I found this study in the journal Psychological Science quite interesting. It's titled "Creating Fair Lineups for Suspects With Distinctive Features" and considers the following scenario:
There's been a crime at your work-place (say, some computers were stolen), and although you did not witness the actual crime, you do remember seeing an unknown person walking the corridors the day before. Since you were busy with other things, and didn't really think much of it, you didn't take a really close look at the person. However you do remember some distinctive features; for example that it was a male, and that he had a peculiar piercing between his eyes.
Let's say, the person who actually robbed you was the one in the picture above,
and now the police goes out to find the criminal, actually succeeds in identifying a number of suspects, and then presents you with the following police lineup on the left.
The actual criminal is not amongst the people in the line-up, but because there is only one person in the lineup (the one at the bottom) with the peculiar trait - namely the eye-piercing - it is likely that you will see more similarity between that person and the robber than you do for any of the other three people in the line-up. Bummer, for the guy with the piercing...
Of course, it is clear that the above situation isn't fair. It's almost as if you knew the robber was female, and were presented a line-up with three men, but only one woman. Obviously this would create a bias for identifying the sole female in the line-up. Hence, the police needs to make a decision: Do they present you with a line-up in which everybody possesses the distinct feature you identified, or do they create a line-up in which the distinct feature is concealed. In our example with the piercing, the decision would be that of choosing between the two line-ups presented below. In the left one the distinct feature was added to all suspects in the line-up, and in the right one it was concealed:
What the researchers were interested in was finding out whether people performed systematically better at identifying the correct suspect in one of the two formats. Their finding, which confirmed a theoretical model of "hybrid-similarity model" of recognition, clearly showed that correct identification is more readily achieved in the lineup in which everybody possesses the peculiar trait (i.e. the left line up) than when the trait is concealed (e.g. by taking out the piercings as in the right line-up).
The actual study had participants view pictures of 32 males, a small number of which possessed peculiar characteristics, such as a scar, a tattoo, piercing, etc. After a short task that featured anagrams as an artistic nod participant's viewed a line-up of six - like the ones featured in the double column picture above - and were asked to identify the one picture they had previously encountered in the studying phase (if at all there was such a picture). The results are neatly summarized in the graph below.
The paper by Theodora Zarkadi, Kimberley Wade and Neil Stewart from the University of Warwick features a couple additional studies, and can be found in the current (January 2010) issue of Psychological Science
Main Reference:
Zarkadi T, Wade KA, & Stewart N. (2009) Creating Fair Lineups for Suspects With Distinctive Features. Psychological science : a journal of the American Psychological Society / APS. PMID: 19883492
