WEST PALM BEACH -- A Palm Beach Post analysis of under-votes from the Nov. 7 presidential election shows why Al Gore pushed so hard to count all the dimpled ballots.
Without including every mark next to a candidate's name -- every hanging chad, pinhole, ding and dimple -- Gore couldn't have found enough votes to win, The Post's ballot-by-ballot review of 9,150 under-votes in Palm Beach County reveals.
Gore could have gained 784 more votes than George W. Bush in the county if all the marks next to their names had been counted, the study found. But the county canvassing board did not count the 5,736 under-votes that The Post found had marks for Gore or Bush.
Without those ballots, Gore still gained 174 votes in the county's 10-day hand recount completed Nov. 26. The county missed a deadline to complete its recount and those gains did not count in Bush's certified 537-vote victory.
For all the talk about hanging chads during the raucous 37-day contest, the under-votes show little evidence of them. And those few punch cards with hanging chads -- at least those remaining after repeated ballot handling -- favored Bush.
Bush would have gained 14 votes had the canvassing board counted only the 62 under-votes with hanging chads. Under-votes are those ballots the canvassing board decided had no clear mark for any candidate.
The former vice president would have picked up a net gain of 25 had the canvassing board accepted the additional 313 ballots in which light could be seen through pinholes or dimples.
But it's not until the much more numerous category of marks with no light are counted that Gore would have realized the 784-vote gain. Overall, of course, the county's vote was never that close; Gore easily carried Palm Beach County, 270,233 to 153,278.
Gore's lawyers, experienced from fights over dimpled ballots in Massachusetts, knew to focus on the dimples from the start and pressed for all of them to count. The Bush team, on the other hand, fought the counting of anything but clear punches.
"We thought all along that those dents didn't get there on their own," Gore lawyer Dennis Newman said. "We knew that Bush would pick up votes also. We just thought that we would pick up more."
"It sounds like more science fiction to me," former Montana Gov. Marc Racicot said Friday, speaking for the Republicans. "It doesn't matter how many times there's a recount, there seems to be a disparity in the analysis. The basis upon which you draw a conclusion is subject to incredible impeachment."
The Post's review was conducted from Jan. 2 to Jan. 29. In the first half of that review, reporters looked at the county's most hotly contested 4,513 under-votes -- ballots that drew challenges from Democrats or Republicans to the canvassing board's determinations.
Of those, The Post found 4,318 ballots with dimples or hanging chads: 2,500 next to Gore's name and 1,818 next to Bush's -- a net advantage of 682 for Gore, which The Post reported Jan. 27.
Those are the ballots the Democrats asked a Tallahassee judge to review -- a strategy that The Post's analysis suggests may have yielded enough votes to overtake Bush. But no judge agreed to review them.
The results of The Post's later review of the remaining 4,637 under-votes show that the pivotal battleground was indeed in the fight over the disputed ballots, but these still helped Gore's cause.
Among those remaining ballots were another 1,418 with dimples or hanging chads, which, when divided among the candidates, yielded 102 to Gore's advantage.
County Canvassing Board Chairman Charles Burton pointed out that any decision to include dimples would have required a recount of all ballots to weed out those that combined a clear punch with a dimple for another presidential candidate, because they would have essentially been over-votes, which are ballots with votes for more than one candidate. The Post did not attempt such a review.
For that reason, however, 130 ballots with dimples for more than one candidate were not included in The Post's totals.
Additionally, 636 ballots with dimples for the eight other presidential candidates did not figure into the final figures.
While the fight over the dimple standard consumed lawyers for the Democratic Party, the three Democrats making up the Palm Beach County Canvassing Board gave in on some dimpled ballots but for the most part refused to count them as votes. The lawyers failed to persuade Burton and Supervisor of Elections Theresa LePore to count all the dimples although the most partisan member of the canvassing board, County Commissioner Carol Roberts, argued they should be counted.
Burton struggled to get a clear definition from Circuit Judge Jorge Labarga, but Labarga's ruling -- by removing a 1990 rule that banned the counting of ballots with indentations -- raised more questions for the canvassing board than it answered.
Rather than set a firm standard, Labarga instructed the board to judge every ballot on its own merit to meet the state standard of determining the intent of the voter.
