Eyewitness Testimony

It's one of the pillars of our criminal justice system but, unfortunately, it's totally unreliable.

One thing in particular that I recall reading about is the shoddy epistemics of the police lineup. Faced with a group of six dudes standing in a room, your would-be witness points the finger at whichever of the six guys he feels is probably the guilty party. The possibility that the guilty criminal simply isn't in the room doesn't "play," pyschologically speaking. The way the game is understood, the police have caught the criminal. The criminal is in the room. The witnesses role in the game is to correctly identify the criminal the police have already caught. Saying "he's not here" doesn't figure into that understanding, so it doesn't happen, and -- bam -- you've got an eyeball witness to your crime.

Comments

looks like you must have been picked out of a lineup.

Posted by melonhead | May 3, 2007 9:03 AM

Yeah, I'm not sure I understand this criticism. I take it the point is that it's unlikely that the witness would pick out the suspect unless the suspect really was at the scene. Even if witnesses always pick someone out, 5/6 of the time it won't lend any credibility to the witness. Of course false positives are possible, but it's hard to see how to avoid them.

Posted by minderbender | May 3, 2007 9:19 AM

Do the police ever run multiple lineups without the suspect involved? It would seem that "calibrating" the witness would be standard.

Posted by Jeffrey Davis | May 3, 2007 9:20 AM

It's worse than that. Lineups often consist of the suspect and five guys who work at the police station. Which means five guys who shaved that morning, dressed to go to work, and combed their hair before the lineup, and one guy who was picked up on the street (possibly because he looked suspicious) and kept in a holding cell for a few hours. Who looks more like a criminal?

Posted by arthur | May 3, 2007 9:41 AM

I don't know about this - "Saying "he's not here" doesn't figure into that understanding, so it doesn't happen,"

Subject to abuse, sure. "Shoddy epistemics" - not so much.

Is there necessarily pressure to pick one out of the lineup? I realize that unscrupulous cops might push in that direction, but the whole purpose of the lineup is to see IF the victim can pick out the ONE suspect the police have among five other "controls".

If the victim happens to choose someone other than the police's suspect, I don't think they're going to arrest the person he picks, who is someone they have no reason at all to connect with the crime.

The police might have various incentives to pick the wrong guy, to make their stats look good or something. But the victim has no such motivation.

Posted by sTiVo | May 3, 2007 9:42 AM

"Even if witnesses always pick someone out, 5/6 of the time it won't lend any credibility to the witness."

You're assuming that the witness picks at random. In practice it's more likely that the witness will pick the person most closely resembling the physical description of the attacker. Since that very physical description may have been used to identify the suspect, you wind up closing the circle neatly around that person. This is leaving aside all of the other subtle clues that could bias the witness, like the appearance of guilt/fear in the real suspect, hints from the investigators, etc.

Posted by MattNotYglesias | May 3, 2007 9:43 AM

This is bullshit. I've seen cop shows. Sometimes the witness says, "He's not here."

That byline should read "Kevin Drum, sponsored by your friendly neighborhood DNA testing lab."

Posted by mike | May 3, 2007 9:44 AM

Just the mere fact that the possibility that the criminal is not in the room is not explicitly presented is enough to bias the average result.

But since (depressingly) the criminal justice system is more about assigning blame than pursuing facts, you're not going to see police improving their lineups or prosecutors throwing out eye-witness testimony anytime soon.

And really, what's the alternative? Police need to be able to make arrests and file charges, and prosecutors need to use evidence in court. The only thing that I could see replacing eye-witness testimony would be omni-present surveillance of the sort they have in England. I'd actually rather take my chances with an eye-witness than be filmed from every angle without my knowledge or explicit consent.

Posted by TW Andrews | May 3, 2007 9:53 AM

I'm not a lawyer, but my understanding is that if you had 5 well-dressed, clean-shaven guys and one rough-looking man who matched the description of the criminal, that would not constitute a valid lineup.

Posted by minderbender | May 3, 2007 10:06 AM

Give me the fucking keys, you fucking cocksucker!

Posted by Stephen Baldwin | May 3, 2007 10:34 AM

I used to be a prosecutor and remember getting ready to start a trial of a woman who had been mugged. She had previously picked the defendant out of a photo line-up. The victim took a look at the guy sitting at the defense table and said "who's that?" We dismissed the case, but the guy had been in jail for almost 11 months waiting for trial. Photo lineups suck.

Posted by exlitigator | May 3, 2007 10:46 AM

minderbender- Lineups generally consist of one suspect and a handful of people from around the station and/or prison. Sometimes this results in "fair" lineups. Sometimes it doesn't. If it doesn't, the defense lawyer can challenge it. This isn't always easy to do. Further, sometimes the cops will subtly hint at who they want the person to select. This is easy to do without necessarily breaking rules. For example, if the witness picks person 3, and the cop knows that person is a prisoner selected to fill out numbers, he might say, "Are you sure? Take all the time you need." That's not an unreasonable statement on its own. But it does increase the likelihood the witness will do what they were told, take more time, and perhaps change their mind.

