Are Offender Profiles Useful in Police Investigations?

In recent years, there are more and more people have become increasingly fascinated by offender profiling because of the influence of media. An example of this is the well-known criminal profiling film The Silence of the Lambs, which is based on the Thomas Harris novel of the same name. Some television shows such as Millennium, Profiler and The X-Files are also based around the premise of criminal profiling. However, Hicks and Sales (2006) emphasize that these portrayals encourage the view of profiling as an art rather than a science though these profilers have academic backgrounds and law enforcement experience. The prevalence of offender profiling has grown over the past three decades (Snook, Eastwood, Gendreau, Goggin and Cullen, 2007), although studies suggest that ‘profiles have been found to be most effective as an additional tool, not as a solution to specific crimes’ (Wilson, Lincon and Kocsis, 1997:2). Using offender profiling would appear to be more beneficial than relying solely on traditional methods of detection (Ainsworth, 2001). In addition, it is important for police detectives to know which crimes are connected in order to collate and compare the information between these related cases (Egger, 1984 as cited in Rossmo, 2000). This essay will discuss the underlying assumptions of both offender profiling and case linkage, and then argue the usefulness of these practices in criminal investigations.

Offender profiling has been defined by different authors using different terminologies such as specific profile analysis, psychological profiling, criminal profiling and criminal personality profiling. However, the underlying concept of definitions remains the same (Jackson and Bekerian, 1997). Beauregard, Lussier and Proulx (2005) state that offender profiling provides some descriptive information about behaviours and personality of an offender by analysing crime scene characteristics, which can narrow suspects and finally solve the cases. Muller (2000) describes the information which the criminal profiler uses is often taken from the scene of crime, and profilers will take into account some factors such as previous crime scene, weapons, behaviours and words to the victim and the geographic patterns of crime. Davies (1992 as cited in Beauregard et al., 2005) emphasizes that offender profiling is an investigative tool which focus on the analysis of offender’s behaviour and accordingly infers some of the criminal’s personal characteristics.

In Europe, offender profiling is defined as ‘attempting to produce a description of the perpetrator(s) of a criminal offence on the basis of analysis of characteristics of the incident and other background information’ (Stevens, 1995: 10), and this definition is adopted by the Association of Chief Police Officers’ Behavioural Science Investigative Support Sub-committee in UK (Ainsworth, 2000). Although the actual process of profiling differs from one profiler to another, the aim remains the same: to deduce the behaviour, personality, and physical characteristics of the offender (Muller, 2000). According to Holmes and Holmes (1996), profiling has the following three main goals: assessing the offender from social and psychological aspects, the evaluating suspected offenders’ relevant possessions, and consulting with law enforcement officials about some strategies when interviewing offenders. The process of constructing a profile of an unknown offender typically includes three stages (Snook et al., 2007). Firstly, collecting crime scene data by police officers is needed and then the data will be forwarded to a profiler; secondly, the profiler starts to analyze the data which stems from the crime scene; and the last step is the predictions about the personality, behaviours, and demographic characteristics of the likely criminal (Snook et al., 2007). Snook et al. (2007) also describe that the processes of analyzing crime scene data can be classified two types: clinical and statistical in nature. Clinically oriented techniques are based on the profilers’ intuition, knowledge, experience, and training, in order to generate predictions. Comparing these with statistically oriented techniques, the latter predictions are based on descriptive and inferential statistical models which derive from the result of analysing offenders’ characteristics who have committed similar crimes previously (Snook et al., 2007).

It would seem that profiling can offer more accurate results in some kinds of offence (Stevens, 1995). Hazelwood and Burgess (1995) claim that some serial rapes, murders, arson, child abusing are extraordinarily appropriate for profiling. Ainsworth (2000) suggests that murder and some serious sexual offences are the most common crimes for using offender profiling. In addition, Ainsworth (2001) comments that such serious, personal contact crimes involve attacks on strangers are the most difficult cases to solve for the police. Wilson et al. (1997) also point out that profiling is most useful in serial offences, because profilers can extrapolate and compare data from the various crime scenes. However, the FBI holds the point that property crimes and robberies probably are not suitable for profiling because sometimes these offences cannot reflect the personality of offenders (Ainsworth, 2000).

