Some sites show huge vote gaps
17 wards have at least 100 more votes than voters; 2 miss by over 500
By GREG J. BOROWSKI  Milwaukee Journal-Sentinel
Posted: Feb. 1, 2005

Record-keeping surrounding the Nov. 2 presidential election in Milwaukee is so flawed that in 17 wards there were at least 100 more votes recorded than people listed by the city as voting there.
In two wards, one on the south side and one on the north side, the gap is more than 500, with fewer than half the votes cast in each ward accounted for in the city's computer system, a Journal Sentinel review has found.

Such gaps were present at different levels in nearly all of the city wards and could hamper the investigation launched last week by federal and local authorities into possible voter fraud by giving an incomplete or inaccurate picture of who actually voted.

They also raise questions about the level of oversight of how the city records who voted in each ward - an important safeguard that, properly done, can be used to spot double voting and other problems.

And unless the gaps can be fully resolved, they leave room for critics to allege that ballot boxes were stuffed in the city, which went heavily to Democrat John Kerry over President Bush in a state with one of the closest margins in the country.

City officials said Tuesday that the large gaps can be blamed on major flaws in how polling-place logbooks were scanned and individual votes recorded in the computer. In some cases, entire pages were bypassed in the process, which involves scanning a barcode for each pre-registered person who votes.

That echoes the "layer upon layer of human error" statement offered last week by Lisa Artison, executive director of the Election Commission, in explaining why some 300 people are mistakenly listed as voting twice. Artison was out of the office Tuesday and unavailable for comment.

After inquiries from the Journal Sentinel about the gaps, officials said election staffers rescanned the logbooks for the 10 most problematic wards. In doing so, they tallied 1,707 votes that, while counted by voting machines, were not recorded properly later.

Officials now plan to rescan all of the books. They also have identified other problems, including new voters who were added to the system without being shown as having voted. They hope those problems will explain the 7,000-vote gap found a week ago by the newspaper in its extensive review of city election records.
Overhaul promised

"When there are mistakes, I want to know about them," Mayor Tom Barrett said. "There are mistakes, and we will clean them up."

Barrett, elected last April, said "there were problems in the department before we got here," but added: "I accept the responsibility (for the problems)."

All together, the Journal Sentinel has found a gap of about 7,000 votes that are unaccounted for. That is based on a difference between the 277,535 ballots cast in Milwaukee and the 269,212 people listed as voting in the city's computer system.

While that leaves a difference of about 8,300, at least 1,300 of that number would represent same-day registration cards that could not be processed for reasons ranging from missing addresses and names to addresses that are outside the city.

Barrett reiterated Tuesday that he welcomes the investigation, which is being overseen by Milwaukee County District Attorney E. Michael McCann and U.S. Attorney Steve Biskupic.

"I don't think we have seen any evidence of fraud," Barrett said. But he acknowledged that vote gaps and other problems shake the confidence of residents and others in the system.

The investigation was launched a day after the Journal Sentinel reported at least 1,200 votes in the Nov. 2 election were cast from invalid addresses. Some are due to transposed digits in addresses, but many others could not easily be explained or the voters readily identified.
Voter ID bill

Meanwhile, a state audit into election problems in Milwaukee, a probe that will include other communities statewide, could get formal approval today. And on Monday, Republicans in Madison renewed a push - which failed two years ago - to require all voters to show photo identification.

On Tuesday, Rick Graber, head of the state Republican Party, challenged his Democratic counterpart to appear at a hearing on the matter Thursday so together they can condemn "the fact that potentially thousands of voters across Wisconsin had their legally cast ballot disenfranchised by fraud and abuse."

Linda Honold, state Democratic Party chair, said she was unsure if she would attend the meeting but added that if she did go, she would do so to oppose the bill.

"If I'm there, I'm not going to be arguing what he wants me to argue," she said.

Others, including the head of the American Civil Liberties Union of Wisconsin and the group Wisconsin Citizen Action, condemned the voter ID proposal.

"The way to prevent fraud is more and better poll workers," said Larry Marx, co-executive director of Wisconsin Citizen Action. "We want to make it easier to vote and harder to cheat. The photo ID bill makes it harder to vote and harder to cheat."

