The Fictional World of Criminal Investigative Analysis
I CHOICE JAMeS Bond
Sherlock Holmes and other fictional characters can draw a considerable amount of information from even the slightest crime scene evidence. They can infer detailed characteristics of the offender and can reach conclusions with a great degree of certainty; and, of course, they always seem to solve the crimes they are investigating.
In this Discussion, you select a fictional sleuth, investigate his or her methods for solving crimes, and compare those methods to those currently used in the role of criminal investigator.
To prepare for this Discussion:
Select a fictional character who investigates criminal behavior or crimes (example: Sherlock Holmes, Harry Bosch, Kay Scarpetta, Philip Marlowe, James Bond, Jack Reacher, Alex Cross, Travis McGee, Jane Rizzoli, or
Describe the fictional crime investigator you selected. Explain a few of his/her methods or signature investigative traits. Compare how these methods may or may not match your perspective of the process and purpose of criminal investigative analysis. Explain what you think modern criminal investigators and criminal profilers should be able to do.
Main Discussion Posting Content
27 (54%) – 30 (60%)
Discussion posting demonstrates an excellent understanding of all of the concepts and key points presented in the text/s and Learning Resources. Posting provides significant detail including multiple relevant examples, evidence from the readings and other scholarly sources, and discerning ideas.
Reply Post & Peer Interaction
9 (18%) – 10 (20%)
Student interacts frequently with peers. The feedback postings and responses to questions are excellent and fully contribute to the quality of interaction by offering constructive critique, suggestions, in-depth questions, use of scholarly, empirical resources, and stimulating thoughts and/or probes.
9 (18%) – 10 (20%)
Postings are well organized, use scholarly tone, contain original writing and proper paraphrasing, follow APA style, contain very few or no writing and/or spelling errors, and are fully consistent with graduate-level writing style.
Bartol, C. R. & Bartol, A. M. (2010). Criminal & behavioral profiling. Thousand Oaks, CA: Sage.
Turvey, B. E. (2012). Criminal profiling: An introduction to behavioral evidence analysis (4th ed.). San Diego, CA: Academic Press.
Inside the mind of the mind hunter: an interview with legendary FBI agent John Douglas
The Forensic Examiner. Spring, 2007, Vol. 16 Issue 1, p10, 4 p.
KSA Media, LLC
Detectives — Interviews
Douglas, John E. — Interviews
Meet Criminal Profiler John Douglas at the 2007 ACFEI National Conference! John Douglas, former head of the FBI’s Investigative Support Unit, has hunted some of the most notorious and sadistic […]
Copyright 2007 Gale, Cengage Learning. All rights reserved.
COPYRIGHT 2007 KSA Media, LLC
Gale Academic OneFile Select
Criminal Profiling Works and EVERYONE Agrees Richard N. Kocsis PhD
To cite this article: Richard N. Kocsis PhD (2010) Criminal Profiling Works and EVERYONE Agrees, , 10:3, 224-237, DOI: 10.1080/15228930903550574
To link to this article: https://doi.org/10.1080/15228930903550574
Published online: 19 May 2010.
Submit your article to this journal Article views: 5313
View related articles
Citing articles: 3 View citing articles
Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=wfpp21
Journal of Forensic Psychology Practice, 10:224–237, 2010 Copyright © Taylor & Francis Group, LLC
ISSN: 1522-8932 print/1522-9092 online
Criminal Profiling Works and EVERYONE Agrees
Richard N. Kocsis, PhD
Forensic Psychologist in Private Practice
This article considers the empirical evidence currently available, which examines the validity of criminal profiling. The manuscript also serves to highlight a third key flaw in Snook et al. (2007) which, in the writer’s view, further renders its conclusions prob- lematic. The phenomena referred to as the nomenclature illusion is discussed to explain its role in hindering the cogent development of scientific knowledge in the area of criminal profiling. The ram- ifications of these findings for competing theoretical approaches to criminal profiling are also explored. Final comments consider the optimal approach for the development of operationally relevant techniques for profiling violent crimes in the future.
KEYWORDS criminal profiling, investigative psychology, meta- analysis, predictive validity, psychological profiling
The manuscript by Snook, Eastwood, Gendreau, Goggin, and Cullen (2007) offers an evaluation of the validity of criminal profiling via meta-analyses of various original data-driven studies. The writer identified three1 key issues that, in his view, rendered claims of Snook et al. (2007) flawed. In the writer’s subsequent manuscript concerning the validity of criminal profiling (see Kocsis, Middledorp & Karpin, 2008), the statistics reported on in Snook et al. were reconfigured, with discussion devoted to two of its key flaws. The first key flaw concerned a contradiction between previously opined criticisms and the analysis of Snook et al. (2007). Specifically, Snook et al. suggested
Since producing this manuscript, the author has identified further issues related to statistical principles in employing meta-analysis that appear to have also been overlooked by Snook et al. (2007). Address correspondence to Dr. Richard N. Kocsis, P.O. Box 662, Dee Why, NSW 2099
Criminal Profiling Works 225
that many of the original studies evaluating profiling were methodologi- cally flawed, ergo invalid. Notwithstanding this, Snook et al. used these same purportedly flawed studies as the basis for their own meta-analyses, the outcomes of which they claimed were robust. Thereafter, Kocsis et al. (2008) sought to highlight the inconsistency in the Snook et al. approach and argued that the use of the original studies necessitated by implication that they were regarded as sufficiently valid to have not been discounted in the meta-analyses conducted by Snook et al. (2007).
The second key flaw concerned an inappropriate combination of data that, in the writer’s view, served to undermine the validity of the Snook et al. (2007) conclusions. Specifically, Snook et al. asserted that samples of police personnel were a representative measure of the capabilities of expert profilers and, based on this, samples of both profilers and police personnel were combined in their analysis. However, as Kocsis et al. (2008) indicated, there is no legitimate reason to equate police personnel with the abilities of profilers. Consequently, the combination of these samples is inherently problematic for the claims advanced by Snook et al. (2007).
After the publication of Kocsis et al. (2008), a rebuttal by Snook et al. (2010) was produced, and this manuscript is a rejoinder. When Kocsis et al. (2008) was prepared there were a number of issues, which the writer wished to discuss, which were foregone in lieu of the need to produce a concise manuscript. In the wake of Snook, Eastwood, Gendreau, and Bennell (2010) however, the relevance of these issues seem especially pertinent and will be canvassed herein in the context of addressing the contentions of Snook et al. (2010).
OBSTACLES IN RESPONDING TO SNOOK ET AL. (2010)
Before proceeding further, two issues need to be identified that, in the writer’s view, impact upon the ability to address some of the matters raised in Snook et al. (2010). The first are contradictions and/or inconsistencies concerning the assertions posited by Snook et al. The second involves disparities between the articulated criticisms of Snook et al. and what the studies of Kocsis et al. posit. That is, the writer is in the unenviable position of having to respond to criticisms that do not appear to be reflective of what appears in the subject research but instead seem to be more premised upon what Snook et al. have synthesized about the writer’s work and/or that of others.
The writer in the limited space provided by this rejoinder cannot detail every perceived contradiction. Consequently, only a small number are referred to
226 R. N. Kocsis
herein to illustrate the flavor and occurrence of such contradictions. One notable example is the following statement: “The issue here is not to rehash criticisms with Kocsis’ research that have been debated elsewhere” (Snook et al., 2010, p. 221).
Despite making this assertion, Snook et al., it appears, do just that, with their criticisms concerning the skill basis of participants and the col- lected samples.2 Indeed, Snook et al. (2010) devote considerable space to questioning the parameters of the collected samples while simultaneously overlooking, it seems, the impact this has upon the earlier Snook et al. (2007) article, the findings of which they maintain are sound. That is, in advancing their criticisms, it seems that Snook et al. (2010) have overlooked that the data set in Kocsis et al. (2008), and Snook et al. (2007) are one and the same. This approach results in the situation whereby the identical data set is sufficiently robust for Snook et al. to conclude “The evidence generated from this research confirms. . . . that profilers do not decisively outperform other groups . . .” (Snook et al., 2007, p. 448) and yet is simulta- neously defective for Kocsis et al. (2008). Logically speaking, if the data set is insufficient for Kocsis et al. (2008) to draw any robust conclusions, then surely it is similarly inherently defective for the purposes of Snook et al. (2007). This same conclusion expressed in Snook et al. (2007) also seems to be unequivocal in its pronouncement, yet Snook et al. (2010) appears to somewhat inconsistently adopt a more equivocal position: “. . . Snook et al.  were clear that the results of their meta-analyses should be treated as tentative . . .” (p. 216). The writer is unable to discern the “tentative” aspect in the aforementioned pronouncement by Snook et al. (2007).
It is not the intention of the writer to chronicle each perceived disparity but to instead highlight examples so readers can perhaps gain an appreciation of the issue.3 Snook et al. (2010) for example, state, “We maintain that there is a rush to judgement by KMK regarding the positive forecasting abilities of profilers” (p. 218). The basis for reaching this conclusion is not clear, as the Kocsis et al. (2008) manuscript to which they refer as “KMK” unambiguously states “The present conclusions can be viewed as inchoate evidence of a
of the criticisms presented by Snook et al. (2010) concerning the skill basis of participants and sample
Readers should refer to the writer’s previous manuscript Kocsis (2006b, p. 466–470) for discussion
Readers can also refer to Kocsis (2006b, pp. 460–462 and pp. 465–466) for other examples of disparities previously identified.
Criminal Profiling Works 227
growing technique . . .” (Kocsis et al. 2008, p. 258).4 The writer fails to see how Snook et al. can interpret “inchoate evidence” as being “a rush to judgment.”
Another conclusion that appears to have been drawn in Snook et al. (2010) is that the writer does not appreciate the importance of holistically for- mulated research and knowledge accumulation. The writer however, would have hoped that his publication record demonstrated a clear commitment to producing research of this nature. Kocsis et al. (2008) in particular rep- resents a holistic evaluation of the validity of criminal profiling which takes into account Snook et al. (2007) and, in the writer’s view, debunks its find- ings. Snook et al. (2010) may have misinterpreted that the issues canvassed in Kocsis et al. (2008) primarily relate to misguided applications of meta- analysis (Ioannidis & Lau, 1999; Esteban, Hernadez, & Kattan, 2008) and other methodological weaknesses.
THE FIRST KEY FLAW: THE CONTRADICTION OF GARBAGE IN, GARBAGE OUT
The first key flaw in Snook et al. (2007) does not, in the writer’s view, lie in the conclusions per se but rather in drawing conclusions based on original studies subject to their meta-analyses, which they maintain were inherently compromised. The dilemma for Snook et al. (2007) is reconciling their conclusions as valid in the face of relying upon original studies that they considered flawed. In an effort to address this Snook et al. (2010) suggest that their use of the studies was to “include all available research (regardless of quality) that has examined the accuracy of criminal profilers in mock pro- filing exercises” (p. 216). Notwithstanding that the notion of “quality” in the absence of articulated criteria is nebulous, such inclusion nonetheless over- looks their assessment of the integrity of the original studies that produced the data and fails to address what Kocsis et al. (2008) highlights: this being the contradiction of using the original studies viewed as methodologically defective to support conclusions that Snook et al. maintain are sound.5
Their arguments, in the writer’s view, when examined under a less- contorted prism, appear to be little more than an unsound compromise and an attempt to retrospectively explain the inherent contradiction. When the
research I have undertaken ever been represented as an authoritative treatise on the topic of criminal
profiling. Rather my findings are clearly reported as tentative observations concerning a practice for
which scientific inquiry is long overdue. Further replication and development in the future is clearly
emphasized” (Kocsis, 2006b, p. 470).
Furthermore, in a previous article the writer has also stated: “. . . importantly, at no point has the
One apparent ambiguity surrounding this issue is the suggested notion by Snook et al. of “quality” and the writer’s concomitant delineation of methodological standards in published studies. Unfortunately, this concept of quality does not, in the writer’s view, appear clear in their prior arguments. Instead, arguably the criticisms by Snook et al. concerning the writer’s studies seem to be expressed in more categorical terms as either acceptable or unacceptable.
228 R. N. Kocsis
layers of sophistry are laid bare, the insurmountable obstacle confronting Snook et al. is, in the writer’s view, reconciling their assertions in the first instance with their actions thereafter, in the face of the more commonly expressed logic of garbage in, garbage out.6
THE SECOND KEY FLAW: WHO ARE NOT CRIMINAL PROFILERS?
The quintessential flaw present in the Snook et al. (2007) article, in the writer’s view, is combining police personnel with profilers as it is the col- lapsing of these distinct sample groups that largely underpin the assertions concerning the lack of proficiency of profilers. As previously argued in Kocsis et al. (2008, p. 251–253), this is an inappropriate combination of data. In all fairness, however, the Snook et al. approach in this regard is not entirely of their own creation but is, in part, attributable to the absence of any uniformly recognized criteria governing who is qualified to operate under the colloquial mantle of profiler(see Kocsis, 2009). Snook et al. are, it appears, in agreement with the writer in noting the lack of consensus of an operational definition for profiler. As a clear demarcation does not exist governing who is and is not a criminal profiler, it seems rather disingenuous, then, to argue that police personnel should be construed in some capacity as reflective of the capabilities of profilers. Nonetheless, in the context of the studies that the writer has been involved in producing, this question can be answered not by debating the ambiguities of who are profilers but rather via a process of elimination in assessing who are not criminal profilers. Common sense dictates that individuals who have not studied criminal profiling, who do not engage in criminal profiling or claim to be criminal profilers, should not be regarded as profilers. Similarly also, police personnel who do not claim to be profilers and have not studied or engaged in criminal profiling should likewise not be regarded as profilers. In this context, therefore, the merging of data pertaining to police personnel with profilers that has been made by Snook et al. (2007) is not, in the writer’s view, appropriate, thus rendering the validity of the conclusions drawn from the assembled data set problematic.
