3. Behavioural profiling

3.1 Background and definition

Several authors (e.g. Ainsworth, 2001) have pointed out the public’s general misconception of the work of profilers. This misconception is supported partly by various television series that give profiling a mythical aura. Even within the rather small circle of profilers there are two main schools of thought: one that considers criminal profiling an art, and the other that considers it a science. Those in the first group place more emphasis on their skills and intuition and maintain that few people have the essential qualifications (Agrapart-Delmas, 2001). Those who consider criminal profiling a science believe, on the contrary, that any person with access to adequate techniques can help conduct a successful investigation (Ainsworth, 2001). Hicks and Sales (2006) mention, however, that most profilers are not prepared to reveal their methods, for fear of being criticized or copied.

The very idea of profiling came about from works of fiction. The first "profiler " reviewed is Dupin in Edgar Allan Poe’s The Murders in the Rue Morgue, published in 1814. The first case of profiling in a real investigation is the analysis voluntarily provided by Thomas Bond, who participated in the autopsy of Mary Jane Kelly in the case of Jack the Ripper in England in 1888. Since the criminal was never identified, the accuracy of the profile cannot be evaluated. More recently, the U.S. army called on psychoanalyst Walter Langer to trace a psychological profile of Hitler in 1943. This profile had correctly predicted, among other things, that Hitler preferred suicide to being captured. Personality tests were also administered to American soldiers, but the results of this research remain secret (Hicks and Sales, 2006).

The first request for profiling expertise by law enforcement agencies dates back to 1956 when New York psychiatrist James A. Brussels was called on to provide a profile of the Mad Bomber (George Metesky). The profile provided by Brussels, which proved to be impressively accurate (down to the clothing worn by the suspect at the time of his arrest), contributed largely to the subsequent popularity of this approach. The FBI began to incorporate profiling into its investigation practices in the early sixties, although the Behavioural Analysis Unit was not founded at Quantico until 1978 (Egger, 1999). In England, interest in profiling soared in the mid eighties, viewed from a psychoanalytical and clinical psychological perspective, after David Canter contributed to the investigation leading to the arrest of the " Railway Rapists," John Duffy and David Mulcahy (Hicks and Sales, 2006; McGrath, 2000). It was following this experience that Canter decided to establish the first university program in investigative psychology at the University of Liverpool in 1994 (Egger, 1999).

Despite the internal disagreements especially relating to profiling methods, it is generally agreed that [translation] "criminal profiling may be defined as a technique promoting the identification of a suspect’s principal characteristics of personality and behaviour, based on the elements of the crime he has committed" (Beauregard and Proulx, 2001, p. 20). According to the authors, criminal profiling may also be called psychological profiling, personality profiling, criminal investigative analysis (FBI/ ICIAF) or investigative psychology (D. Canter) (Copson, 1995). Although this definition is based primarily on criminal profiling activities, the methods and approaches used are highly variable and include, broadly speaking, analysis based on experience and intuition, the clinical approach and the statistical or actuarial approach (Snook, Eastwood, Gendreau, Goggin and Cullen, 2007).

3.2 The practice of behavioural profiling

In Canada, behavioural analysis units perform the following duties: develop profiles of unidentified offenders, analyze crime scenes, reconstruct crime scenes, conduct indirect personality assessments, provide advice on investigations or questioning, assist in the execution of search warrants, analyze statements or testimony, analyze suspicious deaths, conduct threat assessments and present services offered (Lines, 2008).

3.2.1 Who are the profilers?

In its beginnings—and still today in certain countries—there was very little oversight of criminal profiling. Copson (1995) pointed out that in England:

There is no governing body for the regulation of professional or ethical standards in offender profiling. Notwithstanding several postgraduate psychology courses which incorporate some study of it, there is no academic qualification for offender profiling, and there is very little academic literature which deals directly with either the principles or the validity of offender profiling (p. 1).

Traditionally, anyone could call himself a profiler. Consequently, this discipline was practised by "experts" from a range of disciplines (psychiatry, psychology, psychoanalysis, criminology, policing and so on), with or without criminal investigation experience. In North America, however, profilers employed by major police organizations (FBI, RCMP, OPP) are not recruited from among the self-proclaimed experts.

The training of profilers in North America was initially under the responsibility of the FBI Police Fellowship Program. Following the closure of this program, however, the International Criminal Investigative Analysis Fellowship (ICIAF) was created in 1992 and took on the responsibility of providing rigorous, standardized training for profilers, then called criminal investigative analysts (ICIAF, 2005).

Under ICIAF direction, not everyone can become an analyst. In order to apply for the training program, a candidate must meet the following requirements:

• Be a police officer in good standing;
• Possess a minimum three years’ recent experience in the investigation of interpersonal violent crime;
• Possess superior investigation skills, documented in writing, in the area of interpersonal violence;
• Possess a demonstrated ability to articulate thoughts both orally and in writing;
• Speak, write, understand and read English fluently;
• Be approved and sponsored by an ICIAF member in good standing;
• Be recommended in writing by the appropriate official of the agency employing the candidate;
• The agency employing the candidate must agree to cover all training costs;
• The agency employing the candidate must confirm in writing that the candidate will work primarily as an analyst for at least the final year of the training program and three years thereafter.

Once admitted to the roughly two-year program, the candidate must study or obtain training in the following areas: sex offenders and typologies, sexual homicide, legal pathology, crime scene reconstruction, homicide investigation, investigation into suspicious death, child abduction and abuse, interviews and interrogations, normal and abnormal behaviour (psychiatry and psychology), preparation of analyses, threat asses sment, arson and attempted bombings, as well as a professional development course for instructors. The candidate must also familiarize himself with media and public relations strategies, blood spatter analysis, computerized case association systems (ViCAP, ViCLAS), laboratory procedures for criminal analysis and scientific content analysis (SCAN) (ICIAF, 2005).

The candidate must also complete a minimum of six months of investigation work supervised by a member of the ICIAF or the FBI National Center for the Analysis of Violent Crimes (NCAVC), including at least two months of supervised work at NCAVC. At the end of the training, the candidate must pass an examination. The candidate is presented with a case and has thirty days to write up an analysis and prepare an oral defence before the members of an evaluation committee, whose decision must be unanimous. After one year as an associate member in good standing, an application for full fellow status may be submitted to the ICIAF (ICIAF, 2005). Canada currently has four analysts who are full fellows: two employed by the RCMP and two by the Ontario Provincial Police (OPP). Three candidates are currently registered in the training program (two employed by the OPP and one by the RCMP). The Sûreté du Québec employs two analysts, but their status is unknown.