Police forces are always required to focus their investigative efforts to identify crimes which committed by the same offender, because the majority of the crimes come from the minority of offenders (Woodhams and Toye, 2007). Woodhams and Toye (2007) believe that case linkage can be used to examine the likelihood of a series of offences being committed by the same unidentified offender when lacking suitable physical evidence, such as DNA evidence. Crime analysts analyze crime scene actions to consider whether there are some similarities in behaviours which committed by a common offender (Woodhams and Toye, 2007). In addition, Holmes and Holmes (2002) suggest that linkage analysis can identify connections between similar crimes committed by different offenders.

Alison, Bennell, Mokros and Ormerod (2002) state that offender profiling rests on two key assumptions in order to make profiling be possible. The first one is the behavioural consistency assumption, which means if serial offenders commit similar offences, they must exhibit behavioural consistency. According to Alison et al. (2002), the variation in behaviour an offender shows must be less than the variation in behaviour by different offenders. The second assumption relies on what Mokros and Alison (2002) have termed the ‘homology assumption’, which requires that similar offence behaviour of two offenders will match similar characteristics (Goodwill and Alison, 2007). In addition, the process of case linkage also rests on two key assumptions. The first one is the offender consistency hypothesis, the same with the first assumption of profiling. The second assumption of case linkage is that there is variation in the way different criminals commit crimes (Woodhams, Bull and Hollin, 2007). Thus, for it to be possible to link crimes which committed by the same offender, criminals must behave in a stable but distinctive manner (Woodhams et al., 2007).

In terms of the behavioural consistency assumption, it has been hypothesized that each person has a cognitive-affective personality system which would affect the behaviour produced in a given situation (Mischel and Shoda, 1995 as cited in Woodhams and Toye, 2007). This theory implies that when people encounter situations which have a similar psychological meaning to them, they will produce similar behaviour (Woodhams and Toye, 2007). Salfati and Bateman (2005) studied 94 variables relating to serial homicide offences which were obtained from police case files of the Homicide Investigation and Tracking System (HITS) database in Seattle, Washington of USA. They state that serial homicidal offenders are consistent across the series of homicides in their crime scene behaviours. Bennell and Canter (2002) analyzed 86 solved commercial burglaries committed by 43 offenders, they demonstrated that commercial burglary offenders are consistent in their burglary behaviours. Moreover, Santtila, Junkkila and Sandnabba (2005) studied 43 serial stranger rape cases which occurred in Finland during the years 1983-2001, they confirm that rape ‘has a consistent thematic structure’ (Santtila et al. 2005: 102) instead of being a haphazard collection of behaviours.

Whether offenders show more consistency in some criminal behaviour than in others also has received great scrutiny. Researchers group the specific offender behaviours into domains, comparing with others in order to investigate whether offenders show greater behavioural consistency across their series of crimes (Woodhams and Toye, 2007). Grubin, Kelly and Brunsdon (2001) imply that behaviours within the domains of control and escape are less dependent on the situation and the victim’s behaviour, thus, the control and escape behavioural domains show greater behavioural consistency comparing with the sex and style behavioural domains. Besides that, Bennell and Canter (2002) found that target selection and entry behaviours showed relatively consistent comparing with property stolen by studying the sample of commercial burglaries. However, Goodwill and Alison (2007) stress that the situational determinants of an offence can cause enormous variation in the way of committing the offence. In addition, Goodwill and Alison (2007) also hold the point that ‘the behavioural inferences made about an offender across a crime series are especially vulnerable to violating’ (Goodwill and Alison, 2007: 824) the consistency condition. Similarly, Snook, Cullen, Bennell, Taylor and Gendreau (2008) emphasize that offender profiling approaches are based on the lacking empirical support typologies and are often based on an outdated understanding of human behaviour.