THE THIRD KEY FLAW: EXPERT AND TRAINED PROFILERS
Once again, to be fair to Snook et al. (2007), the third key flaw present is not one entirely of their own making but in part appears to stem from a nuance in Pinizzotto and Finkel’s (1990) original research. It needs to
One hypothetical analogy to illustrate the untenable nature of the Snook et al. position may be to formulate conclusions on the nature of criminality drawn from a meta-analysis of studies on phrenology. That is, although we state as researchers that the scientific basis of phrenology is invalid, we will nonetheless utilize a collection of these studies in a meta-analysis, the outcomes of which we then contend are sound.
Criminal Profiling Works 229
be understood that Pinizzotto and Finkel is an amalgam of small exper- iments examining different facets in the construction of criminal profiles. Thus, one experiment examined the content of written profiles, another involved a recognition task, and yet another examined the types of case materials differing participants attenuated upon. The groups that were used in these experiments are typically described in a cursory manner as being composed of four types of participants: profilers, detectives, psychologists, and students. The one sub-component to Pinizzotto and Finkel that has arguably attracted the most attention, and which many of the writer’s stud- ies are modeled upon thereafter, involves a multiple-choice questionnaire in conjunction with case materials as a method of objectively measuring the accuracy of profile predictions.
Though Pinizzotto and Finkel’s (1990) study is composed of numerous small experiments, there were in fact five, not four, groups of partici- pants. This fifth group, importantly however, did not feature in all of the experiments. That is, Pinizzotto and Finkel’s research involved a sample of detectives, psychologists, and students but also contained not one, but two, distinct groups of profiler type participants. The first group was labeled “Group A, Expert/Teacher” and was composed of
“. . . profiling experts [emphasis added] who train police detectives in profiling at the F.B.I. Academy in Quantico, Virginia. Each of these sub- jects is or was an agent with the Federal Bureau of Investigation. They shared a combined total of 42 years of profiling . . . range of 4–17 years experience” (Pinizzotto & Finkel, 1990, p. 218).
In comparison, the second group was labeled “Group B, Profilers” and consisted of
“police detectives from different police agencies across the country who have been specifically trained in personality profiling . . . The course of studies involved 1 year at the Behavioral Science Unit of the Federal Bureau of Investigation in Quantico . . . and 14 years of com- bined experience in profiling (range = 1–6)” (Pinizzotto & Finkel, 1990, p. 218).
What is perhaps overlooked is that the expert profilers of the FBI’s Behavioral Science Unit did not participate in the multiple-choice question- naire sub-component of Pinizzotto and Finkel (1990). The trained detectives who had undertaken a 1-year course in the FBI’s profiling method how- ever, did. Consequently, the findings of the multiple-choice experiment in Pinizzotto and Finkel are not reflective of the FBI’s expert profilers but, rather, individuals who have undertaken a 1-year training course in the FBI profiling method and possessed a modest amount of profiling experience
230 R. N. Kocsis
in comparison to the expert/teacher group.7 As has already been discussed, the classification of who are profilers is a contentious issue. In evaluating accuracy, however, the writer has endeavored to test individuals who might be regarded as suitable professionals. Accordingly, sampled profilers have been described as “. . . well-regarded, senior members of the forensic mental health profession” (Kocsis, 2004, p. 348). The years of formal tertiary educa- tion and training, as well as forensic clinical experience of these subjects, in the writer’s view, surpass and are not thus equivalent to the trained police detectives tested in Pinizzotto and Finkel. Consequently, the third key flaw in the Snook et al. (2007) manuscript is, as the writer sees it, another inap- propriate combination of data pertaining to trained profilers in Pinizzotto and Finkel with profiler subjects collected in the studies by the writer. It was partly owing to this issue that the Kocsis et al. (2008) study concluded by observing
. . . simplified comparisons such as profilers versus nonprofilers are not ideal but have been followed here to make the present research directly comparable with that of Snook et al. (2007). In so doing, it can be seen that the contentions advanced by Snook et al. concerning the accuracy of profilers have been demonstrated to be invalid. Nonetheless, a more incisive depiction of the capabilities of profilers in comparison to a range of other individual skill based groups of participants can be found in Kocsis (2003) (Kocsis et al., 2008, p. 259–260).
This final paragraph of Kocsis et al. (2008) clearly indicates that the reconfigured data are not ideal; however, readers are referred to Kocsis (2003) for a clearer assessment.
MISSING THE POINT: BESD STATISTICS AND CONFIDENCE INTERVALS
As previously indicated, the writer’s capacity to respond to Snook et al. (2010) is hampered by the perceived disparities. What appears to be another manifestation of this problem is a variety of statistical issues raised by Snook et al., as these arguments, in the writer’s view, are suggestive of a funda- mental misunderstanding of the flaws of Snook et al. (2007). For example, it appears that Snook et al. have failed to appreciate that the BESD statistics
The writer wishes to take this opportunity to clarify that these comments should not in any capacity be construed as criticism of the work undertaken by Pinizzotto and Finkel (1990) and/or the activities of the FBI’s BSU or the courses that organization conducts. The focus of the comments is to highlight the distinction in the profiler participants. The significance of this distinction between the differing groups of profilers is fundamentally manifest in the very creation of such a distinction between the two groups of profilers by Pinizzotto and Finkel (1990).
It appears that Snook et al. may be suggesting that they possess a higher standard of objectivity,
Criminal Profiling Works 231
in Kocsis et al. (2008) served as a simple mechanism by which the overall patterns of the meta-analyses in Snook et al. (2007) were displayed. That is, the statistics illustrated how measures of accuracy were influenced by vari- ous combinations of data such as combining police personnel with profilers. The significance of the previous extract from Kocsis et al. (2008) appears to have not been understood, as it indicates that the analysis is not ideal but was adopted to be directly comparable, and thus demonstrate the invalid nature of the assertions in Snook et al. (2007).8 It seems that Snook et al. (2010) have simply missed the point, that the basis to the three key flaws in Snook et al. (2007) do not originate in statistical numbers per se but, rather, the methodology employed to generate those numbers.
Hazards With Self-Appointed Experts
In the writer’s opinion the most troubling statement in Snook et al. (2010) is the claim that
. . .we are in a position to be objective researchers in the field of CP [acronym for “criminal profiling”] because we are not involved in pro- viding profiling advice (and therefore have no stake in how profilers perform) and do not advocate any “approach” to profiling (p. 220).
The writer assumes, in the wake of seminal works on the philosophy of scientific development such as Kuhn (1962) that Snook et al. are not suggesting that academics in university institutions are somehow immune to subjective factors including, for example, career agendas related to their research activities.9 Indeed, deeper consideration of the Snook et al. state- ment seems to imply that anyone, including Snook et al., once involved in providing profiling advice are thereafter somehow impaired with respect to researching objectively that activity.
Alternatively, this statement by Snook et al. (2010) may simply be an admission that they have no expertise in the area as they have no involve- ment in the provision of profiling advice. This being the case then, what does this say about the criticisms posed by them?10 The writer has previously
that readers were instead referred to Kocsis (2003).
It was for reasons such as the third key flaw and the sub-optimal nature of simplified comparisons
as they are not involved in the provision of profiling advice and claim that they do not advocate any
approach. It is not, however, clear how the Snook et al. criteria would reconcile with, for example, the
author’s many collaborators/co-authors who would, following their criteria, also have “no stake in how
profilers perform,” do not advocate any approach, and thus would also qualify as “objective researchers
in the field of CP.”
Their claim concerning their suitability to assess profiling seems analogous to the assertion that a non-practicing lawyer has a more objective understanding of the law than an experienced practicing lawyer.
232 R. N. Kocsis
commented upon a sense of naïvete’ informing proffered arguments by com- mentators in the field of criminal profiling (see Kocsis, 2006b, p. 460–465). This is arguably reflected in observations such as “. . . we believe that police forces should avoid using CP until there is sound empirical data supporting it as an effective investigative tool” (Snook et al., 2010, p. 221).
Confronted by an aberrant violent crime that is not resolvable via conventional investigative methods, police personnel should be reasonably entitled to use criminal profiling as a resource to augment their investigative efforts. As Ormerod and Sturman (2005, p. 171) have noted, there are gen- erally “. . . no rigid legal rules . . .” on what resources police may utilize provided that they “. . . comply with statutory and common-law rules relat- ing to processes of search, arrest and interviewing . . .” When confronted by problems in the real world, potential remedies must be explored. To sug- gest that until a perfect solution is available we should avoid trialing possible solutions is, in the writer’s view, suggestive of a naïve approach to dealing with the capricious vicissitudes of life.
The Nomenclature Illusion
One of the most troubling developments that, in the writer’s view, hinders scientific development in the area of criminal profiling is a phenomenon best described as the “nomenclature illusion.” Briefly, the nomenclature illusion arises from the arguably unwarranted creation of terminology and differ- ences that can serve as mechanisms for generating artificial distinctions that may, in turn, influence perceptions of merit. One example of the nomen- clature illusion can be seen in the study by Torres, Boccanccini, and Miller (2006), who surveyed various forensic mental health professionals concern- ing their perceptions of the scientific basis to profiling. The authors found that fewer than 25% of their respondents believed that “profiling” was sci- entifically valid. However, a simple change in terminology from profiling to criminal investigative analysis saw an increase in the perceived validity among respondents. Through a semantic variation in terminology, therefore, an arguably artificial distinction was created and was accompanied by a cor- responding perceived variation in merit.11 From their research, Torres et al. (p. 55–57) concluded that “. . . profiling is likely to be viewed more favor- ably if it is referred to by another name . . . the definitions provided for profiling and criminal investigative analysis were identical. Only the name of the process varied.”
Another domain wherein the nomenclature illusion appears to have been noted arises in legal contexts in which attempts have been made to admit aspects of criminal profiling as testimony albeit using different terminology. Consistent with the tenets of the nomenclature illusion attempts to admit such testimony have been criticized as “a distinction without a difference” and “a different suit on the same animal” (Grezlak, 1999, p. 2).
Some examples include, but are not limited to, the detection of deceit, eyewitness testimony,
Criminal Profiling Works 233
Within the context of the present manuscript there are, in the writer’s view, two pertinent examples of the nomenclature illusion. The first con- cerns the term investigative psychology (in acronym IP). As indicated in Kocsis (2006b), it is unclear how the topics12 said to comprise IP are not already encapsulated by more established terms such as forensic psychology and/or criminal psychology (e.g., Arrigo & Shipley, 2004; Bull et al., 2006; Healy, 1926; Howitt, 2002; Kocsis, 2009; Pakes & Pakes, 2009; Presetsky, 1986; Toch, 1961, 2000; Wrightsman, 2001). Consequently, beyond acting as a moniker for the collective research activities and university courses loosely associated with Professor David Canter on the topic of criminal profiling, the term investigative psychology appears to be somewhat superfluous (Kocsis, 2006b). Nonetheless, following the tenets of the nomenclature illusion, the invention of the term IP arguably fosters an artificial distinction and, thus, perception that IP is something separate and distinct from criminal profil- ing. This may also even extend to the perception that the methodologies it adopts possess some greater merit over others.
The second pertinent example of the nomenclature illusion relates to the conceptual deconstruction of criminal profiling. Criminal profiling and its various synonyms13 is a colloquially understood term that the writer has used for ease of recognition to describe the assessment of violent crimes as a method of assisting investigators in potentially identifying and thus apprehending offender(s) (Kocsis, 2006a, 2009; Safarik, Jarvis, & Nussbaum, 2000). Thus, the concept of “criminal profiling” extends beyond the predic- tion of offender characteristics and encapsulates numerous other facets such as, but not limited to, the assessment of relationships (if any) between per- petrated offenses, the interpretation of exhibited offense behaviors, and the evaluation of psychopathologies and/or motives (e.g., Douglas & Olshaker, 1995, Hazelwood & Michaud, 1999; Holmes & Holmes, 1996, Palermo & Kocsis, 2005; Ressler & Shachtman, 1992). It appears, however, that over time, attempts have been made to deconstruct this colloquial understanding of criminal profiling by creating subdivisions and artificial distinctions fos- tered by the development of new terminology. That is, by adopting a more constricted conceptualization of criminal profiling as involving only the pre- diction of offender characteristics, concepts, and/or knowledge external to this notion are conceived and labeled as something different to criminal profiling. In the writer’s opinion, an example of this artificial distinction can be observed in the narrative review of research literature presented in Snook et al. (2007). That is, by adopting a limited conceptualization of “criminal profiling,” literature that is arguably rightly inherent to the corpus
investigative interviewing, sexual and violent offending, jury decision making, and expert psychological
Some examples include psychological profiling and personality profiling.
234 R. N. Kocsis
of knowledge concerning criminal profiling can be stealthily, albeit legiti- mately, excluded (e.g., Groth, Burgess, & Holmstrom, 1977; Hickey 1997; Prentky, Burgess, & Rokus, 1989; Proulx, Beauregard, Cusson, & Nicole, 2007; Revitch & Schlesinger, 1989). Consequently, through a process of what, in the writer’s view, equates to effectively quarantining the broader literature available, Snook et al. (2007) can present a characterization of the literature as underdeveloped. One notable example of this, in the writer’s estimation, can be seen in the Snook et al. narrative review, which omits the Crime Classification Manual (CCM; Douglas, Burgess, Burgess, & Ressler, 2006). The CCM arguably represents one of the most significant contributions produced by the FBI’s Behavioral Science Unit.
