A Critical Review of the Literature Examining Profiling
Criminal profiling, as a process of understanding individual’s certain behaviours, has a long history (Woodruff, 1982). The most famous case of Jack the Ripper could be the earliest and most widely known example of profiling in the context of a criminal investigation in relation to the most notorious serial killer from the Victorian period in England. At the time of the killings (1888), physician Dr. Tomas Bond was asked to assist the investigation by London’s Criminal Investigation Division. His report from a crime scene examination identified the behaviours displayed by individual(s) during the offense, for instance, Dr. Bond suggested that all the murders were committed by the same hand, the throats had been cut from left to right and the women had been lying down when murdered (Whittington-Egan, 1975). Furthermore, he made an assumption regarding the offenders’ potential characteristic. He suggested that the individual responsible for these murders probably was a quiet, inoffensive looking man, middle-aged and respectively dressed, eccentric in his habits, most likely without regular occupation, but financially independent (Whittington-Egan, 1975).
It seems that Dr. Bond’s report, similarly to current criminal profiles, challenged the relationship between each of the murders, the crime evidence similarity in addition to the analysing they conducted were indeed committed by the same individual. Further, he point out a series of features that may perhaps characterise the perpetrator including likely age, gender, even psychopathologies (Canter, 1995). Although, example of Dr. Bond criminal case evaluation may be seen as the fundamental concept of criminal profiling, thus as case of Jack the Ripper remains unsolved, the accuracy, validity and reliability of Dr. Bonds profile cannot be measured (Petherick, 2005).
Another example of historical pioneering profile could be psychological evaluation of Adolf Hitler. This case (1943) was carried out by the US Office of Strategic Services with Dr. Langer consultation (Langer, 1972). In the process of Hitler’s psychological assessment Dr. Langer suggested that in circumstances of defeat, Hitler’s most likely reaction would be attempt to suicide rather facing the humiliation of possible capture and trial (Langer, 1972). The history shows that the only one from eight Adolf Hitler evaluation of his future action, constructed by Dr. Langer was accurate, which may question the reliability and validity of profile as an investigative tool.
According to Innes (2003) pioneering profile motivated numbers of scholars to investigate the development of criminal profiling. Inspired by reliability of previous criminal profile a number of researchers conducted their study on its concept and its efficacy as an assisting tool for law enforcement in serial violent cases (Holmes & Holmes, 1996).
As the popularity of offender profiling has increased in recent time due to media attention, massive production of TV forensic crime shows and general fascination of such a topic, the media targeted a potential audience by glamorising and misinterpreting the empirically recognised capabilities’ of the technique (Dowden et al., 2007).
Furthermore, Snook et al. (2008) stated that based on misconception of criminal profile as an investigative tool and overestimation of its results, people believe that criminal profiles can predict a criminal’s characteristics from crime scene. The aim of this paper is to critically review literature of profiling, in consideration to related empirical research, utilisation of offender profiling as investigative tool and its accuracy, reliability and validity in criminal investigations.
According to Dowden, Bennell & Bloomfield (2007) offender profiling is a process by which an offender’s behavioural, socio-demographic and personality characteristics are predicted on the basis of crime scene evidence. Furthermore, Baker & Napier (2003) suggested the purpose of offender profiling as investigative method used by FBI “…is to supply offender characteristics to help investigators narrow the field of suspects based on the characteristics of the crime scene and initial investigative information” (Baker & Napier, 2003).
It can be assumed that profiling has a number of definitions which differ from each other and therefore, may have different meanings as a part of the requests categorisation. For example, psychological profiling involves examination of offender behaviours, motives, emotions and mental background as a further investigation guide (Bennett & Hess, 2001). Conversely, investigative profiling includes investigative outline of examination of criminal case to present a descriptive model of the features that characterise the possible perpetrator(s) of a particular crime(s) under analysis (Kocsis & Coleman, 2000). Criminal profiling is an investigative tool that determines offender characteristics from the crime scene and the behaviour of the offender. It is an inferential process that involves the analysis of offender behaviour, their interactions with the crime scene, the offender and their choice of weapon among other things (Petherick, 2003). Therefore criminal profiling can be defined as an assessment of a crime scene, which may help to understand the behaviours manifested in the order of that crime and from an investigation of those behaviours (Baker & Napier, 2003).