3.2.2 The practice of behavioural profiling

Profiling is used most often in cases where police have few clues that could help solve a case and are not certain what type of person committed the crime. Thus, profiling has been used especially in rape and homicide investigations, particularly with serial crime (Ainsworth, 2001). Profiling would also be particularly indicated in serial crime, ritual crime and particular forms of psychopathological crime (Beauregard and Proulx, 2001). However, the goal of profiling is not to directly identify the person responsible for the crime, but rather to predict the most probable characteristics of the criminal (Douglas, Burgess, Burgess and Ressler, 2006). The practice of profiling is based on certain assumptions: 1) human behaviour is predictable (Latour, Van Allen, Lépine and Nezan, 2007), 2) offenders are consistent in the way they commit their crimes and may be distinguished from other offenders, and 3) the way they commit their crimes is related to their personal characteristics (Goodwill and Alison, 2007). However, certain social psychologists maintain that personality is not an effective predictor of action (Rossmo, 2000) and that the situational context must be taken into account (Bénézech and the Groupe d’Analyse Comportementale de la Gendarmerie Nationale Française [Behavioural Analysis Group of the French National Police], 2007; Homant and Kennedy, 1998).

Profile development is a demanding probabilistic operation requiring the availability of a vast amount of information (Geberth, 1996). In return, a correctly constructed profile can offer a variety of clues concerning, in particular, the criminal’s age, race, sex, socio-economic status, residence, means of transport, level of education, marital status, occupation, criminal and psychiatric history, social and sexual development, military background, physical characteristics, habits, level of organization, pre- and post-crime behaviour and the potential for accomplices (O’Toole, 1999).

As a general rule, when investigators are faced with a crime scene, they will look for three clues: the modus operandi, a signature, and whether or not there is any staging. Modus operandi refers to a set of learned behaviours developed and repeated by the offender in criminal activities because these behaviours proved fruitful in crime. This concept is dynamic and malleable and will evolve according to the offender’s experiences (Douglas, Burgess, Burgess and Kessler, 2006), although relative stability is observed for certain crimes, especially sex crimes (Sjöstedt, Långström, Sturidsson and Grann, 2004). The signature refers to criminal behaviours that go beyond the behaviour necessary to perpetrate a crime and will generally define the criminal’s personality (Douglas et al., 2006). Unlike the modus operandi, the signature will remain stable. The concept of staging is applied when the criminal wilfully alters the crime scene before the arrival of police. According to Douglas et al. (2006), there are two reasons for a person to use staging: in order to 1) distract the investigation from the most probable suspect, or 2) protect the victim or the victim’s family. When the crime is staged, the person responsible is normally someone associated or related to the victim in some way. All these clues will be used to construct a portrait of the potential suspect while allowing investigators to verify whether the crime under investigation could be related to other, similar crimes.

Criminal profiling is currently used in three phases of the criminal justice process: the investigation, the arrest and the trial (Hicks and Sales, 2006). Profiling is used during the investigation phase when traditional methods have failed. At this stage, profiling is especially used to connect serial crimes and identify the physical, psychological and other characteristics related to the criminal’s lifestyle. Profiling is also used during the investigation phase for the following reasons: to suggest certain pre- and post-crime behaviours the criminal is likely to exhibit; to evaluate the possibility that certain crimes evolve into more serious, violent crimes; and to suggest proactive tactics to encourage the criminal to reveal his identity. During the arrest phase, profiling is used to orient searches toward certain sectors or particular elements, to predict the criminal’s behaviour upon his arrest, or even suggest interrogation techniques that are likely to lead to confessions. Finally, during the trial phase, profiling provides the court with expertise that makes it possible to connect various crimes to a single individual and to relate the crime or crimes in question to the individual’s characteristics that were established in the development of the profile (Hicks and Sales, 2006). However, slip-ups observed in certain cases where profiling was used, as in the case of Guy Paul Morin in Canada (Kaufman, 1998) or Colin Stagg in England (Marin, 2003), as well as the lack of science in the methods render the use of profiling during the investigation phase generally difficult (see section 3.6 on admissibility in court).

Little information is available concerning the use of profiling services. Copson (1995) points out that in England, from 1990 to 1994, police services (48 agencies represented out of a total of 56) used a profiler 184 times. The crimes for which profiling was used were homicides (113), rape (40), extortion (12), other sex crimes (10), arson (4), abduction (3) and threats (2). The services most frequently requested were profiling (116) and assistance with comprehension of the crime (112). Requests for profiling services appear to be on the rise in several countries (Copson, 1995; Snook, Eastwood, Gendreau, Goggin and Cullen, 2007), which leads Kocsis (2006) to wonder, "Possibly the greatest mystery surrounding criminal profiling has been its growth despite an absence of robust scientific evidence to validate it" (p. 458).

In Canada, three agencies have a section dedicated to criminal profiling: the RCMP’s Special Services and Behavioural Sciences Branch, the Sûreté du Québec’s Behavioural Analysis Service, and the Ontario Provincial Police Behavioural Sciences Section (Marin, 2003). Criminal profiling services have been offered by the OPP free of charge to police forces in Canada since 1991. Since that time, the Behavioural Sciences Section has dealt with about 3,150 requests for service, while the RCMP responded to approximately 175 requests annually. Of these requests, about 15% were specific requests for a profile of the perpetrator of a crime. For example, in 1996, the OPP received 18 requests for behavioural profiling. In 1997, that figure rose to 33, and to 35 in 1998 (Van Allen, 2008). The services of these agencies were particularly required in cases of homicides, rape, sexual assault and child abuse (Lines, 2008).

In 2006, the Groupe de travail sur le traitement des crimes [Crime Treatment Working Group] prepared a portrait of the various approaches to profiling for a few European countries as well as the United States and Canada. We noted marked differences among the countries.

In the United States, criminal profiling is widely used at the federal level as well as in several states. However, it is reserved for the most serious crimes to facilitate the identification of the perpetrator or perpetrators. Because of the growing number of serial killers since the 1970s, the FBI decided to implement a specific information processing system for murders committed anywhere in the U.S. (Marin, 2003), which led to the implementation of the Violent Criminal Apprehension Program (ViCAP). This system compiles the specific characteristics of all the murders committed throughout the United States and has contributed to the arrest of several serial killers (Marin, 2003). The FBI also has a behavioural analysis service, the National Center for the Analysis of Violent Crime (NCVAC), specialized in the study of criminal behaviour available to all American investigators. The Center comprises four services: (1) the Behavioral Analysis Unit-1 (terrorism and threat analysis), (2) the Behavioral Analysis Unit-2 (crimes against adults), (3) the Behavioral Analysis Unit-3 (crimes against children), and (4) the Violent Criminal Apprehension Program (ViCAP). It should be noted that the FBI’s notion of profiling focuses on the deductive method based on the analysis of crime scenes, psychological analysis and the analysis of behavioural evidence provided by the investigation. In addition, in the U.S. as in Canada, geographic profiling is used to assist investigators in targeting a criminal’s preferred geographic area (National Center for the Analysis of Violent Crime, 2008). Marin (2003) points out that the small size of many European countries may explain why geographic profiling is seldom, if ever, used.

In Great Britain, investigators often use behavioural psychologists who attempt to establish a victim as well as a suspect profile. A criminal’s psychological profile is most often developed by several profilers who are experts accredited by the Behavioural Sciences subcommittee (police officers’ association). According to Marin (2003), it seems that in the particular case of England the police services are quite circumspect as to the usefulness of profiling. Moreover, Marin (2003) points out that some criticism has arisen in England against the potential abuse of profiling. The author cites the case of Collin Stagg, linked to the murder of Rachell Nickell.

