A psychologist's experiment raises questions about the accuracy of eyewitnesses' identification of suspects, which are used more than 75,000 times a year in the United States to charge people with crimes and prosecute cases.

The experiment showed that people who identify a suspect from a police lineup or a group of photos are far more confident of their choice if given positive feedback, even in casual conversation.They become less sure of the identification - and likely to be less-confident witnesses at trial - if given negative feedback or none at all.

In research being published by the American Psychological Association, 352 people were shown a grainy videotape made by a surveillance camera of a person who later shot and killed a store security guard. They were told the person they saw had shot a security guard, and were shown either a lineup or photographs and asked to identify the killer. In fact, he was in neither the lineup nor the photos.

All made a choice - wrong, of course, since they had not seen the gunman.

Those who were told, "Good, you identified the actual suspect," were far more confident in their choices than those told the suspect actually was among the other people shown. Those given positive feedback also were more confident than those who were told nothing.