In terms of the homology assumption, perhaps the empirical evidence to support this assumption is a little weak. Mokros and Alison (2002) studied the cases about the number of 100 British male stranger rapists which had some similarities in the crime scene actions. They tested whether increased similarity in offence behaviour coincided with higher resemblance in characteristics and previous convictions. The characteristics they examined including age, ethnicity, employment, education, marital status, and criminal record. They concluded that in their study, there was no evidence could prove the assumption of homology according to the relationships between crime scene and rapists’ characteristics. Häkkänen, Puolakka and Santtila (2004) analysed the number of 189 arsons in Finland from 1990 to 2001 by using non-metric multi-dimensional scaling procedure. In order to find the associations between crime scene actions and criminal characteristics, they divided the crime scene actions into four types: expressive-object, instrumental-object, expressive-person and instrumental-person. In terms of criminal characteristics, they also divided into four types: adolescent, self-destructive arsonist, serial arsonist and criminal. As a result, they found that offenders who committed different types of arsons with some similar characteristics. Obviously, their study failed to offer strong empirical evidence to support the homology assumption. Similarly, Doan and Snook (2008) examined the number of 87 arsons and 177 robberies, and compared the different types and the various criminal characteristics. They stressed that the evidence is low to moderate to support the homology assumption. Furthermore, Goodwill and Alison (2007) declare that the homology assumption may be in effect in some cases but not others. Furthermore, they argue that the homology assumption may be dependent on the extent to the behaviours which influenced by situational, psychological or interpersonal factors.

In terms of the second assumption of case linkage, some studies have investigated that crimes committed by the same offender can be differentiated from crimes by different criminals. Bennell and Canter (2002) studied 86 solved commercial burglaries, they found that 43 offenders had distinct behavioural features. Bennell and Jones (2005) reported that various linking features exist in different burglaries, and this research also supports the assumption of inter-offender behavioural variation. However, the above studies indicate that the process of case linkage is not perfect and the linkage accuracy appears to vary with crime types (Woodhams et al., 2007). In contrast, Woodhams and Toye (2007) used measures of predictive accuracy which called areas under the curve (AUCs), and they found that there was high predictive accuracy in their study of burglary.

It tends to be the case that both the findings of offender profiling and case linkage will be greatly useful for solving the offences in criminal investigations. Offender profiling can ‘inform the prioritization of suspects possessing specified characteristics’ (Oldfield, 1997 as cited in Woodhams and Toye, 2007: 60), and case linkage can combine police investigative efforts and information from different crime scenes (Grubin, Kelly and Brunsdon, 2001). Besides that, profiles can also be used in providing interviewing skills or approaches for detectives to the potential suspect (Gudjonsson, 1992), and it is important for the trial process, because profiles can provide useful suggestions for prosecutors by assessing the accused, witnesses and juries (Wilson et al., 1997). Furthermore, Holmes and Holmes (2002) stress that linkage analysis systems not only can locate possible suspects from records of similar past offenses, but also can provide maximum information for psychological and geographic profiling efforts through confirming similar crimes.

In spite of the usefulness of profiling and case linkage, Ainsworth (2000) is worried about that people may be over-optimistic for profiling because it has attracted such widespread public interest. Wilson and Soothill (1996) hold the point that a profile will rarely solve a crime or catch a criminal by itself, but it is designed to be an aid to the investigating police. Profiling experts also state openly that offender profiles are merely another investigative aid, they ‘have never meant to lead exclusively to the apprehension of offenders’ (Ressler and Schactman, 1992 as cited in Wilson et al., 1997: 8). Furthermore, Jackson and Bekerian (1997) stress that offender profiles do not solve crimes and the answers they are offered are not solutions, accordingly, profiling should be viewed as a tool which can be extremely helpful in offering some strategies, supporting information management, and improving case understanding. Similarly, Gerberth (1995 as cited in Holmes and Holmes, 2002: 14) asserted that ‘Criminal profiling is an excellent law enforcement tool. However, it is just one of many tools and does not replace good investigative techniques.’

Although psychologists have offered very detailed and accurate profiles in a number of famous cases (Canter, 1994), profiles do not always have dramatic and successful results (Ainsworth, 2000). To date, there have been few accurate and systematic tests to examine the exact usefulness of profiling even though some profiles have been proved to be accurate and have led to the conviction of offenders (Ainsworth, 2000). Studies suggest that although criminal profiling is being utilized by police agencies around the world, there is no strong evidence can show that the profiling is reliable, valid, or useful (Snook et al., 2007). Accordingly, many researchers are reluctant to accept criminal profiling is reliable and valuable. For example, Godwin (1978 as cited in Holmes and Holmes, 2002) asserts that profiling has little effect in solving crimes, and he even describes profiles themselves are dull and tedious. Similarly, Levin and Fox claim that profiles are of little use in identifying the murderers, ‘unfortunately, this tool, no matter how expertly implemented, is inherently limited in its ability to help solve crimes’ (Levin and Fox, 1985 as cited in Holmes and Holmes, 2002: 275). Snook et al. (2007) conducted a narrative review and analysed the published literature, they concluded that the criminal profiling relied on weak standards of proof and profilers were not better than other groups in predicting the characteristics of an unknown criminal, they even declared that the profiling was extraneous and redundant in investigations. Similarly, Snook et al. (2008) argue that criminal profiling has the potential to mislead criminal investigators, hinder the apprehension of guilty criminals and lead to wrongful convictions.

