COLUMBUS, Ohio -- Tanya Thivener's is a tale of two voting precincts in Franklin County. In her city neighborhood, which is vastly Democratic and majority black, the 38-year-old mortgage broker found a line snaking out of the precinct door.
She stood in line for four hours -- one hour in the rain -- and watched dozens of potential voters mutter in disgust and walk away without casting a ballot. Afterward, Thivener hopped in her car and drove to her mother's house, in the vastly Republican and majority white suburb of Harrisburg. How long, she asked, did it take her to vote?
Fifteen minutes, her mother replied.
"It was . . . poor planning," Thivener said. "County officials knew they had this huge increase in registrations, and yet there weren't enough machines in the city. You really hope this wasn't intentional."
Electoral problems prevented many thousands of Ohioans from voting on Nov. 2. In Columbus, bipartisan estimates say that 5,000 to 15,000 frustrated voters turned away without casting ballots. It is unlikely that such "lost" voters would have changed the election result -- Ohio tipped to President Bush by a 118,000-vote margin and cemented his electoral college majority.
But similar problems occurred across the state and fueled protest marches and demands for a recount. The foul-ups appeared particularly acute in Democratic-leaning districts, according to interviews with voters, poll workers, election observers and election board and party officials, as well as an examination of precinct voting patterns in several cities.
In Cleveland, poorly trained poll workers apparently gave faulty instructions to voters that led to the disqualification of thousands of provisional ballots and misdirected several hundred votes to third-party candidates. In Youngstown, 25 electronic machines transferred an unknown number of votes for Sen. John F. Kerry (D-Mass.) to the Bush column.
In Columbus, Cincinnati and Toledo, and on college campuses, election officials allocated far too few voting machines to busy precincts, with the result that voters stood on line as long as 10 hours -- many leaving without voting. Some longtime voters discovered their registrations had been purged.
"There isn't enough to prove fraud, but there have been very significant problems in running elections in Ohio this year that demand reform," said Edward B. Foley, who is director of the election law program at the Ohio State University law school and a former Ohio state solicitor. "We clearly ended up disenfranchising people, and I don't want to minimize that."
Franklin County election officials -- evenly split between Republicans and Democrats -- say they allocated machines based on past voting patterns and their best estimate of where more were needed. But they acknowledge having too few machines to cope with an additional 102,000 registered voters.
Ohio is not particularly unusual. After the 2000 election debacle, which ended with a 36-day partisan standoff in Florida and an election decided by the U.S. Supreme Court, Congress passed the Help America Vote Act in 2002. The intent was to help states upgrade aging voting machines and ensure that eligible voters are not turned away. To a point, it has had the desired effect.
"Viewed dispassionately, the national elections ran much more smoothly than in 2000," said Charles Stewart III, a professor at the Massachusetts Institute of Technology and a specialist in voting behavior and methodology. Because of improved technology "nationwide, we counted perhaps 1 million votes that we would have lost four years ago."
But much work remains. Congress imposed only the minimal national standards and included too few dollars. Tens of thousands of machines -- including 70 percent of Ohio's machines -- still use punch-card ballots, which have a high error rate. A patchwork quilt of state rules governs voter registration and provisional ballots. (Provisional ballots are given to voters whose names do not appear on registration rolls -- studies show that minorities and poor voters cast a disproportionate number of such ballots.) Ohio recorded 153,000 provisional ballots. But in Georgia, one-third of the election districts did not record a single provisional ballot in 2004.
In Florida, ground zero for 2000's election meltdown, professors and graduate students from the University of California at Berkeley studied this year's voting results, contrasting counties that had electronic voting machines with those that used traditional voting methods. They concluded, based on voting and population trends and other indicators, that irregularities associated with machines in three traditionally Democratic counties in southern Florida may have delivered at least 130,000 excess votes for Bush in a state the president won by about 381,000 votes. The study prompted heated critiques from some polling experts.