Sunday, December 18, 2011

Profiling, stereotypes and accuracy

Psychological profiling and body language can be valuable tools for law enforcement so they can identify potential dangerous people or situations.

There are some important issues here, CAN is one, "...can be valuable tools..." and the other is "...identify potential...".

When we do a scientific experiment we have to qualify our measuring devices by checking them against standards. Some equipment can be checked against standards in the lab. Some equipment needs to be calibrated by specialists with specific equipment which has been properly qualified. We keep the qualification records and include copies of them in the results of our experiments.

Typically the equipment will have a specific variability, something like "this tool will measure X units within Y units."

Psychological profiling and body language require a carefully calibrated Observer. Someone who is properly trained. On top of that no one observation identifies a particular type of personality or situation.

The Israeli's supposedly developed a list of criteria that they use to identify potential suicide bombers. There are about 20 items on the list which includes nervous behavior with some sub-items, body language which indicate nervous behavior for example. If someone is displaying 15 of the 20 items the probability of them being a suicide bomber is supposedly about 90%.

So the Israeli's hand this list off to their recruits and then they are capable of identifying the potential bad guys?

No, they spend a ton of time in class rooms, they watch movies, review hundreds of photographs, re-enactments and then finally they go out and actually practice identifying out in the field with someone who is an expert.

After that they have developed a very specific skill that allows them to accurately identify POTENTIAL suicide bombers that they can arrest, search and question. Many times the person is not a suicide bomber, and sometimes they are and lives are saved. How often are mis-identifications made? Israel does not share the profile list or the statistical incidence of incorrect identifications.

On the other hand psychologists go to school for a while, engage in a few experiments and then go out to practice. Anyone good at their job is continually educating themselves. Some of these profilers talk to people involved in crimes and develop some criteria and it becomes popular to believe that these people have become as good at general profiling as the highly and specifically trained Israelis are at specific profiling.

How about police profiling. It typically involves passing around a list of criteria and maybe a couple of hours in profiling classes.

In other words it is crap.

Right now we don't have a specific set of criteria that identifies a person from a chimpanzee. Probably because no one really thinks we need one right? No, not exactly.

In cultural anthropology there is a problem, how do we identify "thinking species" or "civilized" from "non-thinking" or "uncivilized" species. One of the criteria is tool use, of course birds use tools sometimes as well as chimps. Sometimes criteria involving burial rights are used, respect for people who have died. Sometimes familial relationships are used. Right now "civilized" is a politically incorrect as well as a scientifically incorrect term because there is no real, universally accepted criteria which identifies civilization.

The idea in cultural anthropology is to identify the transition from ape-like creatures to people like creatures and we can't do it.

We need multiple criteria. We need specific criteria that when combined are unlikely to occur anywhere but in the group we wish to identify. Incredibly smart, very educated people can't identify "civilization", but, less smart, less well educated people can identify "at-risk" people.

For example, most serial killers are middle age, white men, but, not all middle age white men are serial killers. Most serial killers were abused as children. Not all middle aged white men abused as children are serial killers. Etc, Etc.

Sometimes profiling can help identify a specific suspect from a smaller group of suspects. You can't use the general population as a suspect pool and use profiling to identify a specific person as a suspect no matter how often they do it on television. Sometimes profiling can narrow the suspect pool.

In reality psych profiling is always subjective and can't be effectively qualified. You can't specify how often a profiler is going to be right with any statistical certainty.

Lie detector tests, which uses accurately qualified equipment, can't evaluate specific human behavior with any certainty. How well do you think even the most highly trained people can do? At the very best they will be less accurate than a lie detector test.

In other words, not accurate at all.

No comments: