A Howard University criminologist testified yesterday that the woman who identified Marine Cpl. Lindsey Scott as the man who raped and stabbed her in 1983 did so only after Navy investigators, mistakenly or intentionally, focused her attention on Scott.

However, the testimony came in a procedural hearing with the jury out of the courtroom, and the military judge presiding over Scott's court-martial ruled the testimony irrelevant and inadmissible.

Lt. Col. Eligah D. Clark's ruling was a setback for Scott, whose attorneys have tried to portray Naval Investigative Service agents as bunglers who singled out Scott as a suspect within hours of the attack on the woman at the Quantico Marine Corps base and then fashioned their investigation so it would incriminate him.

Clark sided with prosecutors who argued that Scott, not the NIS, is on trial. He said that if Scott's lawyers wanted to raise questions about the investigation, they could do so in their closing arguments to the jury.

The NIS' "procedural correctness is not an issue" for the jury to decide, Clark said.

Gwynne W. Peirson, director of Howard's Department of Criminology, was the first witness called by the defense during the court-martial, Scott's second since 1983. In the first proceeding, Scott was convicted and sentenced to 30 years at hard labor. That verdict was set aside in July by a military appeals court based on the failure of Scott's previous attorney to prepare a competent defense. The marines ordered a new trial in October.

Peirson criticized several aspects of the NIS investigation, particularly those surrounding the woman's identification of Scott.

He noted that a group of six photographs shown to the woman the day after the attack included four men who had either wire-rim glasses, facial hair or Afros. Only two of the photos met the woman's description of her assailant as a clean-shaven man with close-cropped hair and horn-rim glasses, and one was Scott.

Therefore, Peirson said, the field was immediately narrowed to the two photos; the woman selected both, but said Scott looked more like her assailant.

"It does not appear that it was an impartial, objective" procedure, the criminologist said.

Four days later, the woman was brought before a lineup of six men. Scott was the only man in the lineup who also was in the group of photos. Again, the woman chose two men, including Scott, but she said Scott "scares me the most."

The danger, said Peirson, was that the woman's selection was based not on her memory of her attacker, but on the face she remembered from the photographs.

Peirson also amplified a persistent defense theme: that the woman may have been coached to identify Scott's car, where the rape allegedly occurred, by her husband, a military policeman who worked with Scott and may have known what his car looked like and that Scott was a suspect.

The NIS allowed the woman's husband to accompany her when she picked Scott's car out of a parking lot, a procedure that Peirson scored as "incompetent or with the intent to focus the identification" on Scott.

"There's no valid reason for having the husband there," he said. "It can only injure the identification." He added that the NIS' techniques tended to weaken the woman's identification of Scott, a central element in the government's case.

Marine prosecutors produced a rebuttal witness, Johns Hopkins University psychology professor Michael E. McCloskey, but his testimony seemed to buttress Peirson's when he acknowledged under cross-examination the NIS' procedures raised "the possibility of bias."