Analysis reveals flaws in voting by touch-screen

By Jeremy Milarsky and Buddy Nevins
South Florida Sun-Sentinel Staff Writers
Posted July 11 2004

Florida's relatively new touch-screen voting machines, touted as a solution to the state's 2000 presidential election meltdown, didn't perform as well as machines that use an older technology during a statewide election earlier this year, according to a South Florida Sun-Sentinel analysis.

Records from the March 9 Democratic presidential primary show that votes were not recorded for one out of 100 voters using the new ATM-style machines. That's at least eight times more than the number of flawed votes cast in the same election with pencil marks on paper ballots tallied by an optical scanner.

Experts blame Florida's political leaders for embracing the relatively sophisticated touch-screen machines before they were perfected and made more user-friendly.

Fifteen Florida counties now use touch-screen machines, including Palm Beach, Broward and Miami-Dade.

In December 2001, Broward County chose a $17.2 million touch-screen system over a pencil-and-paper system priced at no more than $5 million. Earlier that year, in May, Palm Beach County agreed to pay $14 million for touch-screens, compared to $3 million for the simpler system.

"Would I have bought them? No," Broward County Elections Supervisor Brenda Snipes said about the touch-screens. She started as supervisor after the machines were in use. "Were we too fast? Yes."

Costly and confusing

In 2000, the nation watched as Florida's arcane voting system became the catalyst for a bitter post-election struggle between Al Gore and George W. Bush over who would become the next president. Butterfly ballots, hanging chads, recounts and court fights to dispute one of the closest elections in state history kept Americans waiting five weeks to learn who would lead the nation. An embarrassment for the state, the battle raged from Tallahassee to Washington.

Pushed into service in the aftermath of the 2000 election, the ATM-like machines have been assailed by politicians, experts and watchdogs as too expensive, confusing to a large segment of voters and packed with flawed software.

If the March undervote rate repeats in the November presidential election and the turnout is the same as in 2000, Broward and Palm Beach counties alone could generate 7,800 flawed votes, a number that worries political leaders who remember the 537 votes by which George W. Bush beat Al Gore.

"That's frightening," Broward County Democratic Party Chairman Mitch Ceasar said. "I thought these machines would correct the incredible situation we had four years ago. I'm angry and disturbed."

The presumptive Democratic Party presidential nominee, John Kerry, already has plans to legally challenge the election based on the questionable reliability of the touch-screen voting systems.

"And you know those machines all those election officials are working on to make them ready on time?" Kerry said at a Fort Lauderdale rally Thursday. "I'll make a deal with them: They fix those machines and we'll fix America."

Primary votes studied

The Sun-Sentinel analysis of the March 9 election reviewed a sampling of nearly 350,000 ballots statewide in which only one choice appeared on the ballot, ion of a Democratic Party presidential nominee. Ballots were not included in the study if they contained other races, referendums or questions.

The study looked for instances in which voters went to the polls and chose no one results known in election parlance as "undervotes." The study then looked at which types of voting machines registered the most undervotes.

The results show:

Undervotes occurred 1.09 percent of the time in counties with touch-screen machines and 0.12 percent of the time in counties that use optical scanning. Optical scan machines counted 12 overvotes (0.01 percent) in the March sample, where voters chose more than one candidate for the party presidential nominee. Overvotes are impossible to cast on touch-screen machines.

Pinellas County registered the highest undervote rate, at 2.87 percent, among touch-screen counties, with 756 undervotes. Only three of the 39 counties using optical scanners performed worse than the best of the touch-screen counties.

Broward County showed a 0.9 percent undervote rate among 18,766 voters, while Palm Beach showed a 0.54 percent undervote rate among 53,059 ballots reviewed, with a combination of 458 undervotes in precincts where there was only once choice on the ballot.

The numbers did not surprise officials of Sequoia Voting Systems and Elections Systems & Software, two companies manufacturing touch-screen machines sold in Florida.

"The most important thing to take from the [Sun-Sentinel] survey findings is that both electronic systems and precinct-based optical scan systems dramatically reduce voter error. ... The Florida numbers demonstrate a substantive improvement over the 2000 presidential election," said Alfie Charles, vice president of business development for Sequoia. Palm Beach County uses machines from that company.

Charles' point was bolstered by a report on the Oct. 7, 2003, California recall election by Henry Brady, a political scientist at the University of California, Berkeley. Brady found that more than 7 percent of the voters using punch card machines cast flawed ballots, such as undervotes and overvotes. About 1.3 percent of those using touch-screen machines cast defective ballots in that California election.

Fiasco elicits changes

The punch card system used in the disputed Florida presidential vote four years ago generated most of the 175,000 disputed ballots.

Gov. Jeb Bush and the Legislature wanted to make sure the chaos didn't occur again.

