Understanding Criminal Behaviour: Beyond ‘Red Dragon’1 Faculty of Science Public Lecture Series 2002-3 University College, Cork John Horgan Few areas of forensic psychology attract more public interest and generate more misconceptions than offender profiling. This lecture, and accompanying paper, attempts to review the area, and considers the nature of approaches to profiling before proceeding to evaluate and discuss their overall validity and reliability. A critical appraisal of the current evidence suggests that although there certainly is a limited place for offender profiling as an ancillary tool to police investigations, albeit for a restrictive range and type of crime, we ought to treat with caution claims to the validity and successes assumed of current incarnations of profiling. Its immense popularity may be welcome in drawing attention to the work of forensic psychologists more generally, not least in Ireland. However, failure to critically consider the practice, extent and accountability of offender profiling and its practitioners may have unfortunate repercussions for the broader development of forensic psychology.
Introduction Since the mid-1990s, forensic psychology has enjoyed an exponential rise in popularity both in academic settings and in the popular media. Indeed, this is such that it is now the fastest growing speciality within the discipline of psychology. One primary driving force behind student and public interest may have been the plethora of popular crime fiction during that period, especially in film and television. Hollywood blockbusters such as “Silence of the Lambs”, and more recently “Red Dragon” have successfully whetted the appetites of would-be profilers drawn to psychology with expectations of becoming the next Will Graham or Clarice Starling. In fact, at the Department of Applied Psychology in UCC, we often receive enquiries from people seeking advice on ‘how to get a career in profiling’. It can come as a bitter disappointment to some then that offender profiling – generally the practice of deducing behavioural characteristics of an unknown offender based on an analysis of aspects of that offender’s crime – is neither transparent in its claimed successes (Silke, 2001), widely practiced (Davis, 1999), or indeed available as an actual career in itself.
1
An updated, developed version of this lecture is published as Horgan, J., O’Sullivan, D., and Hammond, S. ‘Offender Profiling: A Critical Perspective’, Irish Journal of Psychology (forthcoming). Offprints are available directly from the first author. Contact details: Dr. John Horgan, Department of Applied Psychology, University College, Cork, Ireland. Email:
[email protected].
1
In an attempt to dispel such misconceptions it is necessary that we examine the nature and effectiveness of offender profiling, as well as approaches to it, but a prerequisite is that we first consider what exactly is meant by forensic psychology. In general, forensic psychology involves the application of psychological theories, principles and evidence to legal issues at a variety of levels of the criminal justice system. Forensic psychologists, may, for instance, assess a defendant’s pre-trial fitness to plead, advise legal professionals on the nature of interview strategies to be used with particular kinds of witnesses or suspects, deliver expert testimony within the courtroom, and sometimes advise on interview strategies used by the police in the interrogation of suspects and witnesses (Howitt, 2002; Keppel & Walter, 1999). While some forensic psychologists assist in the recruitment and/or training of police officers, others provide counselling and psychotherapy services in the context of occupational management issues, sometimes for problems such as post-traumatic stress or duty-related victimisation. Forensic psychology is still in its infancy in Ireland although it thrives in various jurisdictions including the United Kingdom, several countries in mainland Europe, the United States and Australia. In Ireland, forensic psychologists are currently found working in academic settings engaged in teaching and research (on areas such as the victimisation of children, child pornography, terrorism, delinquency, eyewitness testimony, offender assessment, etc.), or for the Department of Justice, Equality and Law Reform, prison and probation services in the assessment, treatment and management of offenders. The forensic psychologist may also assist police in their enquiries during a criminal investigation. It is in this context that forensic psychologists may be ‘called in’ to provide an offender profile. Offender profiling Within the context of criminological history more generally, the professionalisation of offender profiling is a recent development and its thrust into the media glare has been nothing less than phenomenal. Unfortunately, this is somewhat of a double-edged sword, serving to rather obscure the true extent of the encouraging progress of forensic psychology. Currently, offender profiling refers to an assortment of practices, all of which entail a psychologist or some other ‘external’ agency engaging at some level with the police in order to assist a criminal investigation in some way. While a multitude of definitions exist, a useful summary is provided by Gary Copson (1995), who defines offender profiling as: “Any predictions, recommendations and observations based on the inference of offender characteristics from behaviour exhibited in a crime or a series of
2
crimes, and offered to the investigators as the product of statistical or clinical expertise” (p.v). The logic of offender profiling, in the main, is that behaviour expressed in relation to the commission of a criminal offence (which may be revealed before, during and after the crime itself) enables the ‘profiler’ to draw conclusions about the dynamics of that person's behaviour and personality (Michaud, 1986; Porter, 1983). In other words, features of the crime scene are assumed to reflect features of the offender’s behaviour and personality, such that understanding the nature of the crime scene and offence, we presume to understand something of the offender (Davis, 1999). This may reveal clues as to how offenders behave in everyday life, how they may be ‘distinct’ in some particular ways from others, what may be motivating them to offend in particular ways. A principal task of the profiler then is to dilute such deductions or inferences into a usable form for the police investigators. While the profile will not identify the criminal per se, the profile aims to help police investigators become “more attentive to individuals who fit that profile” (Davis, 1999, p.292). The earliest documented case of forensic analyses involving attempts at profiling perpetrators probably date to forensic analyses of Jack the Ripper’s victims during the Whitechapel murder cases of the late 19th century (Turvey, 1999). Turvey describes how police surgeon Dr. George Phillips began to offer character inferences based on his minute examination of the nature of the wounds inflicted on the killer’s victims and the careful removal of their bodily organs. Profiling also appears to have a basis in attempts to conduct personality assessments of political leaders ‘at a distance’. The predecessor of the US Central Intelligence Agency, the Office of Strategic Services (OSS) requested a profile of Adolf Hitler from psychiatrist William Langer. Langer’s profile was solicited