In Belgium, profiling was introduced in 1996, but really gained popularity in 2001 when the Federal Judicial Police created the Behavioural Science Service (GWSC), which had a staff of 15 in 2008, including four behavioural analysts. The analysts are responsible for profile analysis, hearing support (interrogation), threat assessment, research, and training. In 2007, the service provided support in 133 cases, including 17 profile analyses in 62 cases (Godefroid, 2008).

In 2003, Switzerland adopted the ViCLAS system, which consists of a data bank that collects psychological prints and then prepares psychological profiles of murderers and attackers.

In the Netherlands, Spain, Germany, Italy and the Czech Republic, criminal profiling seems less developed (Serial Crime Working Group, 2006). In the Netherlands, profiling does not exist as such, although certain information related to criminals is gathered and analyzed, notably because of the ViCLAS program. In Spain, it seems that profiling has not extended beyond the confines of universities and symposia, even though Marin (2003) notes several data banks that draw connections in criminal investigations. In Germany, some profiling is done by police departments that develop behavioural profiles with the help of joint teams of police officers and psychologists. However, Germany’s federal structure makes it difficult to quantify the success rate of profiling in that country (Marin, 2003). In Italy, the behavioural analysis service created by the Ministry of the Interior in the forensics unit (Violent Crime Analysis Unit) includes a team of police officers, psychologists and anthropologists who assist in analyzing crime scenes, extrapolating repetitions, developing typical victim profiles and defining the behavioural elements that could help in the search for the criminal. The Czech experience in the field of profiling still seems limited. However, according to Marin (2003), it appears that the Canadian ViCLAS system is gradually being implemented. Its field of application must cover voluntary homicides, assassinations and gross indecency. Loading the database involves 1,000 old cases and an annual volume of 700 cases. There seems to be little information on the profiling methods used in countries other than those mentioned in this section.

3.3 Theoretical framework

Crimes subject to profiling are normally part of a series of similar crimes (Cook and Hinman, 1999). In order to produce a reasonably accurate offender profile, investigators gather clues from the crime scene. According to O’Toole (1999), several behavioural characteristics may be extracted from where the crime took place: 1) the degree of planning involved, 2) the degree of control used by the offender, 3) the escalation of emotions at the crime scene, 4) the level of threat to the victim and the offender and 5) the appearance of the crime scene (organized vs. disorganized). In fact, the premise of profiling is that the more investigators know about the victim, the more they will know about the offender (O’Toole, 1999).

There are several models and typologies for the practice of profiling. Homicide and rape are the crimes most often analyzed, and most of the models are derived from analyses of these crimes. Moreover, "The models […] are labelled non-scientific because, although they may refer to scientific principles to varying degrees, each model relies implicitly or explicitly on an artful component to complete an offender profile" (Hicks and Sales 2006, p. 17). The following sections present the best-known typologies, classified according to the crimes profiled.

3.3.1 Homicides

3.3.1.1 Douglas, Ressler, Burgess and Hartman typology

This first model, which originated in the FBI, makes a distinction between organized and disorganized crime scenes. According to Davis (1999), an organized crime scene indicates planning, premeditation and a conscious effort on the part of the criminal to avoid being caught. A disorganized crime scene indicates spontaneous action and a generally violent assault. The victim is usually chosen at random and the location of the crime is generally the place where the victim and offender met (Davis, 1999). Although several crime scenes may involve both types, the offender classification will be based on the level of organization or disorganization considered predominant at the crime scene. On the basis of this classification, several conclusions have been drawn with respect to the characteristics of the individual (level of intelligence, employment, social adjustment, etc.) (McGrath, 2000).

3.3.1.2 Fesbach’s Typology

A second model, described by Salfati (2000; Salfati and Canter, 1999; Salfati and Park, 2007) but inspired by Fesbach, is based on the function that the homicide serves for the offender. The model distinguishes two types of aggression, expressive and instrumental, characterized by the goals or rewards they offer the offender. Expressive aggression occurs in response to anger provoked by insult, personal attack, humiliation or failure. The goal of this aggression is to punish the victim and make him suffer. The instrumental function comes from a desire for possession or status, such as, for example, to obtain jewellery, money, territory or power. For the instrumental offender, murder is not an end in itself, but may occur if someone comes between him and the achievement of his goal. Attempts to link this typology to elements of crime scenes and personality traits of the offender involve significant methodological weaknesses (e.g. Salfati and Canter, 1999; Salfati and Park, 2007).

3.3.1.3 Typology of Holmes and Holmes

Holmes and Holmes propose a serial killer typology with four categories based on what motivates the individual: visionary, missionary, lust killer and power seeker (Hicks and Sales, 2006).

3.3.2 Sex crimes

3.3.2.1 Typology of Keppel and Walter

An initial typology for sexual homicide comes from an existing typology for rape (Keppel and Walter, 1999). This typology describes the crime according to its function for the offender. Four types of functions may be advanced to describe rape and rape followed by homicide, namely power-assertive, power-reassurance, anger-retaliatory and anger-excitation. The power-assertive offender commits a crime of power where the rape was planned but murder is an unplanned response to an escalation of violence in order to control the victim. The power-reassurance offender also engages in a planned rape in which homicide was not planned. In this type of crime, the offender attempts to demonstrate his sexual competence through seduction. However, when the victim does not cooperate in the offender’s scenarios and fantasies, a feeling of failure and panic pushes the offender to commit the murder. In the case of the anger-retaliatory offender, the rape and homicide are planned. In this type of crime, motivated by anger, the offender tries to avenge himself of a person by attacking a symbolic victim. Finally, for the anger-excitation offender, the goal of the planned rape and homicide is gratification by infliction of pain and terror on the victim. Prolonged torture of the victim will feed the offender’s fantasies and temporarily satisfy his need for domination and control (Keppel and Walter, 1999).

3.3.2.2 Typology of Hazelwood and Warren

This typology comes from Hazelwood and Warren (2000) and describes the offender as impulsive or ritualistic in his actions. According to the authors, the impulsive offender is a common type of sexual offender who generally has little success avoiding identification and apprehension. In fact, this type of criminal acts impulsively, takes little or no action to protect his identity and is apparently unaware of the risks associated with committing a crime. The ritualistic offender is much less common, and unlike the impulsive offender, he is much more successful in his actions and becomes very difficult to identify and apprehend. This type of criminal devotes considerable time and effort to planning and repeating his criminal actions. The validity of this typology has notably received the support of the study by Warren, Reboussin, Hazelwood and Wright (1991).

3.3.2.3 Typology of Beauregard and Rossmo

In a series of articles, Beauregard and Rossmo (Beauregard, Proulx, Rossmo, Leclerc and Allaire, 2007; Beauregard and Rossmo, 2007; Beauregard, Rossmo and Proulx, 2007) attempted to elucidate the hunting scenarios of serial sex offenders. These scenarios were based on four hunting patterns and three attack methods:

• Hunting patterns:

1. Hunter: specifically looks for victims from his place of residence;
2. Poacher: specifically looks for victims based on an area of activity other than his place of residence or travels to another city during the search process;
3. Troller: meets a victim in an opportunistic manner while taking part in non-predatory activities;
4. Trapper: has a certain position or occupation or creates a situation that allows him to meet potential victims in the area he is covering as part of this occupation.

