An unprecedented national survey of pilots by the U.S. government has found that safety problems such as near collisions and runway interference occur far more frequently than previously recognized. But the government is withholding the information, fearful it would upset air travelers and hurt airline profits.

NASA gathered the information under an $8.5 million federal safety project, through telephone interviews with roughly 24,000 commercial- and general-aviation pilots over nearly four years. Since shutting down the project more than a year ago, the space agency has refused to divulge its survey data publicly.

After The Associated Press disclosed details Monday about the survey and efforts to keep its results secret, NASA's chief said he will reconsider how much of the survey findings can be made public.

"NASA should focus on how we can provide information to the public, not on how we can withhold it," NASA Administrator Michael Griffin said in a statement. He said the agency's research and data "should be widely available and subject to review and scrutiny."

Last week, NASA ordered the contractor that conducted the survey to purge all related data from its computers. Congress intervened Monday, saying it will launch a formal investigation and instruct NASA to keep all its data. Griffin said he already was ordering that all survey data be preserved.

The AP learned about the NASA results from a person familiar with the survey who spoke on condition of anonymity because he or she was not authorized to discuss them.

A senior NASA official, associate administrator Thomas S. Luedtke, said earlier that revealing the findings could damage the public's confidence in airlines and affect airline profits. Luedtke acknowledged that the survey results "present a comprehensive picture of certain aspects of the U.S. commercial-aviation industry."

The AP sought to obtain the survey data over 14 months under the U.S. Freedom of Information Act.

"Release of the requested data, which are sensitive and safety-related, could materially affect the public confidence in, and the commercial welfare of, the air carriers and general aviation companies whose pilots participated in the survey," Luedtke wrote in a final denial letter to the AP. NASA also cited pilot confidentiality as a reason, although no airlines were identified in the survey, nor were the identities of pilots, all of whom were promised anonymity.

Griffin said NASA will reconsider its denial.

Among other results, the pilots reported at least twice as many bird strikes, near midair collisions and runway incursions as other government monitoring systems show, according to a person familiar with the results who was not authorized to discuss them publicly.

The survey also revealed higher-than-expected numbers of pilots who experienced "in-close approach changes" -- potentially dangerous, last-minute instructions to alter landing plans.

Officials at the NASA Ames Research Center in California have said they want to publish their own report on the project by year's end.

NASA has a long and storied history of aviation-safety research. Its experts study atmospheric science and airplane materials and design, among other areas.

"If the airlines aren't safe, I want to know about it," said Rep. Brad Miller, D-N.C., chairman of the House Science and Technology investigations and oversight subcommittee. "I would rather not feel a false sense of security because they don't tell us."

Discussing NASA's decision not to release the survey data, Miller said: "There is a faint odor about it all."

Rep. Bart Gordon, D-Tenn., chairman of the House Science and Technology Committee, was expected to announce a formal investigation and will instruct NASA and its contractors not to destroy any data. Gordon said he was disturbed by the report that NASA told its contractor to delete the survey data.

"I cannot imagine any good public purpose being served by destroying records," Gordon said.

The survey's purpose was to develop a new way of tracking safety trends and problems that the airline industry could address. The project was shelved when NASA cut its budget as emphasis shifted to send astronauts to the moon and Mars.

Laura Brown, a spokeswoman for the Federal Aviation Administration, said the agency questioned NASA's methodology. The FAA is confident it can identify safety problems before they lead to accidents, she said.

But aviation experts said NASA's survey results could be a valuable resource in an industry where they think many safety problems are underreported, even while deaths from commercial air crashes are rare and the number of deadly crashes has dropped in recent years.

"It gives us an awareness of not just the extent of the problems but probably in some cases that the problems are there at all," said William Waldock, a safety science professor at Embry-Riddle Aeronautical University in Prescott, Ariz.

Officials involved in the survey touted the unusually high response rate among pilots, 80 percent, and said they think it is more reliable than reporting systems that rely on pilots to report incidents voluntarily.

"The data is strong," said Robert Dodd, an aviation-safety expert hired by NASA to manage the survey. "Our process was very meticulously designed and very thorough. It was very scientific."

Pilot interviews lasted about 30 minutes, with standardized questions about how frequently they encountered equipment problems, smoke or fire, engine failure, passenger disturbances, severe turbulence, collisions with birds or inadequate tower communication, according to documents obtained by the AP.

"I don't believe it's in NASA's purpose and mission statement to protect the underlying financial fortunes of the airlines," David Stempler, president of the Potomac, Md.-based Air Travelers Association, said Monday. "They're to provide safety information, and the consequences will fall where they may. We still believe this is an extremely safe air-travel system, but it could be made even safer."