As mentioned in Kocsis et al. (2008), the proponents of the research paradigm that, for ease of reference, the writer simply refers to as inves- tigative psychology appear to suggest that IP is the only scientifically robust approach to criminal profiling. Therefore, work external to the IP school that suggests that the proficient profiling of violent crimes can be achieved by other means arguably serves to detract from contentions concerning the superiority of IP’s approach. Despite the writer’s views concerning this issue occupying approximately half a page of the entire Kocsis et al. (2008) manuscript, rebuttal of this issue nonetheless seems to be of significant importance to Snook et al. (2010) in terms of the amount of text devoted to discussing it.
The writer in expressing his views concerning the state of criminal profiling does not, however, mean to imply that the topic of criminal pro- filing cannot benefit from further development—quite the opposite, in fact. Any non-partisan review of the writer’s work (e.g., Kocsis 2006a) should suggest that the writer consistently advocates caution in the use of crimi- nal profiling and maintains that it is a technique warranting considerably more development. The pivotal difference between the writer’s perspec- tive and that of Snook et al. is sourced in how genuine progress may be achieved in terms of advancing knowledge in the area. The optimal way forward, in the writer’s view, is through the dispassionate integration of various approaches to criminal profiling. This, to the writer’s mind, neces- sarily involves researchers’ undertaking far more ambitious studies and, in particular, conducting genuinely original data-driven studies that utilize far less accessible but nonetheless more-appropriate populations. It is for this reason the writer previously stated,
Although they seem to echo my sentiments for more empirical research. . . . it seems that rather than conduct their own original data-driven
Criminal Profiling Works 235
experiment to test the findings of my studies or others, they have instead opted for seeking to reinterpret the data collected by myself. . . . This circumstance is not unusual in the field of criminal profiling, for as I keep emphasizing, one of the biggest problems confronting the state of profiling research today is the absence of original experimentation. Researchers in the field seem to favor producing publications that are either not driven by original data or to make use of more readily avail- able forms of data that are not particularly relevant to the practical application of profiling. Clearly, such approaches are less onerous than designing a new experiment and persuading even a modest number of individuals such as profilers to participate in it, which is why perhaps it has proven so popular. As a result, however, very little in the way of new and instructive information about profiling emerges (Kocsis, 2006b, p. 473).
In the view of the writer, researchers need to focus upon collecting their own data and samples, conducting their own experiments, and there- after publishing their findings. It seems, however, that confronted with the logistical encumbrances of such research and the academic imperative of publish-or-perish, some researchers are increasingly utilizing seemingly sub- optimal approaches in an effort to generate publications.14 In this context, the writer notes that other scholars have expressed analogous concerns with regard to research undertaken in other fields: “Researchers are tempted to rely only on college samples simply because of the ease of data col- lection. Ease of research at the expense of clinically relevant research is unwarranted” (Rogers, 2008, p. 413).
Finally, in failing to appreciate the operation of the nomenclature illu- sion, Snook et al., in the writer’s view, foster an artificial distinction between their work and that of others. This, the writer considers to be a great shame, as we share the same ambition of seeking to develop sound techniques whereby crimes may be assessed to develop some understanding of the offender and to assist with the apprehension of that offender. Consequently, the criticisms by Snook et al. are, in the writer’s view, simply consistent with proponents of a competing school of thought in undertaking the same colloquially understood task referred to as criminal profiling (Kocsis, 2009). If Snook et al. did not fundamentally accept the concept of criminal profil- ing, albeit branded with a different suite of names and labels, they would surely not continue to engage in, or even advocate for, further research of this nature. It is with this in mind, therefore, that the writer facetiously concludes that criminal profiling works and everyone agrees!
Examples of this, in the author’s view, include analyses of publication catalogues, the recycling of data generated by others, and the reliance on data that have limited relevance to aberrant violent crime and, in the writer’s opinion, the practical provision of profiling advice.
236 R. N. Kocsis ACKNOWLEDGMENT
I would like to take this opportunity to thank Ms Rowie Daskalakis for her invaluable assistance during the preparation of this manuscript.
Arrigo, B. A., & Shipley, S. (2004). Introduction to forensic psychology (2nd ed.). New York: Elsevier/Academic Press.
Bull, R., Cooke, C., Hatcher, R., Woodhams, J., Bilby, C., & Grant, T. (2006). Criminal psychology: A beginners guide. Oxford: One World Pub.
Douglas, J. E., & Olshaker, M. (1995). Mindhunter. New York: Scribner.
Douglas, J. E., Burgess, A. W., Burgess, A. G., & Ressler, R. K. (2006). Crime
classification manual (2nd ed). New York: Simon & Schuster.
Esteban, W., Hernandez, A., & Kattan, M. (2008). Meta-analysis: Its strengths and
limitations. Cleveland Clinical Journal of Medicine, 75(6), 431–439.
Grezlak, H. (1999, April 12). Profiling testimony inadmissible in murder trial: Too
speculative, prejudicial judge says. Pennsylvania Law Weekly, 1–2.
Groth, N. A., Burgess, A. W., & Holmstrom, L. L. (1977). Rape: Power, anger and
sexuality. American Journal of Psychiatry, 134, 1239–1243.
Hazelwood, R. R., & Michaud, S. (1999). The evil that men do. New York: St Martin
Healy, W. (1926). Pathological lying, accusation and swindling: A study in forensic
psychology. New York: Little Brown and Co.
Hickey, E. (1997). Serial murderers and their victims. Belmont, CA: Wadsworth. Holmes, R., & Holmes, S. (1996). Profiling violent crimes: An investigative tool.
Thousand Oaks, CA: Sage.
Howitt, D. (2002). Introduction to forensic and criminal psychology. Oxford, UK:
Ioannidis, J. P. A., & Lau, J. (1999). Pooling research results: Benefits and limitations
of meta-analysis. The Joint Commission Journal on Quality Improvement, 25(9),
Kocsis, R. N. (2009). Applied criminal psychology: An introduction to forensic
behavioral sciences. Springfield, IL: CC Thomas.
Kocsis, R. N. (2006a). Criminal profiling: Principles and practice. Totowa, NJ:
Kocsis, R. N. (2006b). Validities and abilities in criminal profiling. The dilemma
for David Canter’s investigative psychology. International Journal of Offender
Therapy and Comparative Criminology, 50(4), 458–477.
Kocsis, R. N. (2004). Psychological profiling of serial arson offenses: An assessment
of skills and accuracy. Criminal Justice and Behavior, 31(3), 341–361.
Kocsis, R. N. (2003). Criminal psychological profiling: Validities and abilities. International Journal of Offender Therapy and Comparative Criminology,
Kocsis, R. N., Middledorp, J., & Karpin, A. (2008). Taking stock of accuracy in crim-
inal profiling: The theoretical quandary for investigative psychology. Journal of Forensic Psychology Practice, 8(3), 244–261.
Criminal Profiling Works 237
Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago: University of Chicago Press.
Ormerod, D., & Sturman, J. (2005). Working with the courts: Advice for expert witnessess. In L. J. Alison (Ed.), The forensic psychologist’s casebook (pp. 170–193). London: Willan.
Palermo, G. B., & Kocsis, R. N. (2005). Offender profiling: An introduction to the sociopsychological analysis of violent crime. Springfield, IL: CC Thomas.
Pakes, F., & Pakes, S. (2009). Criminal psychology. Devon, UK: Willan.
Pinizzotto, A. J., & Finkel, N. J. (1990). Criminal personality profiling: An outcome
and process study. Law and Human Behavior, 14(3), 215–233.
Prentky, R. A., Burgess, A. W., & Rokus, F. (1989). The presumptive role of fantasy
in serial sexual homicide. American Journal of Psychiatry, 146, 887–891. Presetsky, F. J. (1986). Legal and criminal psychology. New York: Copley Publishing. Proulx, J., Beauregard, E., Cusson, M., & Nicole, A. (2007). Sexual murderers: A
comparative analysis and new perspective. New York: Wiley.
Ressler, R. K., & Shachtman, T. (1992). Whoever fights monsters. New York: Simon
Revitch, E., & Schlesinger, L. B. (1989). Sex murder and sex aggression. Springfield,
IL: CC Thomas.
Rogers, R. (2008). Clinical assessment of malingering and deceit (3rd ed.). New York:
Safarik, M. E., Jarvis, J., & Nussbaum, K. (2000). Elderly female serial sexual
homicide: A limited empirical test of criminal investigative analysis. Homicide
Studies, 4(3), 294–307.
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 Behaviour, 34(4), 437–453.
Snook, B., Eastwood, J., Gendreau, P., & Bennell, C. (2010). The importance of
knowledge cumulation and the search for hidden agendas: A reply to Kocsis, Middledorp and Karpin (2008). Journal of Forensic Psychology Practice, 10(3), 214–223.
Toch, H. (1961). Legal and criminal psychology. New York: Holt, Rinehart & Winston.
Toch, H. (2000). Legal and criminal psychology. New York: International Thomson Publishing.
Torres, A. N., Boccanccini, M. T., & Miller, H. (2006). Perceptions of the validity and utility of criminal profiling among forensic psychologists and psychiatrists. Professional Psychology: Research and Practice, 37, 51–58.
Wrightsman, L. S. (2001). Forensic psychology. Belmont, CA: Wadsworth.
The Criminal Profiling Reality: What is Actually Behind the Smoke and Mirrors?
Richard N. Kocsis PhD
To cite this article: Richard N. Kocsis PhD (2013) The Criminal Profiling Reality: What is Actually Behind the Smoke and Mirrors?, Journal of Forensic Psychology Practice, 13:2, 79-91, DOI: 10.1080/15228932.2013.765733
To link to this article: https://doi.org/10.1080/15228932.2013.765733
Published online: 21 Feb 2013.
Submit your article to this journal Article views: 8649
View related articles
Citing articles: 3 View citing articles
Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=wfpp21
Journal of Forensic Psychology Practice, 13:79–91, 2013 Copyright © Taylor & Francis Group, LLC
ISSN: 1522-8932 print/1522-9092 online
The Criminal Profiling Reality: What is Actually Behind the Smoke and Mirrors?
RICHARD N. KOCSIS, PhD
Psychologist, Sydney, Australia
Despite developments in both the research and investigative use of criminal profiling over the past four decades, empirical evaluation of the accuracy of the technique, and hence its validity, remains limited. This practice update briefly examines the original empir- ical experiments conducted in the area thus far as well as recent omnibus analyses aimed at assessing accuracy in criminal profiles. Issues that undermine recently promulgated challenges to profiler accuracy are also identified and discussed. Findings evidencing accuracy in criminal profiling from both the original and omnibus studies are also reviewed along with the theoretical implications for differing schools of thought concerning the criminal profiling technique.
KEYWORDS criminal profiling, accuracy, criminal investigative analysis, investigative psychology
The technique commonly and colloquially known as criminal profiling has, over the past four decades, steadily grown both in its use by law enforcement agencies and in terms of research endeavors invested into its development (Kocsis, 2009). Despite these advances in both practice and theory, a relative paucity remains surrounding one of the most critical issues concerning the technique—evaluation of its accuracy. This article provides an overview of the empirically based experiments conducted to date that have attempted to
Address correspondence to Dr. Richard N. Kocsis, P.O. Box 662, Dee Why, NSW 2099, Australia.
80 R. N. Kocsis
quantitatively evaluate accuracy in criminal profiling as well as subsequent omnibus analyses of this research. The findings of the original experiments are reviewed for the purpose of informing the reader’s contemporary under- standing of the accuracy and validity of criminal profiling as well as the implications for the various schools of thought concerning the technique.
THE QUASI-EXPERIMENTAL STUDIES ASSESSING PROFILER ACCURACY
Owing to a variety of logistical and methodological obstacles, robustly measuring accuracy in criminal profiles is not as straightforward or as ele- mentary a task as may first appear (see Kocsis, Middledorp, & Karpin, 2008, pp. 245–248 for full discussion). In light of such issues, what has arguably come to be viewed as one of the optimal methods (in terms of overcoming some of the difficulties for testing accuracy) has been quasi-experimental studies that have sought to measure the capabilities of profilers. Thus, via such experiments, the accuracy of predictions made by profilers has been measured and serves as an indication of the validity of the technique.
These experiments have typically shared a common, yet generic, design. This has been to undertake comparisons of various sampled groups of par- ticipants engaged in a simulated profiling task using case materials from previously solved crimes in conjunction with a multiple-choice question- naire as a means of recording responses concerning predicted characteristics of the likely offender(s). As the cases were previously solved, model answers in the questionnaire concerning the offender’s characteristics were available. The questionnaire facilitated a means by which the answers of respondents (upon examining the presented case materials) could be uniformly quan- tified, objectively scored for accuracy (via the model answers), and then compared among differing participants in terms of relative proficiency in correctly identifying the characteristics (i.e., profile) of the offender(s).1 The participants in these various studies were recruited and grouped based upon commonly shared skill-based designations. Thus, sampled participants typ- ically included groups such as police personnel, psychologists, students, controls, and suitably qualified professionals who commonly operated under the colloquial label of profiler.
The first experiment of this kind appeared as a sub-component of the research reported in Pinizzotto and Finkel (1990) wherein the predictions of trained profilers, psychologists, detectives, and university students were compared after examining a sexual assault and murder case. The results of
It should be noted that several different subscales were typically incorporated into the design of these experiments; however, for brevity, discussion is confined to the measure of overall (i.e., total) accuracy.