It would seem that profilers can provide more accurate profiles than nonprofilers even though some of the above researchers are reluctant to see profiling as a useful tool. Pinizzotto and Finkel (1990) examined profiles which conducted by professional profilers, detectives, psychologists, and students for a series of cases. Their study found that the profiles offered by the profilers were more accurate than by all of the other groups in most cases. In the sex offender case, the profilers significantly offered more accurate items such as the gender, age, and education of the offender. In addition, their study also shows that profilers can offer richer and more detailed reports (Pinizzotto and Finkel, 1990). However, with regard to the homicide case, the detectives were significantly more accurate than the profilers in the items such as the offender’s employment and residence. Kocsis and his research team compared the accuracy of professional profilers, psychologists, self-identified psychics, college students, and various groups of law enforcement officers. They found that the profilers provided the actual offenders in the largest number of correct predictions (Kocsis, 2003), with the accuracy rates ranging from 46 per cent to 70 per cent (Kocsis and Middledrop, 2004). However, there was a notable amount of variability within the profiler group, Kocsis and Middledrop (2004) observed that some profilers were much more accurate than others, even though the cause of the variation was still unclear, whether it was due to overall differences in the profilers’ skills or to specific aspects of the cases.

The popularity of profiling is often seen as a measure of its success (Wilson et al., 1997). Copson (1995) shows that in UK, 29 profilers have been responsible for providing 242 instances of profiling advice between 1981 and 1994. However, Snook et al. (2008) claim that profilers always overemphasize the number of correct predictions rather than the proportion of correct predictions. Although there have been very few pieces of research which have tested both the accuracy and usefulness of profiles in live criminal cases, Copson’s study ‘Coals to Newcastle? Part 1: a study of offender profiling’ is one of the few studies (Ainsworth, 2001). Copson’s (1995) study demonstrates that 82.6% of respondents reported that the advice they received had been useful, though only 14.1% of officers reported the profiling advice had assisted in solving the case and only 16.3% of respondents reported that the profiling advice had opened new lines of enquiry. Besides that, there are more than 50 per cent of respondents said that the advice had offered new information for the case.

In Copson’s (1995) study, 126 officers (68.5%) reported that they would seek profiling advice again definitely in similar circumstances, though less than 50 per cent officers reported that they would definitely use the same profiler again in the similar situation. In addition, Pinizzotto (1984 as cited in Wilson et al., 1997) found that from 192 requests for profiles, actually only 17 per cent were used to help identify the suspects, however, 77 per cent of the respondents reported that the profiles had helped to focus on the investigation. Furthermore, in the earlier survey, Douglas (as cited in Snook et al., 2008) demonstrated that in USA, 46 per cent solved cases of the 192 instances were attributed to the offender profiling which provided by the FBI. Similarly, Jackson, van Koppen, and Herbrink (1993 as cited in Snook et al., 2008) showed that in Netherlands, five sixths of surveyed police officers reported some degree of usefulness about the profiling advice which provided by an FBI trained profiler. Another research shows that in USA, a significant portion of police officers hold the point that offender profiling has its value (Trager and Brewster, 2001 as cited in Snook et al., 2008). Similarly, Torres, Boccaccini and Miller (2006) surveyed forensic psychologists and psychiatrists through the Internet in order to examine their experiences and opinions about profiling. They found that approximately 40 per cent of these professionals feel the criminal investigative analysis is scientifically reliable or valid, and 86 per cent believe that offender profiling is a useful law enforcement tool (Torres et al., 2006).