The biggest flaw, in my opinion, is that afterwards in court the prosecution asks the witness to point to the person in the courtroom who committed the crime. And the witness points dramatically and confidently at the defendant. The witness may have been tentative and unsure during the lineup, when he was connecting the lineup suspect to his memory of the crime, but here in court he's connecting the defendant to his memory of the lineup suspect. And that's pretty easy, he had all the time in the world to examine that guy. By moving the uncertainty to an out of court proceeding, it makes it tougher to evaluate the credibility of the witness' recollection.

Posted by Patrick | May 3, 2007 10:51 AM

I was once jump on and beaten up after a basketball game in high school. It was dark, and I did not get a good look at my assailant. All I could say that it was a black kid about my age, about 5'10" tall (becaue he at least seem taller than I was). After the police report, the detective asked me to go with him to look at some suspects who had been stopped in nearby suburb. I got there and all these kids were put in the line-up, and the police made it clear that I was expected to ID one of them. And I was angry, still scared, and wanted to get even. But for whatever reason, I told them the truth, that I could not definitely ID any of the suspects as the assailant. I have to wonder how many folks succomb to the social pressure and ID someone just for the purpose of pleasing the police.

Posted by Rick Kane | May 3, 2007 10:56 AM

I was mugged once, and got a good look at the leader of the pack. Its amazing, though, how bad my memory for faces is, because I had trouble remembering what he looked like (recall memory). Recognition memory is easier, so maybe if I had seen the guy, I would have been able to pick him out.

Anyway, I had to work with a policeman to do a sketch of the guy, using some lame "Mr. potato-head"-type software for generating a picture of the guy. The policeman didn't know how to use it very well. I'm a good artist, so I drew him a picture, but he didn't care about that...he had to use the lame software. The whole thing was a complete waste of time.

Posted by Jim W | May 3, 2007 11:18 AM

If the victim happens to choose someone other than the police's suspect, I don't think they're going to arrest the person he picks, who is someone they have no reason at all to connect with the crime.

That's correct. The point is that the witness is likely to pick the police's suspect even if it's not the right guy, assuming the police have nabbed someone who generally fits the witness's original description.

If the lineup consisted of 6 guys who looked nothing like the actual criminal, then yeah, the witness would have only a 1 in 6 chance of picking the police's actual suspect. But in most cases, the suspect will be the one who most closely resembles the witness's description, because that's the reason he was arrested.

Posted by Steve | May 3, 2007 11:22 AM

mike: This is bullshit. I've seen cop shows. Sometimes the witness says, "He's not here."

Ah, another Fred Thompson voter heard from. Mike, if you want to be taken seriously, please try to bear in mind the distinction between Things That Are Real and Things That Are Made Up For TV.

Posted by ajay | May 3, 2007 11:29 AM

There's some movement toward a different method of doing lineups, one at a time. The witness is shown one guy, asked, "Is that him?" then another, and another -- they don't know how many people they're going to be shown, so the temptation to settle for the closest resemblance is less. I don't think it's settled that it works more accurately yet, but it seems as if it should.

Posted by LizardBreath | May 3, 2007 11:55 AM

I suspect a tourney method would be the most useful.

Begin with four pairs, eight people in all. The witness rejects one person in each pair. That leaves four people or two pairs. Again the witness rejects one person from each pair. Finally the witness chooses between the two remaining.

Then the witness ranks the identified person on a scale of 1-10 with 10 being extremely likely. That score can be entered as evidence for the jury to ponder.

It is nonsense to ask the witness if he is sure during a lineup.

The real problem is that we want a witness to be certain. That made more sense in older times when most witnesses lived in a small, less diverse, community and saw far fewer people.

Posted by K | May 3, 2007 3:02 PM

When I was mugged, the police caught two suspects about a month later for a crime with a similar enough MO that they were pretty sure that they had the guys who had robbed me. I went through the whole lineup process.

The lineup seemed pretty fair--all twelve of the men looked and dressed reasonably similarly and could have been the perps. I was able with high confidence to identify the person who had actually held the gun in my face, but the other perp was pretty much a blur in my mind and I wasn't able to ID anyone in the second lineup. The cops did pressure me mildly to make a decision, but to their credit they took my statement "I don't recognize any of these men" as a final answer.

Fortunately, there was more than enough "circumstantial" (i.e., non-eyewitness; i.e., reliable) evidence to indict both of them, and they both pled guilty to multiple counts.

Posted by Kevin J. Maroney | May 11, 2007 5:28 PM

Am a nurse doing a criminal justice course with a question? Do police put the victim in the car to take around the neighborhood right after the assault to look for and hopefully identify the suspect? Is that acceptable as eyewitness identification? Thanks for any info

Posted by Edwina Hastings | May 23, 2007 9:36 AM