The Criminal Profiling Reality 81
this experiment found that the profilers were superior to the other compared groups in accurately predicting the characteristics of the likely offender in the sexual assault matter but not the murder case. Thereafter, a series of studies by Kocsis, Irwin, Hayes, and Nunn (2000); Kocsis, Hayes, and Irwin (2002); and Kocsis (2004) undertook similar experiments using various groups such as profilers, police personnel, arson investigators, psychologists, students, controls, and even psychics who examined case materials relating to murder or arson offenses. A gradual pattern began to emerge from these differing studies in finding some evidence to suggest that the proficiency of the sam- pled profilers in predicting the characteristics of the unknown offender(s) was at a superior level of accuracy (on average) in comparison to the var- ious other groups who were likewise tested. The last of these studies was Kocsis (2003),2 which provided an omnibus type analysis that incorporated together the data from the three previous studies by Kocsis and colleagues as well as some further original data. This research yielded a holistic, yet indi- vidualized, group analysis of the performance of profilers in comparison to all other sampled groups of participants in both of the tested crime modal- ities (i.e. murder and arson). The results of Kocsis (2003) found that the sampled profilers surpassed all of the other compared groups in accurately predicting the characteristics of the unknown offender(s).
THE RE-ANALYSIS ARTICLES: CHALLENGES TO PROFILING ACCURACY
Following these experiments, the extant literature in the field began to feature a number of publications that, in the author’s opinion, presented inherently problematic re-analyses of the aforementioned original experi- ments. Unfortunately, this circumstance has contributed to some confusion over the proper interpretation of the findings from the original studies as well as their implications more generally for the validity of criminal profiling and its differing approaches. Moreover, these re-analyses prompted the pro- duction of several articles to address some of the arising misunderstandings (e.g. Dern et al., 2009; Kocsis, 2006b, 2010; Kocsis et al., 2008).
In the author’s view, the confusion can largely be traced to the material germinated in two discrete articles. The first is Bennell, Jones, Taylor and Snook (2006), and the second is Snook, Eastwood, Gendreau, Goggin, and Cullin (2007). Both of these articles share a number of similarities. Neither article features any original data in assessing profiler accuracy. Rather, these re-analyses were generated by synthesizing data taken from Pinizzotto and
Although Kocsis (2003) chronologically precedes Kocsis (2004), this circumstance was due to the publication schedules of the differing journals. Accordingly, the research reported in Kocsis (2004) did, in fact, occur prior to the study published as Kocsis (2003).
82 R. N. Kocsis
Finkel (1990) and/or manuscripts by Kocsis and colleagues. In addition, both of these manuscripts appear to feature a contradiction in the premises adopted in their re-analysis. That is, Bennell et al. (2006) and Snook et al. (2007) both seem to suggest that the aforementioned original experiments are methodologically compromised but thereafter, nonetheless, undertake re-analysis of the data generated from these same apparently compromised experiments. Put simply, both Bennell et al. and Snook et al. appear to contend that the identical data that informed the findings of the original experiments are unsound yet, simultaneously, are sufficiently robust for the purpose of conducting their re-analyses.3 Last, both Bennell et al. (2006) and Snook et al. (2007) present their own versions of omnibus type analyses of the aforementioned original experiments, and the findings of both are said to demonstrate that the profilers are not proficient (i.e., accurate) and thus do not support validity in criminal profiling.
In the author’s opinion, the re-analyses presented in Bennell et al. (2006) and Snook et al. (2007) feature anomalies that serve to undermine their arguments concerning profiler accuracy. The nature of these errors do not appear to be founded in subjective constructs, but are instead based in apparent misunderstandings fundamentally incumbent in the methodology employed in their re-analyses. For brevity and in the context of this prac- tice update only, key points surrounding these errors will be canvassed to provide the reader with an overview of these difficulties.4
THE ISSUE IN BENNELL, JONES, TAYLOR, AND SNOOK (2006)
The misconception in Bennell et al. (2006), it appears, lies in a failure to correctly comprehend the data that were used in the analysis and informed the findings reported in Kocsis (2003). As previously mentioned, the study by Kocsis (2003) represented a combined analysis of the data collected in Kocsis et al. (2000, 2002), Kocsis (2004), as well as some previously unused original data. Description of this methodological parameter is described in Kocsis (2003) first on page 127 and again in similar terms on page 132. For example, “As the purpose of this article is to present a holistic impression of the data from all three of the previous studies as well as to incorporate additional data not previously used. . . (Kocsis, 2003, p. 132).
Unfortunately, it seems that Bennell et al. (2006) have overlooked this as their re-analysis appears to be based on the inaccurate premise that the
et al. (2008, pp. 253–255) and Kocsis (2010, p.227–228).
A detailed examination of the implications from this apparent contradiction is provided in Kocsis
Kocsis et al. (2008), and Kocsis (2010) for a more detailed explanation.
In this context, readers are recommended to consult, in conjunction with this article, Kocsis (2006a),
The Criminal Profiling Reality 83
Kocsis (2003) study represents only a compilation of the three previous publications (i.e., Kocsis et al., 2000, 2002; Kocsis, 2004). As a consequence, the tabulated percentages presented in Bennell et al. (2006, Table 1) reflect an inaccurate representation of the data which was used in Kocsis (2003) and which informed its findings concerning profiler accuracy. Thus, the re- analysis of statistics presented in Bennell et al. to challenge the findings concerning profiler accuracy reported in Kocsis (2003) does not, in real- ity, reflect all the data that were used in Kocsis (2003). The scope of this anomaly was discussed in Kocsis (2006b, pp. 461–465) where it was argued that the criticisms concerning profiler accuracy conveyed in Bennell et al. (2006) failed to account for approximately 27% of the profiler data that was used in the Kocsis (2003) study.
THE ISSUE IN SNOOK, EASTWOOD, GENDREAU, GOGGIN, AND CULLIN (2007)
The findings pertaining to profiler accuracy in the Snook et al. (2007) arti- cle also appear to be conveyed to varying degrees in Snook, Eastwood, Gendreau, Goggin, and Cullen (2008) as well as Snook, Cullen, Bennell, Taylor, and Gendreau (2008). This is significant for two reasons. First, it is important to not confuse the repeated presentation of this material as con- stituting multiple original studies replicating any findings concerning profiler accuracy. Second, any error that is identified and thus argued to debunk the assertions pertaining to profiler accuracy in Snook et al. (2007) logically extend also to all of these other publications.
The misconception in Snook et al. (2007) appears to begin with their claim that the manuscript by Hazelwood, Ressler, Depue, and Douglas (1995) indicates that individuals possessing investigative experience, such as police personnel, are equivalent to the capabilities of profilers. With this postulate, Snook et al. appear to undertake two separate meta-analyses using varying combinations of values derived from the articles by Pinizzotto and Finkel (1990) as well as those by Kocsis and colleagues. Their first meta-analysis adopted the aforementioned premise and combined the values of sampled police personnel with those of profilers and then compared this merged group (which was labeled as Profilers) with the values from all other partic- ipants, who were labeled under the collective rubric of non-profilers. In the second meta-analysis, the values of the sampled profilers were not merged with any other participants (such as the police personnel) and were ana- lyzed in a homogenous stand-alone capacity against all other participants who were, again, collectively described as non-profilers.
Unfortunately, and quite contrary to the assertion noted in Snook et al. (2007), the manuscript by Hazelwood et al. (1995) does not, in any capacity, appear to assert that police personnel are equivalent or can in some way be
84 R. N. Kocsis
equated to the skills, expertise, and proficiencies of profilers.5 Consequently, the rationale for Snook et al. combining profiler scores with police personnel scores is misconceived and the methodological basis for this combination accordingly limited.6 Compounding this is that none of the police personnel in the studies by Kocsis and colleagues (from which approximately 90% or more of all police values originate7) claimed to be profilers or possessed any education, training, or experience profiling crimes (see Kocsis, 2010, p. 228). In light of this, it therefore seems anomalous to expect the variously sampled police personnel, who themselves represented a very diverse skill-leveled group,8 to be commensurate with the profilers. The problem arising from this circumstance is that the challenges concerning profiler accuracy argued in Snook et al. (2007) largely derive from the results of this first meta-analysis wherein the values from profilers and police personnel are combined and compared against non-profilers.
Undermining the Snook et al. assertions further are the results of their own second meta-analysis when duplicated and reproduced in Kocsis et al. (2008). Via this process, it was demonstrated that in the Snook et al. second meta-analysis, the accuracy scores of the profilers surpassed the non-profilers at statistically significant levels.9 Consequently, the challenges concerning profiler accuracy argued in Snook et al. (2007) do not appear to be grounded in a methodologically sound premise (i.e., the first meta-analysis combin- ing profilers with police personnel) and contradicted in the second analysis when a merger of police and profiler scores does not occur.
CRIMINAL PROFILING ACCURACY: THE ILLUSIONS AND THE REALITIES
As previously mentioned, the issues argued to debunk the assertions con- cerning profiler accuracy by Bennell et al. (2006) and Snook et al. (2007)
issue concerning the combination of police personnel with profilers.
Readers are recommended to refer to Kocsis et al. (2008, p. 251–253) for a full examination of this
It should be noted that one of the most common methodological criticisms associated with flawed meta-analyses is the inappropriate combination of data into common variables. Readers may refer to Ioannidis and Lau (1999) and Esteban, Hernadez, and Kattan (2008) for a more detailed statistical discussion of these principles related to problematic applications of meta-analysis.
In this context, the sampled police personnel in the various studies by Kocsis and colleagues
A total sample size of six police participants features in the experiment by Pinizzotto and Finkel
(1990). All other police participant data originates from the various studies conducted by Kocsis and
originated from quite diverse skill levels and were sampled to explore differing facets both quantitative
and qualitative that may be associated with levels of experience in police investigations. For this reason,
the sampled police personnel were, themselves, quite diverse, ranging from minimal levels of experience
and training observed in police recruits, intermediary levels of experience such as with trainee detectives
through to senior detectives with many years of vocational specialization in investigating crimes (e.g.,
Kocsis et al., 2002).
Readers are recommend to refer to Kocsis et al. (2008, Table 1, pg. 256) for a full description of
The Criminal Profiling Reality 85
extend to all other literature that seeks to rely in some capacity on these propositions. Moreover, as indicated in Kocsis et al. (2008) before con- tentions akin to those claimed in such articles can convincingly be made, a substantially larger amount of original data-driven research will need to be undertaken. Such research will require the capabilities of profilers in general and thereafter the capabilities of profilers who adopt differing approaches to be far more comprehensively evaluated in a variety of different crime modal- ities and with a variety of different experimental case simulations. For these precise reasons the conclusions reported in Kocsis et al. (2008, p. 258) were unambiguously described as constituting inchoate evidence concerning a promising forensic psychiatric/psychological technique. It was also for this reason that Kocsis (2010) argued that researchers need to embark upon the far more arduous and more logistically difficult path of generating genuinely original data-driven research with appropriate populations.
Unfortunately, the most concise scientifically robust answer in apprais- ing the accuracy (i.e., validity) of the criminal-profiling technique at this time is that there is no clear or simple answer. What currently exists are a number of productive and very promising beginnings that warrant further develop- ment and measured exploration in the future. Arguments claiming that one approach to criminal profiling has been conclusively demonstrated to be defective or alternatively superior are, in the author’s opinion, unlikely to be helpful or indeed supported by the extant scientific literature on the topic (see Kocsis, 2006b, 2009, 2010; Kocsis et al., 2008).
Additionally, it must also be noted that, akin to previous observations of this nature (e.g., Farrington, 2007; Kocsis, 2003; Kocsis & Palermo, 2007; Kocsis et al., 2008; Pinizzotto, 1984), the research canvassed in this practice update assesses the issue of accuracy and thus validity of the criminal- profiling technique. However, this research has not explored and, therefore, does not extend to the separate issue of utility and in turn evaluating opti- mal methods for the application of criminal profiles in augmenting criminal investigations and their operational impact in terms of resolving crime.
With these various caveats and parameters outlined, what then are the implications for the embryonic, albeit promising, beginnings concerning the accuracy of criminal profiling? Arguably, the first factor is to be aware of the research describing the perceptual illusions surrounding the technique and, in particular, evaluations of accuracy via intuitive means such as anecdotal accounts or repeated applications (e.g. Alison, Smith, & Morgan, 2003; Kocsis 2006a, pp. 13–26; Norris, Rafferty, & Campbell, 2010; Torres, Boccanccini, & Miller, 2006). Additionally, there are also a growing number of useful studies highlighting various inherent skills and attributes that are likely to be asso- ciated and reflected in the proficient construction of criminal profiles (e.g., Gogan, 2007; Hodges & Jacquin, 2008; Pinizzotto & Finkel, 1990; Kocsis, 2006a, pp. 49–64).
With respect to the fundamental question of accuracy in criminal profil- ing, there is now empirical evidence, albeit modest, indicating that suitably
86 R. N. Kocsis
qualified professionals collectively operating under the colloquial man- tle of profiler can decisively surpass the predictions of other compared skill-based groups. There are, however, a number of different dimensions to the aforementioned studies that warrant further consideration in how they may individually inform our understanding of accuracy in criminal profiling.
