The Army today said it had no scientific evidence to back up its claim that many women have been leaving the armed services because they were assigned to tasks that were too physically difficult for them to perform.

That acknowledgment came at the semi-annual meeting of the Defense Advisory Committee on Women in the Services (DACOWITS), whose members include the nation's highest-ranking women soldiers. It fueled speculation among the 200 women present that the Pentagon and Reagan administration are trying to reduce the number of women in the Army and force them into more traditional military jobs.

In August, the Army said it was lowering its recruitment goals for women based on a 15-month study, initiated shortly after Reagan took office, on what role women should play in the Army.

At that time, Lt. Gen. Maxwell R. Thurman, deputy chief of Army personnel, said the study showed women soldiers had a much higher attrition rate than male soldiers because women soldiers often were assigned tasks beyond their physical capabilities.

By requiring all recruits to take a physical test when they entered the Army, he said it would now be able to reduce attrition and make sure soldiers were assigned jobs they could perform.

Later, the Army acknowledged that the physical tests, in effect, will bar most women from 76 percent of Army jobs and limit most female recruits to jobs that traditionally have been held by women.

During often critical questioning from DACOWITS members today, Thurman and William D. Clark, deputy assistant secretary of the Army for manpower and reserve affairs, acknowledged that the study's conclusions about female attrition were based, as Clark put it, on "reasonable judgments" by experienced commanders. He said no scientific studies were done, although the Army did refer to internal reports and a 1976 General Accounting Office study that concluded that the Army needed to improve its system of assigning women soldiers.

"Your study is nothing but a snow job," charged Dr. Margaret Scheffelin, an education researcher for the California State Department of Education and a DACOWITS member. Based on the data described at the briefing, Scheffelin said, the Army appeared to have used a very small sample to make a sweeping judgment -- a charge that the study's director, Col. Charles A. Hines, said was inaccurate.

The Army had promised to make the study available to the public in September, but that deadline, like two others, passed without any document being released. More recently, the Army said it should be available in mid-November but several DACOWITS members said they doubted a thorough report will be released.

"The Army developed its conclusions and then began looking for a rationale to support them," charged an Air Force officer who asked not to be identified but whose comments appeared to summarize the feelings of many present.

Retired Air Force Maj. Gen. Jeanne Holm, who was one of the highest-ranking women in the armed services until she retired, told Thurman that the study had created major morale problems among female troops. She said female soldiers were afraid the military planned to use the study to blame them for its attrition problems. Thurman acknowledged that the Army had done a poor job in presenting the results of the study but he called it "brilliant work."

The harshest criticism of the day came from journalist Sarah McClendon, a former DACOWITS member who served on a committee that was supposed to advise the study team. She said the study's planners never consulted the group and only briefed them after considerable badgering. "We wondered why the hell they even wanted us to help since they never asked for our opinion," she said.