The administration of Illinois Gov. James R. Thompson (R) altered key unemployment statistics last year in the midst of Thompson's close reelection campaign in an effort to prevent loss of extended federal benefits to 50,000 jobless workers in the state, internal government documents show.

The alterations, which occurred at a time when unemployment was a major campaign issue, were made several days after Thompson talked by telephone with Labor Secretary Raymond J. Donovan and requested a delay in submitting the statistics to the federal government.

Donovan granted the delay, but Labor Department officials later questioned the accuracy of the data submitted by Thompson's administration, according to documents obtained by The Washington Post under the Freedom of Information Act.

One Labor Department official said there was a pattern in the statistics that had "never occurred in the history of the UI unemployment insurance program." Another official said Illinois "appears politically arrogant . . . apparently feeling a Republican administration will not cause problems for a Republican governor in a tough reelection race in Illinois November 2."

The alterations in the statistics triggered a criminal investigation by the Labor Department's inspector general. The investigation found evidence that the statistics had been altered improperly, but no evidence linking the alterations to Thompson or Donovan. The probe was then referred to the Justice Department's fraud section, which declined to take further action, officials said.

Neither the governor nor labor secretary was questioned in the probe, officials said, nor was Albert Angrisani, assistant secretary for employment and training, who oversees the program in question.

Thompson said this week through his spokesman, David Gilbert, that he was not aware of any improper alterations in the statistics. Gilbert said the governor's telephone call to Donovan was made strictly to obtain a week's delay in submitting the statistics. Gilbert added that "at this point no one in the governor's office has any information" that statistics were improperly altered.

"If there was any procedure that was taken that was not appropriate, we'll have to take a look at that to see what can be done, who was at fault," said Gilbert. "We'll get to the bottom of it."

Donovan refused to respond directly to queries about the case. Through his spokesman, Mike Volpe, he said he discussed the extension with Thompson, who expressed concern that "a lot of unemployed people would lose their benefits." Volpe said Donovan turned the matter over to his staff, and the delay was granted a few days later. Volpe said Donovan recalls nothing more about the telephone call.

After questions were raised last year about the statistics, the Labor Department sent auditing teams to Illinois. The state was eventually permitted to recount its unemployment claims under federal auditors' supervision. In the recount, the auditors allowed the state to include some claims that had previously been left out. Top department officials then accepted the recalculated figures, so benefits were not cut off.

But "no one paid the piper" for having made the initial alterations in the statistics, said one Labor Department source familiar with the case.

At the time, Thompson was in a hard-fought reelection campaign against Democrat Adlai E. Stevenson III. Thompson won by 5,074 votes out of 3.67 million cast--a 0.14 percent margin. The race was considered the closest in Illinois history.

The Post obtained, under the Freedom of Information Act, copies of interviews and other internal documents produced in the inspector general's investigation. However, some documents were withheld, and others the Labor Department heavily censored.

The documents describe events that began in mid-1982 when the recession was lingering and causing major political problems for Republican candidates. At the time, Thompson was the only GOP governor in the Great Lakes region seeking reelection, and Stevenson was making the economy and jobs a major issue.

Unemployment in Illinois had been rising all year, from 9.2 percent in January to 12.3 percent in October, the month before the election.

Even as the jobless situation was growing more serious, however, the state was in danger of losing its "extended" jobless benefits, which provided 13 extra weeks of benefits in periods of high joblessness.

These benefits were paid after workers had exhausted the regular 26 weeks of unemployment benefits. According to the documents, Illinois was receiving a total of $3.5 million a week in federal extended benefits.

The extended benefits were based on a complex state-by-state eligibility formula. The program was in jeopardy in Illinois because the formula was tightened in one of the federal budget cuts Reagan pushed through Congress in 1981.

If Illinois lost its eligibility, 50,000 jobless workers would lose their benefits and the payments could not be resumed for 13 weeks. The state was thus faced with the paradox of rising unemployment but the possibility of less federal aid to cope with it.