First, contrary to the Gestalt adage of “the whole is greater than the sum of the parts,” the value of the findings obtained through the original indi- vidual studies appear, in the author’s opinion, to have been overlooked in the wake of the various recent omnibus analyses. The qualitative dimen- sion and depth of detail these original individual research efforts offer in seeking to examine how proficient profiling is achieved and in what con- texts are, in the author’s view, likely to be more beneficial to the future development of the technique.10 In many instances, they better illustrate a host of subtle complexities inherent in both the theoretical development as well as operational practice of constructing criminal profiles that often appear lost in the appraisals conveyed in many of the recent omnibus analyses.
With respect to the contribution of the various omnibus analyses, Table 1 summarizes the main results pertaining to total profile accuracy scores from each of the studies and the various compared groups.
In interpreting these findings, it is of equal importance, in the author’s view, to appreciate the context applicable to each of these analyses as well as their results. The first issue is that while three omnibus analyses have been produced, which all derive from the original individual experiments variously conducted by Pinizzotto and Finkel (1990) and/or Kocsis and col- leagues, they each reflect three different sets of data. That is, each of the three omnibus analyses use different combinations of participants variously drawn from the different original experiments and, thus, actually reflect, three dis- tinct comparisons. To assist in clarifying this issue, the key presented in Table 1 identifies in which of the original experiments the omnibus analyses were respectively developed. The second issue to note is that the data used in the two meta-analyses reported in Snook et al. (2007) were duplicated in Kocsis et al. (2008). However, the actual values in the meta-analytic com- parisons between profilers and non-profilers were displayed in Kocsis et al. (2008) as “binomial effect size display” (BESD) statistics followed by some additional analysis to test for statistically significant differences amongst these BESD values. Consequently, the Kocsis et al. (2008; column A) duplicates the analysis wherein profiler and police personnel data were merged in Snook
Citation details can be found in the reference list with some of the key findings also summarised in Kocsis (2006a, pp. 27–64) as, in the author’s view, the optimal development of profiling into the future will come from further replication and development of studies of this ilk.
TABLE 1 Total Accuracy Scores
The Criminal Profiling Reality
Profilers Non-profilers Psychologists Science Students Specialist Detectives Psychics
Non-police Specialist General Police Police Recruits Stereotype Controls
0.31 −0.43 −0.14 0.12 0.07 0.17 −0.36
Kocsis et al. (2008) (A)
Kocsis et al. (2008) (B)
Note: Kocsis (2003): Comparison of sampled skilled-based groups of participants converted into z scores. Value closest to 1.0 denotes level of accuracy. Analysis generated from Kocsis et al. (2000, 2002) and Kocsis (2004) in conjunction with previously unused additional data. Kocsis et al. (2008; column A): Analysis combines profiler and police values compared against all other participants (labeled as non-profilers). Meta-analysis values displayed as “binomial effect size display” (BESD) percentage values. Analysis derived from Pinizzotto and Finkel (1990), Kocsis (2004), Kocsis et al. (2002), and Kocsis et al. (2005). Kocsis et al. (2008; column B): Analysis featuring profiler values only compared against all other participants (labeled as non-profilers). Meta-analysis values displayed as BESD percentage values. Analysis derived from Pinizzotto and Finkel (1990), Kocsis (2004), Kocsis et al. (2000), and Kocsis et al. (2005). ∗Denotes difference at statistically significant level of 0.05.
et al. (2007). Likewise, the Kocsis et al. (2008; column B) duplicates the stand-alone analysis from Snook et al. wherein the profiler values were not merged with any other participants.
With these parameters in mind, the findings of these three differ- ent omnibus analyses can be better understood. It should also be noted that there is broad consensus in the results (irrespective of methodologi- cal considerations) in that the aggregated accuracy scores of the profilers surpass those of all the other compared groups of participants in all of these analyses.11 Despite this preliminary observation, the findings presented under Kocsis (2008; column A) should not be viewed as a compelling assessment of profiler performance in light of the aforementioned method- ological problems apparently applicable to Snook et al. (2007). Instead, only the early analysis in Kocsis (2003) and thereafter Kocsis et al. (2008; col- umn B) should be considered when appraising profiler accuracy in such an omnibus capacity. Additionally, the demarcation between these two studies should be clearly borne in mind, as the Kocsis et al. (2008; column B) anal- ysis does not reflect an ideal assessment of profiler accuracy. Instead, the optimal omnibus analysis at this time remains, in the author’s opinion, the original analysis reported in Kocsis (2003) as it is not imbued with any of the
relevant findings are also found in the subscales that are not reported in this manuscript.
It should be noted that only total accuracy values are canvassed in the present article and that
88 R. N. Kocsis
limitations identified in Kocsis et al. (2008) and in particular Kocsis (2010).12 In this context, the Kocsis (2003) analysis features comparisons of all the individual participant groups present and clearly defined with measures of overall accuracy and all subscales available for consideration. Thus, simplis- tic dichotomous categorizations of profilers versus non-profilers are avoided and, instead, individual patterns in the results among each of the respective skill-based groups and the control participants are clearly discernible. These findings indicate that the profilers achieved the highest level of accuracy in comparison to all other compared groups.
The original experiments such as those undertaken by Kocsis and col- leagues pursued a simple objective of seeking to empirically test the accuracy of predictions made by individuals operating under the collo- quial mantle of profiler. The somewhat rudimentary research question that motivated these experiments was to establish whether any evidence existed that might suggest that criminal profiling actually works in terms of accurate predictions concerning the offender. The results from these early experiments produced some empirical, albeit tentative, findings demon- strating proficiency among the profilers in comparison to various other groups of participants. However, with the wisdom of hindsight, the mod- est ambitions of these early experiments appear to have also led to some unintended and unforeseen implications for the broader theoretical bases underpinning the criminal profiling technique in general. As identified first in Kocsis (2006b, pp. 459–460), some of the differing schools of thought and approaches to criminal profiling (e.g., Palermo & Kocsis, 2005) argue that the only scientifically valid, ergo accurate, method for criminal profil- ing can be accomplished via the adoption of their practices and approach. In this context, research such as that by Pinizzotto and Finkel (1990) as well as by Kocsis and colleagues, which demonstrates that the profi- cient profiling of crimes can be accomplished by individuals who do not employ such practices, arguably serves to undermine the cogency of such arguments.
Readers are recommended to consult Kocsis et al. (2008) and in particular Kocsis (2010, pp. 228–231) for a detailed explanation of these issues. Briefly, the primary purpose of the analysis presented in Kocsis et al. (2008) was to duplicate the analyses presented in Snook et al. (2007) for the purpose of highlighting the methodological flaws inherent in these authors’ assertions concerning profiler accuracy. In this regard, the Kocsis et al. (2008) analysis was not intended as an optimal evaluation of profiling accuracy, as the second analyses (where profiler and police values are not combined) was still likely to suffer some methodological limitation. For example, as explained in Kocsis (2010), there are arguably significant differences between the levels of qualifications and expertise in the sampled individ- uals collectively labeled as profilers between the studies by Pinizzotto and Finkel (1990) and those by Kocsis and colleagues.
The Criminal Profiling Reality 89
In view of the paucity of genuinely original data-driven experimenta- tion testing profiler accuracy, it seems difficult to conceive how any of the recent re-analysis studies can be required in lieu of the arguably far-greater imperative for further original experimentation. In the author’s opinion, this circumstance seems even more perplexing when re-analysis of the original experiments is undertaken by authors who seem to view them as defective. As previously mentioned, the contradiction that seems apparent in Bennell et al. (2006) and Snook et al. (2007) appears difficult to reconcile. The position of these authors appears to be that the original experiments suf- fer from methodological limitations that render them unsound. That being so, it therefore seems inconsistent to undertake re-analysis of experiments that are identified as compromised and to, in turn, contend that the findings developed from this same data are suitably robust for the purpose of fresh conclusions (see Kocsis, 2010, pp. 227–228).13
Moreover, these re-analysis studies appear to feature omissions, misun- derstandings, and inconsistencies as previously reported (see Kocsis, 2006, pp. 460–466; Kocsis et al., 2008, pp. 251–257; Kocsis, 2010, pp. 225–227). However, what is perhaps most disconcerting, as canvassed in this practice update, is the unfortunate circumstance of what are argued to be miscon- ceptions that nonetheless seem to facilitate the finding of results that serve to challenge the proficiency of profilers.
In reflecting upon the empirical research thus far, readers should rec- ognize that while some long-overdue scientific evidence demonstrating accuracy in criminal profiling is available, it remains a technique requiring considerably more development and, in particular, further original data- driven experimentation to test its merits. Moreover, it is also hoped that readers will appreciate that what has come to be known colloquially as the smoke and mirrors surrounding the field is not merely confined to perceptual illusions but is instead a more multifaceted phenomena.
Alison, L. J., Smith, M. D., & Morgan, K. (2003). Interpreting the accuracy of offender profiles. Psychology, Crime and Law, 9(2), 185–195.
Bennell, C., Jones, N., Taylor, P. J., & Snook, B. (2006). Validities and abilities in criminal profiling: A critique of the studies conducted by Richard Kocsis and his colleagues. International Journal of Offender Therapy and Comparative Criminology, 50(3), 344–360.
Dern, H., Dern, C., Horn A., & Horn, U. (2009). The fire behind the smoke: A reply to Snook and colleagues. Criminal Justice and Behavior, 36(10), 1085–1090.
considered unsound would be to not analyze the same material perceived as flawed.
In the author’s opinion, a conventional position to adopt if the original experiments were
90 R. N. Kocsis
Esteban, W., Hernandez, A., & Kattan, M. (2008). Meta-analysis: Its strengths and limitations. Cleveland Clinical Journal of Medicine, 75(6), 431–439.
Farrington, D. P. (2007). Book review: Criminal profiling: Principles and prac- tice, by Richard Kocsis (2006). International Journal of Offender Therapy and Comparative Criminology, 4(51), 486–487.
Gogan, D. (2007). Investigative experience and profile accuracy: A replication study. In R. N. Kocsis (Ed.), Criminal profiling: International theory, research and practice (pp. 365–382) Totowa, NJ: Humana Press.
Hazelwood, R. R., Ressler, R. K., Depue, R. L., & Douglas, J. E. (1995). Criminal personality profiling: An overview. In R. R. Hazelwood & A. W. Burgess (Eds.), Practical aspects of rape investigation: A multidisciplinary approach (2nd ed., pp. 115–126). Boca Raton, FL: CRC Press.
Hodges, E. P., & Jacquin, K. M. (2008). Psychological skills and criminal profile accuracy. In R. N. Kocsis (Ed.), Serial murder and the psychology of violent crimes. (pp. 259–276), Totowa, NJ: Humana Press.
Ioannidies, J. P., & Lau, J. (1999). Pooling research results: Benefits and limitations of meta-analysis. The Joint Commission Journal on Quality Improvement, 25(9), 462–469.
Kocsis, R. N. (2003). Criminal psychological profiling: Validities and abilities. International Journal of Offender Therapy and Comparative Criminology, 47(2), 126–144.
Kocsis, R. N. (2004). Psychological profiling of serial arson offenses. An assessment of skills and accuracy. Criminal Justice and Behavior, 31(3), 341–361.
Kocsis, R. N. (2006a). Criminal profiling: Principles and practice. Totowa, NJ: Humana Press.
Kocsis, R. N. (2006b). Validities and abilities in criminal profiling. The dilemma for David Canter’s investigative psychology. International Journal of Offender Therapy and Comparative Criminology, 50(4), 458–477.
Kocsis, R. N. (2009). Applied criminal psychology: An introduction to forensic behavioral sciences. Springfield, IL: CC Thomas.
Kocsis, R. N. (2010). Criminal profiling works and everyone agrees. Journal of Forensic Psychology Practice, 10(3), 224–237.
Kocsis, R. N., Hayes, A. F., & Irwin, H. J. (2002). Investigative experience and accuracy in psychological profiling of a violent crime. Journal of Interpersonal Violence, 17(8), 811–823.
Kocsis, R. N., Middledorp, J., & Karpin, A. (2008). Taking stock of accuracy in crim- inal profiling: The theoretical quandary for investigative psychology. Journal of Forensic Psychology Practice, 8(3), 244–261.
Kocsis, R. N., Middledorp, J. T., & Try, A. C. (2005). Cognitive processes in crimi- nal profile construction: A preliminary study. International Journal of Offender Therapy and Comparative Criminology, 49(6), 662–681.
Kocsis, R. N., & Palermo, G. B. (2007). Contemporary problems in criminal profiling. In R. N. Kocsis (Ed.), Criminal profiling: International theory, practice and research, (pp. 327–346), Totowa, NJ: Humana Press.
Norris, G., Rafferty, E., & Campbell, J. (2010). Examination of content preference in offender profiles. International Journal of Offender Therapy and Comparative Criminology, 54(3), 412–429.
The Criminal Profiling Reality 91
Palermo, G. B., & Kocsis, R. N. (2005). Offender profiling: An introduction to the sociopsychological analysis of violent crime. Springfield IL: CC Thomas.
Pinizzotto, A. J. (1984). Forensic psychology: Criminal personality profiling. Journal of Police Science and Administration, 12(1), 32–40.
Pinizzotto, A. J., & Finkel, N. J. (1990). Criminal personality profiling: An outcome process study. Law and Human Behavior, 14, 215–233.
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, 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(4), 437–453.
Snook, B., Eastwood, J., Gendreau, P., Goggin, C., & Cullen, R. M. (2008). Taking stock of criminal profiling: A narrative review and meta-analysis. In C. R. Bartol, & A. M. Bartol (Eds.), Current perspectives in forensic psychology and criminal behavior (2nd ed., pp. 183–201). Newbury Park, CA: Sage
Torres, A. N., Boccanccini, M. T., & Miller, H. (2006). Perceptions of the validity and utility of criminal profiling among forensic psychologists and psychiatrists. Professional Psychology: Research and Practice, 37, 51–58.