to attempt to inform the interrogation strategy proposed should Hitler be captured or otherwise defeated (Canter, 1994). Langer successfully predicted that Hitler would choose to take his own life rather than face capture. One of the most celebrated early examples of profiling, however, comes from 1950s New York, and derives from Dr. James Brussel’s psychiatric profile of a serial offender, dubbed the “Mad Bomber” by the press. George Metesky, a disgruntled ex-employee of a local utility company, detonated over 30 explosive devices in locations all over the city. Brussel, a firm advocate of characterology and similar approaches that had dominated 19th century criminology, not only argued that “certain mental illnesses were associated with certain physical builds”, but also advocated during the Mad Bomber case the “diagnosis of an unknown offender’s mental disorders from behaviours evident in their crime-scenes” (Turvey, 1999, p.7). Brussel’s profile became popularised when his prediction that the offender would be wearing a buttoned, double-breasted suit was realized when the police came to question Metesky. Today, offender profiles tend to have one or more of the following objectives:
3
1) The development of a description of an unknown offender, based on an interpretation of crime scene elements of the unsolved offence (this is perhaps the broadest, and most common purpose of profiling); 2) Establishing the probability of a future offence from the same offender based on the crime scene analysis; 3) The linking of a series of unsolved crimes that appear to share distinct features (e.g. often through linking ‘signature aspects’ – essentially personalized offenderspecific behaviours that reflect individual interpersonal behaviour and/or fantasy); 4) The development of an interrogation strategy to assist police in ‘asking the right questions’. Again, this is based on a psychological portrait inferred from an analysis of the crime scene and/or statistical data from offenders thought similar to the unknown subject; The profile then is the production of these types of information for the police in the hope that it will somehow assist their efforts, even if only to help narrow down a list or ‘type’ of potential suspects. In the main, profiling appears to be most useful in cases that suggest some psychological disorder in the offender, especially in attempting to ‘profile’ personality from crimespecific behaviour, or in a series of unsolved crimes that suggest patterned or distinct offence-relevant behaviours. Indeed, given these limitations, current concerns about the broadening remit of profiling may be warranted in light of existing efforts to assert its validity even within the remit of a relatively small number and type of offences. We should realise then that profiling lends itself to use in quite a narrow type of investigation. Profiles appear in cases of child abuse, abduction, but most typically in serious crimes of a serial nature including burglary, rape, arson, and murder. For the most part, the profiler is consulted in the late stages of investigation, and examines the crime scene from a distance via materials provided by the police, and often, but not exclusively, when police leads have been exhausted. It is something of a myth that psychologists regularly visit crime scenes, although it does happen from time to time. Typically, the information made available to a profiler will include case documents on the crime scene, victim-related information, forensic information, including a pathologist’s report, preliminary police reports, and a variety and range of photographs. The profiler will also draw on a degree of personal experience or expertise relating to knowledge of both the type of crime in question as well as the type of criminal suggested from the analysis of the crime. It is important that the profiler should possess criminological knowledge central to suggesting probable demographic attributes. In some cases, the profiler, if a psychologist, will rely on experience from clinical case studies, statistical models of offending and other potentially useful contributions. Overall, however, the profile is basically a summary document, often not more than 6-10 pages,
4
rooted in personal judgements, a degree of educated guesswork, common sense and logic, and statistical and clinical probabilities. The profile is certainly not predicated on any specific ‘absolutes’ (Ainsworth, 2000). Profiles typically contain predictions of such information such as the offender’s sex, approximate age, race and other general characteristics (Ault & Reese, 1980). Usually, profiles contain suggestions related to whether or not the offender is likely to offend again, whether or not the offender may be forensically aware (i.e. perhaps through prior experience of the criminal justice system – e.g. via prior arrest - or through knowledge gained perhaps through textbooks on criminal investigations or forensic analyses, or in fact through previously undetected offences). In cases involving a sexual offence the profile might suggest details of the degree of sexual maturity of the offender or the role of specific fantasy themes in the offending behaviour (Turvey, 1999). Profiles may often also contain information related to the offender’s likely responses to police questioning and can result in the profile providing suggestions about the interrogative strategy to the police (often of particular use in rape cases (Keppel & Walter, 1999)). An important step in the production of a profile entails a reconstruction of the offence. The focus here is on the interaction between the offender and victim. Victim statements can thus be critical in order to ascertain the nature of verbal and physical behaviour displayed by the offender. The profile may then convey some form of classification of the offence. This can be important in indicating the likelihood of a reoffence in the future, so that the police and other agencies involved in the investigation may develop an intervention strategy aimed at warning a potential victim or group of victims (which may be symbolic or otherwise representative of the first known victim of a possible series). The above description of what profiling involves and the kinds of information provided probably represent a non-controversial view of profiling. Whatever the differences that might exist among the offender profiling community about some specific features of profiling, there are some fundamentally critical points we need to bear in mind. Given the multivariate nature of approaches to profiling, it is worth briefly describing the principal approaches currently dominating the practice. These are, in order of their systematic development: the FBI approach, the clinical psychological approach, the empirical datadriven approach (often encompassed under ‘investigative psychology’ or sometimes even ‘statistical profiling’ in the UK), and more recently, the geographical approach. 1. The FBI approach The FBI’s Behavioural Science Unit, now its Behavioural Sciences Investigative Support Services Unit (BSISSU), is the United States’ best-known clearinghouse for investigative queries that merit the production of an offender profile (Davis, 1999; Michaud, 1986; Porter, 1983). Typically, such requests derive from exhausted investigations at local levels, or involve the flagging of a series of unsolved crimes, typically murder, spanning two or more federal jurisdictions.
