Via Radley Balko, Brandon Garrett, author of Convicting the Innocent, looks at wrongful identification by eyewitnesses in the context of wrongful convictions:
The Cotton case fits a larger and troubling pattern among all the DNA cases I studied. In 57 percent of the DNA exoneration trials that included eyewitness testimony, the eyewitnesses had earlier been uncertain, a glaring sign that something was wrong right off the bat. Eyewitnesses identified fillers or other suspects, or no one at all, or they described being unsure. However, almost without exception, those eyewitnesses had become absolutely certain of what they had seen by the time of trial. One eyewitness said "This is the man or it is his twin brother." Another was "one hundred and twenty" percent sure.
Where did this false certainty come from? The trial records I looked at suggest that unsound and suggestive police identification procedures played a large and troubling role. Police used unnecessary show-ups, where they presented the eyewitness with just the defendant. Or stacked lineups to make the defendant stand out. Or offered suggestive remarks, telling the eyewitness whom to identify or to expect a suspect in a lineup. Or confirmed the witness's choice as the right one. Even well-intended, encouraging remarks, like "good job, you picked the guy," can have a dramatic effect on eyewitness memory, as psychologists have shown. Indeed, more than one-third of the cases I looked at involved multiple eyewitnesses, as many as three or four or five eyewitnesses who all somehow misidentified the same innocent person. Further, almost half of the eyewitness identifications were cross-racial. Psychologists have long shown how eyewitnesses have greater difficulty identifying persons of another race.
People cite My Cousin Vinny ad nauseum, but one of the things it got absolutely right is the way eyewitnesses work in court. Remember when Joe Pesci cross examines the elderly woman, then proves she lacked the clarity of eyesight to actually identify the defendants? Eyewitnesses are problematic because they can be completely wrong and totally convincing at the same time, precisely because they sound so certain. The Innocence Project found that eyewitness misidentification plays a role in more that three-quarters of exonerations through DNA testing. Cross-racial eyewitness identification is particularly unreliable, which is depressing but not exactly surprising.