January/February 2001

January/February 2001 | Contents

AGGRESSIVE NEWSPAPERS WERE KEY FLORIDA PLAYERS

BY JAMES HARPER

"It's not our job to help select the president of the United States. But it is our job to explain what's going on."

Martin Baron, executive editor of The Miami Herald, was talking about a provocative story his newspaper published on December 3. If Florida's election machinery had functioned perfectly, the story said, and if all the votes had been counted as people intended when they entered the polls, Al Gore would have won Florida -- and therefore the presidency -- by more than 23,000 votes.

At that moment, the fate of Florida's vote was still being argued in court. Republicans dismissed the paper's analysis as "statistical voodoo." Democrats hailed it as "compelling evidence" in support of their man.

Now, with the presidential race finally over, at least half a dozen papers -- including the Herald -- have begun an effort to count all the disputed ballots in Florida. It was just under way as cjr went to press, and promises to be at least as controversial as the Herald's December 3 story.

That story, based on a university professor's analysis of each of the state's 5,884 precincts, didn't change the outcome, of course. Nor was it expected to.

"It's not evidence that could be introduced in court, and we didn't expect it to be introduced," Baron says. "We did expect it would have an influence on political conversation."

Indeed it did. The story was picked up by newspapers all over the country and was a topic on the Sunday morning talk shows. Curtis Morgan -- the Herald reporter who wrote the piece -- received 634 e-mails, almost evenly divided between Democrats and Republicans.

The Herald's story was the most striking example of enterprise journalism from Florida's newspapers, which found themselves, once again, at the center of an international spectacle. The five-week rollercoaster ride between Election Day and Gore's concession speech was especially challenging, even in a state long accustomed to boatlifts and custody battles, hurricanes and celebrity murders.

"We've covered presidential politics before," says David Dahl, the St. Petersburg Times's state editor and former Washington bureau chief, "but we never had to do it with such a national audience. Everybody had to stretch their game."

While the Florida story had papers around the state scrambling to analyze and explain, ground zero happened to be the home turf of three savvy and competitive newspapers: the Herald, The Palm Beach Post, and the South Florida Sun-Sentinel, based in Fort Lauderdale.

Within a few days of the vote, all three newspapers began a series of articles attempting to define the precise impact of flawed election machinery. When one paper thought of analyzing the election data one way, its competitors would follow with a more sophisticated or wide-ranging analysis -- a leapfrogging series that culminated in the Herald's statewide analysis declaring Gore's "virtual" victory in Florida.

"And then we did the whole country," says Ellen Soeteber, the Sun-Sentinel's managing editor, referring to a December 8 report that detailed how punch card ballots were responsible for significantly higher ratios of uncounted votes throughout the U.S. in 1996.

"This is one of the most competitive newspaper battlegrounds in the country," says Soeteber, who spent six-and-a-half years at the Sun-Sentinel before moving on to be editor of the St. Louis Post-Dispatch this month. "So this is typical."

And so, while the lawyers for the respective political parties squared off over recounts and pregnant chads, the battle among the newspapers to explain what happened intensified.

November 8 -- Both the Sun-Sentinel and the Post had first-day stories describing widespread voter complaints over Palm Beach County's now infamous butterfly ballot. This, of course, was before anyone knew that the glitch might change the outcome of a national election.

November 10 -- The Sun-Sentinel offered a precinct-by-precinct analysis of Palm Beach County that showed uncounted presidential votes coming most often from Gore strongholds. "Substantial numbers of seniors and African-American voters double-punched their ballots," the story reported.

November 11 -- The Herald followed up with a more detailed story showing that forty-four Palm Beach precincts had thrown out more than 10 percent of their ballots -- 2,500 lost votes in those precincts alone. Democrats were a majority in thirty-eight of those precincts, the Herald reported; twenty-four of the precincts had black majorities, and eight were composed predominantly of voters over sixty-five.

November 12 -- The Post was able to report that it had talked to 1,172 voters in one suburban Boca Raton enclave -- well over half of the 2,042 people who had cast ballots in that precinct the previous Tuesday. The official count had recorded thirty votes for the Reform party candidate, Pat Buchanan. But none of the people Post reporters talked to acknowledged voting for Buchanan on purpose; five said they knew they had done so by mistake.