The canvassing board looked for dimples throughout the ballot, including other races, to show if the voter intended to vote that way, a standard that ignored Democratic arguments that design flaws in punch-card voting machines resulted in more dimples on the ballot's presidential column.
"I never maintained that we had a correct standard," Burton said. "I simply thought we had a reasonable standard and it's one we tried to employ reasonably."
The Post found more than 400 ballots that did not count even though the voter dimpled the ballot in at least one other race.
The under-vote review also confirmed what other Post analyses have previously shown -- that people voting at polling places using Data Punch voting machines were more likely to under-vote than those using the county's more prevalent type of machine, the Votomatic.
While less than a third of the votes Nov. 7 were cast on the newer Data Punch machines, 55.8 percent of the 8,090 under-votes cast on voting machines were done on Data Punch, compared with 44.2 percent on Votomatics. Absentee ballots accounted for the other 1,060 under-votes.
Data Punch machines will not be used during Tuesday's municipal elections, LePore has said.
On Dec. 8, when the Florida Supreme Court ordered 64 counties to recount under-votes by hand, it offered little direction on how to accomplish that. Counties began counting but the U.S. Supreme Court halted the recounts the next day.
Now, newspapers making public records requests to review ballots are learning how difficult it is to single out under-votes for examination.
The Orlando Sentinel reported, for instance, that Seminole County could produce just 210 of the county's 219 under-votes.
Palm Beach County's LePore says the under-votes and over-votes can't all be accounted for without counting all 462,644 ballots cast in the county.
Reporters from The Palm Beach Post counted 9,150 under-votes but the county recorded 9,251 under-votes during its hand recount. In 145 of the county's 637 precincts, the number of ballots counted by The Post did not match the official tally. In 12 of those precincts, the discrepancy was greater than 10 ballots but no more than 25. In 102 precincts, the differences were three or fewer. In some cases, The Post found more under-votes than the county, and in others, it found fewer.
In one precinct, 135B, the canvassing board incorrectly characterized 23 under-votes as over-votes. LePore has refused several requests to review that error.
State law requires that only elections officials handle the ballots. Lepore's office charged the four organizations a combined $125 an hour for the ballot review. The Post paid for 93 hours.
Aside from The Post, reviews were conducted by The Miami Herald, the Republican Party of Florida and Judicial Watch, a conservative government-watchdog group.
Elsewhere in the state, The Post also found 73 uncounted marks for Bush and 54 for Gore on absentee ballots in Martin County, a net gain of 19 for Bush. In St. Lucie County, The Post found 190 marks for Gore and 79 for Bush, a net gain of 111 for Gore.
A media consortium that includes The Palm Beach Post, The Wall Street Journal, The Washington Post, The New York Times, the Los Angeles Times, CNN, The Associated Press and the Chicago Tribune is examining under-votes and over-votes in all of Florida's 67 counties.
A separate review of under-votes only, by The Miami Herald and USA Today, is nearly complete.
Experts say no count -- whether by hand or machine -- will ever be 100 percent certain. Computer industry consultants estimate the error rate for counting punch cards could run as high as 1 percent and varies with the number of times the cards are handled.
Palm Beach County's ballots underwent three machine recounts, one hand recount and a sample hand recount of four precincts. Each time the result was different.
During the hand recount, the canvassing board converted about 1,000 under-votes into votes, most of them for Bush or Gore.
In the 37-day election contest of Florida election results, Gore had hoped to find a mother lode of votes through hand recounts in heavily Democratic South Florida. A manual recount in Broward County added 567 votes for Gore. The Palm Beach County hand count, if it had been done on time, would have added another 174.
Democratic efforts to get a hand recount in Miami-Dade County were stymied. A Palm Beach Post analysis of the Miami-Dade under-vote showed Bush would have gained 6 votes. The Miami Herald found Gore would have gained 49.
In the central and north part of the state, The Orlando Sentinel and two other newspapers counted a net gain of 366 marks for Gore in 15 optical-scan counties, most of which Bush carried in the Nov. 7 election.
Staff writer Noah Bierman and staff researcher Monica Martinez contributed to this story.
Copyright © 2001, The Palm Beach Post