• Attack methods:

1. Raptor: attacks a victim upon encounter;
2. Stalker: first follows a victim upon encounter and then attacks her;
3. Ambusher: attacks a victim only after she has been enticed to a location controlled by the offender.

Empirical studies based on this typology have made it possible to identify three scenarios with a total of five variations, based on the observed combinations of hunting patterns and attack methods (Beauregard, Proulx, Rossmo, Leclerc and Allaire, 2007; Beauregard and Rossmo, 2007; Beauregard, Rossmo and Proulx, 2007):

1. Coercive scenario

1.1 Intrusive
1.2 Outdoor

2. Manipulative scenario

2.1 With sophistication (trick)
2.2 By infiltration

3. Non-persuasive scenario: direct action

3.3.2.4 Typology of Blanchette, St-Yves and Proulx

Blanchette, St-Yves and Proulx (2007) have proposed a typology for rapists and pedophiles. In both cases, there are three types: festive, organized and isolated. Their empirical research suggests the following characteristics:

Typology of a rapist

1. The festive rapist:

• resembles an offender in general: antisocial;
• interpersonal relationships marked by hostility, lack of empathy and immediate satisfaction of needs;
• sensitive to influences extolling the domination of men over women;
• regular contact with immediate family (87%), good hygiene (82%), single (82.2%), consumes alcohol regularly (82.6%) and has at least one close friend (95.7%);
• consumes alcohol prior to the offence (78.3%), coercive approach to committing the offence (87%) and does not masturbate the victim (0%).

2. The orderly rapist:

• motor vehicle in excellent condition (85.7%), good hygiene (100%), lives with someone (87.5%), owns his own home (75%), owns a motor vehicle (87.5%), is not single (0%) and does not regularly frequent erotic bars (0%);
• does not consume pornographic material prior to committing the offence (0%), premeditation (87.5%), no anal penetration (0%), cunnilingus (0%) or masturbation of the victim (0%), coercive approach in committing the offence (87.5%), does not mutilate the victim (0%).

3. The isolated rapist:

• non-existent social life;
• does not frequent restaurants (0%), motor vehicle in excellent condition (75%), good hygiene (100%), does not own his own home (0%), does not frequent bars (0%), single (83.3%), has a job (83.3%), has at least one close friend (83.3%) ;
• does not consume pornographic material prior to committing the offence (0%), premeditation (83,3%), vaginal penetration (100%) but not anal (0%), asks for fellatio (83.3%), coercive approach (100%), no mutilation (0%), does not attack handicapped victims (0%), does not come from a poor or dysfunctional environment (0%), no deviant sexual fantasies prior to committing the offence (0%), forces the victim to perform sexual acts (83.3%) and genital touching (83,3%).

Typology of a pedophile

1. The festive pedophile:

• resembles a general offender: antisocial;
• regular contact with immediate family (85.2%), involved in sports (79%), good hygiene (82.8%), lives with someone (75.9%), regularly consumes alcohol (86.2%) and drugs (82.8%);
• member of the victim’s family (75.9%).

2. The organized pedophile:

• regular contact with immediate family (88.2%), good hygiene (76.5%), lives with someone (88.2%), owns his own home (100%), owns a motor vehicle (100%);
• does not consume medication/solvents prior to committing the offence (0%), the victim is a close acquaintance (76.5%), member of the victim’s family (88.2%), premeditation (94.1%), the victim is not under the effects of a drug or alcohol (0%).

3. The isolated pedophile

• non-existent social life;
• regular contact with immediate family (90.5%), single (90.9%);
• does not consume medication/solvents prior to committing the offence (0%), premeditation (90.9%).

3.3.2.5 Typology of Fortin and Roy

Fortin and Roy (2007) are interested in consumers and producers of child pornography on the Internet. They describe three types of users:

1. recreational (satisfaction of sexual curiosity);
2. sexually compulsive (unconventional sexual behaviours: consumers of pornography, sexual relations with several partners, use of prostitutes’ services, and so on);
3. the at-risk user (no previous history of compulsive sexuality, but online habits have a depressive or reactionary impact on his life).

These authors also identify four types of collectors:

1. Secret (secret, commercial sources, no previous history);
2. Isolated (sexual aggressor);
3. Family (seeks validation of his behaviour);
4. Commercial (profit seeker, also aggressor).

3.3.2.6 Typology of Holmes and Holmes

Holmes and Holmes essentially return to the typology of Keppel and Walter (1999), with a few nuances: 1) power reassurance, 2) anger retaliation, 3) anger exploitive and 4) sadistic (Hicks and Sales, 2006).

They also propose a typology for child molesters, which they divide into situational molesters (regressed pedophile; morally indiscriminate; sexually indiscriminate; or naive/inadequate) and preferential pedophiles (sadistic pedophile; seductive molester or fixated molester) (Hicks and Sales, 2006).

3.3.3 Arson

3.3.3.1 Typology of Rider

According to Rider, the arsonist may be motivated by jealousy, a need for recognition, a search for strong sensations, or compulsion (pyromania) (Hicks and Sales, 2006).

3.3.3.2 Typology of Douglas, Burgess, Burgess and Ressler

Douglas, Burgess, Burgess and Ressler (2006) present a range of motivations that is both broader and more precise than Rider: 1) revenge, 2) excitement (strong sensations, attention, recognition or sexual excitement), 3) vandalism, 4) concealment of a crime and 5) profit. Note that Holmes and Holmes also submit a typology of the arsonist that is in fact a collage of the characteristics identified by Rider and Douglas, Burgess, Burgess and Ressler (Hicks and Sales, 2006).

3.3.3.3 Typology of Fritzon, Canter and Wilton

The article by Fritzon, Canter and Wilton (2001) describes a classification system based on four models of action (expressive, integrative, conservative and adaptive). The adaptive mode refers to cases of vandalism where the person takes advantage of the opportunity present to commit his crime. The target is less important than the desire to change. The expressive mode refers to the external manifestation of an internal process. This form of pyromania suggests that likely targets are those that will allow the individual to get attention (for example a hospital or a large building). The integrative mode refers to the act of setting fire to oneself or surrounding objects in what appears to be suicide. This act results from a state of personal distress and is directed toward himself. Finally, the conservative mode generally results from a need to avenge someone or something.

3.3.4 Terrorism

3.3.4.1 Hacker’s typology

Hacker is the first to propose a typology for terrorism. He distinguishes three types: 1) the crusader, ideologically motivated and generally responsible for recruitment and planning; 2) the criminal, a violent individual seeking a pretext, who carries out his acts without becoming ideologically involved, and 3) the crazy, mentally vulnerable or disturbed, attracted by philosophical certainty (Miller, 2006).

3.3.4.2 Strentz’s typology

Strentz’s typology also includes three categories: 1) the leader, egocentric, paranoid and charismatic; 2) the activist, antisocial or psychopathic, sometimes a former inmate or mercenary, not truly engaged ideologically, and 3) the idealist, devoted to building a "better world," hopeless and dependent (Miller, 2006). This typology has certain similarities with Hacker’s.

