Offender profiling is practiced by many individuals and organisations as a forensic tool to aid the investigation of serious crimes that often occur in a series. The aim of profiling is to identify or locate an offender based on their “likely socio-demographic characteristics” (Alison, 2005, p. 3). It has been stated that “there is a fundamental disagreement between those practitioners who consider profiling to be an art, and those who wish to portray it as a science” (Ainsworth, 2001, pp. 5-6). The purpose of this paper is to investigate the various methods employed by these practitioners and compare them to various psychological methods to identify if profiling can be recognised as scientific or otherwise.
In order to test whether methods used to profile offenders are scientific, it is important to define what constitutes science in the domain of psychology. Field and Hole (2003) state that fundamentally science is based on empirical research, the observation of one entity or phenomenon’s effect on another. This effect can be observed naturally, or the cause can be controlled or manipulated, bringing light to the correlational and experimental research methods (Field & Hole, 2003). This approach, based on observing cause and effect (causality), is formalised as positivism, a term devised by nineteenth-century philosopher Auguste Comte (Robson, 2002).
Positivists make several assumptions about science and how facts are scientifically discovered (Robson, 2002). It is assumed that we can only form new scientific knowledge from facts, these facts can only be derived from direct observation, and hypotheses can be tested against these facts in order to prove a causal relationship between one fact and another (Robson, 2002). Another prominent concept in science is inductive reasoning, coined by John Stuart Mill, another philosopher from the nineteenth-century. Inductivism claims that once a hypothesis has been tested and confirmed, and nowadays reviewed and published, the resulting theory can be used by future researchers to prove their theories, and so on (Field & Hole, 2003).
These traditional views have made way for, or at least accommodated, variant theories and paradigms (coined by yet more philosophers) like Relativism, Post-positivism, Constructivism and Feminism (Robson, 2002). In psychology, and other social research, one cause for debate is the credibility of qualitative data in social science (in contrast to the fact-based, positivist approach that relies heavily on quantitative data) (Armstrong, Gosling, Weinman, & Martaeu, 1997). Another concern is sample sizes and the distribution of participants used in social studies (Streiner, 2006). Malim and Birch (1998, p. 62) suggest some caution when any studies are undertaken as “psychologists can apply scientific method as a means of gathering information, but this is not a problem-free approach”, as is demonstrated in the following review of the various approaches of offender profiling.
The FBI developed their Behavioural Sciences Unit in the late 1970s at the FBI Academy in Quantico, Virginia (USA) (Jackson & Bekerian, 1997). The unit was formed following the successes of others, for instance Dr James Brussel’s accurate psychological profile of the New York bomber George Metsky, and was the first attempt at formalising the process of “profiling for investigative purposes” (Jackson & Bekerian, 1997, p. 4). The BSU’s focus was initially on Sexual Homicide and their first study is detailed in Ressler, Burgess and Douglas’ (1992) book of the same name.
For the study a sample of thirty-six incarcerated sexual murderers was used, and data were collected between 1979 and 1983 (Ressler, Burgess, & Douglas, 1992). Quantitative and qualitative data were taken from multiple sources in order to collate a large amount of information about each individual, his crimes (they were all male), and his victims. Sources followed two veins: “official records (psychiatric and criminal records, pretrial records, court transcripts, interviews with correctional staff, and/or prison records) and interviews with the offenders” (Ressler, Burgess, & Douglas, 1992, p. xi). It proved difficult for the agents to interview the offenders, with offenders demonstrating a range of intentions for accepting the interview; for instance, whilst some wished to contribute, others used them to protest their innocence, and others used it as an opportunity to brag to the agents about the nature of their crimes (Ressler, Burgess, & Douglas, 1992).
The interviews consisted of a series of structured questions and the responses were used to obtain a statistical picture of various categories of characteristics; for instance formative events (e.g. family background and childhood sexual experiences) were noted, but additionally performance in school and in the workplace, in order to see if there was a correlation. Preoccupation with murder and finding out about antecedent behaviours (i.e. fetishism and animal cruelty) were also categories of questions that the agents asked the offenders (Ressler, Burgess, & Douglas, 1992).
Initially the FBI used the data from this small study to investigate what motivates sexual homicide, and this was formalised in the Motivational Model (Ressler, Burgess, & Douglas, 1992). The model suggests that a negative upbringing and certain formative events that occur during development may lead to negative self perception, fantasies and deviant behaviour. As the child experiments with these behaviours (such as animal cruelty and fire-setting) and realises that they satisfy his fantasies, they can develop into serious crimes as they are practiced; this phenomenon coined as a ‘feedback filter’ because as the behaviours cause satisfaction the individual feels compelled to do it again. Together with the qualitative data from the study, the model is intended to help investigators identify offender traits when encountering similar crimes in the future.
The next stage was adding offenders convicted of rape and arson to the study and creating the CCM – the Crime Classification Manual (Douglas, Burgess, Burgess, & Ressler, 1997). This exhaustive handbook, intended as a guide for other law enforcement agencies, divides the three main themes (homicide, rape and arson) into categories, suggesting to investigators the psychological traits of the offenders that commit them. Finally, the “Criminal Profiling and Consultation Program” was formed as a structured way to provide advice and physical profiles to agencies that request it (Ressler, Burgess, & Douglas, 1992, p. 110). A six-stage method to profiling named the Criminal Profile Generating Process was created, starting with the gathering of as many data as possible from many inputs, putting the crime into one of their classifications (e.g. from the CCM) and also classifying the offender type and motivations (using the Motivational Model) before producing the profile with the intention for apprehension (Ressler, Burgess, & Douglas, 1992).
The FBI provide us with their own take on science in the preface of the CCM: “The nature of science requires that one first observe and then attempt to categorize, compare, and classify observations.” (Douglas, Burgess, Burgess, & Ressler, 1997, p. ix). This provides us with grounding on their assimilation, classification, and reconstruction approach to generating profiles, but is their approach actually scientific? Critics of the initial study often state that the size of the sample of offenders that were interviewed was far too small to be scientifically reliable (Jackson & Bekerian, 1997). It has been argued that as this is the case the study failed to provide the reliable data necessary in the creation of the crime classification system, the psychological methodologies, and therefore the FBI’s profiles (Ainsworth, 2001).
The interviews themselves have been questioned; in some cases interviewees did not give their consent but the interviews were conducted anyway, raising ethical concerns, plus the offenders often had different motives for agreeing to the interviews. The rooms and environments in which the interviews were conducted varied greatly, and in Sexual Homicide Ressler et al. (1992) even admit that “the lack of secure facilities for [the] interviews and the nature of violent offenders are not conductive to the exchange of information” (Ressler, Burgess, & Douglas, 1992, p. xiii). Ainsworth (2001) believes that the interviewing of convicted offenders is itself a flaw, questioning the reliability of research conducted in this environment. As well as the interviewees being selective in what they choose to reveal, he suggests that as serial killers, “many of whom will have been classified as having antisocial personality disorder or as being psychopaths, are the least suitable candidates for research using retrospective self-reports” (Ainsworth, 2001, p. 112).
Ainsworth (2001) also questions the subjective nature of the methodologies that were developed, suggesting that the way FBI profilers come to their conclusions is not based on an objective, scientific model. This has led to some raising concerns as to how FBI profiles are created, questioning the lack of clarity as to how the FBI’s inferences are formed (Howitt, 2002). The problem with this subjective interpretation, Ainsworth (2001) suggests, is that there is the possibility that different profilers would come to different conclusions when presented with the same data. Rossmo (1996, cited by Jackson & Bekerian, 1997) states that without providing more transparency (and consistency) as to how their profiles are drawn up there is no way to replicate their work and test their hypotheses, which some argue is one of the cardinal rules of science. In order to satisfy this rule, a theoretical basis to their research would have been necessary (Jackson & Bekerian, 1997; Ainsworth, 2001). However the FBI do note that the nature of their approach is not reliant on the initial study alone, as stated by Ressler et al. (1992, pp. 135-136): “Investigators traditionally have learned profiling through brainstorming, intuition, and educated guesswork. Their expertise is the result of years of accumulated wisdom, extensive experience in the field, and familiarity with a large number of cases”.
The next area for review is the work of clinical psychologists and psychiatrists in profiling. Instead of making inferences between historical or statistical data to draw up profiles, these practitioners use their expertise in the diagnosis and treatment of abnormal behaviours or mental illness and apply it in a case-by-case manner in order to “ascertain the underlying motives between an offender’s actions” (Ainsworth, 2001, p. 16). Profilers from a clinical background often tend to be “forensic psychologists, working in the overlap between criminal justice and mental health, within prisons or secure units” (Gudjonnsson & Copson, 1997, p. 69) , for instance those l ike Paul Britton who started his profiling career whilst working at a psychiatric treatment unit (Britton, 1997) .
Clinicians may each take different paths to drawing up profiles of offenders, often relying on their own judgement to come to their conclusions. For instance Richard Badcock considers the offender’s developmental issues and has done a lot of work on violent sexual assaults and “stranger rapes” (Ainsworth, 2001, p. 143) , whilst Paul Britton typically likes to collate as much information about the case as possi ble before making his judgement (Britton, 1997).
Britton has amassed his own controversies within the profiling community, and one case is mentioned more than most: the murder of Rachel Nickell (Britton, 1997; Alison & Canter, 2005). After being called in by investigators to help find the offender Britton started drawing up a profile of the murderer. The profile generated was mostly based on his clinical experience, stating that the suspect would suffer from disorders like sexual dysfunction and discomfort relating to women (Britton, 1997). A handful of random statistics were also added to the equation, for instance the fact that “the offender would be aged between twenty and thirty ... most sexual attacks are committed by young men” (Britton, 1997, p. 246) was also predicted. After a Crimewatch appeal a man called Colin Stagg was arrested but released. Stagg fit Britton’s profile almost exactly, but the police did not have any sound evidence to keep him detained. Intent that they had found their killer, the police and Britton introduced a female undercover officer posing as a ‘lonely heart’ to Stagg with the hope of him confessing to the murder. With Stagg revealing details about his fantasies, some which bore close resemblance of how Rachel Nickell was treated, he was put on trial for the murder (Britton, 1997).
The judge threw the case out of court as the evidence was believed inadmissible, and years later Paul Britton faced charges of professional misconduct by the British Psychological Society (BBC News, 2002). Alison and Canter (2005) have provided a critical review of some of the evidence used in the trial, stating that certain ‘rhetorical tactics’ were used in the investigation. They believe the use of “real-world data ... not collected for the purposes of research” (Alison & Canter, 2005, p. 199) when formulating the structure of interpersonal relationships between the undercover officer and Stagg manipulated him for the purposes of the investigation, and therefore could not be credible evidence.
Those that take the clinical approach receive a lot of criticism from others questioning their methods. It is believed that this main critique is due to of a lack of co-ordination between these practitioners who each take their own approach to the discipline; therefore, “the failure to publicly articulate even a rudimentary account of what they do in common has left them vulnerable to ridicule as unscientific readers of imaginary runes” (Copson, Badcock, Boon & Britton, 1997, p. 13). These authors argue that attacks upon them are unfounded as over half of the profiling work undertaken in Britain has used the clinical approach.
In fact, they have gone part the way to defining a scientific methodology in their ‘procedural model’, which identifies the common steps taken when a clinician forms a profile (Copson, Badcock, Boon, & Britton, 1997). The first three steps of the model involve gathering information from the investigators and from a visit to the crime scene, and these allow the profiler to reconstruct the events surrounding the crime – the what, how and ‘to whom’. The fourth and central step to the model is using the case information and imported research to infer a motive for the crime. At this stage it must be noted that this model is very similar to the FBI’s Criminal Profile Generating Process reviewed earlier, with the first four steps of the clinical model almost identical to the first three of the FBI’s. Copson et al. (1997, p. 15) agree on the likeness, “but a clinical background should enable them to identify a wider and more sophisticated range of subliminal behavioural signals”, employing their practical knowledge of cognitive dynamics to make their inferences. The remainder of the clinical model takes a different approach, considering the psychological constructs and demographic and social factors before discussing findings with the investigating officer and producing the final report.
