To Catch a Criminal, It's All About Presentation

Dallas’s police department changed the way it lines up its suspects for identification. Instead of the common “six pack” method where the victim looks at six photos at once, detectives (blind to who the suspect is) started showing the photos one at a time, reports the Associated Press. This small change, according to the AP, can lower misidentification rates by 39 percent. (HT: Daniel Lippman) [%comments]


This topic has been widely and deeply studied in experimental psychology, and was well understood when I was an undergrad in the mid-1980's.

See, for example:

John Doe

When the police shows six pictures at the same time, the victim chooses the one with more similarity with the suspect.

On the other hand, when only one picture is shown at one time, the phenomenon above does not occur. This implies in less false positive.


Social / forensic psychologists have known about and been advocating for this for years. About time some police depts started listening to the science!
Identifications are much more reliable if you're making an absolute judgement (is this the person you saw?) rather than what amounts to a relative judgement (which of these?) - in the latter, people are likely to choose the person who looks the closest out of the bunch.


I recall learning about this at psych 101 4 years ago. It was presented as something well and long established among psychologists that this was a more accurate way to identify people.

Glad to see that at least one police department has caught on.


It definitely gets rid of position effects, so that should help.


This is a statutory requirement for photo lineups in NC.


Is this legal though? I know there are tight restrictions on how identifications must be conducted in order to ensure the evidence will stand in court. Does this meet the requirements? Mind you, I'm not asking if it *should*, but rather if it does.


This strikes me as biased reporting. The post reports, correctly, that the sequential method reduces false positives by 39%, but it fails to mention that it also *increases* false negatives by 15%. Both statistics appear in the linked article. Further, it's far from clear that we should favor reduction of false positives to reduction of false negatives.

Eric M. Jones

There would still be a bias caused by the order of presentation. Perhaps it should be secretly guaranteed that the first X pictures and the last Y pictures shown are dummies and the real suspects are buried somewhere.

It's still a crummy system...your chance of being accused at random are too great.


Is this economics?


Yes, Eric@6:21pm, you're right. It's far better to imprison 1000 innocents than to let even 1 criminal roam free. I wonder what the stats are with false positive/false negative rates are for the "innocent until proven guilty" standard in American criminal law.

But in all seriousness, we have to look at the absolute numbers. If, for example, there are 100 false positives a year and 10 false negatives per year, you're probably making a good trade by decreasing false positives 10% for a 10% increase in false negatives.

And I think we really should do everything reasonable to avoid wrongful imprisonment.


I knew about this study back in the 90's. Unfortunately the only time I had the chance to put it into practice, police handed me a sheet with six photos on it -- I asked to see them one at a time, but they refused.

I ended up not being able to identify the guy who jumped me, so he escaped charges.

All was not lost though. He did get bitten by a police dog when he was arrested ;)


Not really a new idea.... Worcester, Mass. police had me view a sequential photo lineup in 1990. (I did untimately identify the rigth suspect).