According to Holmes & Holmes (2002) profiling has four main goals. Firstly, provide the criminal justice system with a social and psychological assessment of the offender. Secondly, reduce range of perpetrators under the investigation. Next, to provide the criminal justice system with a psychological evaluation of belongings found in the possession of the offender, such as souvenirs, photos, pornography and so on. Finally, provide interviewing suggestions and strategies.
Most currently, Turvey (2002) identifies two main points of profiling, divided by their goals and main concern. The first is the investigative phase, which involves perceptive features of the unknown offender for the known crime. He suggested that in the investigative stage there are five primary goals; to reduce the viable suspect pool and to help prioritise the investigation into those remaining suspects, to assist in the linkage of potentially related crimes by identifying crime scene indicators and behaviour patterns (i.e., MO and signature), to assist in assessing the potential for escalation of nuisance criminal behaviour to more serious or more violent crimes (i.e., harassment, stalking, voyeurism), to provide investigators with investigative relevant leads and strategies, and to help keep the overall investigation on track (Turvey, 2002). Furthermore, Hicks & Sales (2006) stated that constructing a profile of an unidentified offender involves three stages; collecting evidence from a crime scene by police officers, constructing behavioural, demographic and behavioural offender profile based on those evidence, and compile report which may assist to criminal investigation (Hicks & Sales, 2006). However, Turvey (2002) argued that characteristics of age, sex, race and intelligence could not be a part of the profile, because these are typically assessed investigative processes inductively and they not based on physical evidence. He stated that instead of that, the focus should be put on knowledge of the victim, knowledge of the crime scene, and on knowledge of methods and physical evidence (Turvey, 2002).
A series of studies were conducted to investigate the reliability, validity and accuracy of a criminal profile as a tool designed to assist law enforcement to prioritise search areas (Ratcliffe, 2004), and the most fundamental question in regards to profiling is whether the technique actually works, and were the predictions of profilers of the unknown offender accurate.
Snook et al. (2008) stated that current profiling practices involve typology to categorised crime scene and offender background characteristics, such as organised-disorganised dichotomy. To investigate this statement, Canter, Alison and Wentink (2004) analysed 39 crime scene behaviours that had been categorised as organised or disorganised in the FBI’s Crime Classification Manual. They stated that the analysis of these cases did not expose any distinct subtest of organised and disorganised behaviours. Therefore, they stated that this typology could not be easily used in serial homicides cases (Canter et al, 2004).
Another study to examine the assumption that offenders who display similar crime scene behaviours are likely to have similar personal and socio-demographic characteristic was conducted by Mokros & Alison (2002). They investigated this hypothesis on a sample of 100 British stranger-rapist whose crime scene behaviour was categorised according to occurrence of variety of variables (use of weapon, use of cover up etc) and index according to their similarity with one another. Results of the analysis revealed that rapist who display similar offending behaviour were no more similar in their socio-demographic variables (occupation, living situation, ethnicity etc), criminal history or age at the time of offence.
However, limitations such as outsized sample and some unrelated crime scene variable tested data limitation due to its disregard for any situational factors that can interact to affect crime scene behaviour (eg. intoxication), opportunity sample that may not necessarily be representative of the wider offender population could jeopardised final outcome of those study. However, the implication of the findings for the offender profiling are extreme, especially in relation to the reliability of utilising previous case data to create current profile (Markos & Alison, 2002).
