The trouble with eyewitness testimony

More than 1,400 people convicted of serious crimes have been proved innocent over the past 25 years, according to the University of Michigan Law School’s National Registry of Exonerations. But why were these people wrongly convicted? In a great many cases, one significant factor was faulty eyewitness identifications.

Eyewitness testimony can be extremely powerful. When a witness with no motive to lie swears under oath that he or she personally saw a defendant commit a crime, it is hard not to believe the testimony. But in recent decades, extensive scientific research—which we reviewed while co-chairing the National Research Council committee that wrote the recent report “Identifying the Culprit: Assessing Eyewitness Identification”—have identified a number of factors that can lead an eyewitness to make a mistake.

When judges and juries lack awareness of the factors that can affect the accuracy of eyewitness testimony, they can make mistakes, too.

In 1967, for example, before most of this research, the Supreme Court ruled, in Manson v. Braithwaite, that the reliability of an eyewitness’ identification, and hence its admissibility as evidence, depends in part on the level of certainty that the witness expresses when identifying a defendant in court.

But the court was wrong. Careful studies have demonstrated that, regardless of the level of certainty an eyewitness expresses at the time of original identification, a witness’ confidence in the correctness of the identification steadily increases over time. This occurs for psychological and cognitive reasons that have nothing to do with the accuracy of the identification, such as reinforcement of a witness’ beliefs by law enforcement and accounts of events promulgated by attorneys and news media.

If we want to avoid making the same tragic mistakes again and again, all of us—police, prosecutors, judges, jurors, public officeholders and the public—must act on this research. Here is how: First, lineups should be administered according to the same kind of “double-blind” standards used in scientific testing. For example, the police officer administering the lineup should not be involved in the investigation. An eyewitness should be told that the lineup may include a suspect and that the investigation will continue regardless of whether someone is identified.

Second, the level of confidence the eyewitness expresses at the time of the initial identification should be documented and made available at trial.

Third, judges should be educated about the well-documented failings of lineups.

Fourth, through expert testimony or judicial instructions, jurors should be educated about the many factors that can affect eyewitness identification and that those need to be considered in evaluating the reliability of an eyewitness’ testimony.

With these reforms, the number of mistaken eyewitness identifications would be greatly reduced, and judges and jurors would be able to make more accurate assessments of eyewitness testimony. This would go a long way toward ensuring that our criminal justice system gets to the truth.