Discrepancies found in votes, signatures
A study on the November general election shows thousands of discrepancies between the number of votes cast and signatures collected by poll workers at the end of the day.
BY NOAKI SCHWARTZ AND JASON GROTTO    Miami Herald    07 May 2005

A study by a member of the Miami-Dade Election Reform Coalition found frequent discrepancies between the number of votes cast and signatures collected by poll workers in the November general election.

The study, expected to be released next week, found that workers at dozens of polling places submitted counts of signatures to elections officials that did not match the number of votes recorded on the touch-screen machines.

Most of the discrepancies appear to have been caused by poll workers miscounting signatures. Others may have been caused by voters signing in but walking away before voting or glitches in at least one machine.

NO PAPER BALLOTS

Still, because the iVotronic touch-screen machines do not use paper ballots, elections officials may never know for sure highlighting a continuing issue with the machines.

''I think it's significant that we can't be sure that things can't be counted, compared and investigated all the time,'' said the study's author and University fo Miami Professor Martha Mahoney, who also is a member of the coalition.

The news comes weeks after Elections Supervisor Constance Kaplan resigned amid revelations that human error led to the county's iVotronic touch-screen machines tossing out hundreds of votes. The $24.5 million machines have become such an issue that the county manager has asked the elections department to advise him on whether to keep them.

The study's findings were a surprise for the elections department, which had considered the November election a success.

''This seems to be an area where we have to strengthen our training,'' said Seth Kaplan, a spokesman for the department.

On Election Day, registered voters sign a paper before voting, signatures that are supposed to be counted by poll workers and compared to votes recorded on the machines at the end of the day. Large discrepancies indications of a problem are supposed to be reported to elections officials.

The study found that there were 5,917 cases where there were more votes than signatures. Of those cases, 4,353 are expected to be clerical errors where poll workers miscounted the signatures, the study found.

For example, in one precinct 590 ballots were cast but only three signatures were counted.

MACHINE FAILURE

In another 2,208 cases, there were more signatures than votes. Some of the cases may have been voters who signed in and then left before voting.

In at least one case, the problem appears to have been the machine. At precinct 816, the Church of the Ascension, the number of signatures was 945, a figure verified on the machine. But the number of reported votes that day was 1,116.

After testing the machine, its maker, Election Systems & Software, told the state Division of Elections that an internal memory bank failed but the votes were recorded.

In a memo to County Manager George Burgess, elections chief Lester Sola said he expects state elections officials to review ES&S's findings with him on Monday.