Just months after that election, Florida outlawed the punch card system. Counties were given two choices of new systems to buy before 2002: the older optical scan where voters use a pencil to fill in a bubble on a page of paper next to their choice, similar to a standardized test in schoolor the new touch-screen machines. The state certified specific systems and machine makers that could be used.

Some counties, such as Orange and Leon, already had optical scan machines and kept them, choosing not to spend tens of millions of tax dollars on the newer devices. Many of the other counties that use punch cards, including a majority of the most populous in Florida, chose the touch-screen technology.

Lobbyists pushed for the sale of the more expensive machines, and county commissions agreed to make the purchases, spending nearly $57 million combined in Miami-Dade, Broward and Palm Beach.

Broward bought its 5,000 machines before the state had even certified them. Purchase contracts in Broward, Miami-Dade and Palm Beach spell out machine standards and require vendors to fix problems. The contracts do not specifically mention undervotes, however.

Election experts now say the purchases of the touch-screens were rushed.

"It was like Florida was trying to change a tire on a car going 100 miles an hour," said Kurt Browning, elections supervisor of Pasco County.

The machines were widely touted by vendors and their representatives for having advantages over the optical scan devices. The sales pitch said ballots on touch-screen machines could be easily converted into multiple languages, an important feature in diverse South Florida. Many organizations representing the blind also pushed for these machines because they come with an audio feature that can read a ballot aloud.

Lobbyists said little about how complicated the machines were.

"The information about the machines was all coming from the vendors. There was no independent study about the efficacy of these machines. Every supervisor who purchased one was taking the vendors' word for it on how well the machine worked. And what did the vendors say? `Trust us. Trust us.'" said Ion Sancho, elections supervisor in Leon County.

Sancho stuck with his older optical scan machines. Leon County reported zero undervotes in the March sample.

South Florida counties' first big test of the touch-screen machines was in September 2002. From the outset, glitches occurred where voters complained of wrong ballots and other mistakes that were attributed to training and the setting up of machines.

In January, 137 undervotes surfaced in the House District 91 special election covering coastal areas of Broward and Palm Beach. The victor, now state Rep. Ellyn Bogdanoff, R-Boca Raton, won by 12 votes over Mayor Oliver Parker of Lauderdale-by-the-Sea.

Some political insiders suggested voters deliberately cast undervotes. However the District 91 outcome has led to demands by U. S. Rep. Robert Wexler, D-Boca Raton, that the machines have a paper record of the votes. Several states and Palm Beach have announced plans to buy printers, which are under development but state officials and manufacturers say it is unlikely they will be ready for the Nov. 2 presidential election.

Seniors find difficulty

Meanwhile, some segments of the voting population seem to struggle with touch-screen devices.

In Broward's March election, the Sun-Sentinel identified precincts where the most undervotes occurred, generally in senior communities.

Seniors can have trouble with touch-screens, said Douglas W. Jones, a University of Iowa computer scientist and an electronic voting machine consultant to several governments, including Miami-Dade.

He recounted a story told to him of an elderly voter at a touch-screen machine in Iowa.

"She mistakenly cast her ballot prematurely and grew increasingly frustrated in the voting booth, trying to fix the situation. She was a victim of not knowing how to use the machines," Jones said.

"These machines need to be studied by human behavior specialists. They need to be designed to make voting as easy as putting an X in a box," Jones said.

He also said, "All this was jumped into too quickly. I don't believe this change to new machines needed to be done as if it was an emergency."

Meghan McCormick, spokeswoman for ES & S, whose machines are used in Miami-Dade and Broward, said voters are given reminders that they failed to cast a vote on a touch-screen machine and some simply choose to cast blank ballots.

"We have safeguards in place. In our experience, some people choose not to vote," McCormick said.

Pasco Supervisor Browning, who wrote the Florida State Association of Supervisors of Elections' white paper a year ago supporting touch-screen voting, said the machines bought in 2001 have been improved with d software and possibly printers in the near future.

"Was the [ES&S touch-screen] ready? Who knows? But I do know that the system I bought in 2001 is not the same system we are using today," Browning said. "Its better, and it will continue to get better."

A recent issue has emerged concerning the software used in the ES&S machines used in Miami-Dade and Broward. Officials are scrambling to resolve a software error that hinders efforts to audit and double-check the accuracy of vote totals. Nicole de Lara, a spokeswoman for Florida State Secretary Glenda Hood, described the problem as minor and said Friday she expects it to be fixed within a week.

Theresa LePore, the Palm Beach elections supervisor, said the only way to cut down on the number of undervotes would be to give voters the choice of casting a ballot for "none of the above." The Legislature would have to approve that change.

Short of that, LePore said it is impossible to eliminate undervotes because some people will choose not to vote for any candidate or will make mistakes.

"There is only one perfect voting system," LePore said. "That's the one that doesn't involve humans."