Perceptions of the Validity and Utility of Criminal Profiling Among Forensic Psychologists and Psychiatrists
Angela N. Torres, Marcus T. Boccaccini, and Holly A. Miller Sam Houston State University
Criminal profiling is the process of using crime scene evidence to make inferences about potential suspects, including personality characteristics and psychopathology. An exploratory Internet survey of forensic psychologists and psychiatrists was conducted to examine their experiences with and opinions about profiling and to determine whether referring to profiling as “criminal investigative analysis” had any impact on these opinions. About 10% of the 161 survey respondents had profiling experience, although more than 25% considered themselves knowledgeable about profiling. Fewer than 25% believed that profiling was scientifically reliable or valid, and approximately 40% felt that criminal investigative analysis was scientifically reliable or valid. Although the scientific aspects of profiling lacked support, respondents viewed profiling as useful for law enforcement and supported profiling research.
Keywords: profiling, criminal investigative analysis, general acceptance, admissibility, forensic
The image of a criminal profiler sorting through crime scene evidence and definitively identifying a guilty suspect is popular in novels, television shows, and movies. However, this popular im- age is more fiction than fact, and the process and limits of real profiling work are often misunderstood. Criminal profiling is the process of using behavioral evidence left at a crime scene to make inferences about the offender, including inferences about person- ality characteristics and psychopathology. In its most basic form, profiling is simply the postdiction of behavior; an action has taken place that allows investigators to make inferences about the person responsible (Davis & Follette, 2002). Despite popular images of criminal profiling, the main goal of profiling in real investigations is to narrow the scope of a suspect pool rather than to identify a single guilty criminal (Douglas & Olshaker, 1995).
Mental Health Practitioners and Criminal Profiling
Psychiatrists and psychologists made significant contributions to the early development of criminal profiling. The profiles created
ANGELA N. TORRES is currently a student in the clinical psychology PhD program at Sam Houston State University. Her areas of interest in research include criminal profiling; sexual offenders; eating disorders; body image; female offenders; lesbian, gay, bisexual, and transgender issues; and advocacy. MARCUS T. BOCCACCINI received his PhD in clinical psychology from The University of Alabama. He is an assistant professor in the Department of Psychology at Sam Houston State University and conducts research in the areas of witness preparation, expert testimony, and psychological assess- ment practices.
HOLLY A. MILLER received her PhD in clinical psychology from Florida State University. She is an associate professor in the College of Criminal Justice at Sam Houston State University. She teaches, consults, and conducts research in the areas of malingered psychopathology, assessment and treatment of offend- ers, psychopathy, and law enforcement training, assessment, and evaluation. CORRESPONDENCE CONCERNING THIS ARTICLE should be addressed to Marcus T. Boccaccini, Psychology Department, Box 2447, Sam Houston State University, Huntsville, TX 77341. E-mail: [email protected]
by these practitioners were based largely on clinical judgment and prevailing theories of personality and psychopathology. For exam- ple, many early profiles from mental health practitioners were rooted heavily in psychoanalysis (Grubin, 1995). These profiles focused on the probable interpersonal functioning and psychopa- thology of the person responsible for committing the crime (Wil- son, Lincoln, & Kocsis, 1997). Psychiatrist William Langer devel- oped one of the most famous early profiles— a profile of Adolf Hitler for the U.S. Office of Strategic Service during World War II (Pinizzotto & Finkel, 1990). Langer’s psychiatric approach led to a profile of Hitler’s personality and mental disorder as well as to Langer’s prediction of suicide as Hitler’s response to defeat (Pin- izzotto, 1984). The most lauded and cited example of early pro- filing comes from the work of another psychiatrist. In 1956, David Brussel constructed a profile of New York’s “Mad Bomber” that was accurate in many details (see Weinerman, 2004a).
Although some criminal profiling work is done by mental health professionals (e.g., psychologists and psychiatrists), most of it is done by trained law enforcement agents. It is difficult to get a clear estimate of how many mental health professionals are involved in profiling. The best data come from a survey of 152 police psy- chologists, who reported spending approximately 3% of their time profiling offenders (Bartol, 1996). However, Bartol did not report a standard deviation value for this finding, making it impossible to infer whether most psychologists spent 3% of their time conduct- ing profiles or whether a few professionals spent a large proportion of their time conducting profiles while others avoided this work completely. Although Bartol’s published survey results cannot help researchers determine the number of respondents who did and did not participate in profiling, Bartol did find that 70% of the police psychologists “seriously” questioned the validity of profil- ing work (p. 79). This combination of findings from the Bartol study suggests that many respondents were not actively involved in generating profiles. It is possible, but unlikely, that the same psychologists who expressed negative views about profiling were actually participating in a significant amount of profiling-related
This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
TORRES, BOCCACCINI, AND MILLER
activities. The applicability of Bartol’s findings to other groups of mental health professionals is not clear. Larger groups of profes- sionals have not been surveyed about their participation in profil- ing work.
The pattern suggested by the Bartol (1996) survey—that is, that a small number of psychologists participate in profiling work and the vast majority avoid it because of concerns about validity—fits well with our perception of the relationship between psychology and criminal profiling during the past 20 years. However, this pattern may be changing. A recent issue of the American Psycho- logical Association’s (APA) monthly professional magazine, Mon- itor on Psychology, highlighted the role of psychology and psy- chologists in criminal profiling (Weinerman, 2004a, 2004b). Several psychologists who are active in profiling work were in- terviewed, and, although the articles clearly acknowledged that many concerns about the scientific validity of profiling still exist, the reader is left with the impression that interest in profiling work by psychologists is on the rise. This trend is also evident in the recent publication of profiling research by psychologists (e.g., Canter, Alison, Alison, & Wentink, 2004; Canter & Wentink, 2004; Kocsis, 2004; Salfati, 2003; Salfati & Canter, 1999) and coverage of profiling in recent editions of psychology–law text- books (Bartol & Bartol, 2004; Constanzo, 2004; Wrightsman & Fulero, 2005), suggesting that professional interest and practice in this area also may be on the rise. Indeed, the use of psychological profiles in criminal investigations has been described as “burgeon- ing” (Kocsis & Hayes, 2004, p. 149).
When reviewing profiling research, one must recognize that multiple approaches to profiling exist, each with different under- lying theories, methods, and assumptions (see Homant & Kennedy, 1998, for a concise overview; see also Turvey, 2002). Most profiling studies examine the scientific merit of aspects of these specific approaches as opposed to the profiling process in general. These studies can be thought of as examinations of con- struct validity in that they gauge support for the theories and concepts underlying various components of profiling approaches. For example, one focus of recent profiling research has been the validity of the offender typologies underlying some profiling ap- proaches. For instance, in several studies Canter and colleagues coded information from police files and found limited support for typologies sometimes used to classify serial killers, such as the classification of crime scenes as organized or disorganized and use of the Holmes and Holmes system that includes visionary, mission, hedonistic, and power or control killers (see Canter et al., 2004; Canter & Wentink, 2004). These findings raise serious questions about the validity of approaches that rely on these types of dis- tinctions for generating profiles. However, researchers also have used data from real cases to develop more empirically supported typologies of murderers and rapists (Canter, Bennell, Alison, & Reddy, 2003; Salfati, 2003; Salfati & Canter, 1999; see also Kocsis, Cooksey, & Irwin, 2002). Future research in this area clearly is needed to determine whether psychologists can use these typologies to accurately classify offenders in future cases and to provide useful information for investigators.
A second line of profiling research has focused on the reliability and accuracy (criterion validity) of professional profilers. Formal
examinations of the reliability of profiling are limited to a study of six Federal Bureau of Investigation (FBI) profilers’ agreement in classifying crime scenes as organized, disorganized, mixed, or unknown (Ressler & Burgess, 1985, as reported in Homant & Kennedy, 1998). The identification of a crime scene as organized or disorganized has important implications in some profiling ap- proaches. Agreement among the agents in classifying crime scenes in 64 cases was reported to be 74%. The researchers concluded that the overall level of agreement found in the study indicated ade- quate reliability; however, there was notable variation in the ac- curacy of the agents. For instance, the classification accuracy of one rater was only 51.7%.
Several studies have examined the accuracy of profiles gener- ated by professional profilers. The general design used in these studies has been to compare profiles generated by profilers with those generated by nonprofilers (e.g., students, psychologists, po- lice detectives). One important limitation of this research was that most of these studies used fewer than five profilers because the researchers had difficulty finding profilers willing to participate in the research (see Kocsis, 2004; Pinnizzotto & Finkel, 1990). Nev- ertheless, the existing studies in this area seem to converge in finding that trained profilers write longer and more detailed reports that contain more information and predictions about the offender (Kocsis, 2003b; Pinizzotto & Finkel, 1990).
Existing accuracy studies also provide some support for profil- ers making predictions that are more accurate than those of non- profilers, although results are somewhat mixed and are based on a small number of profilers and studies (see Kocsis, 2003a, for a review). For example, Pinizzotto and Finkel (1990) examined the profiles generated by four current or former FBI profilers, six police detectives with profiling training, six police detectives with no profiling training, six clinical psychologists with no profiling training, and six undergraduate students with no profiling training. Each participant generated profiles for one murder case and one arson case on the basis of actual police investigation materials from the two cases. Participants also answered 15 multiple-choice questions about the suspect, including age, race, occupation, the relationship between the victim and the offender, and the likeli- hood that the offender had committed similar crimes in the past. Researchers knew the correct response to each question. In the profiling of the murder case, all the participants performed better than chance, with the average number of correct responses for each group ranging from 5.3 (35%) to 7.0 (47%); however, no signif- icant differences were observed between the groups of partici- pants. Thus, the accuracy of the professional profilers was similar to that of untrained psychologists and students. In the profiling of the sexual assault case, the 10 profilers performed significantly better than the other three groups of participants. Again, all groups performed at an above-chance rate, but the average number of correct responses for profilers was 10.0 out of 15.0 (67%), com- pared with correct response averages ranging from 5.5 (37%) to 8.5 (57%) for the other three groups.
Kocsis (2003a) summarized the results of three studies con- ducted by his research team comparing the accuracy of profes- sional profilers (N ???? 11, across studies), psychologists, self- identified psychics, college students, and various groups of law enforcement officers. In each study, participants reviewed crime descriptions and case materials from actual solved cases (i.e., murder, arson) and completed multiple-choice questionnaires
This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
about the suspect. Across these studies, the profilers provided the largest number of correct predictions about the actual perpetrators, with accuracy rates ranging from 46% (Kocsis, Irwin, Hayes, & Nunn, 2000) to 70% (Kocsis, 2004). However, Kocsis (2003b) cautioned that a notable amount of variability was observed within the profiler group, indicating that some profilers were much more accurate than others. It is unclear whether this variation was due to overall differences in the profilers’ skills or to specific aspects of the cases.
Profiling Admissibility in Court
When profiles are created, both the prosecution and the defense may attempt to have the profiler testify as an expert witness. Profiling evidence has been admitted and denied admission into court. Many professionals who engage in profiling work believe that profiling testimony is more likely to be admitted into court when it is called something other than profiling (Cooley & Turvey, 2002; Risinger & Loop, 2002). For example, in United States v. Meeks (1992), testimony that was based on motivational analy- sis—the name sometimes used by the Federal Bureau of Investi- gation (FBI) for profiling—was admitted because it was found to be a generally accepted investigative technique that is based on specialized knowledge (Cooley & Turvey, 2002). Profiling also has been referred to as behavioral evidence analysis, linkage analysis, signature analysis, and investigative analysis. The FBI, which originally coined the term profiling, now uses criminal investigative analysis to refer to its profiling work (Turvey, 2002).
Although motivational analysis testimony was allowed in United States v. Meeks (1992) because it was found to be a generally accepted investigative tool, profiling testimony has been found inadmissible because it was not perceived as being generally accepted (see Grezlak, 1999). A defense attorney in one case in which a psychologist’s profiling testimony was barred from being entered into evidence openly referred to profiling as “voodoo” (Grezlak, 1999, p. 3).
One likely reason for the discrepant decisions concerning the admissibility of profiling testimony is that there is little documen- tation one way or the other about whether profiling, regardless of what it is called, is a generally accepted investigative technique. Just as perceptions of profiling are often based on lore, so are opinions about its acceptance. The general acceptance of profiling has not been systematically examined among mental health pro- fessionals, and arguments that support the acceptance of profiling in the law enforcement community typically are based on the rationale that profiling must be accepted if it continues to be used (see Kocsis & Hayes, 2004).
In all states and federal jurisdictions, general acceptance is a criterion that judges consider when making decisions about the admissibility of expert testimony. General acceptance in the field is the main criterion on which decisions are to be based in states relying on the Frye (Frye v. United States, 1923) standard for determining the admissibility of expert testimony. In Daubert v. Merrell Dow Pharmaceuticals (1993), the Supreme Court ruled that the Federal Rules of Evidence, and not the Frye v. United States standard, should be used for determining the admissibility of expert testimony in federal courts. Many states have since adopted Daubert-like standards for admitting expert testimony.