Another study conducted to investigate whether it was possible to predict whether homicide offenders had a previous criminal record on the basis of their victim’s characteristics was performed by Santtila, Runtti, & Mokros (2003). Santilla et al. (2003) compared crime scene and offender characteristics with victim variables for 502 solved single victim/single offender homicides that had occurred in Finland between 1980 and 1994. The data within each case was coded according to the presence or absence of 8 victim variables (victim criminal record, victim ‘violent’ criminal record, relationship status, alcoholism, intoxication, psychiatric disorders, unemployment, housing status) and 2 offender variables (offender criminal record, offender ‘violent’ criminal record). All victim variables that occurred less than 5% of the time were discarded; the remaining victim variables that showed significant association with offender variables were subjected to multivariate analysis. Results indicated that while most offenders (71%) had no previous criminal records, there was a significant association between victims having a criminal record and the perpetrator having a criminal record. Results also pointed to a strong relationship between increased anti-social victim variables and situation characteristics (such as intoxication) and the offender having a previous violent criminal conviction. Authors acknowledge that the study is limited by the “present vs. not present’ coding of variables such as employment or relationship status which does not allow for any recognition of differences within categories. When trying to assess a construct such as anti-social tendencies, qualitative differences between consistently unemployed individuals and someone who has just been made redundant would be advantageous. The use of exclusively Finnish cases may also limit the ability to generalize the findings to other populations. These research findings provide support for the relationship between some crime scene / victim variables and certain offender characteristics which is inherent to the profiler’s efficacy.
Furthermore, Snook et al (2007) suggested that the use of criminal profiling (CP) in criminal investigations has continued to increase despite limited empirical evidence that it is effective. To support his statement Snook et al (2007) conducted a narrative review and a 2-part meta-analysis of the published CP literature. Narrative review results suggest that the CP literature rests largely on commonsense justifications. Results from the first meta-analysis indicate that self-labelled profiler/experienced-investigator groups did not outperform comparison groups in predicting offenders' cognitive processes, physical attributes, offense behaviours, or social habits and history, although they were marginally better at predicting overall offender characteristics. Results of the second meta-analysis indicate that self-labelled profilers were not significantly better at predicting offense behaviours, but outperformed comparison groups when predicting overall offender characteristics, cognitive processes, physical attributes, and social history and habits. Therefore, that could be assumed that the profiler’s ability to compose accurate profile of unknown offender do not rely on profiler specialised knowledge (Snook et al, 2008).
To investigate the accuracy of criminal personality profiles provided by profilers in contrast to non-profilers Pinizzotto & Finkel (1990), conducted a study on a sample consisting of 28 individuals divided into five sub-groups. Group A contained 4 profiling experts, who taught police detectives profiling for the F.B.I. Group B consisted of 6 police detective profilers who had completed a one-year course at the Behavioural Science Unit of the F.B.I. Six police detectives untrained in offender profiling made up Group C, while Group D contained 6 clinical psychologists without experience in criminal investigations or offender profiling. Lastly, Group E was made up of 6 university undergraduates who were paid $10 each for their participation in the study. Participants were provided with information and materials from two separate crimes that had taken place and were subsequently solved by police. Of these investigations, one was a homicide and the other involved a sex offense. Participants were required to complete a multitude of tasks for this study testing their ability to determine what information was most relevant and their ability to write a detailed profile of the offender. Participants were also required to complete a multiple-choice questionnaire on different offender variables (e.g. gender, age, race etc) and were asked to rank a list of possible offenders in order of most likely to least likely to have committed the offence.