3.3.4.3 U.S. Secret Service typology

The U.S. Secret Service distinguishes between five types of terrorists: 1) the crusader, ideologically motivated by his political or religious convictions; 2) the ultraconservative political terrorist, who believes in individual rights in a regime seen as repressive, is active in a quasi-military organization that is ultraconservative, authoritarian and extremist; 3) the political anarchist is an ultra-left-wing activist who sees the government as racist, elitist and economically oppressive; 4) the religious terrorist, who does not answer to anyone except God and who kills in his name; and 5) the criminal terrorist who is more opportunist than idealist and acts for his own benefit (Miller, 2006).

3.3.4.4 Miller’s typology

Finally, Miller (2006) proposes his own typology, which associates the type of terrorist with psychopathological characteristics: 1) the leader (narcissism and paranoia), 2) the believer (antisocial and borderline), 3) the soldier (avoidance and dependence), and 4) the public danger (histrionic and schizoid).

There are several other typologies and theoretical approaches, including Turco’s psychoanalytical approach (based on borderline, narcissist disorders), Turvey’s inductive approach and Canter’s psychodynamic model (Hicks and Sales, 2006). Levi-Minzi and Shields (2007), as well as Salfati, James and Ferguson (2008), have attempted to develop a profile of the serial prostitute killer.

3.4 Criticism of existing models

Hicks and Sales (2006) engaged in a systematic criticism of theoretical models intended to guide the practice of behavioural profiling. These authors pointed out the lack of standards and empirical studies aimed at evaluating the effectiveness, reliability, validity and usefulness of models and typologies. In addition, they note some confusion attributable to a terminology that is sometimes ambiguous and generally variable from one model to another. The models themselves combine general approaches to typologies and taxonomies that are sometimes intrinsically inconsistent or have significant category overlap. Finally, significant methodological shortcomings often cast doubt on the validity of the studies claiming to support these models.

It seems that, despite the claims of certain authors, notably Canter, these models cannot claim to be "scientific." In general, it is agreed that the current practice of profiling depends on the profiler’s intuition and professional experience. The procedures for using these models, making decisions when faced with ambiguous data and developing profiles also remain somewhat inexplicit. Lastly, these approaches do not all culminate in observable characteristics that could assist investigators.

In short, Hicks and Sales (2006) point out that "None of the models has provided any evidence that profiling, as currently practised, has any substantial investigative value" (p. 65). Few models have any empirical support whatsoever.

3.5 Empirical support

It would seem relevant to point out from the beginning that it is [translation] " …  difficult to measure the effectiveness of criminal profiling in crime solving. Most of the time, the success of an investigation depends on all the factors as a whole" (Latour, Van Allen, Lépine and Nezan 20 07, p. 529). In addition, as Hicks and Sales (2006) point out, "…because F.B.I. profilers are explicitly trained not to put profiles in writing, the extent to which profiles can be systematically or scientifically reviewed is also limited" (p. 121). Conducting rigorous empirical studies to support or refute the effectiveness of profiling would therefore pose a challenge.

3.5.1 General evaluations

One of the only large-scale studies directly related to the effectiveness of criminal profiling was conducted by Copson (1995) in England. However, this study is almost fifteen years old and was conducted when Canter founded his Investigative Psychology program in Liverpool. Copson first covered four previous studies. The FBI investigation conducted by Douglas in 1981 (unpublished) revealed that profiling would have helped target the investigation in 77% of the cases where the accused was identified and would have allowed identification in 15 cases (8%). According to the FBI investigators, profiling would have saved a full-time investigator the equivalent of 594 days, and they unanimously supported maintaining the service.

In 1992, Britton conducted similar research in England for the Home Office (unpublished). Copson (1995) noted that, using the contribution of profiling to the arrest of suspects as a criterion for determining its effectiveness, "Judged on this stark criterion, and contrary to popular perception, little evidence was offered that profiles were either accurate or had contributed to any arrest"  (p. 6). Nevertheless, profiling was deemed viable and has been continued in England.

The same year, a student (Goldblatt) wrote a thesis (unpublished) on David Canter’s profiling program at Surrey University (precursor of the Liverpool program). Based on the information provided by Canter himself, out of 57 profiles, one suspect had been accused in "at least twelve cases," but it was difficult to judge the exact contribution of the profile. The analysis of twelve solved cases suggested that of 114 pieces of information included in the profiles, 72% were correct, 19% incorrect and 9% undetermined. Note that in both cases, the profile was produced after the suspect’s arrest. Despite everything, the program was considered a success.

Finally, the Dutch Consumer Satisfaction Survey was conducted by Jackson and his colleagues in 1993 with the Scientific Research Advisory Unit in the Netherlands. This survey covered twenty cases over a two-year period. Although the opinion of the profilers constituted a formal profile in only six cases, the majority of the detectives considered it satisfactory. Copson (1995) concluded from an examination of these four investigations that "What each of the four previous studies has in common is that they all rely to a large extent on the opinions of detectives who have used offender profiling advice in live investigations. […] None of the four, however, can be regarded as definitive" (p. 7).

Copson then presented the results of his own investigation conducted in England, which involved 48 of 56 police forces in 184 profiling cases. He noted that although the profilers’ opinions helped solve a criminal case in only 14.1% of the cases, 82.6% of the investigators stated that profiling proved operationally useful. According to the respondents, profiling led to the identification of the offender in only 2.7% of the cases, allowed for a better understanding of the crime or criminal in 60.9% cases, confirmed the investigator’s judgment in 51.6% of the cases, and helped structure the interrogation in 5.4% of the cases. In 2.3% of the cases, profiling served other purposes and proved to be useless 17.4% of the time.

This research demonstrated that the main variable affecting investigators’ perception of the usefulness of profiling was the identity of the profiler himself, which led Copson (1995) to state that, "It appears that, at this stage of the development of profiling in Britain, approaches to profiling are so idiosyncratic as to be indivisible from the identity of the profiler"  (p. 29). In addition, it seems that few investigators acted directly on the advice of profilers. Given the small contribution of profiling to the identification of offenders, Copson (1995) concluded that it would not be of added value to conventional methods of investigation: "If […] profiling is to be judged valid on its own terms then its success ought to be based upon telling officers something of the type of person who has committed the offence under consideration, so that the conduct of the investigation – and its outcome- might be influenced by advice based on those inferences. The respondents in this survey perceive that profiling does not succeed on those terms" (p. 31).

More recently, Snook, Eastwood, Gendreau, Goggin and Cullen (2007) conducted a meta-analysis of empirical research on the effectiveness of profiling. Only four studies were analyzed, and although the profilers seemed to have done marginally better than other groups, the authors concluded that "The evidence generated from this research confirms the perceptions of those who have concluded that the criminal profiling field relies on weak standards of proof and that profilers do not decisively outperform other groups when predicting the characteristics of an unknown criminal." (p. 448) However, two important points must be made. First, the profilers involved in the analyzed studies were self-proclaimed profilers, in the sense that they did not necessarily have formal training in the field and were not members of the ICIAF. Next, the meta-analysis largely covered the work of Richard Kocsis and his team, work that was harshly criticized because of major methodological deficiencies.

