NASA's overhaul of the way it manages risks connected with flying the space shuttle is "going in the right direction" but needs more sophisticated methods for assessing the dangers, an independent panel reported yesterday.

Before the shuttle was first launched in 1981, the agency had developed complex procedures, generating mountains of paper, to figure out each thing that could go wrong on the enormously complex vehicle and decide how to prevent or minimize that risk. In the wake of the Jan. 28, 1986, Challenger disaster that killed seven astronauts, the agency is changing those procedures, automating the data, reviewing every potential hazard and reclassifying many.

However, the panel said, NASA still needs to hire more experts in the use of modern statistical analysis, essential for such a complex system, and more staff for its new safety office, which should be given more weight in decisions about flight hazards.

George Rodney, safety office chief for the National Aeronautics and Space Administration, said yesterday that the agency and contractor safety work force has more than doubled in the last year and is still growing.

The agency also has set up numerous post-accident advisory boards and panels to make sure all problems are communicated to decision-makers. However, the panel expressed concern that the notion of individual responsibility often gets lost.

"When we ask a NASA manager about how a decision is made, often we are told that it is made by such-and-such a board," said the eight-page interim report by the National Research Council panel, signed by chairman Alton D. Slay, a former Air Force general and now an aerospace consultant.

"There may be a tendency for those involved in the multilayered review and decision process to hide in the anonymity of panels and boards . . . . The committee recommends that {NASA Administrator James C. Fletcher} periodically remind all of the NASA organization of the specific individuals by name and position who are responsible for final decisions."

Before Challenger, 748 items on the shuttle were classified as "Criticality 1" on NASA's critical items list. This meant that if any one of them failed, the result could be loss of vehicle and crew. There are some 2,000 critical items under categories of declining risk.

Someone at a high level must sign a waiver for each high-risk item, approving it for flight based on a specific rationale.

"It is encouraging," the panel said, that failures of Criticality 1 items have occurred during flight which did not result in the worst-case effect -- loss of crew and vehicle.

In the past, NASA's procedures have provided no way of evaluating how likely it is that the worst-case effect is likely to result from a given failure, said David S. Johnson, the panel's project director, even though some risks are much higher than others. The panel, in this report as well as the last, encourages NASA to "assign priorities to critical items" taking into account such probabilities. Though NASA's top officials have indicated a "desire to defer" doing so, the panel urges they find a way to concentrate only on the highest risk items before the next flight and delegate the rest to subordinates.