To continue qualifying for the benefits, the state submitted weekly reports to the Labor Department. These reports calculated a mathematical "trigger" that, if hit, would end the program in Illinois.

The trigger worked this way: the benefits would continue as long as 5 percent or more of all workers covered by unemployment insurance were receiving regular unemployment benefits. If the figure fell below 5 percent, the state would "trigger off" the extended benefits program.

The actual unemployment rate was higher than the 5 percent trigger level because some workers had exhausted their jobless benefits altogether, while some others were not covered by unemployment insurance.

The documents obtained by The Post, as well as interviews with some key officials, detail how the statistics were altered during several weeks in late July and early August, 1982.

In a staff memorandum on July 19, 1982, Thompson was warned that the extended benefits were "likely to trigger off by the end of this week." The memo added that if this happened, Aug. 7 would thus be the last week benefits could be paid and the program could not begin again until mid-November.

Two days after that memo was written, a similar conclusion was reached by officials in the Illinois Bureau of Employment Security, the state agency that administered the unemployment insurance benefits.

In a memo analyzing trends in the jobless data, the Illinois officials wrote July 21 that "it is not very likely" the state would remain on the extended-benefits program. Already, for the week ending July 17, Illinois had hit the 5 percent trigger rate, the memo said. Given the downward trend in the data, ". . . We can reasonably expect the EB program to trigger OFF" the next week, they said.

It is not possible to determine who wrote the memo since Labor Department officials deleted all names in releasing the documents.

The next week, Thompson's staff urgently attempted to arrange a telephone call to Labor Secretary Donovan. A July 27 memo prepared for Donovan by his staff said that Illinois was about to trigger off extended benefits. However, the governor "believes these figures are incorrect" and wanted a week's grace period in submitting the data, according to the memo.

Donovan was told by his staff that the "only solution you can offer" was to grant a week's delay. The delay was granted.

The same day, Thompson held a news conference in Chicago. According to an account in the July 28 Chicago Tribune, Thompson appealed to Illinois jobless workers who had not applied for benefits to file claims so the state would not "trigger off" extended benefits.

Thompson also said the state would use the week-long grace period to "recheck figures" that might allow Illinois to remain on extended benefits.

Labor Department investigators later interviewed officials in the Illinois Bureau of Employment Security, which prepares the data, to determine what happened in the days before and after Thompson's call.

The reports of these interviews were made available to The Post. All names and titles, as well as other information, were deleted, making it impossible to determine who was responsible for the alterations.

One official told the investigators that on July 26 a tentative computer run had shown the trigger rate was 4.997 percent. Since the Labor Department in the past had not permitted rounding off such a figure to 5 percent, Illinois would be forced off the extended benefits. Maryland had just lost extended benefits under similar circumstances.

The Illinois official said the 4.997 percent figure was immediately reported to his supervisors. But he said he was told not to send it to the Labor Department "until further notice."

The following week--after Thompson's call to Donovan--the official said he was ordered to prepare the report by adding into the statistics some 1,500 unemployment claims from an earlier week. This would be just enough to boost the trigger rate to exactly 5 percent, the official said.

The official told investigators that he "questioned the propriety of this adjustment" but was told it was an attempt to compensate for claims that hadn't been counted earlier.

That report was submitted Aug. 2, showing a 5 percent trigger rate. It was the second time in a row Illinois hit 5 percent exactly.

The Illinois official said he used a similar method for preparing the next week's report, this time adding 4,754 extra claims, "just enough to reach the 5 percent trigger rate."

This was the third consecutive week the state hit precisely the 5 percent rate.

For the next week, the official said the report was not altered but the 4,754 claims that had been "borrowed" the previous week were not subtracted as they should have been because of a "human error."

For the fourth time, Illinois reported a trigger rate of 5 percent.