Daubert v. Merrell Dow Pharmaceuticals (1993) departs from Frye v. United States (1923) by asking judges to consider the scientific merit of the theories, procedures, and research on which potential experts will base their testimony. The Supreme Court noted in Daubert v. Merrell Dow Pharmaceuticals that general acceptance is one criterion that can be used for evaluating scien- tific merit but also recommended that judges consider other sources of evidence, including whether the theory or technique on which the testimony is based has been subjected to peer review, whether that theory or technique can be falsified, and whether it has a known and acceptable error rate. Under Daubert v. Merrell Dow Pharmaceuticals, any theory or procedure upon which ex- perts rely should be based on a falsifiable theory that has been tested, be subjected to peer review, be published, have a known error rate, and be generally accepted in the field (Goodman- Delahunty, 1997).
Survey Results as Evidence for General Acceptance
Error rates of profiling methods, the falsifiability of profiling theories, and the peer review status of profiling theories and methods, are, for the most part, objective criteria that can be identified at any point in time through the current scholarly liter- ature. Although arguments concerning the general acceptance of profiling work can be informed by findings from existing empir- ical research, general acceptance can be examined directly by surveying the opinions of those in the relevant profession or field. For example, researchers have used surveys to systematically examine general acceptance within the psychology field of the validity of polygraph testing (Honts & Quick, 1995; Iacono & Lykken, 1997) and factors influencing the accuracy of eyewitness memory (Kassin, Ellsworth, & Smith, 1989; Kassin, Tubb, Hosch, & Memon, 2001).
The purpose of the current study is to use a survey approach to explore perceptions of the validity and utility of criminal profiling among forensic psychologists and psychiatrists. This study also seeks to provide information about the proportion of mental health professionals in these fields who are active in profiling work, to explore similarities and differences in opinions about profiling from forensic psychologists and psychiatrists, and to examine whether changing the term criminal profiling to criminal investi- gative analysis has an impact on perceptions of its acceptance and validity.
Exploratory Profiling and Criminal Investigative Analysis Survey
Identifying Appropriate Mental Health Professionals
Because there is no professional society or organization dedi- cated to mental health professionals involved in profiling work, we asked members of professional organizations with a more general focus on mental health and law issues to complete an Internet survey about the scientific merit and utility of profiling. These organizations were Division 41 (American Psychology–Law So- ciety [AP-LS]) of the APA, the American Academy of Forensic Psychology (AAFP), the American Academy of Psychiatry and Law (AAPL), and the Police and Public Safety section in APA Division 18, Psychologists in Public Service. AP-LS member
This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
TORRES, BOCCACCINI, AND MILLER
e-mail addresses were obtained from the list of attendees posted on the 2004 AP-LS conference Web site. We cross-checked names from the AP-LS Web site with the APA membership directory to ensure that those asked to participate were psychologists, as op- posed to graduate students, attorneys, or other affiliate members. We e-mailed AAFP members directly, using addresses available through the organization’s Web site. The presidents of AAPL and of the Police and Public Safety section of APA Division 18 requested that we first send e-mails soliciting participation to them, which they would then forward to their respective members.
E-mails soliciting participation were distributed to 1,637 foren- sic mental health professionals: 175 AP-LS members, 147 AAFP members, 840 AAPL members, and 475 members of the Police and Public Safety section of APA Division 18. Only 9.9% (N ???? 161) of those who received an e-mail solicitation completed the study. Although this response rate was admittedly lower than we had hoped for and clearly places limits on the applicability of the findings to forensic mental professionals as a whole, the completed surveys provide information about the perceptions of profiling by a sizable group of forensic mental health professionals (N ???? 161). Moreover, a low response rate was anticipated for several reasons. Few practitioners in these four organizations were expected to be active in profiling, and many were expected to avoid answering questions about profiling because they felt it was outside the scope of their field. Although the Internet can be a cost-effective method for collecting survey data, response rates for Internet surveys are often markedly lower (ranging from 7% to 44%) than those for mail and telephone surveys (Schonlau, Fricker, & Elliot, 2002). In addition, response rates for Internet surveys of academic and professional groups tend to be lower than those of other groups (Cook, Heath, & Thompson, 2000). The expected low response rate was the main reason for soliciting participation from such a large group of forensic mental health professionals (N ???? 1,637).
Although the survey data reported in this article provide infor- mation about the practices and beliefs of 161 forensic mental health professionals, the low response rate raises legitimate ques- tions about whether the sample is representative and, thus, whether the findings can be generalized to forensic mental health profes- sionals as a whole. The low response rate also suggests that it is unlikely that respondents were able to accurately express the amount of variation in beliefs and practices that is present among forensic mental health professionals as a whole. It may be that those who responded were strong proponents or advocates of profiling and that the survey findings represent either an overly positive or an overly negative view of profiling. For this reason, we strongly caution readers against interpreting the findings re- ported here as clear evidence either for or against the acceptance of profiling in the field. We consider this study to be an early exploratory step in moving the fields of psychology and psychiatry toward a more empirically based understanding of the acceptance or nonacceptance of profiling. Readers must interpret the survey findings with this important limitation in mind.
Of the responding forensic mental health professionals, 44.7% (n ???? 72) reported AP-LS membership, 19.3% (n ???? 31) reported AAFP membership, 44.1% (n ???? 71) reported AAPL membership, and 8.1% (n ???? 13) reported Police and Public Safety section membership. Many participants reported membership in multiple organizations. With respect to profession, 50.9% (n ???? 82) identi- fied themselves as psychologists, 43.5% (n ???? 70) identified them-
selves as psychiatrists, 3.1% (n ???? 5) reported having a doctorate in medicine and a doctorate in philosophy, and 2.5% (n ???? 4) did not provide information about their profession.
Survey Format and Questions
A major goal of this study was to examine whether perceptions of the validity, utility, and acceptance of profiling methods varied depending on whether these methods were described as “profiling” or “criminal investigative analysis.” Two versions of the study survey were created that were identical in all aspects except one: one version used the term profiling in the instructions to partici- pants and in the study questions, whereas the other version used criminal investigative analysis. The latter is the term currently used by the FBI to refer to profiling. The definition of profiling provided to mental health professionals was “Profiling is the process of using physical and behavioral evidence available at a crime scene to make inferences about the personality characteris- tics of the individual(s) responsible for committing the criminal act.” In the alternate version of the survey, criminal investigative analysis was substituted for profiling, but the definitions were otherwise identical.
The first five survey questions asked participants to provide yes or no responses to questions about their training, knowledge, and experience with profiling methods or criminal investigative anal- ysis (see Table 1). The second five questions asked participants to provide yes or no responses to questions about the scientific and professional status of profiling or criminal investigative analysis (see Table 2).
The final five questions assessed respondents’ knowledge of standards used for admitting expert testimony into court, including Daubert v. Merrell Dow Pharmaceuticals (1993), Frye v. United States (1923), and the Federal Rules of Evidence.1 Each item was scored as correct or incorrect. We totaled the scores on the five items to form an admissibility knowledge composite score, which we found to have acceptable internal consistency (Kuder– Richardson 20 ???? .72).
For each professional organization, mental health practitioners with last names beginning with A–M were directed to one version of the survey, and those with last names beginning with N–Z were directed to the other version. The order was counterbalanced for the different organizations. For two organizations, those with names beginning with A–M were directed to the profiling survey. For the other two organizations, those with names beginning with A–M were directed to the criminal investigative analysis survey.
Experiences and Attitudes of Psychologists and Psychiatrists
Preliminary analyses revealed only one significant difference between psychologists and psychiatrists on the 10 main survey questions concerning training, knowledge, experience, and atti- tudes about profiling. This one effect revealed that psychiatrists were somewhat more likely to say that profiling was a useful tool for law enforcement (95.4%) than were psychologists (84.9%), ????2
1 A copy of the five-item questionnaire assessing knowledge of admis- sibility standards is available from Marcus T. Boccaccini.
Do you consider yourself knowledgeable about the field of
Profiling (n ???? 92)
31.5% 16.3% 10.9%
CIA (n ???? 69)
26.1% 20.3% 10.1% 10.1% 13.0%
Percentage of Surveyed Forensic Mental Health Professionals With Profiling Knowledge, Training, and Experience
Do you have any training in ?
Have you ever generated a for an investigation? Have you ever testified in court about a ?
Have you ever been asked about your opinion of
Note. CIA ???? criminal investigative analysis. Percentage values in this table reflect the proportion of “yes” responses to survey items. All differences between profiling and CIA failed to reach statistical and clinical significance, p ???? .05, phi coefficients (????) ???? .06.
(1, N ???? 172) ???? 4.12, p ???? .05, ???? ???? .17. Although this difference is large enough to reach statistical significance, the majority of respondents from both professions felt that profiling was a useful tool for law enforcement. Because the overall pattern of findings across the 10 main survey questions was similar for psychologists and psychiatrists, findings are reported below for the entire sample of forensic mental health professionals.
Training, Knowledge, and Experience With Profiling
Table 1 provides information concerning the proportion of fo- rensic mental health professionals who reported having training, knowledge, and experience with profiling and criminal investiga- tive analysis. A similar pattern of findings was observed for participants responding to questions about profiling and criminal investigative analysis. Although a sizable minority of forensic mental health professionals reported having knowledge of profil- ing and criminal investigative analysis (31.5% and 26.1%, respec- tively) and training in these techniques (16.3% and 20.3%, respec- tively), only about 10% reported ever having generated a profile or criminal investigative analysis for an actual case. Similarly, only about 10% reported having testified in court about their use of these techniques or their opinion of these techniques in general.
Attitudes About Profiling and Criminal Investigative Analysis
Forensic mental health professionals’ attitudes about the scien- tific merit and utility of profiling and criminal investigative anal- ysis are reported in Table 2. Fewer than half of the forensic mental health professionals felt that either profiling or criminal investiga- tive analysis was reliable, was valid, or had enough scientific support to be admitted into court. Perceptions were especially negative among those who received the version of the survey using the term profiling. These forensic mental health professionals rated profiling to be significantly less reliable and valid than those who were given the term criminal investigative analysis for the same process. These findings support the position that profiling is likely to be viewed more favorably if it is referred to by another name. Recall that the definitions provided for profiling and criminal investigative analysis were identical. Only the name of the process varied.
Although forensic mental health professionals questioned the scientific merit of profiling and criminal investigative analysis, there was general agreement that these techniques were useful tools for law enforcement. In addition, nearly all forensic mental
Percentage of Surveyed Forensic Mental Health Professionals Who Identified Profiling and Criminal Investigative Analysis (CIA) as Being Valid and Useful
Is scientifically valid for linking a defendant to a crime?
Is scientifically reliable?
(n ???? 92) 23.0%
CIA (n ???? 68)
43.3% 40.0% 31.1%
????2(1, N ???? 160) ???? 6.84* .22
9.69** .26 0.31 .05
3.03 .14 0.10 ????.02
Do you think that any circumstances?
Do you think that Do you think that
is scientifically supported enough to be admitted into court under
is a useful tool for law enforcement? should be researched empirically?
Note. Percentage values in this table reflect the proportion of “yes” responses to survey items. ???? ???? phi,
as a correlation coefficient. In this table, ???? provides an index of the strength of the association between the label given to the profiling process and positive perceptions of scientific merit and utility. Larger positive values of ???? indicate more positive evaluations for CIA as opposed to those evaluations for profiling.
*p ???? .05. **p ???? .01.
effect size for ????2 analyses that can be interpreted
TORRES, BOCCACCINI, AND MILLER
health professionals felt that profiling and criminal investigative analysis should be the focus of empirical research.
Because many forensic mental health professionals reported that they did not consider themselves knowledgeable about profiling and criminal investigative analysis, it can be argued that only the perceptions of those who are knowledgeable should be considered for gauging general acceptance. To address this concern, we com- pared beliefs about the validity, reliability, and scientific support for profiling and criminal investigative analysis for those who did and did not consider themselves knowledgeable about these meth- ods. Small cell sizes precluded separate analyses for profiling and criminal investigative analysis. No significant differences between those with and without knowledge were revealed, with effect sizes (????) ranging from .01 to .05. For the 47 forensic mental health practitioners who reported being knowledgeable about profiling or criminal investigative analysis, 27.7% (n ???? 13) felt that these procedures were scientifically valid, 25.5% (n ???? 12) felt that they were reliable, and 25.5% (n ???? 12) felt that they had adequate scientific support for being admitted into court.
Knowledge about standards for admitting expert testimony into evidence was assessed to examine whether those with higher levels of admissibility knowledge were more likely to see profiling and criminal investigative analysis as lacking adequate scientific merit. The average number of correct responses to the admissibility questions across the 158 mental health professionals who com- pleted all five questions was 3.65 (SD ???? 1.48). Psychologists (n ???? 80, M ???? 3.99, SD ???? 1.35) exhibited significantly more knowledge about admissibility standards on the five-question scale than psy- chiatrists did (n ???? 69, M ???? 3.26, SD ???? 1.46), t(147) ???? 3.15, p ???? .02, Cohen’s d ???? 0.52.
A scientific merit composite score was created by totaling the “yes” responses to the survey questions about the reliability, va- lidity, and scientific support for profiling (see Table 1). Scores on this composite could range from 1 to 3, with high scores reflecting positive attitudes about the scientific merit of profiling. These scores were then correlated with scores on the admissibility knowl- edge scale using Spearman rank-order correlations (rs). Among forensic mental health professionals as a whole, admissibility knowledge was negatively related to beliefs about the scientific merit of profiling and criminal investigative analysis, rs ???? ????.16, p ???? .05. Those with higher levels of admissibility knowledge tended to view profiling as having less scientific merit. However, further examination revealed that this trend was almost completely accounted for by psychologists (rs ???? ????.26, p ???? .05). Admissibility knowledge and beliefs about the scientific merit of profiling were not significantly correlated for psychiatrists, rs ???? ????.03, p ???? .05.
