A Common and Exasperating Fallacy about Eyewitness Identifications

"Inaccurate eyewitness identifications are involved in (80%, 90%, pick your percentage poison) of wrongful convictions. This demonstrates that eyewitness identification evidence is very unreliable and that it is the primary cause of wrongful convictions."


I have protested against this kind of reasoning before. See, e.g., this post and this post. But my protests haven't stopped people -- defense lawyers, newspaper reporters, judges, and even academics -- from making this kind of argument. I find this extremely exasperating.
It may be true that eyewitness identifications are frequently or very frequently inaccurate. But the sort of statistic quoted above does not show (A) that they are such (i.e., frequently or very frequently inaccurate), or (B) even if they are frequently inaccurate, that inaccurate eyewitness identifications are the primary cause of wrongful convictions of innocent people.
Why not?
1. It is not enough to know the number of cases in which eyewitness identifications are inaccurate. It is necessary (but not sufficient) to know the number of cases in which eyewitness identifications are accurate. If eyewitness identifications are accurate far more often than they are inaccurate, the discovery that a "large" number of eyewitness identifications are inaccurate does not demonstrate that eyewitness identifications are frequently or very frequently inaccurate.
2. The fact (if it is a fact) that incorrect eyewitness identifications were "involved" in wrongful conviction of innocent people in a high percentage of cases does not demonstrate that inaccurate eyewitness identifications are the primary cause of wrongful conviction of the innocent.

• I'll bet that courtrooms and judges and jurors and defense counsel and oxygen were "involved" in the vast majority of the cases in which defendants were tried and wrongfully convicted of a crime. It doesn't follow that either courtrooms, judges, jurors, or oxygen molecules are the primary cause of wrongful convictions of innocent defendants. For example, it would be useful -- and probably essential -- to know how often other inaccurate evidence of guilt -- evidence other than eyewitness identifications -- was used in trials that ended with the conviction of the innocent and how much the use of such other evidence accounted for the wrongful convictions that have been observed.

• It is often said, almost in the same breath, that inaccurate eyewitness identifications and false confessions are each involved in, say, 80% [or some other such number] of wrongful convictions of innocent people. Does it follow that (A) the primary cause of wrongful conviction of the innocent is inaccurate eyewitness identifications and (B) the primary cause of wrongful conviction of the innocent is false confessions? No: the two propositions are inconsistent.

Of course, it is possible that (B) eyewitness identifications are very frequently inaccurate and that (C) they are the primary cause of wrongful convictions of innocent people. But the statistic that (A) "inaccurate eyewitness identifications are involved in X% of cases in which innocent people have been shown to be innocent" doesn't get you from A either to B or to C.

&&& The dynamic evidence page

Evidence marshaling software MarshalPlan