WASHINGTON, DC - Voter identification requirements designed to combat fraud can reduce turnout, particularly among minorities, a new study shows.

A study by the Eagleton Institute of Politics at Rutgers University shows turnout in 2004 was about 4% lower in states that required voters to sign their name or produce documentation. Hispanic turnout was 10% lower; the difference was about 6% for blacks and Asian-Americans.

The study, presented to the U.S. Election Assistance Commission this month, comes as voter identification requirements are being subjected to increased scrutiny. Two researchers last year told the Election Assistance Commission that they found little evidence of voter fraud at polling places.

Thomas O'Neill, who led the Eagleton research team, said the new study shows that "the harm may be as great as the benefit."

The commission's chairwoman, Donetta Davidson, called the study "premature." More work needs to be done, she said. "You can't make determinations based on one year," she said. "We have new states that have ID requirements now that weren't in that review."

In 2005, Indiana enacted the nation's stiffest photo ID law. Yet turnout increased 2% in November compared with 2002, the previous non-presidential federal election year.

In local races before the November elections, "we could not find one instance where a legitimate voter could not vote," said Indiana Secretary of State Todd Rokita, a Republican.

Under federal law, first-time voters must verify their identity. Nearly half the states go further, requiring some identification from all voters - ranging from a signature to a government-issued photo ID, which is required only in Indiana.

Photo ID laws in Georgia and Missouri have been blocked in court. A photo ID law in Florida allows voters to choose from many forms of identification.

Liberal groups such as Common Cause have fought the measures on claims that they disenfranchise the poor, minorities and seniors.

Several states are considering photo ID laws this year, including California, Colorado, Kansas, Mississippi, Oklahoma, Texas and Washington. Most of the efforts face substantial opposition. "We've been beating that horse for two or three years now," said Tim Storey, an elections expert at the National Conference of State Legislatures. "The states that were going to do it have mostly done it."

The Eagleton Institute research relied on data from 2004, before some of the toughest ID laws were passed. It shows that as requirements stiffen - from stating a name to signing it to presenting documentation - turnout drops slightly. The researchers considered other factors, including contested elections, registration requirements, race, ethnicity, age, education and income. They used a 2004 Census Bureau survey to determine racial and ethnic breakdowns.

"It validates some of the things that have been said all along about the problems of voter ID," said Kimball Brace of Election Data Services. Taken together with the earlier research by Tova Wang and Job Serebrov that found little evidence of voter fraud at polling places, it shows voter ID laws can have more of a negative than a positive impact, Brace said.

Advocates of tougher ID requirements cite Indiana. The state did extensive voter education before the election and allowed voters without ID to cast provisional ballots, which are less likely to be counted.

"It wasn't the horrible thing that some feared," said Thor Hearne of the conservative American Center for Voting Rights. Hearne predicted continued support for state ID laws if they are carefully drafted and implemented to avoid a drop in turnout.