Implications for Practice
The purpose of conducting this exploratory survey was to pro- vide an estimate of the acceptance of profiling among forensic mental health professionals. Although the low response rate nec- essarily limits the strength of the conclusions that can be drawn from the survey findings, this survey should be considered an early step in moving the field toward a more empirically based under- standing of the acceptance of profiling.
Mental Health Professionals’ Involvement in Profiling
Few forensic mental health professionals responding to the survey had ever conducted a profile or criminal investigative analysis, yet a somewhat larger minority considered themselves knowledgeable about these techniques. Why would those who claim to be knowledgeable about profiling not conduct profiles? Three possible interpretations for this pattern of findings are of- fered here: (a) one that argues against mental health professionals’ involvement in the practice of profiling, (b) one that is more positive about involvement in profiling, and (c) one that is open minded about involvement in profiling.
The first interpretation is that the knowledge possessed by some respondents suggests to them that profiling lacks scientific merit. Support for this interpretation is provided by the finding that both profiling and criminal investigative analysis were seen as lacking adequate reliability, validity, and scientific support by more than half of all respondents. This pattern of findings was similar for those who did and did not consider themselves knowledgeable about profiling, suggesting that many with knowledge about pro- filing felt that profiling and criminal investigative analysis lacked scientific merit. This perspective may be especially common among psychologists who are active practitioners and who are knowledgeable about standards for admitting expert testimony into court. Psychologists who said that they were more knowledgeable about admissibility standards tended to have the most negative views about the scientific merit of profiling and criminal investi- gative analysis.
The second, more positive interpretation is that some forensic mental health professionals have become familiar with profiling methods or research (or both) because they have been asked about profiling by attorneys, journalists, or students. Some of these forensic mental health professionals may be willing to undertake profiling work but have never been hired to do so.
The third and more open-minded interpretation is that some forensic mental health professionals are keeping a watchful eye on developments in profiling but are not yet comfortable practicing in this area. During the past 5 years, many of the top peer-reviewed psychology–law journals have published articles concerning pro- filing. These journals include Psychology, Public Policy, and Law (Alison, West, & Goodwill, 2004; Canter et al., 2004), Law and Human Behavior (Davis & Follette, 2002; Wells, 2003), Behav- ioral Sciences and the Law (Canter et al., 2003; Salfati & Canter, 1999; Snook, Canter, & Bennell, 2002), and Criminal Justice and Behavior (Canter & Wentink, 2004; Kocsis, 2004). Those involved in forensic practice or research may have seen these articles or been given information about the latest developments in profiling through continuing education outlets. This interpretation also is supported by the finding that nearly all forensic mental health practitioners believed that profiling was worthy of scientific study. Many respondents who questioned the scientific merit of profiling still believed that more research should be conducted.
These three interpretations are not mutually exclusive. All or none of these rationales may apply to a single forensic mental health professional who is knowledgeable about profiling but is not involved in profiling work. Those individuals who are consid- ering becoming involved in profiling work should be aware that some forensic mental health professionals who are knowledgeable about profiling avoid working in this area. At this point in time,
profiling does not appear to be generally accepted among forensic mental health professionals as a valid and scientifically supported technique. Whether research advancements in this area will even- tually improve the opinions of profiling among mental health practitioners is not clear. The finding from the current study that approximately 70% of mental health practitioners believed that profiling or criminal investigative analysis was not a valid tech- nique is nearly identical to Bartol’s (1996) finding from nearly 10 years ago that 70% of police psychologists “seriously questioned” the validity of profiling (p. 79).
However, findings from the current study do suggest that some forensic mental health professionals harbor negative views about the term profiling rather than, or in addition to, the profiling process. Survey participants who were asked to express their opinions about criminal investigative analysis (another name for profiling), expressed much more positive views. On the basis of this finding, we recommend that those individuals who are asked to comment on or testify about profiling think carefully about the basis for their opinion of profiling before making any statements. Beliefs about the validity of profiling should be based on existing evidence rather than on negative attitudes about the term profiling. Many of the recent scholarly articles addressing the current status of profiling research are cited in this article. We encourage those individuals who are asked to comment or testify about profiling to become familiar with this growing body of research literature.
Limitations and Conclusions
Although findings from the current study show that many fo- rensic mental health professionals have concerns about the scien- tific merit of profiling, these findings alone are insufficient for arguing that profiling evidence should not be admitted into court. Views about the acceptance of profiling as a valid technique may be higher among other professionals who do profiling work, in- cluding law enforcement agents and criminal justice experts. In addition, most participants felt that profiling was a useful tool for law enforcement, suggesting that they believe someone should be doing this work.
The seemingly counterintuitive pattern of respondents reporting insufficient evidence for the validity of profiling but endorsing its use for law enforcement potentially could be explained in multiple ways. The most likely explanation is that these forensic mental health professionals think about profiling in a similar way to that of professionals in the law enforcement community by reasoning that profiling must be at least somewhat useful if it continues to be used (see Kocsis & Hayes, 2004). Indeed, it is possible that these forensic mental health professionals have seen, or have at least heard of, the process being effective for law enforcement and are willing to admit that it may be useful in some cases. A second explanation is that respondents may have believed that more information about the scientific merit of profiling is needed before they are willing to assert that it is not at all useful and should be discontinued. In other words, given the small amount of profiling research available, they may have believed that the existing evi- dence against profiling is not strong enough to argue that it should never be used. A third explanation is that the respondents to the survey represent a biased sample of forensic mental health profes- sionals. It may be that a large portion of the responding sample was made up of individuals who have some interest or training in
profiling, and, thus, they were likely to view its use within law enforcement as important. However, this explanation seems less plausible given that the overwhelming majority of the respondents viewed profiling as invalid and unreliable and did not rate profiling positively, overall.
One final limitation of this survey is that participants were provided with a general description of profiling. There are many different approaches to profiling, with different underlying theo- ries, methods, and assumptions. For those knowledgeable about profiling, responses may have varied if we had asked about a specific approach. Major approaches to profiling include the FBI’s criminal investigative analysis (Douglas & Burgess, 1986), behav- ioral evidence analysis (Turvey, 2002), investigative psychology (Canter, 1995), and geographic profiling (Rossmo, 1999). It may be that support for some of these approaches is stronger than others, and future studies may seek to examine support for differ- ent profiling approaches.
Alison, L., West, A., & Goodwill, A. (2004). The academic and the practitioner: Pragmatists’ views of offender profiling. Psychology, Pub- lic Policy, and Law, 10, 71–101.
Bartol, C. R. (1996). Police psychology: Then, now, and beyond. Criminal Justice and Behavior, 23, 70–89.
Bartol, C. R., & Bartol, A. M. (2004). Psychology and law (3rd ed.). Wadsworth: Belmont, CA.
Canter, D. (1995). Criminal shadows: Inside the mind of the serial killer. London: HarperCollins.
Canter, D. V., Alison, L. J., Alison, E., & Wentink, N. (2004). The organized/disorganized typology of serial murder: Myth or model? Psy- chology, Public Policy, and Law, 10, 293–320.
Canter, D. V., Bennell, C., Alison, L., & Reddy, S. (2003). Differentiating sex offenses: A behaviorally based thematic classification of stranger rapes. Behavioral Sciences and the Law, 21, 157–174.
Canter, D. V., & Wentink, N. (2004). An empirical test of Holmes and Holmes’s serial murder typology. Criminal Justice and Behavior, 31, 489 –515.
Constanzo, M. (2004). Psychology applied to law. Wadsworth: Belmont, CA.
Cook, C., Heath, F., & Thompson, R. L. (2000). A meta-analysis of response rates in web- or Internet-based surveys. Educational and Psy- chological Measurement, 60, 821–836.
Cooley, C. M., & Turvey, B. E. (2002, June). Reliability and validity: Admissibility standards relative to forensic experts illustrated by crim- inal profiling evidence, testimony, and judicial rulings. Journal of Be- havioral Profiling, 3(1). Retrieved August 12, 2002, from http://www .profiling.org/journal/index.html
Daubert v. Merrell Dow Pharmaceuticals, 113 S. Ct. 2786 (1993).
Davis, D., & Follette, W. C. (2002). Rethinking the probative value of evidence: Base rates, intuitive profiling, and the “postdiction” of behav-
ior. Law and Human Behavior, 26, 133–158.
Douglas, J. E., & Burgess, A. E. (1986, December). Criminal profiling: A
viable investigative tool against violent crime. FBI Law Enforcement
Bulletin, 55(12), 9–13.
Douglas, J. E., & Olshaker, M. (1995). Mind hunter: Inside the FBI’s elite
serial crime unit. New York: Mindhunters.
Frye v. United States, 293 F. 1013 (D.C. Cir. 1923). Goodman-Delahunty, J. (1997). Forensic psychological expertise in the
wake of Daubert. Law and Human Behavior, 21, 121–140.
Grezlak, H. (1999, April 12). Profiling testimony inadmissible in murder
trial. The Pennsylvania Law Weekly, p. 3.
TORRES, BOCCACCINI, AND MILLER
Grubin, D. (1995, September). Offender profiling. The Journal of Forensic Psychiatry, 6, 259–263.
Homant, R. J., & Kennedy, D. B. (1998). Psychological aspects of crime scene profiling: Validity research. Criminal Justice and Behavior, 25, 319 –343.
Honts, C. R., & Quick, B. D. (1995). The polygraph in 1995: Progress in science and the law. North Dakota Law Review, 71, 987–1019.
Iacono, W. G., & Lykken, D. T. (1997). The validity of the lie detector: Two surveys of scientific opinion. Journal of Applied Psychology, 82, 426 – 433.
Kassin, S. M., Ellsworth, P. C., & Smith, V. L. (1989). The “general acceptance” of psychological research on eyewitness testimony: A sur- vey of the experts. American Psychologist, 44, 1089–1098.
Kassin, S. M., Tubb, V. A., Hosch, H. M., & Memon, A. (2001). On the “general acceptance” of eyewitness testimony research: A new survey of the experts. American Psychologist, 56, 405–416.
Kocsis, R. N. (2003a). Criminal psychological profiling: Validities and abilities. International Journal of Offender Therapy and Comparative Criminology, 47, 126–144.
Kocsis, R. N. (2003b). An empirical assessment of content in criminal psychological profiles. International Journal of Offender Therapy and Comparative Criminology, 47, 37–46.
Kocsis, R. N. (2004). Psychological profiling of serial arson offenses: An assessment of skills and accuracy. Criminal Justice and Behavior, 31, 341–361.
Kocsis, R. N., Cooksey, R. W., & Irwin, H. J. (2002). Psychological profiling of sexual murders: An empirical model. International Journal of Offender Therapy and Comparative Criminology, 46, 532–554.
Kocsis, R. N., & Hayes, A. F. (2004). Believing is seeing? Investigating the perceived accuracy of criminal psychological profiles. International Journal of Offender Therapy and Comparative Criminology, 48, 149– 160.
Kocsis, R. N., Irwin, H. J., Hayes, A. F., & Nunn, R. (2000). Expertise in psychological profiling: A comparative assessment. Journal of Interper- sonal Violence, 15, 311–331.
Pinizzotto, A. J. (1984). Forensic psychology: Criminal personality profil- ing. Journal of Police Science and Administration, 12, 32–40.
Pinizzotto, A. J., & Finkel, N. J. (1990). Criminal personality profiling: An outcome and process study. Law and Human Behavior, 14, 215–233.
Ressler, R. K., & Burgess, A. W. (1985, August). Violent crime. FBI Law Enforcement Bulletin, 1–32.
Risinger, D. M., & Loop, J. L. (2002). Three card monte, Monty Hall, modus operandi, and “offender profiling”: Some lessons of modern cognitive science for the law of evidence. Cardozo Law Review, 24, 193–285.
Rossmo, D. K. (1999). Geographic profiling. Boca Raton, FL: CRC Press. Salfati, C. G. (2003). Offender interaction with victims in homicide: A multidimensional analysis of frequencies in crime scene behaviors.
Journal of Interpersonal Violence, 18, 490–512.
Salfati, C. G., & Canter, D. V. (1999). Differentiating stranger murders:
Profiling offender characteristics from behavioral styles. Behavioral
Sciences and the Law, 17, 391–406.
Schonlau, M., Fricker, R. D., & Elliot, M. N. (2002). Conducting research
surveys via e-mail and the Web. Santa Monica, CA: Rand.
Snook, B., Canter, D., & Bennell, C. (2002). Predicting the home location of serial offenders: A preliminary comparison of the accuracy of human judges with a geographic profiling system. Behavioral Sciences and the
Law, 20, 109–118.
Turvey, B. E. (2002). Criminal profiling: An introduction to behavioral
analysis (2nd ed.). San Diego, CA: Academic Press.
United States v. Meeks, 35 M. J. 64 (CMA 1992).
Weinerman, L. (2004a). Criminal profiling: The reality behind the myth.
Monitor on Psychology, 35(7), 66–69.
Weinerman, L. (2004b). Does profiling really work? Monitor on Psychol-
ogy, 35(7), 67.
Wells, G. (2003). Murder, extramarital affairs, and the issue of probative
value. Law and Human Behavior, 27, 623–628.
Wilson, P., Lincoln, R., & Kocsis, R. (1997). Validity, utility and ethics of
profiling for serial violent and sexual offenders. Psychiatry, Psychology,
and Law, 4, 1–11.
Wrightsman, L. S., & Fulero, S. M. (2005). Forensic psychology (2nd ed.).
Wadsworth: Belmont, CA.
Received August 9, 2004 Revision received June 15, 2005
Accepted June 16, 2005 ????