On Nov. 2, election workers in 260 polling places submitted data to the Miami-Dade Supervisor of Elections office that did not match up with the total number of touch-screen ballots reported by the canvassing board, according to a study conducted by the Miami-Dade Election Reform Coalition, a nonpartisan watchdog group.

Out of those 260 polling places, 23 submitted totals that were off by more than 50 voters, while 68 submitted totals that were off by more than 10 voters. In one polling place, the difference was 1,284. That polling place was one of five that showed very large discrepancies. The coalition is studying those five for possible clerical errors.

Not including those five polling places, the percentage difference between the reported signature totals and the machine ballot counts varied from less than 1 percent to 34 percent. The differences included polling places where there were more reported signatures than ballots cast, and others where ballots cast exceeded reported signatures.

A Daily Business Review inspection of voter logs showed that in some polling places where there were discrepancies, the totals reported by election workers equaled the total number of voters in the precinct, including absentee and early voters. But the county reporting form for signature totals, Certificate No. 2, asks workers to fill in the total of polling place signatures only, not the total of absentee voters.

The reform coalition said the discrepancies cast doubt on the county’s ability to check the accuracy of the controversial iVotronic touch-screen machines. Since there are no paper records votes cast, critics argue that it’s essential for election officials to carefully reconcile the total of voter sign-in signatures with the electronic tallies on the machines. To ameliorate this problem, critics have urged that the machines be outfitted with printers to produce backup paper records of individual votes.

"The counting of signatures and reporting of discrepancies to the canvassing board are fundamental to counting votes correctly" said Martha Mahoney, a University of Miami law professor and member of the reform coalition who led the analysis. "It’s really important to do this on election night. How do we know otherwise whether the machines are correctly reporting every vote?"

Seth Kaplan, a spokesman for the Miami-Dade supervisor of elections office, said the discrepancies do not necessarily indicate voting machine malfunctions. Human error in counting and the lack of a policy for the inclusion of absentee and early voters could account for many of the mistakes, he said.

Kaplan said his department recognized that the signature counts were off and said it was something the supervisor of elections office would seek to improve in the future. "Are we batting 1.000 on them all being correct? No," Kaplan said. "It’s a training issue. Whenever we become aware of issues, we re-emphasize those certain issues in training."

Kaplan said that under the department’s new leadership, the discrepancies would likely be looked into. Former Supervisor Constance A. Kaplan, who is not related to Seth Kaplan, resigned after irregularities were found in the March 8 slot machine referendum.

"All of these procedures are under review, and these are the kind of things we want to tighten up," Kaplan said. "This is something that we will be looking at."

Miami-Dade County Commissioner Katy Sorenson, who served on the canvassing board in the 2004 elections, declined to comment on the discrepancies, saying she was not familiar with the coalition’s study. "Things can happen where people decide not to vote, they can sign in and leave, so I don’t really know if that’s a problem," she said.

Jenny Nash, a spokeswoman for Florida Secretary of State Glenda Hood, said the issue of reconciling signature totals and machine counts was a local one and is not the concern of her office. "Each supervisor has their own process for how they reconcile the numbers," Nash said.

The discrepancies between the signatures and the ballots are not necessarily indicative of iVotronic machine problems or missing votes. In some instances the differences were due to sloppy counting of signatures by poll workers.

For example, Precinct 11 recorded 693 votes and 849 signatures. But a Daily Business Review inspection of the voting log found 694 signatures. Another precinct with a major difference was Precinct 362, which recorded 583 votes and 859 signatures. A review of that precinct’s signature log found 580 signatures.

Nevertheless, inaccurate signature counts could interfere with the ability of the supervisor of elections office and the county canvassing board to promptly identify problems in a close election, such as machine malfunctions and election fraud, before the election results are certified.

Latest glitch

Miami-Dade and Broward counties have experienced a series of embarrassing election glitches since they adopted touch-screen voting systems in 2002. In March, Constance Kaplan resigned as elections supervisor after reporting to the county manager that in the vote on slot machines, nearly 1,300 fewer votes were recorded than there were voters who showed up to vote in the single-issue election. County leaders are considering switching to a paper-based optical scan system.

In an audit of the 2002 general election, the Miami-Dade Audit and Management Services Department found discrepancies between signatures and ballots cast in 48 randomly ed precincts. The audit found that the discrepancies did not affect any election outcomes. But it recommended greater attention to reconciling signature and ballot counts. It called this an "an important audit control, as well as a positive tool for promoting precinct worker corrective action."

Last October, the election reform coalition wrote to Secretary of State Hood, expressing its concern about the absence of a statewide system for reconciling voter signatures and electronic ballots.