5
The BSISSU, and the FBI’s entire approach to offender profiling developed primarily from research conducted from the 1970s through to the early 1980s that saw FBI agents interview and conduct detailed research on 36 imprisoned serial murderers (Ressler, Burgess & Douglas, 1988) and build on that data through a mixture of personal experience, intuition and educated guesswork. The serial killer research represents the data pool from which the logical underpinnings and statistical assumptions made about unidentified offenders (i.e. the unknown subject of the profiles) were derived. All were male, and while 29 of the 36 were immediately classified as serial killers, all of the offenders were suspected of having killed multiple other victims. The nature of these offenders’ crimes was what allowed them to be typified as a relatively homogenous group: often their murders involved elements of sexual sadism, mutilation and often bizarre and/or ritualistic imagery or behaviour. The relevance of these features, apart from suggesting broad psychological ‘problems’, was unclear to investigating officers and the FBI’s unit was turned to in order to decipher possible clues as to the ‘personality’ and ‘mental make-up’ of suspects in these types of criminal cases. In general, the FBI dichotomised the crime scenes of this group of offenders into two broad categories – organised and disorganised (Ressler et al., 1988; Turvey, 1999). Organised crime scenes reflected an overall level of control and planning, with offenders bringing their own weapons with them and generally coming prepared. The organised offenders carefully planned their offence, often using restraints in subduing the victim. Again reflecting an overall level of organisation, the organised offender would take care to transport the victim to a predetermined location both in the execution of the offence or sometimes post-mortem to hide the location of the body. The victim of the organised offender would be ‘personalised’, in that while the victim may have been a stranger, they were seen to represent a vital element of the offender’s fantasy (Ressler et al., 1988). Disorganised crime scenes on the other hand reflected polar opposites. The crimes often appeared the result of unplanned, spontaneous ‘blitz’-style attacks, often subsequently resulting in greater levels of exhibited violence reflecting an overall lack of planning and forethought. The overall crime scenes were ‘chaotic’, with little effort made to conceal either the victim, or materials used in the offence (e.g. weapons). Both the organised and disorganised crimes were then correlated with characteristics of the offender sample. Offenders with organised crime scene behaviour were found to possess an average or above average intelligence level, were socially and sexually competent, often living with a partner, were mobile (with a car), and often had a degree of negative stress precipitating the crime itself (e.g. through personal turmoil at home, at work, etc.). The organised offenders typically not only follow the reported progress of the criminal investigation through the media, but would even change jobs or leave town as a result. Offenders who displayed disorganised crime scene behaviour, on the other hand, were found to have below average intelligence levels, were socially, sexually and
6
occupationally unskilled, were reported to be quite anxious during the commission of the crime, lived alone, and would not bother to ‘follow’ the news reports or make subsequent efforts to purposely avoid detection (Ressler et al., 1988). While the organised offenders’ behaviour and personality appear to reflect psychopathic behaviour (with the offender very much aware of their behaviour), the disorganised offenders were more typically representative of the presence of some psychotic disorder (Turvey, 1999). Howitt (2002) praises the FBI research for essentially producing a valuable body of knowledge from such a small sample size of offenders. The FBI model served as the blueprint for the hundreds of profiles constructed by the Unit annually. However, their findings and practices on which they were based have not escaped criticism. Canter (1994, p.10) reminds us that the term ‘offender profiling’ was created by the FBI, but that in doing so, they had inadvertently “created the impression of a package, a system that was sitting waiting to be employed, rather than the mixture of craft, experience and intellectual energy that they themselves admit is at the core of their activities”. It is generally recognised that the FBI approach does not easily conform to the rigours of scientific research (Godwin, 1998), and the dangers of both overgeneralisation and oversimplification are clear from the nature and extent of the research from which this dichotomy arose. In particular, trenchant criticism is offered by Turvey (1999), who asserts that the organised-disorganised dichotomy is essentially a false one, arising from “mistaken ideas about the developmental nature of criminal behaviour and the value of crime-scene reconstruction” (p.148). This, Turvey argues, can lead to confusion during the process of profiling as well as a failure to understand how a particular offender’s behaviour may develop over time. The problem of the categories not being mutually exclusive is also an issue raised by Godwin (1998), with the FBI developing a third category (‘mixed’) into which some offenders who display elements of both ‘organisation’ and ‘disorganisation’ are placed, or sometimes when the likelihood of the presence of more than one offender exists (Davis, 1999). To this day, the FBI (currently referring to their approach as ‘criminal investigative analysis’) do not make any aspirations to the development of profiling as a scientific endeavour, instead emphasising the primary role of basic ‘on the job’ law enforcement experience as the best predictor of success in dealing with such crimes. 2. The clinical psychological approach This form of profiling typically is performed by clinical psychologists or psychiatrists who rely more heavily on clinical data and models of offending, although as Silke (2001) asserts, the clinician will prefer to regard each case as unique thereby deserving a detailed case-specific analysis and classification. In contrast to the FBI approach, clinical profiles suggest that less effort is made to fit the offender into what essentially may be an unsuitably rigid classification type, and rather than relying on investigative experience,
7
the clinician relies on drawing inferences from expertise or experience with similar crimes and offenders, often in forensic and/or mental health settings. Given the increasing specialisation engaged in by psychologists (and forensic psychologists in particular), this form of profiling may on the surface appear quite potent for the police investigators: if the police consider consulting a psychologist on a rape investigation, there are several ‘rape experts’ available for consultation, whereas if the case involves serial arson attacks, one of several ‘arson experts’ are consulted, and so forth. In sum, individual skills and specialities are available for exploitation to their fullest potential. Clinical profiling is not without its faults however, and despite the overtly ‘psychological’ base to this offender profiling approach, clinical profiling is not a uniformly discrete process. For example, clinical profilers could easily arrive at different profiles from the same profile inputs. This can reflect differences in the professional backgrounds and training of the clinicians as well as differences in their theoretical orientation within the discipline of psychology. Unfortunately, for the nature of the information provided to the police, what can follow from this is that different importance can be attached to interpreting aspects of the crime scene. In the context of efforts to establish the scientific credibility of offender profiling, the clinical approach could then be criticised as conceptually weak (i.e. we may in fact be simply identifying differences in the styles and procedures of individual profilers). A related limitation here is that in dealing with such an individualistic approach, there may be little sense of transparency regarding how the profiler arrives at his/her conclusions. From the police perspective, however, a traditional difficulty with clinical profiles is that much of the information provided could be interpreted as of little practical use to them. Some clinical profiles highlight issues to do with the significance of the murders to the offenders, “internal conflicts and fantasy life” (Canter, 1989, p.12), all of which Canter argues “are hidden from view and therefore not readily open to conventional police investigation”. 3. The empirical data-driven approach Given that a major focus in offender profiling has been the classification of offences it is natural that a number of statistical approaches would emerge to address the problem. In fact, such developments were inhibited in the early years of offender profiling due to the fact that raw data was sparse and statistical treatments tended to be rather descriptive in tone (Perkins, Hilton & Lucas, 1990). However, with the advent of efficient database development the scope for statistical treatment of behavioural offence data has broadened and become well established. The CATCHEM database was probably one of the first in the UK developed by Derbyshire Police in response to the murder of three young girls (Aitken et al., 1995; Smith 1990). Other databases quickly emerged (Davies, Wittebrood & Jackson, 1997; Grubin & Gunn, 1990; Heritage, 1992) and the UK Police now typically use the VICAP (Violent Criminal Apprehension Program) system originally developed in Canada.
8
The statistical procedures used vary enormously from Linear Discriminant Function Analysis and Logistic Regression (Davies et al., 1997; Smith 1990) through Multidimensional Scaling and Cluster Analysis (Canter & Heritage 1990; Grubin, Kelly & Brundson, 2001) to the more statistically sophisticated Baysian Belief Networking (e.g. Aitken et al., 1996). One of the more prolific advocates of the data driven approach is Professor David Canter at the University of Liverpool; he is one of the leading advocates of the development of ‘investigative psychology’ in the UK. Canter (1989; 1994) encourages an empirical approach to profiling based on the collation of offence and offender characteristics and their subjection to statistical modelling in order to support any inferences of profile characteristics. The technique primarily employed by Canter and his colleagues (e.g. Canter & Heritage 1990) was multidimensional scaling (MDS) in which offences and/or their features were represented as points in space. The closer two points are found to be the more similar they are assumed to be. The specific techniques employed by Canter and his team derive from the work of Louis Guttman who developed the principle of ‘smallest space’ (Guttman, 1968). Thus the MDS methods typically used are usually known as ‘smallest space analysis’ (SSA). Such an approach is essentially a way of developing a categorisation system of offender behaviours (e.g. violent behaviour, impersonal behaviour, intimacy behaviour etc.) thought to be crucial to understanding the interaction between offender and victim. As such it is a procedure for identifying and recognising patterns in large, complex datasets (Salfati, 2000). A major advantage of this data analytic approach over many of the others is that MDS is very forgiving of sparse and ‘muddy’ data, making very few assumptions about distributions or sample size. The data from each offence are inputted into the model and the profiler may be able to ascertain the likelihood of a series of offences. Where some surface elements may ‘flag’ a possible series, SSA attempts to assess the statistical probability of the offence being committed by the same offender, owing to similarities and trends in offender-victim interaction styles, before, during, and after the commission of the offence (also see Canter & Larkin, 1993). On the surface, the fact than an approach to profiling has proceeded with due caution paid to developing a theoretical basis is regarded as a positive development. Such an approach, if valid, would see the relative subjectivity of clinical judgement removed from the profiling process. This is because the profile would be rooted in empirical research and statistical probabilities deriving from a datapool of known offenders. What distinguishes the statistical approach however, and this is indicative of how openly critical Canter is of the FBI approach, is explained by Howitt (2002): the objectivity of such an approach may indeed represent a favourable feature, but the conclusions from such profiles may “not necessarily [be] more helpful to the investigators” (p.219), with the possibility of missing “salient features of individual cases” (Silke, 2001, p.245) an inevitable result of generalising across such large datasets (also see Turvey, 1999).
9
One weakness that the data-driven approach has when compared to the clinical approach is that the latter may be of use in bringing experience to bear on a number of wider, yet equally pertinent issues. The statistical profile in isolation will not be able to offer inferences about the perpetrator’s background or other predisposing influences (and how these might influence the nature and/or extent of the offence in question). However, when considered with the geographical approach, to be described next, such data-driven approaches to profiling still represent the most encouraging progress made to date to support the assumptions and assertions inherent in profiling construction. 4. The geographical approach This approach is probably the most recent, and is most commonly associated with the work of Kim Rossmo (1995, 1997). The basis for geographical profiling is that it may be possible to locate the probable residential or employment base of a criminal. This is based on an analysis of apparent patterns in the criminal’s offence characteristics. The theoretical underpinnings of this approach essentially borrow, as above, from mathematical and statistical models, but are interpreted on an individual basis through the filters of environmental and cognitive psychology. The approach involves assumptions about the spatial dimensions to behaviour (not necessarily ‘criminal’), and as such Rossmo’s method is typically associated with the need to understand the relevance of factors such as routine behaviour patterns of victims, the location of the offence, demographic attributes of the surrounding environment, etc. (Woodhams, 2002). Encouragingly, geographical profiling does incorporate findings from other approaches. Canter and Larkin’s (1993) work on serial rapists for instance suggests that offenders operate within distinctive “offence regions”. Their analysis of 45 male serial rapists illustrated that some serial offenders (the ‘marauders’) tend to hunt in areas with which they are familiar: 87% of the sample move outwards from their home base to conduct their attacks. Canter and Larkin also proposed the existence of ‘commuters’, thought to travel to a ‘target location’ both to hunt for and victimise targets. Canter and Larkin’s analysis demonstrated that offenders not only operate within distinctive regions but that the distance travelled tends to correlate directly with the distances between offences. The authors argue for the existence of ‘domocentric’ behaviour in rapists, and suggest that the principles offered may be useful for testing in other, non-criminal behaviours (e.g. shopping, the search for new homes, etc.) (also see Brantingham & Brantingham, 1981; LeBeau, 1987). There are now a variety of computer software packages that will produce a colourful geospatial map of an area once the profiler has inputted the locations of a series of crimes. One of these is the Rigel program (Woodhams, 2002), which takes the profiling input details of several crimes and presents then a 2D or 3D surface map, displaying the
10
“probable location of offender residence or employment using different colours” (Woodhams, 2002, p.36). Apart from the program taking, at times, up to two weeks to deliver its surface map, Turvey (1999) chastises the method for being about as “enlightening” as sticking pins on a big map and points to the perpetual error made by this and other forms of profiling in that an attempt is made to infer “meaning” from behaviours which are often taken “out of the overall behaviour and emotional context that it was produced in” (p.262). Nevertheless, geographical profiling represents an interesting development, and will continue to be tested in combination with data-driven approaches. Does profiling work? Current approaches to offender profiling do not easily lend themselves to objective, empirical evaluation. In some cases, this is due to the wide variety of practices and procedures (let alone the assumptions underpinning the appropriateness or otherwise of subsequent deductive/inductive decision-making styles with which to reach conclusions) that may constitute ‘offender profiling’ per se, as we will see below. It may well be the case that an attempt to evaluate the validity of profiles may tell us more about differences between the training, background and professional orientation of the profilers themselves than any firm conclusions about assumed uniformity in the practice of profiling. Given this, we ought to state that any profile must exist solely as an ancillary tool to a police investigation: it is essential that the profile be viewed as such and its limitations recognised both by profilers as well as the police. This question ‘does profiling work?’ is not as easy to answer as we might imagine, and indeed we may be forced to consider this question from a variety of perspectives, diverging agendas and mixed outcomes. A direct answer assumes that profiles have explicit objectives and that there is a sense of rigour associated with the refinement and development of profiling practices. Unfortunately, neither represents the current reality of offender profiling and numerous practical, conceptual and theoretical issues contribute to the complexity of any attempt to evaluate the effectiveness and utility of offender profiling. To begin with, we ought to perhaps reiterate the immense task faced by the very prospect of profiling. The profiler is essentially constructing a set of relatively subjective judgements and offering hypotheses about a person whom the profiler has never met. The potential for failure is therefore quite high (failure to reliably predict behaviour is already a challenge in everyday clinical assessment cases when the psychologist has access to the client). Some profiles are so vague as to be practically useless to the police, while others are so specific as to invite a complete lack of corroboration on key features. In less fortunate cases, this may end up skewing the nature of a police investigation, as the examples to follow illustrate. A significant problem in assessing profiling methodology involves lack of transparency. In cases in which psychologists contribute, these are usually high-profile criminal
11
investigations with an unrelenting glare of publicity on any technique used by the police, especially anything remotely novel: the need for transparency and clarity is therefore of paramount importance for all parties involved. It may not be so difficult to answer why there are so few profiling cases made available for subsequent dissection and analysis – we rarely hear of or see the failures. It is essential for the development of profiling that the nature of the items provided in profiles must be individually assessed following the conclusion of an investigation. This would allow us to check for instance, which aspects of the profile have been correct, and which have been incorrect. It is inevitably the ‘hits’ we hear about through the media – often specific items of information provided in the profile will be over-exaggerated in their importance (as evident in the Brussel case), but the ‘misses’ still tend to be shrouded in embarrassment and acrimony, with no clear sense of what was ‘right’ or ‘wrong’ about the profile. Relying on simplistic ‘accuracy rates’ to ascertain whether or not profiling works can thus be misleading. Without access to failures as well as successes, our conceptual development of this area will remain limited, particularly regarding how we might refine the nature of profiling objectives to more discrete, and possibly as a result, more achievable goals. While most profilers operate alone in the construction of their profile, ‘group’ profiling is not necessarily