in the percentage of votes they had to throw out. They ranged from 0.18 percent discarded in Leon County to 12.4 percent in Gadsden County. In other words, per capita Gadsden's voters cast 69 times more unreadable votes than Leon's voters did per capita.

Both Leon and Gadsden gave their nods to Gore. Both used the vaunted optical ballot reading system rather than the much criticized punch card system.

There were two key differences between them in how difficult it was to vote successfully. First, Gadsden's ballot layout may have been more confusing, which apparently caused caused more people to pick two candidates for President, thus rendering their ballot invalid.

Second, Leon was one of 23 counties that had optical readers in each precinct. This allowed voters to test-run their ballots through the machine to check for errors before handing them in. Gadsden was one of 16 counties that kept all its optical readers at headquarter, with the result that voters couldn't use the machines to double-check their ballots.

Gore's lawyers have made great efforts to show that Democrats were victimized by a tendency to live in counties with old-fashioned punch card systems. In truth, there was little advantage to optical systems, unless they were stationed in the voting area. Counties with optical systems that lacked instant error checking disqualified 3.8 percent of all votes. Counties with the much-demonized punch cards systems, with all their hanging and dimpled chads, threw out a virtually identical 3.9 percent. Bush was more popular in counties with optical systems without error checking. Gore did better in counties using punch cards.

Overall, there was little detectable anti-Gore bias at the county level. f the 51 counties that Bush won, 31 had throw-out rates above the state average. In comparison, only five of Gore's 16 counties had higher than average disqualification rates.

A more sophisticated method leads to the same general finding. Assume, for the moment, that in each county the voters who failed to cast a readable ballot had intended to vote for candidates in the same proportions as their more competent neighbors in the county. For example, say there were 1000 uncounted ballots in a particular county where the 10,000 countable votes split 6,000 for George W. Bush and 4,000 for Al Gore. Thus, assign 600 of the uncountable ballots to Bush and 400 to Gore.

This model suggests Gore would have gotten 49.6 percent of the disqualified votes versus 48.2 percent for Bush. That's not, however, a very good estimate of what voters who fouled up actually intended. What it's useful for is showing how little evidence there is that disparities in voting technology and ballot design cost Al Gore much.

This method is not decisive, though, because voting procedures are not the only factor that causes rejection rates to vary among counties. Voters differ as well. Some counties' voters appear cleverer than other counties' voters at grasping whatever the local rules of voting are.

For example, in Gadsden County, about one of every eight would-be voters fouled their ballot irretrievably by voting for two different candidates for President. Gadsden's ballot confused many locals by listing eight Presidential candidates in one column, plus the Constitution and Worker's World candidates in another column. About twelve percent voted Chinese-menu style, picking a Presidential candidate from each column.

Certainly, the election officials in this poor, rural, undereducated, county should have tried harder to devise a more foolproof ballot. Still, it's hard to imagine that the rocket scientists and other highly educated citizens who live in Brevard County near NASA's Cape Canaveral would have been quite so baffled. (Brevard's discard rate was only 0.27 percent).

Each county's bad ballot percentages stem from two separate causes: different voting conditions and different levels of voter competence. This leads to an important implication. The UPI model estimated that there was a virtual tie in bad ballots when calculated at the county level. This equality between the candidates could have come about in two ways. First, the supporters of both candidates could live in counties with equally challenging voter environments, and could both be equally competent at working with the various counties systems. Or second, and more likely, one candidate's voters might be more competent but tend to live in counties with harsher processes for casting a ballot correctly.

Fortunately, a recent Miami Herald precinct-level study cast some light on this question of whose backers tend to be more competent -- and thus presumably who was more discriminated against by county to county differences. The Herald's team dug up election results and bad ballot totals for each of Florida's 5,885 precincts. Stephen Doig, an Arizona St. journalism professor employed by the newspaper to build a huge spreadsheet out of the data, then guesstimated for each precinct what percentage of the non-voters would have gone for each candidate. He assumed that the failed voters had split among the candidates in exact same manner as the rest of the precinct. (This is the same method as in the UPI study. The Herald just focused at a more detailed level.)

Under this simplistic assumption, which Doig readily admits is "extreme," 56 percent of Florida's 185,000 disqualified ballots were intended for Gore versus only 42 percent for Bush. Following this logic, Gore would have won a 23,000-vote victory among disqualified ballots.

So, Gore may indeed have represented "the will of the people" in Florida, just as he has claimed.

The Vice-President's problem in turning the Herald's educated guess into a compelling legal theory for why he should win the election ought to be that elections aren't traditionally decided by journalists' statistical estimates of what the semi-mythical Will of the People might be. Instead, winners normally are chosen by actual counts of however many voters managed to show up and successfully cast a ballot according to pre-established rules.

On the other hand, if counties did not apply these rules equally -- as they did not -- then one candidate might be argued to have lucked into an unfair advantage because his voters happened to live where it was easier to vote successfully. Surprisingly, the candidate who appears to have gotten this fortunate break was the Vice President.

The Herald's estimated 14-percentage point bad ballot margin for Gore in the precinct level analysis is ten times his margin in the county level analysis.

Why the difference? In Doig's precinct analysis, variation among counties in voting conditions, while still affecting the results, are less important than in the UPI county-level model. The reason is because there are an average of 88 precincts per county. Although some precincts were probably harsher toward voters than other precincts within a county, the big objective differences in voting procedures are found between counties, not within them. Thus, much of the variation in this precinct data stems not from objective differences in how counties run elections, but from variations in the type of voter within each county.

The disparity between the precinct and the county estimates imply that Gore's backers were substantially less competent than Bush's. In turn, that suggests that Bush's voters were more likely to live in counties where the voting systems and ballot designs tended to be less forgiving of voter confusion. Otherwise, the county level estimates would not have come out virtually the same for the two candidates.

Republicans have rightly pointed out that undecided voters must have intentionally left blank some fraction of the 185,000 unreadable ballots (2.9% of the total). However, as Doig observed, it's hard to imagine that the None-of-the-Above vote exceeded the 0.8% bad ballot rate found in counties with error-checking readers in the precincts. Since the throw-out rate was almost five times higher in counties where voters didn't get machine assistance in reviewing their ballots, this suggests that voter mistakes caused most unreadable ballots.

By the way, Nevada is the only state that offers it citizens a None-of-the-Above choice on the ballot. In November, 0.54% punched that choice.

Doig said, "I'm guessing that voter error was the main cause of these uncountable ballots."

What's more significant for the question of which candidate suffered more in Florida from unequal application of voting laws is that Doig probably understated Gore's share of unintentionally messed up ballots. He noted, "There could be an even higher percentage that went for Gore."

The Miami Herald explained why. The "formula probably was conservative, awarding Bush too large a share of the pie. The biggest problems with rejected ballots were in low-income, mostly minority neighborhoods statewide -- areas that voted heavily Democratic. That could suggest that the same group, which included a larger percentage of first-time and less educated voters, might have made similar errors in all precincts."

It's unfortunate that one candidate's enthusiasts turned out to be somewhat more bungling on average than the other candidate's. But ineptness is not proof of unequal protection of the laws. Paradoxically, here it appears that the opposite was true.