“It was probably 8.5 percent if you count the undervotes [ballots with no candidate selected], because I count the undervotes as a perfectly valid choice,” Carlberg said.

Carlberg said between 6,000 and 7,000 voters who cast their ballots in Duval County in 2000 did not select a candidate for president. He also said that there were an “unusually high number” of overvotes, or ballots with more than one candidate selected for a single office.

Carlberg said these overvotes and undervotes may have contributed to Duval County’s ranking. He declined to comment on the report’s finding that race affected ballot spoilage.

John Sullivan, an official for the Registration and Elections Department in Fulton County, Georgia—which was ranked second worst by the report—echoed Carlberg’s criticism.

“I would assume that the report is very incomplete because there are a lot of counties that are a lot worse than us,” said Sullivan, referring to a state report on the November 2000 election, which ranked Fulton County more favorably.

Sullivan said the policy of reporting overvotes and undervotes in Fulton County, which has a black population of 44.6 percent, might have disadvantaged the county in the report.

“We have always reported overvotes and undervotes. Most counties do not,” he said.

He also declined to comment on the effect of race, though he said the most important factors in improving results were voter education and voting technology.

While Ancheta acknowledged the merit of both Carlberg and Sullivan’s claims, he said that different methodologies lead to different definitions of spoiled ballots.

“There is always the possibility that [an undervote] was a perfectly legitimate ballot,” Ancheta said. “That is a legitimate concern to be raised.”

He added that the report’s methodology may have penalized some counties, but he maintained that differences over the definition of a spoiled ballot did not fundamentally alter the survey’s validity.

“It doesn’t negate the fact that the percentages are much higher than one would like to see,” he said.

Carlberg also said Duval County voters will cast their vote this November not by punch card, but by manually selecting candidates using a pen. The ballots will then be tabulated using an optical scan system similar to Scantron forms used in exams.

“We have a whole new voting system,” he said, adding the change was mandated by state law.

He said they chose the manual system over a touch screen ballot because it was more voter friendly and economically feasible.

Sullivan also said that in the wake of the November 2000 elections Georgia has gone to touch screen voting machines.

Ancheta said that achieving this type of reform—focusing on solving ballot spoilage issues at the state and local level—was one of the ultimate goals of the report.

The report is by Professor of Law Christopher F. Edley Jr., co-director of the Civil Rights Project, Professor Philip Klinkner of Hamilton College and research assistants Jocelyn Benson and Vesla Weaver.

An updated report that includes data from the 2002 elections will be released this winter.

—Staff writer Christopher M. Loomis can be reached at cloomis@fas.harvard.edu.