more reliable: the team of psychologists and psychiatrists established during the 1960s recommended to the police that the murders later associated with Albert deSalvo, the “Boston Strangler”, were actually committed by two men, “each of whom lived alone and probably worked as a schoolteacher” (Porter, 1983, p.3) and one of whom was thought to be homosexual. In fact deSalvo lived with a wife, two children, worked in the construction industry and as Porter reminds us, “would never have been found by any of the policemen assigned to search the city’s homosexual community” (p.3). With this in mind, we should also note that the simplicity of definitions of profiling can be deceptive in evaluating the effectiveness of the practice(s) of profiling (Howitt, 2002). After all, all of the practices described in these forms of profiling reflect extremely complex psychological issues that relate to behavioural consistency (i.e. incorporating notions that we can actually derive any meaningful deductions or inferences at all from crime scene ‘behaviours’ that enlighten us as to what ‘type’ of person the offender might represent and the assumption that there is consistency in certain behaviours and personality traits more generally), the homogeneity of types of offenders (a point certainly represented in the FBI approach), and a host of other issues that, without empirical research and hypothesis testing, leave aspirations to the scientific bases of profiling wide open to easy criticism. The central hypothesis of profiling, according to Canter and Heritage (1990) remains that “offenders differ in their actions when committing a crime and that these differences reflect (and therefore correlate with) overtly available features of the offender” p.187) – but this is certainly a hypothesis open to empirical testing. If profiling is to be scientific (hence empirical), Boon (cited in Woodhams, 2002, p.34-35) argues that we must accept that profiles must be falsifiable and be “aiming to make a model of reality”. Boon argued that these difficulties could be overcome in approaches to profiling, but that true empiricism is probably too great a goal to aspire to. For present purposes, however, we can agree that the self-correcting nature of the scientific method is absent from most approaches to profiling, and the
12
unfalsifiability of both the methods and principles underpinning the conclusions of profiles is akin to the conceptual problems that have plagued psychoanalytic analyses. In fact, it might come as no surprise to its detractors that psychoanalytic notions remain common in clinical profiles, and despite their unfalsifiability, their persistence owes much to historical precedents: in Brussel’s profile of the Mad Bomber, the psychiatrist deduced that Metesky hated his father but would love his mother - this derived from Brussel’s interpretation of the way in which Metesky “rounded out the sharp points in his W’s so that they resembled cartoon versions of a woman’s breasts” (in Porter, 1983, p.2). Even if such assumptions are empirically tested, there are a variety of practical matters we must consider in relation to strengthening the reliability of the methods used in profiling and the processes than underpin them. For a start, the quality of the information provided to the profiler (aside at all from any conceptual issues arising from inferring motive or deducing behavioural characteristics from behavioural ‘indicators’) is absolutely crucial to the process of producing a reliable profile. And, as with the criticisms levelled at the clinical approach, it is inevitable that some data sources will play a more significant role for individual profilers. It is not inconceivable that single pieces of profiling input data may serve to skew the direction, nature and results of any particular profile, regardless of what ‘kind’ of profiling produced it. Evaluations of the effectiveness of profiling continue to be very rare, with literally less than half a dozen such studies in existence (Copson, 1995; Jackson et al., 1993; Kocsis, 1995, Pinizzota, 1984). Perhaps the most enlightening appraisal of offender profiling is that produced by, helpfully, a police officer. In 1995, Gary Copson, of the Metropolitan Police Service in London, conducted a ‘consumer survey’ of profiling activities. Copson (1995) scrutinised over 180 cases in which police investigators made use of 29 offender profilers, all psychologists or psychiatrists. In general, just over half the sample of police officers felt that they had received ‘some additional information’, although when pressed, this amounted to 14% of the sample agreeing that the profile had actually contributed in a practical way towards solving the crime in question. Corroborated in Copson’s analysis is the fact that profiles therefore rarely lead the police to the direct or immediate identification of an offender: although this remains one of the core myths of profiling, Copson reported only 3% of cases in which this appeared to happen. The results are not all so negative however, and we have some reason for considering other yardsticks for measuring the effectiveness of profiling. One positive finding from Copson’s study was that over 60% of police officers surveyed felt that it was helpful to have the profiler present. Many of these officers reported having been given a sense of reassurance and encouragement in the formulation of their own opinions and decisions, and having ‘experts’ on hand to reinforce those judgements felt beneficial. Consequently, about the same number of officers reported that they would certainly consult a profiler in future.
13
The issue of difficulty in establishing the criteria for success was certainly illustrated in the Copson study. However, the profilers’ information (most of whom would have identified with the ‘clinical’ approach) appeared largely correct in over 2/3 of the data provided, but unfortunately, and again owing to the lack of empirical work on such cases following the conclusion of investigations, we are not clear on what kinds of information (and what patterns if any exist within and between them) the 1/3 of ‘incorrect’ data consisted of. Conclusions According to Silke (2001, p.243), offender profiling currently suffers from an identity crisis, with descriptions synonymous with ‘offender profiling’ including: “psychological profiling, criminal personality profiling, criminal personality assessment, criminal behaviour profiling, offender profiling, criminal profiling or investigative profiling”. Taken together with the multitudinal range of objectives expected (and proffered) within and from offender profiles, it is little wonder then that offender profiling remains “neither a readily identifiable nor a homogenous entity” (Gudjonsson & Copson, 1997, cited in Ainsworth, 2000, p.103; also see Homant and Kennedy, 1998 for suggestions to encourage distinctions between profiles in relation to their objectives). Although a sense of diversity in the area is important for theoretical progress, Bekerian and Jackson (1997) warn that too much diversity can lead to – and in the case of offender profiling to date, has led to – unintentional fragmentation and confusion. Given the relatively mixed results of profiling to date, we need to point to other influences to explain why offender profiling still occupies its distorted reputation and rather undeserved recognition. Part of this identity crisis is grounded in confusion about whether profiling necessarily requires a scientific basis. In asking whether offender profiling currently ought to be considered more an ‘art’, or soft science than anything remotely resembling sound empiricism, the current reality of