The procedure was repeated by the participants for both cases. Results indicated that profilers provided significantly more detailed profiles, were more accurate in their written profile of the sex offender and were more accurate in their rating of the possible offenders. However, there were no differences in profile accuracy found for the homicide case as a function of expertise, with expert profilers not significantly more accurate than university students. The authors suggested that this outcome of no difference might be accounted for by the higher censorship of information in the homicide case that may have hindered the ability to make more accurate predictions and by the fact that in homicide cases there can be no victim statement to provide insight. There are methodological shortcomings which limit the overall validity and generalisability of these results obtained in this study. These include the very small numbers of participants in the sub-groups and the overall sample, the use of multiple choice questionnaires which may provide cues to non-profilers regarding areas to be considered and possible answers and the use of only 2 cases to investigate accuracy. This article ultimately suggests that profilers are more accurate and detailed in the information they provide regarding offender characteristics than other groups of people both involved and naïve to criminal investigations.
Another study to measure profile accuracy and investigative experience as a factor for the accurate construction of a criminal profile was conducted by Kocsis et al (2002). They compared a group of Irish police officers with two control groups of university students in a simulated profiling experiment. The results of this experiment showed no significant difference between any of the groups in the number of characteristics correctly predicted, and also suggests that investigative experience may not be a necessary factor for the accurate construction of a criminal profile.
Furthermore, Alison, Smith & Morgan’s’ (2003) two studies investigated the assumption that individuals are prepared to perceive artificial statements when offered as a relatively accurate description of complete strangers in form of an 'offender profile' with a real case, and two distinctly different offender outcomes (one genuine, one fabricated) given to two groups of police officers (n = 24, n = 22). The result shows that the over half of both groups classified the profile as accurate and, regardless of distinct differences between the offenders, there were no differences in accuracy ratings of the genuine offender and fabricated offender.
Second study examined whether this effect transferred to a genuine profile, again using professional groups (senior police officers, n = 33; forensic professionals, n = 30). Regardless of receiving different offenders, over 75% of each sample rated the profile as at least somewhat accurate and over 50% as a generally or a very accurate assessment. Furthermore, the finding indicated that mean ratings of the genuine offender did not differ from ratings of the fabricated offender. The majority of individuals rated the profile as useful. These studies lend preliminary support to the hypothesis that individuals tend to construct meaning around unclear statements about a third party within the context of offender profiling.
Another study related to the question of profiling validity and accuracy was the study conducted by Kocsis & Middledorp (2004). The aim of their research was to investigate whether there was a relationship between the perceived accuracy of information contained in an offender profile and one’s belief in profiling. The 353 participants in this study were Australian university undergraduates (52% male, 48% female) with a mean age of 19.7 years, without specific knowledge of criminal profiling or forensic psychology. Each of the participants was provided with a 3 section survey. The first section of the survey was made up of cover page and a ‘belief in profiling’ scale. There were 3 different versions of the cover page, one which promoted the effectiveness and value of profiling, one which devalued and criticised it, and one cover page which contained information completely irrelevant to profiling. After reading these cover articles the participants completed a questionnaire rating their belief in profiling. The next section of the survey provided participants with a description of a murder (crime scene, victim wounds) that had previously taken place and an offender profile that had been constructed for this case. At this stage participants were told one of two things, (a) that the profile they had was written by a professional profiler for the police, or (b), that the profile had been written for the police (without specifying who wrote it). The last section of the questionnaire required the participants to rate how accurate they thought the profile may be in relation to 39 offender characteristics.
Obtained results indicated that while estimated accuracy of the profile was not affected by whether the author was specified as professional or not, it was significantly related to scores on the ‘belief in profiling scale’ and the cover information provided. This relationship showed that increased belief in profiling was significantly correlated with increased perception of profile accuracy irrespective of actual profile content. The study was limited by a small degree of variance in participant age that may make generalizing results to the larger population problematic and by the potential for participants’ answers to be affected by demand characteristics of the study. The results of this study are highly relevant to the question of profiling validity as it highlights the implicit problems associated with relying on perceived accuracy rather than empirical evidence to assess the accuracy and validity of offender profiling.