Another Illinois official whose name was also deleted in the documents is quoted as having "readily acknowledged" to the investigators that "had the claim reporting system not been altered they would have triggered off and would have lost the extended benefit eligibility for 13 weeks."

This would have meant that workers in Illinois would be without extended jobless benefits until after Election Day.

Several of the officials interviewed by the Labor Department investigators denied that Thompson's staff had ordered the changes, according to the documents.

However, one of these officials, under investigators' questioning, "conceded the political realities of a large number of individuals going off UI compensation in an election year."

The unemployment compensation program is run by the Employment and Training Administration, while the Bureau of Labor Statistics, a separate Labor Department unit, compiles and publishes federal jobless statistics.

According to the documents, one Illinois official told the investigators that the Employment and Training Administration had given the state data a "clean bill of health" in early August. The official did not say who provided it, however.

Angrisani, the assistant secretary for employment and training, did not return a reporter's telephone queries last week about the Illinois case. A spokesman for him issued a statement saying Angrisani had "simply followed" the recommendations of career professionals under him.

The Illinois data was immediately questioned by two career officials, William (Bert) Lewis, administrator of the Office of Employment Security, and Carolyn Golding, administrator of Unemployment Insurance, each of them said in an interview last week.

Lewis said the trigger statistics are "volatile" and thus unlikely to come in at the same figure two or three weeks in a row. Illinois had reported hitting the 5 percent trigger precisely in four consecutive weeks.

"It didn't take us four times" to question the data, he said. Lewis said he had never seen a comparable situation.

Another official told the Labor Department investigators that Illinois might legitimately have reported the 5 percent level the first time. But, the official said, when the state reported 5 percent the second time--in the delayed report with the 1,500 additional claims--"knowledgeable UI program people could not believe it."

Still another Labor Department official, identified in the documents only as a specialist on unemployment insurance, said his "immediate reaction and conclusion were that the figures were bad or fixed." This official said Illinois "appears politically arrogant . . . apparently feeling a Republican administration will not cause problems for a Republican governor in a tough reelection race in Illinois November 2."

In Illinois at the time, the explanation given publicly for the continuation of extended benefits was that enough claims had been found in a "recount" to prevent the state from "triggering off" the program.

Agaliece Miller, director of the Illinois Bureau of Employment Security, told the Chicago Sun-Times Aug. 2 that "her staff in 60 field offices had searched their mail bins and other places and had found about 2,000 additional unemployment applications." The newspaper quoted her as saying, "We squeaked through." Miller, who has retired, could not be reached for comment last week.

But the Labor Department career officials had doubts about how the state had "squeaked through."

On Aug. 12, Lewis filed an "incident report" with the inspector general's office, triggering an investigation. The report noted that Illinois officials had expected disqualification for extended benefits, but it did not occur. The report suggested that an "abnormal pattern has developed."

Lewis said last week that Illinois had, in effect, double-counted the unemployment claims for two weeks, and in the third week failed to adjust for this double-counting.

The investigation in August and September included sending a team of auditors to Illinois. On Sept. 14, the state was requested to recalculate the statistics for the weeks in question under the direct supervision of the Labor Department auditors, and in the recount the auditors allowed the state to include some unemployment claims that had previously not been included in the statistics.

Lewis said the recount was unusual in that the department had never before sent auditors and staff to oversee a state's trigger calculations.

A few days later, in a letter to Golding, the Illinois officials promised not to alter the statistics as they had done before. The recalculations showed that Illinois still qualified for extended benefits. The result was that in late September, Labor Department officials decided to approve the data.

Later, the inspector general's office forwarded the results of this investigation to the Justice Department. Officials there declined to carry it any further, according to a Justice Department spokesman who could not explain why.

"You can look at it two ways," said a Labor Department official. "The system worked" in that the altered statistics were detected. But, he said, "no one paid the piper" for making the alterations in the midst of Thompson's reelection campaign.