Criminal Profiling : Is It Science?

A break from the religious stuff and onto the other half of my life, science! This is an essay I had to write up for one of my subjects and I received pretty good feedback on it so I thought I would share. We had to choose a forensic procedure and critically analyze it, so let’s see how well Criminal Profiling stands up against criticism. For anyone who would like my references I would be happy to email them to you, there’s a list of around 15.

Criminal Profiling (CP) has become some-​​what prominent in the public due to popular films such as Criminal Minds and Silence of the Lambs. These media however, almost never deals with the scientific validity of such a technique pertaining to investigating serious crimes, crimes namely of a sexual/​violent nature, resulting in ubiquitous unsubstantiated beliefs among the public regarding the efficacy of CP. As much as CP has evolved over the years the core focus still involves inferring unknown characteristics such as demographic, physical characteristics, emotional and behavioral characteristics (Inductive Profiling) of an unknown offender based on information shared by other criminals of a similar crime in the past. This is different to Deductive Profiling which is defined as interpreting forensic evidence, including photographs and reports, and a study of victimology, to reconstruct offender crime scene behavior patterns, and from those patterns infer motivation, emotion and other characteristics (Turvey, 2002).

Inductive Profiling is some-​​what a problematic approach to crime investigation as it mixes objective information, such as gender and age of the victim; with subjective inferences about the perpetrator such as behavioral and emotional status at the time of the incident. The likelihood for biased opinions creeping into an investigation may be increased by the subjective and inferential nature of such a system.Some basic aspects of the crime scene usually taken into account by the criminal profiler at the beginning of the profiling assignment are; the state of the crime scene, what weapons were used if any, geographical pattern of the crime(s) and the criminals access to and from the crime scene (Holmes and Holmes, 1998). Criminal profilers then collaborate with law enforcement officials by providing a psychological and social assessment of the offender(s); along with a psychological evaluation of personal effects found in the possession of the offender(s), they also provide consultation with law enforcement officials on the approach that would be most effective when interviewing the offender(s) (Holmes & Holmes, 2002). However, Holmes and Holmes state not all crimes are suitable for profiling; profiling is only appropriate when the crime suggests the unknown offender is intensely violent, ritualistic and exhibits signs of psychopathology such as psychopathy and sadism. An assumption which criminal profilers make is the behavior of criminals across their crime scenes is determined by traits of the offender rather than situational factors of the crime (Homnant & Kennedy, 1998). Along with that, this violent nature tends to escalate the degree to which offenders behave across their crimes and other aspects of their lives (Turvey, 2008). CP is intended to serve as an investigative aid to the criminal justice system and will seldom by itself solve a crime, its purpose is to narrow down the list of probable suspects (Kocsis, 2007). While CP may seem to have some useful features can it be shown to be scientifically reliable as a probability based subjective and inferential technique?

In order for CP to be considered scientific, any inferential hypotheses put forward requires a falsifiable characteristic. It must be said that inferential evidence is very important in validating scientific hypotheses. To say that science cannot be established through inferential data is rather fallacious, take for example a Carbon atom which had never been directly observed, up until recently (DOE/​LBNL, 2008), but the scientific consensus was they existed. This is not an argumentum ad populumas the scientific consensus was based on a multitude of inferential evidence and that evidence is the bases of such an argument not the consensus. Take for example the atom which was objectively established through empirical inferential evidence; the trouble with CP is the room for biased and subjective conclusions which results from a lack of objective methodology. Scientists go to great measures to ensure objectivity in their experiments by applying double even triple blinded, placebo controlled methodology which also minimizes or eliminates any occasion for biased methodology and interpretation of results. However, this well established method of science cannot be so readily applied to the subjective aspects of offender profiling. How can an emotional inference pertaining to an unknown suspect be falsified? At best such an inference is probabilistic guess work. Whether or not CP is falsified is irrelevant, to be scientific CP needs to be falsifiable, and any hypothesis which is not able to be falsified is simply not scientific. Falsifiability is an indispensable philosophy of science (Popper, K. 1959). It may be argued that a hypothesis in CP can be falsified when and if the right suspect is apprehended however, there is an important distinction to be made. Even though the atom hadn’t been directly observed, the opportunity to falsify the existence of it along the way was a real possibility; this is only possible with CP at the end of a case when and if the criminal is apprehended. This hardly provides a reliably falsifiable aspect to CP, and only increases the potential for inherent problems; time may be wasted following leads gained from an incorrect profile if there is no way to falsify parts of the profile along the way.