Furthermore, Dabney et al. (2006) postulated that the viability of offender profiling, and more specifically the utilization of archival data to correctly identify offender characteristics, is reliant upon the assumption that the archival information is unbiased. The authors examined data from a previous observational study of shoplifting in order to assess whether shoppers were over-sampled in accordance with offender stereotypes. Researchers began recording the observers’ behaviour and results indicated that when allowed to deviate from specific sampling protocol and use individual discretion, the observers over-sampled shoppers on the basis of perceived age, gender and race. Dabney et al (2006) stressed that the research assistants in the study had been specifically trained and directed to ignore shopper characteristics, but even still there behaviour was affected by previous ideas regarding likely offender characteristics. The authors suggested that this type of expectation bias is likely to be found in police practices concerning the detection and apprehension of people and may skew the data used to create offender profiles. The relatively small number of observers (N = 10) and design of study meant that the results weren’t as reliable as desired. Dabney et al (2006) concluded that the use of statistical probabilities to identify and investigate possible offenders can lead to the over-sampling of particular populations. The inherent danger of over-sampling is that it leads the reinforcement of disproportionate statistics and construction of inaccurate offender profiles.
Kocsis (2003) sought to examine and compare the content of written profiles as provided by professional profilers and non-profilers. To affect this comparison Kocsis recruited 5 groups of participants, including 5 professional profilers, 29 psychology graduates, 34 police officers, 27 biology undergraduates and 19 self-proclaimed psychics. Participants were given detailed case information (including crime scene photographs and autopsy report) from an actual solved murder investigation and directed to write a detailed description of the offender. Content analysis was conducted, coding profile data on the basis of its relation to information regarding (a) offender’s physical characteristics, (b) offender’s non-physical characteristics and (c) aspects of crime or offender’s behaviour before, during or after crime. Results indicated that the groups differed significantly in the total number of words and total number of features used in the profiles. While the validity of this research is somewhat weakened by use of such a small sample of profilers, the major limitation of this study is that it makes no comment on the accuracy on the information provided. The results suggested that professional profilers provide richer, more detailed information regarding offender characteristics. However without verifying the accuracy of the information, it says little about the actual efficacy of such a practice.
In conclusion that can be assumed that review of profiling literature suggests that most authors believe profiling has the potential to work. Although development of the research conducted into the efficacy of offender profiling has been completed by a select group of people, with individual researchers such as Kocsis, who repeatedly endeavour to empirically examine offender profiling, there a number of limitations which may contribute to discredit criminal profiling as investigative technique for law enforcement. For example, validity of the results of many of the studies utilising profilers is questionable because the ability to see real differences, patterns and relationships is fundamentally damaged when using such small samples. Whilst researchers continue to investigate diverse features of profiling as an investigative technique, limitations such as a lack of active profilers’ participation in these studies negatively impact the results. Furthermore, even scholars contribution to current body of knowledge in area of criminal profiling as investigative tool has been widespread; numerous studies have been limited by methodological and conceptual shortfall. For example, although literature shows that the empirical investigation of offender profiling performed by authorities in this areas is based on numerous features, approaches, and categorisations, is hard to find the study which was conducted out from laboratory sitting. Therefore, to explore the reliability, accuracy and credibility of profiling it is essential to subdue profilers’ secrecy and obtain their full co-operation on the field of criminal profile as an investigative tool examination.
However, a review of available research articles have provided partial support for a relationship between particular crime scene variables and offender characteristics (Mokros et al., 2002; Santtila et al., 2003) and have quite consistently found that professional profilers predict offender characteristics with greater overall accuracy than non-profilers (Kocsis et al., 2002; Pinizzotto et al., 1990; Snook et al., 2007). It is relevant; however, to note that actual rather than just comparative accuracy rates have been identified and that profilers’ predictiveness has been less than perfect. Although the experimental evidences are still not overwhelming, studies such as those conducted by Kocsis (2003) promote the idea that profiling might have some validity, efficacy and accuracy.
