Tuesday, December 01, 2009

More on the fallibility of eyewitness testimony

Mark Pryor over at the new blog DA Confidential supplies this example from personal experience demonstrating the fallibility of eyewitness testimony and the risks of making a case based primarily on eyewitness evidence:
In a recent trial, our main witness was a self-confessed drug addict, a man aged about 50 with (it turns out) poor eyesight. During direct examination, and in the usual manner, the ADA paused then asked him:

"And the man who fired the gun, do you see him in the courtroom today?"

The witness took his time looking around the court, eying the judge, the jury and then finally... a wavering finger pointed towards the defense table:

"There, I think that's him."

The finger had shifted and a surprised deputy looked up, identified as the man firing off shots the night of the murder.

A surprised female deputy, in full uniform.
From the research I've seen, eyewitness errors primarily occur because our eyes capture less than we imagine and much of what we "see" is actually filled in by our memory. So witnesses who knew an offender prior to the incident are much more likely to ID the right person. That's why, in cases where witnesses are ID'ing strangers, I'm increasingly convinced such testimony cannot legitimately be relied upon "beyond a reasonable doubt." That's particularly true in light of scientific findings that certainty by eyewitnesses doesn't necessarily correspond to the accuracy of their testimony. To prevent ever-more false convictions based on mistaken eyewitnesses, IMO corroboration should be required in such cases that independently ties the defendant to the crime.


Anonymous said...

As a recent Texas Monthly pointed out, a small number of white women have difficulty identifying the black men who raped them. While the percentage of these witnesses with this difficulty may be small the frequency with which this crime ocurrs causes these cases to accumulate.

Gritsforbreakfast said...

FWIW, there's no evidence to my knowledge that the percentage of witnesses with this difficulty is "small," nor that it's limited to white women or cross-racial IDs.

Check out this example for mostly white witnesses, a white perp, and an absurdly low rate of correct identification. Cross-racial IDs have a slightly higher error rate, but from the research I've seen, misidentifications can and do occur regardless of race and gender.