profiling, and ‘current’ is emphasised, probably reflects an uneasy and perpetually shifting combination of both – the amount of each tends to alternate, sometimes dramatically so, from profiler to profiler and case to case. For a person to claim competence at any activity that might constitute the description of ‘profiling’, an in-depth knowledge and understanding of the specific type of offender under scrutiny, as well as issues relating to the offence, is essential. Boon (cited in Woodhams, 2002) asserts that offender profiling contains elements of ‘art’ in that experience and creative talent play strong roles in the process of deducing a profile. The profiler should certainly clarify to the investigation team exactly what information he or she is likely to be able to offer and what the implications of that information might entail. This assumes, of course, that the profiler is very clear on what information is likely to be of use to the police. Sometimes profiles contain such jargon as to be of little practical relevance for the investigation team. The importance of recognising and acknowledging one’s limitations as a potential contributor to a police investigation cannot be overstated. According to Turvey (1999, p.266), investigators should only “enlist the services of competent professionals who are willing to engage in frank discussions about the limitations of their method.” Similarly, the police need to realise that it is unrealistic for any person, psychologist or otherwise, to possess an expert level
14
of knowledge and understanding of every type of offence and offender. There have been past instances where profilers have been quick to offer assistance to any number and variety of offences, and the consequences have been damaging to existing attempts to develop the scientific qualities of profiling. According to Silke (2001), less than about 2% of all forensic psychologists will ever engage in offender profiling. The reality of profiling remains such that although Ainsworth (2000) describes the rate of profiling as increasing on an annual basis, it is still not nearly as common as we might like to imagine. Although the FBI consistently and systematically deliver profiles, UK and European practice is somewhat more contained and restricted. And while it may be at least of some comfort that the majority of the media-friendly profilers offering their unbridled assessments on the recent Washington snipers case were not psychologists, it can only take one public relations disaster to call into question not only offender profiling, but the parent speciality or even discipline espoused by the profiler. This was probably the only clear outcome of a lengthy and misguided police investigation in the UK that began with the murder of a young woman and ended with calls for offender profiling to be dismissed outright. Rachel Nickell was walking on Wimbledon Common with her young son, and in broad daylight was stabbed to death in a frenzied, vicious attack. Forensic psychologist Paul Britton was asked to help the investigation, and was specifically asked to provide a profile of the likely offender. Before long, the police identified nearby resident, Colin Stagg, as a possible fit to Britton’s profile, but had no incriminating evidence to ensure his conviction. The police found a sexual fantasy letter in Stagg’s apartment that Britton subsequently took as support for his profile theory that the killer would possess violent, sexualised fantasies. Britton subsequently advised the police to pursue a ‘honey-trap’ operation to ensnare Stagg (Wilson & Soothill, 1996). Using an undercover female police officer to engage in a relationship with Stagg, the officer (who used the pseudonym “Lizzie James”) would offer Stagg sex in return for his admission that he was capable of the kind of crime that fit the profile of the Nickell case. During the six-month undercover operation, however, Stagg held firm and although he eventually admitted to “Lizzie” that he had been quizzed by police investigating the murder, he adamantly denied he was responsible in any way for it. After a lengthy investigation, the police team was still no closer to identifying Nickell’s killer, and following a hearing at the Old Bailey, Britton’s role in leading the investigation was “strongly condemned” by the presiding judge (Wilson & Soothill, 1996, p.13). The police were also openly pilloried for uncritically accepting the guidance of the psychologist and allowing him to apparently exceed the boundaries expected of any external support provided to the police. Although the lessons were learned by the police as far as maintaining an objective distance with psychologists was concerned, it is not merely the interests of individual profilers that can suffer (after all, Canter, the most vocal critic of Britton, argued that profiling itself was not at fault in this case, but that Britton himself had been unscientific: “what we’re dealing with here [in Britton’s analysis] is opinion presented as if it’s based on a great deal of expertise” (Wilson & Soothill, 1996, p.13).
15
Fortunately, one thing that unites all critics (constructive or otherwise) of offender profiling is an acknowledgement of the lack of empirical research and the need to attempt to develop a sense of rigour throughout offender profiling practices (Salfati, 2000). This is inevitable if profiling itself is to develop, but is equally valid if forensic psychology should continue to be associated with profiling in any way. Through extensive hypothesis testing of the assumptions that informally underpin approaches to profiling, we should be able to move profiling beyond its depressingly introspective equation to “the Rorschach Test in the hands of the skilled technician” (Turco, 1990, p.150). That such empirical testing is possible is evident from some imaginative work to date (e.g. Kocsis, Irwin, Hayes, & Nunn, 2000). Given the infancy of forensic psychology in Ireland, confusion and false expectations in relation to the potential development of forensic psychology represent possible realities that neither the police nor psychologists can afford to entertain. We only do so through lack of openness, and the danger of any police force uncritically accepting psychological services are obvious in light of the Rachel Nickell case. As Ainsworth (2000) reminds us, although there is strong evidence to support the increasingly steady relationships between psychologists and the police, the expectations of police investigators of offender profiling, as well as the psychologists who engage in it, must be realistic, the process and its conclusions must be clear, intelligible, and perhaps most of all, open to scrutiny. Anything less than this would represent a disservice to both psychologists and the police. We will conclude with mention of an issue, identified by Canter (1989), but which may certainly be relevant to the Irish context. Canter asserts: “police officers are unlikely to admit psychologists to their investigations unless some mutual trust and reciprocal benefit is expected. This is a tricky cycle to break into, because it is difficult to make a contribution until some experience has been gained, yet difficult to gain experience until some contribution can be offered” (p.13). If offender profiling is ever to be considered in Ireland should the need arise, this fundamental challenge must be openly and successfully addressed by both the police and forensic psychologists. For the moment, however, the initial onus may well be on psychologists to demonstrate the relevance of their contributions to forensic settings. That this is already happening (e.g. Taylor & Quayle, 2003) represents encouraging progress.
