But the 101 participants we asked to view this lineup did not.
Here are their lineup choices:
34 |
7 |
3 |
20 |
33 |
4 |
Bias: .198 / .168
Size: 3.75
Again, the bias figure means that 19.8% picked the suspect, in relation to 16.67% expected by chance (1/6). If the suspect does not stand out then people who know only the verbal description should not be able to pick him out at better than chance rates. In this case, he does not stand out.
And once again, the size figure means that the lineup effectively contains 3.75 people: the suspect plus 2 and 3/4 alternatives. There are supposed to be at least 5 "fillers" in this photo lineup and all of them should be appropriate alternatives to the suspect. In this lineup there are only two other members who are a fully appropriate alternatives to the suspect.
By substituting a different photo of the suspect, one in which he is not squinting for the photographers flashgun, the lineup is actually seen as unbiased by our mock witnesses, even though there are just less than 3 useful fillers. But even though the lineup is not biased against the suspect, because three of the fillers are ineffective the risk of false identification for an innocent suspect is about 1 in 3, or 33%. The 3 ineffective fillers in this lineup fail in their role as a safeguard against false identification.
The moral of this story is that if you want an accurate idea of whether the suspect is the person the witness saw, it is important which photo goes into a lineup.
It reminds me of an aphorism a professor of mine used to quote:
Quality (lineups) are a lot like buying corn.
If you want nice fresh corn
you have to pay a fair price.
But if you can be satisfied with corn
that has already been through the chicken,
THAT comes a little cheaper.