Underlying the model are the authors’ three principles of clinical profiling – that advice given to investigators should be custom made, interactive, and reflexive. The authors also warn of several dangers, one being to caution profilers working too closely with investigators as this may lead to allegations of collusion “such as tailoring a profile to fit a known suspect, or devising some interviewing strategy which is unethical or even unlawful” (p. 16); this hint may have been influenced by Britton’s negative experience of the Nickell enquiry. Another criticism from the Nickell case has been listed as a danger and is the fact that if data is summarised before being provided to the profiler this may cause misrepresentation (Alison & Canter, 2005).
The final area for review is the statistical approach to profiling, developed initially by David Canter. Canter was the Professor of Psychology at the University of Surrey when invited to co-operate on his first case of profiling (Canter, 1995) and his academic background reflects in his approach to the discipline. He believes in the scientific principles that theories are developed and tested, results can be replicated if research is repeated, and grounds his profiling method on “clarity and accessibility to facts and concepts” (Canter, 1995, p. 110). Underpinning his approach is the belief that similar to how science tests developed theories to predict future relationships, crime investigators must analyse cases solved in the past to solve cases in the future (Canter, 1995).
Canter is a critic of the FBI’s work as their claim to the use of psychology goes against what he believes is scientific, stating that “talking to criminals about their crimes is not psychology, and is certainly not science” (Canter, 1995, p. 106). Paradoxically Canter’s work is similar to that of the FBI’s because he also uses historical data to establish future facts, but his statistical approach removes the qualitative data and investigative intuition of the FBI’s (Ainsworth, 2001). The statistical approach is an entirely opposite methodology from the clinical psychologists’; relying solely on statistical data from the crime scene to map offender behaviours that may allow them to identify the type of offender they are looking for (Howitt, 2002).
Statistical profilers first examine crime scene records and note certain behaviours and features associated with the crime, for instance Howitt (2002) identifies several features and behaviours that may be associated with rapes including: Binding of limbs, forced fellatio, sodomy, torture and the offender apologising (to name just a few). A comprehensive selection of behaviours is collated into a master list (for that type of crime) and then the instances in which they occur together are taken from a selection of cases (Howitt, 2002). This method enables us to gain a greater understanding of which behaviours occur together, and also allows new cases to be linked with those that display similar traits allowing the possibility to associate several offences with one individual (Howitt, 2002).
There has also been further research in the use of statistics in profiling, and this is looking at patterns of where crimes are committed in order to locate where perpetrator(s) reside. There are two veins to this type of study: firstly, the ability to locate the type of people that commit crimes in certain areas – the demographic or environmental aspects; and secondly, the investigation into where a serial criminal resides in relation to the location of their crime scenes.
The demographic approach is known as crime mapping and allows the discovery of valuable information like crime concentration figures (Ainsworth, 2001) which can identify areas with high- and low-crime rates. The discipline concerned with the location of serial criminals is known as geographical profiling, and Rossmo (1997) believes it is most useful in the quest to locate the criminals committing multiple instances of murder, rape and arson. The main advantage to geographical profiling is the ability to narrow down lines of enquiry by focusing on the more statistically likely locations in which the offender will reside, with the aim to substantially cutting time and costs (Rossmo, 1997).
The statistical approach to profiling does appear to be rooted in science, with Canter using a good sample size for his studies and the opportunity to test his hypotheses in several cases (Ainsworth, 2001). Ainsworth (2001) notes that although Canter’s work is impressive, many of his methods are very technical in nature and are not “easily digestible by the lay-person” (p. 133), namely the police detective or investigator charged with implementing them. Britton (1997) criticises Canter’s staunch opposition to the use of the case-by-case approach when he believes that a more open mindset, with profilers considering each others’ professional input, would work better.
