Tuesday, March 10, 2009

Eyewitnesses in staged test only 8% accurate

In a test performed by a Lubbock TV station, only 8% of Texas Tech students could correctly identify the perpetrator from a photo array in a staged purse snatching:
The students entered the lecture hall one by one, unaware they were about to become eyewitnesses to a crime. We told them we were there covering a story about what it's like to be law students. We decided it was best not to hire a real bad guy, so we got our news director, Benji Snead, to do the dirty work.

We wanted to make this test as bona fide as possible, so we went to visit the professionals. We sat down with investigators with the Lubbock County Sheriff's office to see how they build a suspect lineup and created one for our investigation.

Next it was time for the crime. Our suspect walks into the classroom, steals the purse and flees the room. During the experiment one brave student sitting in the back of the class even grabbed Benji's arm, attempting to catch the purse snatcher. We then asked the class to write down everything they could remember about what went down.

The results showed an outstanding 32 students chose suspect 2, and 20 of them were more than 50% certain. Four students chose suspect 1, three of whom more than 50% sure. Four chose suspect 3, nine students thought it was suspect 4, and eight thought it was suspect 6. And 10 students either couldn't identify the perp or believed he was not in the lineup. As for the actual suspect, only six students chose our news director, suspect 5, and not one of them was more than 50% sure of their answer.

"The interesting thing is number 5, who it ending up being, I said wasn't it for sure. I probably had the best view out of anybody in the classroom," Taylor explains. Even the professor was incorrect, "truthfully he was right next to me and I had no clue which one it was. I thought it was number 6."

So how accurate is eyewitness id? In this case only 8 percent accurate - not good odds when your freedom is at stake.

17 comments:

Anonymous said...

That's some scary numbers.

Anonymous said...

I have seen this in action myself in more dramatic fashion. During my freshman semester at SHSU, our Intro to CJ professor (Dean of the CJ College) outlined his expectations (list of no-no's) for student behavior. Within a couple of weeks, a student violated several of those no-no's by coming into class late, wearing a hat and munching on a donut. The professor asked the guys three time to get rid of the hat and food, to which he didn't respond. The fourth time, the guy stood up, pulled out a chrome plated pistol and fired three shots at the dean who fell instantly. The guy ran while the majority of the class ducked for cover. Several minutes later, the dean got up and asked us all what the guy looked like and had us pick out the guy from a lineup. The results were comparable to the exercise you just described.

Anonymous said...

not good at id'ing suspects but can sure play football, 39-33.

plato

Gritsforbreakfast said...

Yup Plato, that 15-42 record Tech has accumulated against UT sure is impressive. ;)

Anonymous said...

Did anyone go get Lubbock County D.A. Matt Powell's comment?

I bet not...

Anonymous said...

Uh oh, forgot your oppositional research skills. Low blow, IMO.

Plato

Anonymous said...

And these weren't just regular ol' college kids either, this was an evidence class at the law school. Pretty amazing.

Anonymous said...

45-35

What happened at Tech was no fluke. They always forget.

Anonymous said...

DOing that test at Tech is sorta like doing it at the State School or at the ACLU.

Teachers in CJ programs across the country have been doing this trick for at least the last 30 years. Good scoop, gritso ratso.

Anonymous said...

You almost make Limbaugh seem to be reasonable.

Gritsforbreakfast said...

12:41- ironic, isn't it, that crimjust programs have been doing this demonstration for 30 years, everyone knows eyewitnesses are notoriously unreliable, yet the law enforcement lobby STILL fights eyewitness ID reforms?

Anonymous said...

My dad's a pretty conservative fellow, but once told me he'd never convict someone based on an eyewitness alone. Why? Once at a party for his work the secretary's sister called him by another name. Turns out, dad looks just like this woman's husband. From across the room, she mistook him for her own spouse.

Anonymous said...

Perfection in an imperfect world. Imagine all the people...

DefenderWill said...

Law enforcement will always oppose reforms designed to reduce false convictions until the block grant system is eliminated. District Attonrey's offices get thier budgets done by the number of cases filed, plead or taken to trial. Any political crime (DWI, Sex-anything, or welfare cases) they get money for in block grants or flat out payments. Many District Attorney's offices has contracts with DHS that pay out for each case. None of this money comes with the condition that the person charged or convicted is actually guilty. There is presumption that all people intend the natural consequences of thier actions. If the government is held to the same standard as the rest of us, the presumption is that the government intends that as many people are convicted as possible without regard to actual guilt. These abuses will continue as long as we keep paying out cash to keep them comeing.
In my own humble opinion...
DefenderWill

Anonymous said...

I remember a test similar to this one while I was in the police academy. Most of us were wrong, however the duration of our exposure to the suspect was seconds. Due to this we were not able to identify him.

It should be noted that eye witness testimony is made more accurate by exposure to a suspect. For instance, a subject that is raped for hours by an unmasked subject would most likely have no problem identifying a perpetrator.

Eye witness testimony should be viewed like any other type of evidence, with common sense.

Anonymous said...

Another thing is that we (police) ask if the witness is 100% sure of the ID. If they say no, then that witness ID is not used. You have to be 100%. I've discarded multiple line-ups after I hear, "I'm pretty sure that's him."

Unknown said...

..unless that person had it coming, like Shaka Sankofa.