Baker, K., & Napier, M. (2003). “Criminal Personality Profiling,” in James, S., and Nordby, J. (Eds.), Forensic Science: An Introduction to Scientific and Investigative Techniques, Boca Raton, FL: CRC Press.
Bennett, W.W., & Hess, K.M. (2001). Management and supervision in law enforcement. 3rd ed. Belmont, CA: Wadsworth-Thomson Learning.
Brussel, J. (1968). Casebook of a criminal psychiatrist. New York: Bernard Geis.
Canter, D. (1994). Criminal shadows. London: Harper Collins.
Davis, J. A. (1999). Criminal personality profiling and crime scene assessment. Journal of Contemporary Criminal Justice, 15, 291-301.
Dowden, C., Bennell, C., & Bloomfield, S. (2007). Advances in offender profiling: A systematic review of the profiling literatures published over the past 30 years.
Journal of Police and Criminal Psychology, 22, 44-56.
Hicks, S. J., & Sales, B. D. (2006). Criminal profiling: Developing an effective science and practice. Washington, DC: American Psychological Association.
Holmes, R.M. and Holmes, S.T. (1996). Profiling violent crimes: An investigative tool, 2nd ed. Thousand Oaks, CA: Sage.
Kocsis, R.N. and Coleman, S. (2000). The unexplored ethics of criminal psychological profiling. In: Godwin, M.G., ed. Criminal psychology and forensic technology: A collaborative approach to effective profiling. New York: CRC Press, pp. 323–338.
Kocsis, R.N., Hayes, A.F. & Irwin, H.J. (2002). Investigative experience and accuracy in psychological profiling of a violent crime. J Interpers Violence, 17(8), 811–823.
Langer, W. (1972). The mind of Adolf Hitler. New York: New American Library.
Mokros, A., & Alison, L. (2002). Is offender profiling possible? Testing the predicted homology of crime scene actions and background characteristics in a sample of rapists. Legal and Criminological Psychology, 7, 25-43.
Petherick, W.A. (2003). “What’s in a Name? Comparing Applied Profiling Methodologies,” Journal of Law and Social Challenges, June, 173-188.
Petherick, W.A. (2005). Criminal Profile: Into the Mind of the Killer, Sydney: Hardie Grant Books.
Pinizzotto, A.J. and Finkel, N.J. (1990). Criminal personality profiling: An outcome and process study. Law of Human Behavior, 14, 215–233.
Ratcliffe, J.H. (2004). “Crime Mapping and the Training needs of Law Enforcement,” European Journal on Criminal Policy and Research, 10, 65-83.
Santilla, P., Häkkänen, H., Canter, D. V., & Elfgren, T. (2003). Classifying homicide offenders and predicting their characteristics from crime scene behavior.
Scandinavian Journal of Psychology, 44, 107-118.
Snook, B., Cullen, R.M., Bennell, C., Taylor, P.J., & Gendreau, P. (2008). The Criminal Profiling Illusion: What’s behind the smoke and mirrors? Criminal Justice and Behavior, 35(10), 1257-1276.
Snook, B., Eastwood, J., Gendreau, P., Goggin, C., & Cullen, R. M. (2007). Taking stock of criminal profiling: A narrative review and meta-analysis. Criminal Justice and Behavior, 34, 437-453.
Turvey, B. (2002). Criminal profiling: An introduction to behavioural evidence analysis. San Diego, CA: Academic Press.
Whittington-Egan, R. (1975). A casebook on Jack the Ripper. London: Wiley.
Woodruff, P. (1982). Hippias major/Plato. Indianapolis: Hackett.
The effectiveness of psychological profiles
Journal Journal of Police and Criminal Psychology
Publisher Springer Boston
ISSN 0882-0783 (Print) 1936-6469 (Online)
Issue Volume 16, Number 1 / March, 2001
Subject Collection Behavioral Science
SpringerLink Date Thursday, January 03, 2008
Jennifer Trager1 and JoAnne Brewster1