Bennell, Jones, Taylor and Snook (2006) conducted an analysis of Kocsis’ work and expressed a number of concerns, particularly with respect to the subjectivity of the measures used and the classification of several groups, including psychics, and one used as the control group. They noted no significant difference between the performance of the self-proclaimed profilers and the other groups of professionals as individuals. In addition, although Kocsis compared the accuracy of the profiles produced by various groups of individuals, he did not focus on the accuracy of the profile itself. Thus, although the profilers sometimes produced better results than comparison groups in a normative way, the absolute accuracy of the profile was often not very impressive. Finally, Bennell, Jones, Taylor and Snook (2006) pointed out the extremely small size of the profiler samples (three to eleven persons in all the studies combined), which made any statistical analysis doubtful, and the fact that the suggested task for evaluating profiler performance was not suitable for the profiling work performed by police. Kocsis (2006) responded to these criticisms, albeit not very convincingly.

3.5.2 Homicides

Kocsis, Middledorp and Try (2005) attempted to compare the capacities of various groups, including a group of five self-proclaimed profilers, to provide a file in a homicide case. The methodology used produced results with weak credibility. This article lends no serious support to the hypothesis that profiling is effective in homicide cases.

3.5.3 Sexual assault

In their study, Goodwill and Alison (2007) were interested in predicting the aggressor’s age based on the victim’s age in stranger rape. The relationship between the ages of the individuals involved appears to be influenced by the degree of planning and aggressiveness in the perpetration of the crime. They concluded that " …in cases where there is evidence of planning and offender over-aggressiveness the offender’s age can be predicted by the victim’s age within less than 3 years" (p. 833). However, where there is nothing to suggest that a process was used to select and target the victim and plan the crime, it would be extremely difficult to predict the age of the aggressor.

Certain researchers promote the use of databases and software that link previously documented criminals to new crime scenes based on relative similarities of the modus operandi or signature. Yokota, Fujita, Watanabe, Yoshimoto and Wachi (2007) applied similar software to the profiling of sex offenders in Japan. This approach directly identified the aggressor 24 out of 81 times, meaning that the aggressor was ranked as the first possible suspect 29.6% of the time. Overall, the offenders were ranked fourth on average. However, by limiting candidates to those residing in the prefect where the crime was committed, the success rate (offender ranked first) rose to 55.6%. This approach proved promising in cases where crimes were committed by repeat offenders and supports the hypothesis of a certain consistency in behaviour when an individual commits several crimes. The success of such method is, however, based on the size and quality of the database.

3.5.4 Arson

In one of his articles, Kocsis (2004) attempted to see whether there was any difference in profiling capacity among several groups for a series of arson incidents. However, the methodology was questionable and the results could not be considered credible. Overall, the results show that self-proclaimed profilers tended to have somewhat better results than non-practitioners of profiling, but these results were not significantly superior to a group taken individually, such as science students. Moreover, in addition to homicides, the study by Kocsis, Middledorp and Try (2005) also involved a profiling task in an arson case. The aforementioned concerns remain, and support for the effectiveness of profiling is also questionable in arson cases.

3.5.5 Burglary

Oatley, Ewart and Zeleznikow (2006) applied automated methods to a large profiling database for burglars based on crime scene characteristics. An initial approach, based on association rules, classification rules and decision trees, did not provide the anticipated results: "The data mining technologies of association rules and decision trees/classification rules did not produce operationally useful knowledge. The association rules we derived were of poor quality, either indicating no significant relationships or relationships that are too complex to determine using this method" (p. 73). However, by associating new crimes with spatiotemporal and behavioural data, the matching of crimes to criminals proved to be accurate 24% of the time. In 59% of the trials, the offender was among the ten most probable suspects identified by the software, 77% were among the 30 most probable and 94% were among the fifty most probable. These researchers also attempted to identify clues as to whether a crime scene would be revisited by the burglar within the same year. It seems that the research method, type of stolen property, method of entry and use of trickery would allow for this distinction, but the authors do not specify the success rate.

3.6 Admissibility in court

3.6.1 Admissibility criteria

The admissibility criteria for expert testimony in court, including those related to new scientific techniques or theories, have evolved since the early 20th century. These criteria are used as a guide for judges who must determine the admissibility of testimony referring to various aspects of criminal profiling.

In the United States, several states still rely on Frye v. US (1923), where the District Court of Columbia examined the admissibility of polygraph evidence. The court barred this evidence and ruled that:

Just when a scientific principle or discovery crosses the line between the experimental and demonstrable stages is difficult to define. Somewhere in this twilight zone the evidential force of the principle must be recognized, and while the courts will go a long way in admitting experimental testimony deduced from a well-recognized scientific principle or discovery, the thing from which the deduction is made must be sufficiently established to have gained general acceptance in the particular field in which it belongs.

Thus, following Frye, the criterion for admissibility became acceptance of the technique or theory in question by the scientific community from the relevant discipline. However, there were several criticisms of this decision, such as that simple acceptance by the scientific community could not possibly constitute a sufficiently strict criterion (as the judge stated, it was already generally agreed that the earth was flat).

Since 1993, several U.S. states have adopted the standards set out by the U.S. Supreme Court in Daubert v. Merrell Dow Pharmaceuticals (1993), which supposes that:

1. The theory can be and has been tested;
2. The theory has been subjected to peer review;
3. The theory is reliable and its error rate is known;
4. The theory is generally accepted by the scientific community.

The publication of results in refereed reviews would provide some guarantee that the theory is not completely erroneous. If perfect reliability is not required, it is up to the court to judge the acceptable level. To do this, the probability of error for the proposed theory or technique must be known and presented in court.

In Canada, the admissibility of expert testimony depends on four criteria, established by the Supreme Court of Canada in R. v. Mohan (1994):

1. relevance (including the relationship between the case heard and scientific validity)
2. necessity (the testimony is necessary for the judge or jury to understand all the elements of evidence);
3. the absence of any exclusionary rule;
4. a properly qualified expert.

The Ontario Court of Appeal, in R. v. Clark (2004), adopted similar criteria:

1. necessity (the testimony goes beyond the normal scope of knowledge and experience of the average jury);
2. reliability (the testimony is anchored to the facts and is not limited to speculation);
3. The testimony is not immeasurably impressive (to prevent the jury from giving it more weight than is appropriate, which is liable to transform the trial by peers into a debate by experts).

The criteria of scientific validity mentioned in Mohan were specified following the Supreme Court of Canada ruling in R. v. J.-L. J. (2000), which refers to essentially the same standards as those adopted in the United States following Daubert (cf page 32).

3.6.2 Profiling before the courts

The variety of tasks accomplished by profilers has an impact on the type of evidence or testimony they are called upon to present in court. The courts’ position seems to depend in part on the type of evidence brought forward.

Few testimonies related to profiling seem to have been admissible in Canada since Mohan. In general, a distinction should be made between two types of expertise: 1) crime scene analyses and 2) behavioural analyses.

