When “I don’t know” improves the accuracy of eye-witness identification
The inaccuracy of eye-witness identification is well documented. Eye witnesses make mistakes and those mistakes grow in size over time as memory fades. These researchers were interested in knowing what would happen to the accuracy of eyewitness identifications if they allowed eyewitnesses to say they simply didn’t know when exposed to a lineup.
The researchers compared three conditions: witnesses given an explicit I don’t know response option; witnesses given a spontaneous, free report option where they could choose to write in an I don’t know response; and a forced choice option wherein the participant did not have the option of an I don’t know response. And here are some of their (only somewhat surprising) results:
Participants in the spontaneous/free-response condition were told to identify whether a photograph depicted a previously seen perpetrator. They were free to offer any written response but were not given specific response choices such as Yes, No, or I don’t know. Without the explicit presentation of an I don’t know response option, the participants in the spontaneous/free-response condition offered an I don’t know response only 2% of the time [3 of 139 participants]. In other words, there was not a stampede to the I don’t know response when it was not explicitly offered.
When explicitly given an option to check ‘I don’t know’, about 20% of the research participants did so. This is substantially higher than those in the free-response condition, but reflects their inability to express an opinion. That is, those that said I don’t know when given the opportunity, really didn’t know which photograph depicted the actual perpetrator. In other words, witnesses tend not to spontaneously report that they do not know the identify of the perpetrator, but if they’re asked, a substantial number admit they really don’t know. And they should.
When participants were forced to choose either yes or no–they were right just as many times but they were wrong two times as often. That is, they said the photograph depicted the perpetrator when it was really the innocent foil–these were false positive identifications. Taking away their opportunity to be unsure– forces errors.
The researchers say that what this shows us is that the inclusion of the I don’t know response is useful in increasing accuracy of eye witness identifications. They cite the realities that if a witness identifies a subject but really isn’t sure about the identification and then hears the subject will be prosecuted–they are likely to have increased confidence in their initially shaky identification. When we instead pay no attention to the low confidence or, more accurately, the I don’t know identification–the cascade of errors shown to exist in wrongful conviction cases will not include errors in eyewitness identifications.
While the inclusion of the I don’t know category does result in fewer identification decisions overall, (about 20% less based on this research) “those that are lost are uninformative and arguably better not rendered, because investigators may give them more weight than they merit and witnesses may forget how uncertain they originally were”.
It’s a good tradeoff. Decrease quantity of confirmations, but increase accuracy. And if you missed Tim Perfect’s earlier work on increasing the accuracy of witness testimony by simply having them close their eyes–it’s well worth your time to review it.
Weber N, & Perfect TJ (2012). Improving eyewitness identification accuracy by screening out those who say they don’t know. Law and Human Behavior, 36 (1), 28-36 PMID: 22471383