Ainsworth (2000) points out that many senior detectives do not trust the usefulness of offender profiling which provided by profilers. Copson (1995) explains that the negative views taken by senior detectives may come from the misunderstanding of profiling. Copson and Holloway (1997 as cited in Ainsworth, 2000) show that profiling helped to solve only 16 per cent of the crimes in which it was used, and identified the offender in less than 3 per cent of cases. In contrast, Ainsworth (2001) emphasizes that we should look into these figures in a little more detail because normally profiling would only be considered when the case was too difficult to be solved. Another reason is that organizations such as the FBI are reluctant to offer the figures about the successes and failures of the profiles they provide (Muller, 2000). Britton’s implication that a large amount of cases have been solved by routine police work, not by offender profiling (Britton, 1997 as cited in Ainsworth, 2000). However, Canter argues that profiling has advantages in terms of the resources especially for some particular crimes comparing with that the police throw more and more resources at the crime hoping that ‘something will turn up’ (Canter, 1994: 21). When evaluating the usefulness of profiling, it is necessary to take into account if ‘the profile is seen in isolation or merely as one part of more general guidance which a psychologist might provide to investigators’ (Ainsworth, 2000:119). Muller (2000) also emphasizes that although some profiles may lead the wrong directions to the investigation, ‘this may only be a problem if the police place a greater amount of faith in the profile than they do in their own investigative skills’ (Muller, 2000: 259).

Another problem should be concerned is that when evaluating the accuracy of a profile, police officers are excessively subjective (Kocsis, 2003). Smith and Alison (2001 as cited in Kocsis, 2003) found that police officers were unable to discern the difference in the amount of accurate information in two profiles, and they tended to interpret ambiguous statements by their own subjective intentions. Kocsis and Hayes investigated the perceptions of police officers concerning the utility and accuracy of the profile, also found that ‘perception of the accuracy of a profile is quite likely to be associated with the reader’s perception regarding the identity of its author’ (Kocsis and Hayes, 2003 as cited in Kocsis, 2003: 129). Kocsis and Middledorp (2004) examined a sample of 353 participants in order to explore Kocsis and Heller’s findings that the relationship between one’s belief of a profile and the perceived accuracy of that profile. Kocsis and Middledorp (2004) found that there is a positive relationship between belief and perceived accuracy. They point out that the extent accuracy which participants perceive depending on their beliefs in profiling. Ainsworth (2001) also suggests that the dangers of the self-fulfilling prophecy should be concerned because police officers have the cognizant sense that any psychological profile may fit some individuals, which leads to the inaccuracy. Psychologists will tend to be focus on the probabilities of offence; in contrast, the police officers tend to operate in terms of guilt or innocence (Ainsworth, 2001).

It would seem that it is difficult to assess the accuracy and the value of profiles. One important issue should be considered is that whether profiling can bring new information to an enquiry or whether the profiling can confirm those cases which the police already suspected (Ainsworth, 2001). Ainsworth (2001) implies that in many cases, it is difficult to discern the new material because there may be an overlap between the information which the profiler offers and that which has already been collected by the police. In some cases, although the profiler’s advice may not be seem as the new information, profiler’s input may help police prioritize the information which they are considering (Ainsworth, 2001). Moreover, the value of profiling may depend on different cases. For example, profilers can offer deeper insights of some more serious forms of mental illness and some sexual assault cases than non-experts and some police who have little previous experience about those cases (Ainsworth, 2001).

In conclusion, the future of psychological profiling and case linkage appear promising, research seems to be supporting the underling assumptions of these practices even though there have been few systematic and accurate studies to test the exact usefulness of offender profiling and case linkage. In terms of the homology assumption, the empirical evidence is especially weak to support it. Profiling as an additional tool plays an important role in criminal investigations, especially in serial offences. Studies suggest that profilers can provide more accurate profiles than nonprofilers. Case linkage provides a good way to collate and compare the information between related cases, thus, investigative efforts can be combined and officers can avoid the repetition of investigations. Most police officers would seek profiling advice and case linkage because of the agreement of their usefulness. However, over-optimistic vision of profiling and case linkage should be concerned because they do not always have dramatic and successful results even though many solved cases were attributed to offender profiling and case linkage. Finally, profiling and case linkage are useful tools and cannot replace the other good investigative techniques.