When math professor Earl Whitehead was denied a promotion at the University of Pittsburgh nearly four years ago, he asked to review his personnel file to figure out what had happened.

The school agreed to provide him part of the file, but it refused to make available letters of reference which had been requested of colleagues outside of Pitt, saying they were exempt from review and should be to protect the integrity of the letters.

"Without confidentiality, the author of a reference letter may be influenced by the candidate's possible future retaliation, or, alternatively, special treatment, in response to a reference letter," Pitt attorneys wrote in a court brief.

Mr. Whitehead appealed the university's decision to the state Department of Labor and Industry, saying he had a right to know what was said about him so that he could respond to it, said John Lucas, Mr. Whitehead's attorney.

"His expertise is a very specialized part of mathematics, and there aren't very many people in the field that could evaluate him," Mr. Lucas said.

Mr. Whitehead wanted to know if the people Pitt chose to review his work were qualified to do so, and the Department of Labor, which sided with him, agreed that he should.

But Pitt appealed that decision and won in a recent opinion by Commonwealth Court.

Mr. Whitehead is expected to file a petition with the state Supreme Court, asking for permission to appeal, Mr. Lucas said.

In the Commonwealth Court decision, the judges cited the Personnel Files Act of 1978, a state statute which dictates what is or is not open in a personnel file. Amended in 1990, the statute excludes the review of letters of reference. But it allow for inspection of performance evaluations.

Mr. Whitehead, who is still an associate professor at Pitt, based his case on the fact that the statute does not define either term. He said the letters of reference written about him by outside colleagues were evaluations.

And even though Pitt officials used the term "evaluation" in their requests for the letters, the court disagreed with Mr. Whitehead's interpretation.

This idea of confidentiality in personnel files raises questions about accountability versus candor, experts say, which can be difficult to balance.

Robert Hill, a spokesman for Pitt, refused to comment on the issue other than to say he was pleased with the court's ruling in the case. He would not discuss why letters of reference should be kept confidential or what harm could be caused if they were not.

But Irene Frieze, president of the Pitt faculty and a professor of psychology, agreed with the court's decision that letters of reference must be kept confidential.

"The problem is, if you want frank evaluations, you can't let people see them," Ms. Frieze said.

The purpose of the outside letters, she said, is to get a sense of the candidate's reputation in other parts of the country and world. The writers of the letters are asked to evaluate a candidate's reputation in the field, his or her body of research and stature as a scholar.

"They're not going to give any kind of honest letter if they think their letters are going to be read," she said.

But the American Association of University Professors, a nonprofit organization with 45,000 members nationally, believes that all records in a personnel file should be open to review, including reference letters.

"Openness works like the proverbial sunshine," said Jonathan Knight, the director of the program in academic freedom and tenure.

He believes having open files holds people accountable for their opinions and helps eliminate the potential for discrimination.

In 1999, AAUP released a report in which the organization formally endorsed an open review system.

"It was, at the time, and is still today, a tough issue," Mr. Knight said. "We took a position that was not in keeping with policy of many institutions."

Academics are, or, at least, should be, used to receiving constructive criticism, he said, and letters regarding promotion or tenure are a part of that.

"The knowledge that one's assessment will be shielded from the scrutiny of the faculty member being discussed could encourage careless and unsubstantiated commentary," the report said.

Some have suggested a compromise in which negative letters are summarized, or the names of the person writing them are redacted, but that doesn't foster openness and accountability, Mr. Knight said.

It's important for a faculty member to know what was said for several reasons, he said, including to form the basis for any appeal of the decision.

At the University of Rhode Island, faculty members have complete access to their personnel files, including outside letters of reference, as part of their collective bargaining agreement.

The person writing those letters, though, is warned of that in the beginning, said Louis Kirschenbaum, a chemistry professor there and past president of the local AAUP chapter.

The warning comes in the request letter, which states: "I should mention that, although our committee will hold your letter in confidence, URI faculty members are entitled to consult their own personnel files."

Though it's rare, Mr. Kirschenbaum acknowledges, some colleagues, when they learn that the candidate may see their letters, refuse to write them.

But, he said, having an open policy all around -- even department votes are recorded by name -- promotes fairness and prevents back-stabbing.

"We expect people to be honest," he said. "People really should be able to stand up for their opinions."

First Published: April 24, 2006, 4:00 a.m.