3.6.2.1 Crime scene analyses

Crime scene analyses include inferences based on observation of the state of the crime scene and the victim. It is generally a matter of reconstructing the crime as it occurred and, sometimes, evaluating whether the crime scene was staged to launch the investigation on the wrong track.

In R. v. Ranger (2003), the Ontario Court of Appeal noted with respect to testimony that a crime scene had been staged:

As noted by the Crown at trial, the fact that the crime scene may have been staged to look as if the house had been burglared (sic.) is a piece of circumstantial evidence that may provide some insight into the perpetrator’s motivation and, in turn, his or her identity. No issue has been raised with respect to the relevance of this aspect of the expert evidence. Similarly, no real issue is raised with respect to the reliability of the evidence on this narrow point. The reliability of any opinion that a crime scene was staged would be very much a function of the particular witness’s experience with scenes of break and enter (p. 14).

This point of view was also upheld by the Ontario Court of Appeal in R. v. Clark (2004), while reconstruction by a well-qualified expert and the demonstration that the crime scene had been staged (admitted as evidence in this particular case) were qualified as potentially admissible. This opinion is also shared in the U.S. (see notably US v. Meeks, 1992).

As a result of the development in the Klymchuk case (2005, 2008), there could be restrictions placed on the admissibility of this type of evidence in Canada. In 2005, the Ontario Court of Appeal decision based on both Ranger and Clark mentions the "established reliability" of the evidence based on the observation and reconstruction of crime scenes. However, since the court decided that the expert testimony had gone far beyond this area, it had been declared inadmissible. In 2008, the Ontario Superior Court ruled that another expert’s testimony regarding the fact that the crime scene had been staged was inadmissible, based on two arguments: 1) the expert presented by the Crown had participated in the original investigation and developed a profile of the person who had committed the crime; the conclusion was that it would be impossible for him to detach himself entirely from this mode of reasoning if he limited himself to the observations made on the crime scene to support his testimony, and 2) he failed to convince the court that the criteria of necessity stated in Clark was achieved. For the court, the description of the crime scene and its comparison with normal burglary could have been done by the police officers who led the investigation, and this would have been sufficient for the jury to draw the necessary conclusions. In other words, the knowledge required to understand the facts presented as evidence did not go beyond the normal scope of the average jury, and consequently, expert testimony was not necessary.

3.6.2.2 Behavioural analyses

The objective of crime scene analysis as proposed by Ormerod (1996) is to determine what happened (the "what"), while behavioural analysis, or what the courts consider criminal profiling, attempts to identify the motive for the crime (the "why") or trace a profile of the person likely to have committed it (the "who") –or not to have committed it, according to the rule of exception stated in Mohan (who stipulates that where a crime or its perpetrator has particularly distinctive characteristics, an expert can testify that the accused does not have these characteristics). Behavioural analysis includes the attribution of certain crimes to the same person based on an examination of the elements of the crime constituting the criminal’s modus operandi or signature (linkage analysis).

To our knowledge, this type of testimony has always been considered inadmissible in Canada, as the Ontario Court of Appeal explained in Ranger:

…attempts to adduce expert opinion evidence about WHY an offence was committed in a particular manner and, more particularly, about WHO is more likely to have committed the offence, that is, the kinds of evidence that I have labelled more particularly as criminal profiling, have generally not met with success, either in this jurisdiction or elsewhere (p. 19).

Such testimonies have been considered inadmissible by the Supreme Court of Canada (in Mohan and J.-L. J.), the Ontario Court of Appeal (in Ranger, Clark and Klymchuk), the Ontario Superior Court (in Klymchuk) and the Quebec Superior Court (in R. v. Croteau, 2004). In Ranger, the Ontario Court of Appeal explained that:

Criminal profiling is a novel field of scientific evidence, the reliability of which was not demonstrated at trial. To the contrary, it would appear from [the expert’s] limited testimony about the available verification of opinions in her field or work that her opinions amounted to no more than educated guesses. As such, her criminal profiling evidence was inadmissible (p. 22).

This is currently the unanimous position of Canadian courts on the behavioural aspects of profiling, which also corresponds to our review of the scientific literature: the scientificity of criminal profiling is not always established so as to allow it to meet the requirements established in Mohan and J.-L. J.

Béliveau and Vauclair (2007) also approached two elements related to behavioural profiling: propensity and the evidence of similar facts. The use of expert testimony to establish the propensity of an offender to commit a crime is generally inadmissible:

[translation]
…in Morin, a psychiatrist was deemed not to be able to testify regarding the propensity of the accused to commit the crime, since this type of evidence was also inadmissible. Similarly, the accused could not have an expert testify to show that, because of his mental state, he would be incapable of committing the crime of which he was accused. This would constitute evidence of a good reputation, which is normally given by persons testifying about the opinion of community members and by the accused who may refer to acts of good behaviour (p. 373).

However, an expert could testify if the accused and the perpetrator of the crime shared distinctive traits that were so unusual as to be like a signature. Consequently, according to the rule of exception in Mohan, when a crime or its perpetrator have distinctive characteristics, an expert may testify that the accused does not have the corresponding characteristics (Béliveau and Vauclair, 2007). However, these characteristics must prove to be completely distinctive and not just "abnormal."

In the case of evidence of similar facts, if it is relevant in principle, it will generally be inadmissible because its prejudicial effect frequently surpasses its probative value (Béliveau and Vauclair, 2007). This type of evidence is based primarily on the improbability of a coincidence of the crime elements being sufficiently distinctive to link the accused: [translation] " …  when the aim of this evidence is to prove the identity of the perpetrator of the crime, there must be a high degree of similarity between the acts in order to demonstrate that the accused is, not the type of person to have committed the crime, but the very person who committed it" (p. 231, emphasis is ours).

In the United States, however, the situation is slightly different, especially because of the differences in admissibility criteria between the states. It is clear that criminal profiling does not meet the requirements of Daubert or even Frye and that, when these criteria are applied, the testimonies based on the behavioural aspect of profiling are deemed inadmissible. This is notably the case for rulings by the Superior Court (State v. Fortin, 1999) and the Supreme Court (State v. Cavallo, 1982) of New Jersey, the Ohio Court of Appeal (State v. Lowe, 1991; State v. Roquemore, 1993) and the Tennessee Court of Appeal (State v. Stevens, 2001). In addition, in cases where experts are called to testify, based on psychometric tests or penile plethysmography, about the concordance or non-concordance of the personality of an accused with the "typical" profile of a pedophile (as was the case in Mohan in Canada), Peters and Murphy (1992) observed that, with the exception of California, all the American courts that examined this question deemed the testimony inadmissible. The five types of objections raised are as follows: 1) the practice is not scientifically reliable, 2) the testimony would not be relevant insofar as it would not constitute added value for the rest of the elements of evidence, 3) the risk of usurping the jury’s role of deciding the verdict of innocence or guilt is too great, 4) the grave risk of prejudice surpasses the probative value of the testimony, and 5) the reputation of the accused may be established without resorting to expert testimony.

