Spotlight Is on Accuracy of Vote Tally

The 2000 presidential election exposed a widely held myth: When you cast a vote, it counts, no exceptions.

Now the phenomenon of uncounted ballots, long known to election officials, is the subject of furious debate. Judges on a panel of the 9th U.S. Circuit Court of Appeals -- reconsidering the decision last week by three of their colleagues to halt the Oct. 7 recall election -- on Monday asked lawyers for both sides how many uncounted votes were too many.

One of the crucial issues the judges tried to evaluate was whether punch-card ballots, made notorious by the Florida recount in 2000, really are so inferior to other systems that using them would disenfranchise voters.

“How much of a noncounting of ballots are counties allowed to do before they run into constitutional problems?” asked Judge Alex Kozinski.

State election officials, who widely denounced the possibility of delay and have asked that the ruling be reversed, argue that no system is infallible. Some of the nonvotes or over-votes that studies of the machines consider to be errors, they say, may in fact be intentional -- the result of voters’ deliberate choices not to vote for a candidate or to vote twice.

In any case, election officials say, creating the perception in the public that a mistake-free voting system can be built is irresponsible. “Nobody suggests anything will be perfect,” appeals court Judge Margaret McKeown noted during Monday’s hearing. “How do you distinguish between the punch-card ballot and the errors that will also be inherent in the new scanning system?”

Punch cards, optical scanners and touch-screen systems -- the three voting systems used in California’s 58 counties -- have produced disputes and allegations of widespread errors.

After the 2000 election debacle in Florida, California election officials decertified punch-card machines, effective Jan. 1

But optical scanners, election officials warn, do not prevent all of the voting problems associated with chads, the tiny bits of paper punched out of the ballot.

As for touch-screen voting, critics have filed suit in Riverside County, which in 2000 became the first in the state to use the machines, saying that their rights as voters were violated by a system that is unreliable and easily manipulated.

“Truly, all voting systems can fall apart,” said Conny B. McCormack, Los Angeles County’s registrar-recorder. “A system could work terribly in one place and well in another.”

Since the problems in the last presidential race, officials nationwide have grappled with the problem of how to conduct the fairest possible election. How many uncounted votes are tolerable?

And which system produces the fewest so-called residual votes -- ballots that fail for one reason or another to register results in each race?

The three-judge panel of the 9th Circuit that ordered the election postponed found that punch-card systems used in Los Angeles, San Diego and four other counties would fail to count about 2% of the votes.

That means the votes of about 40,000 voters would not count on punch-card ballots if turnout were about 8 million, as it was in last year’s gubernatorial election.

Other voting systems, research indicates, produce errors at half that rate, meaning votes in counties using other methods would be less likely to be thrown out.

In California, counties use 10 variations of the three main voting methods -- punch card, optical scanners and touch screens.

Los Angeles County, for example, had been set to use optical scanners next year, as a stopgap before implementing a $100-million upgrade to touch-screen machines, planned for 2005.

The optical scanners would be acceptable under the appeals court’s decision. But the temporary switch worries McCormack.

“We’ll have 100% of the voters who have never seen this system before,” she said, noting that voter error caused by using a new system might outpace the problems with punch cards.

McCormack compared last year’s gubernatorial race with the 2000 presidential election to suggest that some uncounted votes might not have been caused by machine errors. In Los Angeles County, fewer than half of 1% of the ballots cast for president were disqualified because of over-voting -- choosing more than one candidate for a single office. The under-vote -- no vote was cast for president -- was 2.1%.

Last November, in the gubernatorial race between Gov. Gray Davis and Republican Bill Simon Jr., the over-vote among county voters was 0.61% and the under-vote was 3.8%, despite a substantial education effort urging voters to make sure they had cleanly punched their ballots.

“Is that an error, or is that people skipping the election because they didn’t like their choices?” McCormack asked.

A joint MIT/Caltech study cited last week by the 9th Circuit panel estimated that only about 0.5% of under-votes nationwide are intentional, a figure McCormack and other election officials have questioned.

Orange County election officials sidestepped the punch-card controversy for the recall election, planning instead a pen-and-paper ballot originally intended only for absentee voters. Instead of punching out chads, voters in Orange County will use black pens to mark a rectangle next to the name of their candidate of choice.

Still, simple as it sounds, this system has the potential for error, officials said. If a voter circles the rectangle instead of filling in the box completely, then the ballot will have to be reviewed individually by election officials, said Angela Burrell, spokeswoman for the county registrar of voters.

“You can’t eliminate mistakes, but you can certainly minimize them,” said David E. Hart, chairman of Hart Intercivic, the Austin, Texas, company that designed the system for Orange County.

In Riverside County, the touch-screen voting machines eliminate over-votes because they will not allow a ballot to be cast if more than one candidate has been selected for an office. The system also asks voters to review their ballots before confirming the choices.

However, critics of the machines sued Riverside County. Susan Marie Weber, the Palm Desert resident who filed the suit in August 2001, said she was concerned about a lack of a paper trail. “We have the government creating the ballot, counting the ballot and telling us how the vote came out,” said Weber, a member of the Desert Area Libertarians.

The system is particularly vulnerable to fraud, she said, “because it’s so easily manipulated and so difficult to determine how it was manipulated.”

Mischelle Townsend, Riverside County’s registrar of voters, disagreed. Townsend said the machines had proved reliable in five recounts. Additionally, election officials test the machines before and after elections, and count 1% of the precincts by hand to make sure the hand-count results match the computer tallies. U.S. District Judge Stephen V. Wilson dismissed the lawsuit last September.

He ruled that the machines were an improvement, eliminating votes not counted because of damaged ballots or improper markings.

Weber has appealed the ruling. A three-judge panel from the 9th Circuit will hear oral arguments Oct. 8 in Pasadena.

Times staff writer Stuart Pfeifer contributed to this report.