Keeping in mind CP is dealing with the probability of certain characteristics of offenders, many researchers have effectively used Toulmin’s philosophy of argument to establish whether probable characteristics in offender profiles can be based on substantive arguments which support or refute various inferred claims about the offender. Toulmin’s philosophy of argument allows an in depth analysis by a breakdown of more complex arguments into their elemental parts. One such paper detailed;

Of the nearly 4,000 claims studied [by criminal profilers], nearly 80% were unsubstantiated. That is, they contained no grounds, warrant, backing or rebuttal. Moreover, less than 31% of the claims were falsifiable… only 25% of statements in profiles were inferences about offender characteristics’

(Alison, L. J., Smith, M.D., Eastman, O., & Rainbow, L. 2003)

Even though the paper convincingly argues the potential importance of CP in law enforcement it concludes by conceding there is a need for a ‘careful, systematic evaluation of profiling advice’. It would seem that 3200 unsubstantiated claims about unknown offenders is rather hit and miss, basically what might be expected from chance and much like the guess work of a psychic detective (Joe Nickell, 1994). Turvey provides details on a case in which the profiler admits his conclusions were probabilistically manufactured, the court stated;

‘Although we do not doubt the usefulness of behavioral analysis to assist law enforcement officials in their criminal investigations, we cannot allow an individual’s guilt or innocence to be determined by ‘the ipse dixit’ [asserted but unproved opinion^*] of the expert [criminal profiler^*]

(2008, page. 708)
Turvey goes on to explain that according to the court, the profiler ‘acknowledged that his analysis involves some degree of speculation, and he further negated the sufficiency of his own analysis when he conceded that each case is ‘unique’ and that criminals are often driven by any number of motives’. Perhaps a reason for such a lack of support from this scientific study and the court is the range of false dichotomies drawn from a crime scene pertaining to an offender. One such dichotomy can be found in the Holmes and Holmes typology of crime scenes, the organized/​disorganized typology in serial Murder;

‘The crime scene is also very disorganized. In this respect, the crime scene reflects the personality of the killer… The method of murder for the power/​control type of killer is described as ‘simple and organized”
(Holmes, R.M. & Holmes, S.T. 1998)

According to Jackson and Bekerian (1998) such a dichotomy is widely accepted,pointing out that many countries and organizations such as the FBI, Canada, U.K and the Netherlands use the organized/​disorganized typology as a theoretical tool at the heart of offender profiling. However, a 2004 study An Empirical Test of the Holmes and Holmes Serial Murder Typology) of such a typology shows there seems to be little evidence supporting any interpretations of the concepts and theories underlying such an ambiguous twofold model. It further details a study aimed at the etiology of serial killing sating this typology ‘provides no indication at all there is a dichotomy that might relate to how organized or disorganized the offender is’ (Stone, 2000). Stone’s paper shows once again that a careful evaluation of profiling advice must be central to the field of CP if it is to be taken as a legitimate science. Along with that, the definition of disorganized and organized is too arbitrary and subjective; it could mean anything to anyone. The link that Holmes and Holmes propose ignores the multiple points between these two extremes, hence the false dichotomy. Instead what they have is better described as a spectrum of possibilities and the very inferential nature of this spectrum makes it difficult to assign personality traits so readily to a crime scene.

The difficulty with CP and its trait based characteristic being considered an empirically supported theory, are the documented flaws in any trait based model theory (TBM) which claims to have a scientific basis in its predicting ability (Mischel, 1996). Mischel et al. have shown that TBM are not a reliable source for predicting personality and behavioral predispositions, TBM typically have low correlations, are unable to predict behavior very well and all this leads to the conclusion that behavior is not consistent across situations; the paper shows that situational factors have just as much of an affinity to the prediction of personality and behavioral propensity of an unknown suspect. In any case the predictive ability of CP is lacking as published in a 2007 meta-​​analysis, which synthesizes results by combining the results of a number of independent studies, performed by Snook, Eastwood, Gendreau, Goggin and Cullen (2007). Possibly the most damaging conclusion in their analysis of the predictive accuracy between experienced profilers and college students/​psychologists, showed there is no statistical difference in predicting behavior between the two groups. If there is no difference in predicting ability between so called ‘professionals’ and un-​​professionals then it would seem that all probative value is lost. The meta-​​analysis did show that professional profilers were only slightly better at predicting overall offender characteristics but this is to be expected as this is the profilers supposed expertise, they should be much more familiar with the CP literature providing them with a wider range of options, ideas and hypotheses. If criminal profiling is greater than chance, there should be a statistically significant different between the two groups and there simply is not.

The scientific literature does not seem to largely support criminal profiling as an empirically verifiable technique. Mischel’s research shows CP relies on an outdated theory and this conclusion can be supported independently by its disappointing profiling accuracy expressed in the literature. CP desperately lacks a fundamental aspect which can be found in any valid science, a worthy falsifiable feature to its methodology. Criminal profiling has a well founded scholarly frame work but the profilers conclusions are opinions not facts and this seems to be supported in the scientific literature, these opinions must have reliability otherwise there is little probative value to such a technique. The profiling process at best maximizes the potential of existing evidence but should not be used as the first and/​or main approach to crime scene investigation, there is just not enough empirical support for this.