In certain cases, however, the criteria adopted in Frye or Daubert are not considered applicable to the profiler testimony. For example, in Simmons v. State (2000), the Alabama Court of Appeal decided that deduction of the motive based on a crime scene examination constituted specialized knowledge not covered by Frye. It was therefore maintained that:

Whether the offender received sexual gratification while committing the offence was a critical issue of the case, and [FBI expert] Neer’s testimony was probative on that issue. Inferences had to be drawn from the physical evidence presented at the crime scene (p. 16).

Because this testimony was not subject to the criteria in Frye, it was not necessary to demonstrate that the method of inference used (to deduce from the crime scene and victim examinations that the criminal had committed this crime initially to satisfy a sexual desire) was widely accepted by the scientific community. The Court also mentions that Officer Neer’s testimony did not, in his view, constitute a case of profiling, which is limited to the attempted application of the general characteristics of serial killers to a given individual. It has been advanced that this type of testimony is injurious and of little probative value. In light of other cases studied, this appears to be an exceptional decision in terms of the flexibility given to the expert and the nature of the testimony admitted as evidence.

In two other cases, the Delaware Supreme Court (Pennell v. State, 1991) and the Louisiana Supreme Court (State v. Code, 1993) admitted an analysis of the modus operandi and signature linking a series of homicides as evidence. Here again, the application of criteria established in Frye was explicitly avoided:

[FBI] Agent Douglas […] was providing an expert opinion based upon his knowledge and experience in the field of crime analysis. This Court has held that when an expert’s opinion is based solely upon his own knowledge and experience, the Frye test has no application (Pennell v. State, 1991, p. 7).

Note that the definition of profiling adopted in Pennell is the same as the one retained later in Simmons (see previous).

Elsewhere in the world, although France recognizes that certain expert opinions would be potentially admissible, the courts’ reaction would currently be characterized by a certain degree of defiance, as shown by at least two quashed judgments following the admission of profiling testimonies (Marin, 2003). In England, Copson pointed out in 1995 that " …  it is made abundantly clear that there are great and potentially insurmountable difficulties in introducing profiling evidence in British courts…" (p. 27). Ormerod (1996) added that there were no known cases where psychological profiles were admitted as evidence in England and France. Following a study of the legal implications of criminal profiling, he concluded that "The prejudice contained in a profile will in almost all cases exceed the limited probative value of such an opinion" (p. 877). Finally, Woskett, Coyle and Lincoln (2007) state that no testimony resulting from criminal profiling has yet been introduced in Australia and that the general opinion of Australian lawyers would be largely negative.

3.6.3 Implications

Given the previously mentioned admissibility criteria and Canadian case law, in order for criminal profiling to effectively contribute to the investigation phase, research must be done to establish a scientific method for doing so (Mohan and J.-L. J). This would involve 1) the development of theories to generate testable hypotheses; 2) submission of these hypotheses to factual testing within the framework of empirical research; 3) submission of the findings of this research to examination by the scientific community by publishing them in refereed journals; 4) ensuring replication of these results in order to establish, after meta-analysis, the reliability and margin of error in theoretical predictions, and 5) eventual recognition of the theory’s validity by the scientific community.

This is no simple matter and would require years of concerted effort as well as a means of making criminal profiling a science.

3.7 Conclusions

3.7.1 Summary

We cannot logically conclude from our review of the research that behavioural profiling functions in a systematic manner. However, there is anecdotal evidence that profiling can work: we need only think of Brussels and the case of the Mad Bomber.

The literature is replete with approaches and typologies, but as Hicks and Sales (2006) point out, for the most part these models suffer from a lack of theoretical bases and empirical validation that could confirm and explain the links between crime scene elements and the everyday behaviours and personality of offenders. For these reasons, none of the proposed models can be considered "scientific" in the strict sense of the term, as confirmed by the courts on several occasions. Finally, several models fail to provide investigators with operational characteristics to describe potential suspects, so that the contribution to the investigation work is often limited. There are, however, certain typologies that are satisfactory in this respect (for example, the organized/disorganized dichotomy of Douglas et al., 2006).

Few empirical studies meet the rigorous criteria of credible scientific research. For example, Gray, Watt, Hassan and MacCulloch (2003) note the regular absence of a control group. In addition, much of the research, including Kocsis, cannot be applied to the Canadian or American context for the simple fact that it involves self-proclaimed profilers. Although rigorous enough, it could not reveal anything about the performance of profilers from the ICIAF training program. If we are to believe data from Copson (1995), profiling would lead to the arrest of offenders only 3% of the time in England and 8% by the FBI. This information dates back 15 years and was accumulated at the start of Canter and the ICIAF training program before the first profiles were produced. The situation has very likely evolved considerably since then.

In short, based on published research, while we are of the view that profiling can possibly contribute to police investigations, it is more of an art than a science. It has not yet been demonstrated, in our view, that profiling can systematically provide conclusive results. The use of databases to identify repeat offenders seems particularly promising, however. We also note that the practice of criminal profiling seems reasonably well supported in Canada; the ICIAF selection and training program should be able to keep the practice of profiling safe from charlatans. We are, however, of the opinion that profiling methods should be formalized, performance criteria should be developed, and empirical research should be undertaken to measure the true effectiveness of criminal profiling in Canada.

3.7.2 Limits

There are several limits to behavioural profiling research. First, each study contains a set of variables specific to the researcher’s orientation and there is little overlap. In addition, because of strong competition among individuals calling themselves profiling experts, there is little exchange of technique. As a result, there is very little progress in the development of this discipline despite the growing interest of researchers. According to Muller (2000), "As long as the FBI has a monopoly on profiling (which it does in most western nations except Britain) and they decline to share any information, it will be very difficult to prove that it is worthwhile" (p. 260). We can subscribe to this position since, although we benefited from valuable collaboration from the ICIAF and the OPP, the FBI has indicated to us that they would not share any internal information.

In addition, there are several studies on profiling effectiveness formulated as internal surveys. For example, Kocsis and Hayes (2004) studied whether police had a positive preconception when they assessed a profile created by one of their counterparts compared to a profile created by a person whose training was not mentioned. In addition to revealing nothing about the effectiveness of profiling, these studies lack "naturalistic validity," in that they take place in a context that is different from the one in which profiling is normally practised.

These limits impact our evaluation of profiling insofar as they are principally guided by published research that is rare, often not very rigorous and conducted with self-proclaimed profilers.

3.7.3 Recommendations

[R1] – Inferential methods in behavioural profiling should be formalized and recorded (which does not mean, we should point out, that they must be made public, as criminals would then devise a method to defeat them).

[R2] – Performance criteria should be developed to evaluate the true effectiveness of behavioural profiling.

[R3] –Research should be undertaken to empirically evaluate the effectiveness of behavioural profiling in the Canadian context. This research should cover three particular aspects: 1) the performance of profilers compared to that of detectives who do not have such training (in order to establish the added value of profiling for conventional investigation methods), 2) profile accuracy (by comparing profiler predictions to offender characteristics in solved cases), and 3) the actual contribution of profiling to suspect identification and arrest.

Previous Page Table of Contents Next Page