https://www.sciencedaily.com/releases/2023/02/230201084403.htm

During a typical eyewitness lineup, police display a "six pack" of photos. One is the police suspect. The other five are "fillers;" they fit the description of the culprit, but the investigators know they are innocent. The witness selects the person who best matches their memory or rejects the whole lineup if they don't think the culprit is present. Investigators following best practices then ask the witness to rate how confident they are in their decision.

Previous studies have found a high confidence rating from eyewitnesses (i.e., 90% or more) implies greater accuracy. But when an eyewitness picks a filler or rejects the whole lineup, which happens in an estimated 24% and 35% of lineups, respectively, investigators miss out on potentially valuable evidence.