It appears profilers cannot agree on how to draw up their profiles. Even if some may disagree, each method employs some grounding in science, for instance the FBI’s rudimentary studies and the statistical hypotheses, and even the clinical approach is based on practitioners’ experience in diagnosing disorders that have been discovered using scientific methods. The FBI’s initial approach is now dated and although their theories are based on the initial study the founders are now gone and new computer systems assist agents in a similar way to the new statistical approach, whilst still retaining an investigative focus (Wilson, 2007).
Canter’s statistical method certainly takes a formal, academic route, ticking the right boxes in the science department. With hypotheses testing in mind it should be replicable, but is it actually replicable for the detectives that need to implement and understand it? Profiling seems to find itself sitting in an awkward position between the police investigator and the psychologist and often this relationship can be fraught with problems. It seems that as each method of profiling has proved to be successful in some cases and destructive in others the debate as to whether a scientific approach has been taken is not as important as learning the benefits and limitations of each method. This will prove their efficacy in the capture of criminals, with the priority as to not jeopardise investigations (and the reputation of profiling) in the process.
Ainsworth, P. B. (2001). Offender Profiling and Crime Analysis. Cullompton: Willan.
Alison, L. (2005). From trait-based profiling to psychological contributions to apprehension methods. In L. Alison (Ed.), The Forensic Psychologist's Casebook (pp. 3-22). Cullompton: Willan.
Alison, L., & Canter, D. (2005). Rhetorical shaping in an undercover operation: the investigation of Colin Stagg in the Rachel Nickell murder enquiry. In L. Alison (Ed.), The Forensic Psychologist's Casebook (pp. 197-234). Cullompton: Willan.
Armstrong, D., Gosling, A., Weinman, J., & Martaeu, T. (1997). The place of inter-rater reliability in qualitative research: an empirical study. Sociology , 31 (3), 597-602.
BBC News. (2002, October 30). 'Cracker' misconduct charge dropped. Retrieved March 27, 2008, from BBC News Web site: http://news.bbc.co.uk/1/hi/england/2377235.stm
Britton, P. (1997). The Jigsaw Man. London: Corgi.
Canter, D. (1995). Criminal Shadows. London: HarperCollins.
Copson, G., Badcock, R., Boon, J., & Britton, P. (1997). Editorial: Articulating a systematic approach to clinical crime profiling. Criminal Behaviour and Mental Health , 7 (1), 13-17.
Douglas, J. E., Burgess, A. W., Burgess, A. G., & Ressler, R. K. (1997). Crime Classification Manual. San Fransisco: Jossey-Bass.
Field, A., & Hole, G. (2003). How to Design and Report Experiments. London: SAGE Publications.
Gudjonnsson, G. H., & Copson, G. (1997). The Role of the Expert in Criminal Investigation. In J. L. Jackson, & D. A. Bekerian (Eds.), Offender Profiling: Theory, Research and Practice (pp. 61-76). Chichester: Wiley.
Howitt, D. (2002). Forensic & Criminal Psychology. Harlow: Pearson.
Jackson, J. L., & Bekerian, D. A. (1997). Does Offender Profiling Have a Role to Play? In J. L. Jackson, & D. A. Bekerian, Offender Profiling: Theory, Research and Practice (pp. 1-7). Chichester: Wiley.
Malim, T., & Birch, A. (1998). Introductory Psychology. Basingstoke: Palgrave.
Ressler, R. K., Burgess, A. W., & Douglas, J. E. (1992). Sexual Homicide: Patterns and Motives. New York: The Free Press.
Robson, C. (2002). Real World Research. Oxford: Blackwell Publishing.
Rossmo, K. (1997). Geographic Profiling. In J. L. Jackson, & D. A. Bekerian (Eds.), Offender Profiling: Theory, Research and Practice (pp. 159-175). Chichester: Wiley.
Streiner, D. L. (2006). Sample Size in Clinical Research: When is Enough Enough? Journal of Personality Assessment , 87 (3), 259-260.
Wilson, C. (2007). The Serial Killers: A Study in the Psychology of Violence. London: Virgin Books.
