What if almost everything you knew about criminal profiling was wrong?
What if criminal profilers were more like psychics making vague claims based on base rates than hard-core FBI agents with guns? What if criminal profiling was based on an outdated and flawed psychological theory of personality and offender typologies did not accurately reflect reality? What if criminal profilers were not more accurate in predicting offender characteristics than non-profilers? What if there was no evidence supporting the claim that profiling training improves predictive accuracy? What if the popularity of criminal profiling is just due to various cognitive errors such as confirmation bias, post hoc thinking and mistaking correlation for causation? What if social psychological mechanisms are more important than evidence for explaining the beliefs about criminal profiling of law enforcement personnel and psychologists?
A crucial review article by Snook, Cullen, Bennell, Taylor and Gendreau (2008) called “The criminal profiling illusion: What’s behind the smoke and mirrors?” (full text) was published in the Criminal Justice and Behavior.
It successfully subjects criminal profiling to a corrosive acid bath.
Common criminal profiling typologies are invalid
Criminal profiling typologies are kind of like archetypes. They are systems of classification used to classify and categorize crimes and criminals. Common profiling typologies involve categorizing crimes and criminals as disorganized or organized and classifying rapists as power-assurance, power-assertive, anger-retaliatory and anger-excitation rapists. However, scientific research has shown that these are not valid typologies. For instance, such studies have failed to find any group of behaviors that could be classified as organized or disorganized.
Criminal profiling is based on an outdated psychology theory of personality
The review explains that criminal profiling is based on the assumption of trait theory, a psychological model that came under heavy scientific criticism in the 1960s. It more or less overestimate the effects of offender characteristics and undervalues the effects of environmental factors in accounting for offender behavior. This would seem to suggest that offender behavior should be consistent across crime scenes and that this behavior should be related to behavior displayed in the offender’s personal life.
However, research has shown that environmental and situational factors are just as important as offender characteristics in accounting for offender behavior. Only a subset of offender behavior is consistent across crime scenes, whereas others are not. Those behaviors that strongly depend on environmental/situational factors were often not consistent across crimes.
Criminal profilers rarely outperform non-profilers
To test if criminal profilers do any better than non-profilers, groups of profilers and non-profilers are given details of already solved crimes and asked to predict offender characteristics in a multiple-choice test. Their answers are compared with the actual case and this determines accuracy. In some areas, such as physical attributes, the profilers managed to out-compete the non-profilers. In other areas, such as cognitive professes, offense behavior and social history, the non-profilers performed better than profilers. Overall, profilers had a slight advantage, but the 95% CI error bars were extremely wide (2-5 times the limit of r = .1) and frequently overlapped zero. For individual studies, overlapping zero might not be a big deal, but wide 95% CIs were even found in the meta-analyses.
When the non-profiler group included experienced criminal investigators (who were not criminal profilers but presumably had more knowledge of base rates than non-profilers), the meta-analyses showed a slight advantage to the profilers, but again with very wide 95% CI error bars. The researchers argue that even if this result can be replicated, experiences criminal investigators should be able to perform roughly equal to criminal profilers.
Why is the belief that criminal profiling is effective so widespread?
The authors of the paper outlines several different reasons for why people believe that criminal profiling works despite the fact that it does not. Some of them are more related to information about criminal profiling being supplied by the environment (e. g. media attention, papers and so on)
: the paper discusses a study were researchers found that 60% of reviewed papers about criminal profiling appeal to one or more anecdotes in their attempt to evaluate criminal profiling. Anecdotes are very psychologically powerful but have almost no scientific value. The problem is not that proponents of criminal profile use of case studies, but that they are used in a deceptive manner: the vast majority of case studies discussed in the criminal profiling literature is positive and neutral or negative case studies are rarely presented.
: the more you repeat a message, the more likely it is to be believed. Proponents of criminal profiling like to assert that more and more people are becoming criminal profiling and that criminal profiling is useful for solving crimes and catching perpetrators. Despite the lack of evidence for effectiveness, repeating those assertions over and over again contributes to the spread of false beliefs.
: by shifting focus from the proportion of correct predictions to the absolute number of correct predictions, proponents of criminal profiling are able to maintain a deceptive shroud of efficacy. For instance, a study done in the mid 80s by Douglas and colleagues only looked at hits and correct rejections, while ignoring misses and false rejections. Once all data had been analyzed, the accuracy was not even over 40%.
: a long list of cognitive biases reinforce the myth of criminal profiling as effective. Confirmation bias leads to people remembering the times when criminal profiling works and forget when it did not work. Self-serving bias make profilers take more credit for a successful case than they deserve whereas bystanders may attribute more responsibility to the profilers than contributing factors from the environment. Mistaking correlation for causation and post hoc rationalizations may make profilers and law enforcement personnel attribute success to the profiler when in reality there was no causal link and so on.
: it is easy for humans to interpret general and vague statements as if they were specific to their situation. This is an important contributing factor for why people believe in astrology. Similarly, criminal profilers make a lot of vague statements in their profiles. The paper discusses a study where researchers looked at over 800 statements made by criminal profilers about the criminal and found that there was no evidence for over 80% of them, over 50% were not even falsifiable and almost 25% were too vague. Just being exposed to a vague profile can increase the perceived credibility of criminal profiling, even when the method used to generate the profile is known to be flawed.
: criminal profiling has gotten a lot of media attention: newspaper articles, books, films like Silence of the Lambs and TV-series like Criminal Minds facilitate the confusion between the scientific results of criminal profiling and the criminal profiling fantasy.
Three major conclusions presented by the paper are that:
(1) criminal profiling lacks scientific support and has the ability to interfere with criminal investigations. Therefore, it should not be used as an investigative tool.
(2) it is extremely problematic that the false belief that criminal profiling works is so widespread, both among the public as well as among law enforcement and psychologists.
(3) criminal profilers and other proponents of the effectiveness of criminal profiling should meet their burden of evidence if they want to attain scientific credibility.
Snook, B., Cullen, R. M., Bennell, C., Taylor, P. J., & Gendreau, P. (2008). The criminal profiling illusion: What’s behind the smoke and mirrors? Criminal Justice and Behavior, 35 (10), 1257-1276 DOI: 10.1177/0093854808321528