"Despite the statutory requirement to account for voted ballots, the Polling Place Procedures Manual does not include a procedure for accounting for electronic ballots, counting signatures of voters, or explaining any gaps between these figures," the letter to Hood said.

Despite the letter, Hood did not put any such system in place in the touch-screen counties.

Machine malfunction?

The voting process works like this: After entering the polling place but before casting a ballot on an iVotronic machine, Miami-Dade voters must sign next to their printed name in an election log of registered voters. At the end of Election Day, poll workers are responsible for counting the signatures and recording the total on Certificate No. 2, a paper document.

Poll workers are told prior to the election to compare the number of signatures to the number of votes tallied on the iVotronic machines to find discrepancies. Any differences could indicate vote fraud, election machine malfunction, poll worker error or other problems and should be reported to the department.

Seth Kaplan said that poll workers in the November election were told to visually check the signature count against the machine totals and to report any large differences to the department. However, that instruction was not written in the procedures guide for poll workers.

The reform coalition’s analysis and a Daily Business Review spot check found numerous polling places where there were large discrepancies including polling places where there were more signatures reported than electronic ballots cast, and ones where there were more ballots cast than signatures reported.

At Precinct 124, the certificate documented 17 signatures, while the machines counted 1,301 votes. That precinct has been earmarked as a probable clerical error in the coalitions study. A check of the voter log by the Daily Business Review confirmed that there were far more than 17 signatures at the precinct.

It’s the smaller differences that will be scrutinized by the coalition. For example, in Precinct 41 there were 910 votes and 844 signatures, a difference of 62. At combined Precinct 117/166, the ballots totaled 995 and the signatures numbered 1,276.

In some of these polling places, Mahoney speculated, the differences may have been due to clerical errors by poll workers.

For Precinct 816, in the Church of the Ascension at 11201 SW 160th St., a Review inspection of the voter log showed 945 signatures, while the iVotronic computer tape showed a count of 1,032. But the individual machine counts are listed on the tape as well, and they add up to 945. But the number certified by the canvassing board came in at 1,116 votes.

In this case, the discrepancy may have been due to an iVotronic machine malfunction. Lynn Kaplan, a volunteer observer for the reform coalition who was at that polling place on Nov. 2, said in an interview that as a poll worker was closing down one of the iVotronic machines at the end of the day, an error message popped up on the machine’s digital screen saying: "Internal malfunction/unit closed to save data/vote data corrupted."

Kaplan said that while a control number called a public count said that 84 voters had voted on that machine that day, the computer had tallied no results The public count shows up on the exterior of a machine and keeps tabs on how many ballots have been cast each day. The public count number should match the number of votes recorded on the machine’s memory.

Kaplan also said she witnessed a poll worker take the Personal Electronic Ballot cartridge from a nearby machine that was in the process of downloading election data, and the PEB into the malfunctioning machine.

The PEB is used by poll workers to control the machines. It starts up the machines at the beginning of the day. At poll closing time, PEBs are ed into machines to download the election results. After the PEB is used to gather all the data, it should contain the results from the precinct.

After the PEB switch, Kaplan said, the poll worker did another computer printout of the results. Both machines showed zero votes.

A county computer specialist was summoned. After several printouts of election results including all machines in the precinct, the final printout showed 84 votes on both machines.

That left significant discrepancies between the signature total for the polling place, the control number showing votes on the machines and the computer printouts of voters.

The machine’s manufacturer, Omaha-based Electronic Systems & Software, did not respond to request for comment.

Lynn Kaplan said election workers at Precinct 816 had lots of trouble at the end of the Election Day reconciling the conflicting iVotronic numbers with the total number of signatures. "I shudder to think what goes on in all the other precincts if people aren’t keeping up with these things," she said.

The reform coalition’s Mahoney said she wants the supervisor of elections’ office to investigate what happened that night at Precinct 816.

"It shows a lack of transparency in the system," she said.

The Supervisor of Elections office said it is looking into what happened with Precinct 816 in response to a Daily Business Review request for an explanation.

Mahoney said the situation in Precinct 816 highlights the need for better procedures and training for reconciling signature and ballot totals. She also stressed that the county must investigate discrepancies to see if they resulted from fraud or equipment malfunction.

She noted that unless the numbers are inspected immediately after the election, mistakes would not be caught in time for certification, which must be complete in the 48 hours after the elections.

"It’s a very important comparison," Mahoney said. "With electronic voting, how can you be sure you’ve got the right number of ballots? The benchmark has to be the number of voters that came into the polling place."