However, one major issue with it is that it is unclear, aside from the original study

conducted by interview with serial sexual murderers, how these associations between crime scene characteristics

and offender characteristics have actually come about.

Traditionally, the FBI in their accounts have placed great emphasis on their investigators'

field experience and they cite this as the key factor that underpins the associations that they assert.

So a key question is how well do these profilers actually perform?

Although there have been many anecdotal examples and unpublished reports claiming the success

of the FBI's method of profiling, there are, in fact, very few published empirical

analyses and each of these are controversial in their own way.

However, as an illustrative study, we will go back to one of the first. And these were

conducted by Anthony Pinizotto with the collaboration of the FBI. In a paper published in 1990,

Anthony Pinizotto and Norman Finkel sought to determine whether professional profilers

were more accurate than non-profilers in generating personality profiles and correctly identifying

offender features from crime scene details.

They also wanted to explore whether the processes used by the profilers was qualitatively different

from the processes used by the non-profilers. Now in total they had 28 participants who

were categorised on the basis of their profiling expertise. We are going to concentrate on

just 24 of those participants. These were 6 police detectives who made up the PROFILERS

group. That is, they had all been specifically trained in profiling. They had completed a

1-year program through the FBI Behavioral Science Unit and they had a range of 1-6 years

experience in actually profiling.

The second group were the group of DETECTIVES. This group was made up of 6 detectives from a large metropolitan

city police department. Now these officers had no training in profiling, but they were

experienced police investigators in homicide and sex offence cases

(which were the two crimes chosen for this study).

The third group were a group of clinical PSYCHOLOGISTS and the fourth group was a group made

up of STUDENTS. In each of these were 6 participants who met the criteria and they were selected

as non-police participants who were both naive to profiling and criminal investigations.

All participants were given materials from two closed cases; that is these crimes

had actually already been solved. One case involved a homicide and the other a sex offence.

In both cases participants were provided with information about the victim

and also the crime scene reports.

So for the sexual offence, it detailed the events leading up to the offence, what actually

occurred in the offence itself, and the events that followed the offence. In addition,

in this case, participants also read a detailed victim statement, what happened and what

the offender did and said.

Now in the homicide case, participants were given 14 black and white crime scene photos.

They were also given autopsy and toxicology reports. Now these detailed the cause,

the manner and the mode of death. The toxicology report showed whether there were any chemical

agents present in the victim's system at the time of death that may have caused or

contributed to the death.

Now in both cases, and in order to protect the identity of the victim, some information

that would normally be available to profilers, such as maps of geographical areas where the

crimes took place, were actually absent.

So all participants read both cases and they balanced the order, so some read the homicide

case first and the others read the sexual offence case first. Now after they read each case,

they were asked to cover the material and write down as many details about the case

that they were able to remember.

They then had to use this to note down all the details from the crime scene that they

felt were necessary and important to be used in writing a profile about the type of person

of would commit such a crime, and also note down why they felt these facts were important.

They were also then asked to write a profile of the type of person who committed the crime,

and to give as much detail as possible, and then they had to record this onto audiotape.

Finally they were given a multiple choice question sheet. This consisted of 20 questions

about the suspect (questions about their age, their gender, their residence, their employment).

Now 15 of these could be scored as being correct or not.

And as a very last step they completed a line up task. Participants were given five descriptions

of possible suspects and they had to rank order them from 1 (the suspect they thought

who was most likely to have committed the crime) down to 5 (the person least likely to have

committed the crime).

So what did they find on these measures? Well, the analysis indicated that profilers provided

richer and longer reports than the non-profilers in terms of number of the predictions, for example.

Critically, however, they also provided significantly more accurate predictions in both cases.

Profilers were significantly more accurate than the non-profiler groups.

However, when their responses to the multiple choice questions were analysed, profilers

only scored significantly better than the other three groups combined for the sexual

offence case.

Similarly when looking at the line up results, the superior ability of the profilers to identify

the correct offender only emerged in relation to the sexual offence. It didn't emerge

on either measure in the homicide case.

Finally when looking at how profilers used the information they were given, to see if

the way they used it differed from how other participants used the information, they found

no evidence that the profilers assessed the information in qualitatively different way

from the non profilers.

So this study gives us a mixed picture of profilers of how profilers actually perform. Professional profilers provided

richer profiles, they made more accurate predictions in their written profiles than participants

belonging to the other groups.

Further, professional profilers were more accurate in terms of multiple choice responding

and in picking out the perpetrator from a line up for the sex offence case than non

profilers, but those accuracy differences were not seen in the homicide case. Here profilers

performed just as well, or rather, just as poorly as detectives, psychologists and students.

So why? Was the version of the homicide case just too sanitised? All the profilers noted that

information they'd expected to see when constructing a profile was actually missing in this study.

Or was it more to do with the fact that the homicide offender did not fit the base rates

in all ways? And this shows an inherent problem in profiling. What if your offender

isn't a statistically normal or a 'typical' offender?