November 15 -- The Herald published a map of Miami-Dade County that showed that errors ran highest -- as much as 10 percent -- in the county's black precincts. Three days later, it reported the unusually high rate of spoiled ballots statewide, and, based on a more detailed analysis of the state's four most populous counties, said the highest rates came in Democratic and minority neighborhoods.

That got some of the Herald staff thinking: what if you could call up every precinct in Florida and see where all the uncounted ballots were from? It took Herald reporters Anabelle de Gale and Lila Arzua two weeks to gather the numbers from Florida's sixty-seven counties. Among other questions, they wanted to distinguish between undervotes -- where no preference had been registered on the presidential portion of the ballot, and overvotes -- where two or more preferences had been marked.

November 28 -- Six days after Miami-Dade's canvassing board had given up on its manual recount because it knew it couldn't meet the Florida Supreme Court's deadline, the Sun-Sentinel -- which had already beaten the Herald by three days in reporting the number of discarded ballots statewide, which was 50 percent higher than it had been in 1992 -- told readers that 58 percent of Miami-Dade's undervotes came from precincts that Gore carried.

December 1 -- As the U.S. Supreme Court heard its first oral arguments in the case, the Sun-Sentinel published an analysis and a detailed map of all three southeast Florida counties: "One-third of the disqualified votes, 22,807 in all, were concentrated in mostly black zones where at least 8 percent of the votes for president went uncounted . . ." staff writers Stacey Singer, John Maines, and Scott Wyman said.

The story went on to suggest that 17,900 of those uncounted ballots probably belonged to Gore, leaving only 4,474 for Bush.

Provocative stuff, although the numbers applied only to three Democratic-leaning counties in south Florida. The table was set for the Herald's statewide stab at divining the true intentions of Florida voters.

Stephen Doig, the Arizona State University professor who crunched the numbers of the Herald story, is a former Herald staff member who helped pioneer computer-assisted reporting in the late 1980s and early 1990s. To calculate where all the state's uncounted votes might belong, he used a well-established principle of polling that says a large group's opinions will be distributed pretty much the same as in smaller, representative samples.

Since the overwhelming majority of votes in each precinct had been counted, Doig had some unusually large and accurate samples to work with. He simply distributed the rest of the votes in each precinct -- all 185,246 of them throughout the state -- according to the same percentages that each candidate won in that precinct.

But what if many or most of those uncounted ballots were never intended as votes for any candidate?

Doig ran the numbers again, this time eliminating varying percentages of the uncounted ballots: 10 percent, 20 percent, 30 percent. He found that even if he ignored 90 percent of the uncounted ballots, Gore theoretically would have won by 1,443 votes -- a slim margin indeed, but almost three times as big as the 537-vote cushion that finally sealed the election for Bush.

Doig's analysis did more than cast doubt on the official result. The numbers also offered strong proof of trends that other journalists and political observers could only generalize about.

For example, Doig demonstrated what difference the type of voting machine made in the number of ballots thrown out: Florida had fifty-one precincts in which more than 20 percent of the ballots were rejected; forty-five of those precincts, or 88 percent, used punched cards.

The analysis also underlined the unfairness of recounting the discarded ballots only in selected counties. In areas that Bush carried, such as Naples, Fort Myers, and the county just north of Jacksonville, Doig showed that Bush would have padded his lead over Gore by thousands of votes.

The most obvious criticism is that statistical analysis is no substitute for the actual counting of votes. It relies on too many assumptions -- that every voter followed directions correctly; that every machine worked perfectly -- which could never be duplicated in the real world. And as the U.S. Supreme Court ultimately decided, there is no provision in law to guess at votes that haven't been properly, and promptly, counted.

But to Judy Miller, the Herald's city editor who supervised the story, its news value is clear. "People deserve some idea of what is in those ballots," she says.

Her boss, Mark Seibel, the assistant managing city editor who oversaw the paper's post-election coverage, says the story wasn't designed to support or discredit either candidate, but to underscore "the failure of our elections equipment to accurately count the vote. It's clear to me that voting equipment determined who won this election."

James Harper retired from the St. Petersburg Times in May, after working twenty-four years as a reporter, editor, and columnist.