The criminal investigation into possible fraud appears to be focusing, at least initially, on the 1,305 registration cards that could not be processed, as well as some 2,800 verification cards that have been returned as undeliverable.

Under state law, those undeliverable cards already must be forwarded to the district attorney - though the newspaper has found spotty compliance on this statewide.
Problem wards cited

The size of voter gaps in some wards, often discrepancies of 10% or more, surfaced as the Journal Sentinel examined the 7,000-vote gap and whether it is isolated in some troublesome wards or reflected across the city.

In a pool of more than 277,000 votes, a gap of 7,000 represents about 2.5%. If it were spread evenly between the city's 312 voting wards, it would mean about 22 votes in each.

Instead, the review found that the gap in 17 wards topped 100 votes, while in other wards there was, in effect, a surplus of voters listed. That is, city records list more people voting than the number of ballots cast.

This is likely a result of a problem, also identified by the Journal Sentinel, in which hundreds of individuals - usually already-registered voters who changed addresses or completed a new form - are mistakenly recorded with two votes in the city's computer system.

In all, 82 wards faced this problem. Ward 189, voting at Lyons Park Pavilion, 3301 S. 55th St., listed 262 more people as having voted than ballots cast. That is off by 25%.

Only nine of the wards matched perfectly.

The remainder, 221, listed fewer voters than ballots cast, in a range from 1 to 596.

The newspaper also attempted to match the number of ballots counted in each ward to the last recorded voter at each ward, but incomplete information made this difficult.

The 17 wards with more than 100 more votes than voters account for 4,083 of the votes in the original gap.

It is unclear why some of the problems were not apparent as city workers transferred voter information from polling-place logbooks to the computer. This is done by scanning a barcode next to the name of each pre-registered voter.

Same-day registrants are recorded later, by typing in information from their registration cards. In reviewing the work Tuesday, election officials identified cases in which new voter registration cards were entered into the system, but those voters were not recorded as voting.

That, too, would account for a gap, though it's unclear how many votes this affects and from which wards.

While the process is extremely tedious and prone to human error, it is a critical one. Mistakes and oversights can obscure any existing cases of fraud and open the door wider for new ones in the future.

For instance, voter information incorrectly recorded in the computer system can mean a newly registered voter is not listed on the rolls for the next election, on Feb. 15.

If a voter changes addresses and the computer does not remove the bad address, it leaves an opening in the system where someone bent on fraud can readily vote twice. Such duplicate registrations happened hundreds of times, due to what Artison described as a computer "glitch."

And, of course, if hundreds of names appear incorrectly in the records as duplicate voters, it can be almost impossible to identify and weed out those who actually vote twice.

The voter gaps do not alone point to fraud. Nonetheless, they are a measure of the depth of problems of a system that has few safeguards in place.

At the top: Ward 93, which voted at the Cooper Park Pavilion, 8701 W. Chambers St. City records showed 923 ballots cast, but listed only 398 people as voting, a difference of 525, for a discrepancy of 57%.

On Tuesday, the city rescanned the logbooks for the site, and identified 515 votes that had been missed. Incidentally, Ward 94, which also voted at the Cooper Park site, was only off by 8 votes, or 1% of its total.

In Ward 93, Bush topped Kerry, 510-403.

Next in line: Ward 229, which voted at Garland School, 1420 W. Goldcrest Ave., on the city's far south side. City records showed 1,134 ballots cast, but listed only 538 people as voting, a difference of 596, for a discrepancy of 53%.

The city rescanned the logbooks for that site and identified 593 unrecorded votes.

In that ward, Kerry topped Bush, 608-516.

In several other cases, though, the rescans added only a handful of votes to wards the Journal Sentinel found were off by 100 or more.

On Monday, the Journal Sentinel filed an open records request seeking copies of election-day logs or other records for each ward that would show the number of the last voter in line when polls closed.

If everything is handled perfectly, the three numbers - voters at the polls, ballots counted and voters recorded - would all match. While such a match would not rule out fraud, for now even that most basic check cannot be made by the newspaper or even the Election Commission staff.

Using election-day "incident logs," which are supposed to include this number, the newspaper found many cases where the information was missing and others where apparently no log was kept or filed with the city.