The data that the Office of Personnel Management are using to justify a federal pay cut and pay freeze has been questioned by the government's own numbers-collecting agency, the Bureau of Labor Statistics.

In a memo to Commissioner Janet Norwood, the Bureau of Labor Statistics (BLS) staff said it didn't know how OPM reached its conclusions that U.S. workers are overpaid and said OPM appears to be comparing statistical apples and oranges to make its points.

Patrick Korten, OPM's executive assistant director, said his agency sticks by its figures and conclusions. "The data is there. We took the most conservative figures on private-sector quit rates , and . . . the real bottom line is that the private-sector quit rate is much higher than the government's." He said the bureau's memo questioning OPM's data is "statisticians who are nit-picking . . . if something doesn't match up at the fifth decimal point, they say it isn't right. But our data is right."

Last month OPM Director Donald Devine approved a report that said the federal pay system must be overhauled because it has produced inflated salaries for most workers. It said that the president's plan to cut federal pay 5 percent is justified and if Congress rejects the pay cut, the only way to make similar dollar savings would be to fire 125,000.

OPM said the surest way to measure federal vs. private pay is to compare voluntary quit rates. It said the government quit rate is 3.8 percent, compared with voluntary turnover in industry of about 13 percent. The Bureau of Labor Statistics memo, however, said the staff questioned the wisdom of basing federal pay policy on a comparison of its quit rate with that of industry. Because of budget cuts made by the Reagan administration in 1981-82, the bureau no longer monitors the quit rate for manufacturing industries.

On the day OPM issued its pay study report, top Labor Department officials asked the bureau to prepare a paper on the quit rate. That paper is the memo to Norwood, prepared by the Division of Monthly Industry Employment Statistics. It says in part:

"In making its case that quit rates for the federal government were low relative to the private sector, OPM cited BLS manufacturing rates of 14.4 percent and 13.2 percent for 1980-81. . . . We do not know how OPM derived these data. The BLS survey was terminated with the release of December 1981 data because of budget cuts. However, prior to termination of the survey, monthly data on quits were published for manufacturing each month as a rate per 100 employes." It said the last month showed a quit rate in industry of 0.7, and "the annual average of the monthly rates for 1981 was 1.3."

Bureau officials said they thought OPM took a portion of its survey -- made monthly to monitor the economy as reflected by turnover in manufacturing industries -- and "annualized" it by multipling a monthly number by 12. They said turnover in manufacturing industries (about 20 percent of the labor force) is always higher than in nonmanufacturing (about 80 percent of the work force).

"Regardless of how OPM computed the quit rates in its report," the bureau report said, "we feel that there are serious definitional problems in comparing a quit rate for the federal government to a quit rate in manufacturing" because the "occupational mix" of the two sectors is vastly different.

A bureau official said that the turnover rate OPM cited for larger nonmanufacturing industries of between 18 percent and 19.2 percent included separations for all reasons, including voluntary quits, retirement, firings and deaths, more extensive criteria than that used for figures on government workers.