16
REFERENCES Ainsworth, P.B. (2000). Psychology and crime: Myths and reality. London: Longman/Pearson. Aitken C., Connolly T., Gammerman A., Zhang G. & Oldfield D. (1995). Predicting an offender’s characteristics: An evaluation of statistical modelling. Police Research Group Special Interest Paper 4. London: Home Office. Aitken C., Gammerman A., Zhang G. Connolly T., Bailey D., Gordon R. & Oldfield D. (1996). Bayseian belief networks with an application in specific case analysis. In A. Gammerman (Ed.) Computational learning and probabilistic reasoning. Chichester: Wiley. Ault, R.L. & Reese, J. (1980). Profiling: A psychological assessment of crime. FBI Law Enforcement Bulletin, 49, 22-25. Bekerian, D.A. & Jackson, J.L. (1997). Critical issues in offender profiling, in J.L. Jackson & D.A. Bekerian (Eds.). Offender profiling: Theory, research and practice. Chichester: John Wiley. Brantingham, P.J. & Brantingham, P.L. (1981). Notes on the geometry of crime. In P.J. Brantingham & P.L. Brantingham (Eds). Environmental criminology. Beverly Hills: Sage. Canter, D. (1989). Offender profiles. The Psychologist, 2, 12-16. Canter, D. (1994). Criminal shadows. London: HarperCollins. Canter, D. & Heritage, R. (1990). A multivariate model of sexual offence behaviour. Journal of Forensic Psychiatry, 1, 185-212. Canter, D. & Larkin, P. (1993). The environmental range of serial rapists. Journal of Environmental Psychology, 13, 63-69. Copson, G. (1995). Coals to Newcastle? Police use of offender profiling. Police Research Group Special Interest Paper 7. London: Home Office. Davies A., Wittebrood K. & Jackson J.L. (1997). Predicting the criminal antecedents of a stranger rapist from his offence behaviour. Science and Justice, 37, 161-170. Davis, J.A. (1999). Criminal personality profiling and crime scene assessment: A contemporary investigative tool to assist law enforcement public safety. Journal of Contemporary Criminal Justice, 15, 291-301.
17
Godwin, M. (1998). Reliability, validity and utility of extant serial murderer classifications. The Criminologist, 22, 194-210. Grubin, D. & Gunn, J. (1990). The imprisoned rapist and rape. London: Institute of Psychiatry. Grubin D., Kelly P. & Brundson, C. (2001). Linking serious sexual assaults through behaviour. Home Office Research Study 215. London: Home Office. Guttman L. (1968). A general nonmetric technique for finding the smallest coordinate space for a configuration. Psychometrika, 33, 469-506. Heritage R. (1992). Facets of sexual assault: First steps in investigative classification. MPhil. Thesis. University of Surrey. Homant, R.J. & Kennedy, D.B. (1998). Psychological aspects of crime scene profiling. Criminal Justice and Behaviour, 25, 319-343. Howitt, D. (2002). Forensic and criminal psychology. London: Prentice Hall. Jackson, J., Van Hoppen, P.J. & Herbrink, J. (1993). Does the service meet the needs? Netherlands Institute for the Study of Criminality, mimeographed report. Keppel, R.D. & Walter, R. (1999). Profiling killers: A revised classification model for understanding sexual murder. International Journal of Offender Therapy and Comparative Criminology, 43, 417-437. Kocsis, R. (1995). Offender personality profiling: A viable investigative supplement. Bond University, mimeographed report. Kocsis, R.N., Irwin, H.I., Hayes, A.F. & Nunn, R. (2000). Expertise in psychological profiling: A comparative assessment. Journal of Interpersonal Violence, 15, 311-331. LeBeau, J.L. (1987). The journey to rape: Geographic distance and the rapist’s method of approaching the victim. Journal of Police Science and Administration, 15, 129-161. Michaud, S.G. (October 26 1986). The FBI’s new psyche squad. New York Times Magazine, 1-5. Perkins, D.E., Hilton, M. & Lucas M. (1990). A study of serial offenders in special hospital and prison. Report of the Home Office Science and Technology Group. London: Home Office. Pinizzotta, A.J. (1984). Forensic psychology: Criminal personality profiling – An outcome and process study. Law and Human Behaviour, 14, 215-233.
18
Porter, B. (April 1983). Mind hunters. Psychology Today, 1-8. Ressler, R.K., Burgess, A.W. & Douglas, J.E. (1988). Criminal homicide: Patterns and motives. Lexington, MA: Lexington. Rossmo, K. (1995). Place, space and police investigations: Hunting serial violent criminals. In J. Eck & D. Weisbard (Eds.) Crime and place: Crime prevention studies Vol. 4. New York: Mouncey. Rossmo, K. (1997). Geographic profiling. In J.L. Jackson & D.A. Bekerian (eds). Offender profiling: Theory, research and practice. Chichester: John Wiley. Salfati, C.G. (2000). The nature of expressiveness and instrumentality in homicide: Implications for offender profiling. Homicide Studies, 4, 265-293. Silke, A. (2001). Chasing ghosts: Offender profiling and terrorism. In D.P. Farrington, C.R. Hollin & M. McMurran (Eds.) Sex and violence: The psychology of crime and risk assessment. London: Routledge. Smith V.O. (1990). Statistical profiling (CATCHEM): Report of confirmatory analysis. Report of the Home Office Scientific Research and Development Branch. London: Home Office. Taylor, M. & Quayle, E. (2003). Child pornography: An internet Crime. London: Routledge. Turco, R.N. (1990). Psychological profiling. International Journal of Offender Therapy and Comparative Criminology, 34, 147-154. Turvey, B. (1999). Criminal profiling: An introduction to behavioral evidence analysis. London: Academic Press. Wilson, P. & Soothill, K. (1996). Psychological profiling: Red, green or amber? The Police Journal, 69, 12-20. Woodhams, J. (2002). Conference report – Offender profiling: Current practices and future directions. Forensic Update, 68, 34-37. (This lecture was delivered at the UCC Science Faculty Public Lecture Series 2002-2003, on February 11, 2002.)
19