Hi all,
The recent discussion on behavioural profiling has just led me to recheck to see what I can find on it's efficacy. Came across a simple article in Psychology Today (not the highest prestige journal out there, but contains articles that don't expect one to be an expert in the field).
In it, they mention how profiling approaches break down (broadly) into
1) statistical/analytical approaches
2) clinical interpretive approaches (i.e. of the Douglas type - experience, intuition, educated guesswork)
It's the latter that most people hear about, and yes, it's the latter that performs more poorly. While they don't go into details, the article does mention an experimental study. They trained one police force to use analytic approaches and another they left to use their usual approaches. After a year, the one trained with analytic profiling tools had solved over 260% more crimes. They didn't have a group trained in the clinical/interpretive approaches, which is unfortunate as I would have liked to see that outcome.
They also mention a study which reports that profile predictions were accurate 66% of the time, but don't provide enough details in this article to know how accuracy was judged, or what the chance rate would be. If it is scored such that the chance rate by random guessing would be 50%, then 66% is better than chance, but hardly something to bank upon. But if the chance rate of guessing was 10%, then that's a very big improvement.
The police themselves don't view profiles as if they're manna from heaven, but egos run high in police forces and profiles come from outsiders, so opinion polls might not be a great measure of their efficacy. Other similar values they mention are fairly low rates (like in only 2.7% of cases did the profile lead to an arrest) could suffer from the same problem - a reluctance to credit the outsider with contributing to the arrest, particularly as profiles are intended to make suggestions about what to look for, and an arrest only occurs after you've looked and found. But maybe the study is able to account for this problem, I'm just thinking of questions after having found this one.
anyway, it looks like there are some studies out now that are starting to look at behavioural profiling. Personally, I'm more interested in reading up on the statistical/analytical approaches (hardly a surprise there). I still haven't found a good study on the more "famous" version of profiling, but then, it's always the problem child everyone knows and not the quiet good one.
Here's a link to the article if anyone's interested: Behavioural Profiling
- Jeff
The recent discussion on behavioural profiling has just led me to recheck to see what I can find on it's efficacy. Came across a simple article in Psychology Today (not the highest prestige journal out there, but contains articles that don't expect one to be an expert in the field).
In it, they mention how profiling approaches break down (broadly) into
1) statistical/analytical approaches
2) clinical interpretive approaches (i.e. of the Douglas type - experience, intuition, educated guesswork)
It's the latter that most people hear about, and yes, it's the latter that performs more poorly. While they don't go into details, the article does mention an experimental study. They trained one police force to use analytic approaches and another they left to use their usual approaches. After a year, the one trained with analytic profiling tools had solved over 260% more crimes. They didn't have a group trained in the clinical/interpretive approaches, which is unfortunate as I would have liked to see that outcome.
They also mention a study which reports that profile predictions were accurate 66% of the time, but don't provide enough details in this article to know how accuracy was judged, or what the chance rate would be. If it is scored such that the chance rate by random guessing would be 50%, then 66% is better than chance, but hardly something to bank upon. But if the chance rate of guessing was 10%, then that's a very big improvement.
The police themselves don't view profiles as if they're manna from heaven, but egos run high in police forces and profiles come from outsiders, so opinion polls might not be a great measure of their efficacy. Other similar values they mention are fairly low rates (like in only 2.7% of cases did the profile lead to an arrest) could suffer from the same problem - a reluctance to credit the outsider with contributing to the arrest, particularly as profiles are intended to make suggestions about what to look for, and an arrest only occurs after you've looked and found. But maybe the study is able to account for this problem, I'm just thinking of questions after having found this one.
anyway, it looks like there are some studies out now that are starting to look at behavioural profiling. Personally, I'm more interested in reading up on the statistical/analytical approaches (hardly a surprise there). I still haven't found a good study on the more "famous" version of profiling, but then, it's always the problem child everyone knows and not the quiet good one.
Here's a link to the article if anyone's interested: Behavioural Profiling
- Jeff
Comment