Does the three-factor model of psychopathy identify a problematic subgroup of young offenders?

https://doi.org/10.1016/j.ijlp.2008.04.003Get rights and content

Abstract

The present study tests the utility of the personality-based three-factor model of psychopathy according to the Hare Psychopathy Checklist: Screening Version (PCL:SV). This model of psychopathy excludes aspects of criminal behavior as opposed to other models of psychopathy. The main research question was to what extent the three-factor model of psychopathy can identify a problematic subgroup of young offenders. The sample consisted of 148 incarcerated young male criminal offenders, between 15 and 25 years of age (mean = 19.07, SD = 2.11) who were recruited from a central detention center for young offenders in a northern German county. Model-based cluster analysis of the three psychopathy factors showed that three different clusters labeled: (i) Unemotional/Impulsive–Irresponsible, (ii) Low traits, and (iii) Psychopathic personality, had the best fit to the data. The psychopathic personality cluster with high scores on all three factors of the PCL:SV exhibited, as expected, a significantly higher prevalence of conduct disorder and substance use problems, but was not significantly different from the other clusters on past criminality and previous incarcerations. In conclusion, the results showed that the three-factor model of psychopathy can be useful in identifying a problematic subgroup of young offenders.

Introduction

It can be argued that psychopathy should be defined and measured in a more personality-based way, excluding actual criminal behavior (Cooke and Michie, 2001, Cooke et al., 2006, Cooke et al., 2004) which today is included in the most commonly used psychopathy assessment instrument, the Psychopathy Checklist (Forth et al., 2003, Hare, 2003a, Hart et al., 1995). Instead, criminal behaviors could be considered as external correlates or consequences of the more basic personality disturbance. The psychopathy construct, including items measuring criminal behavior, has proven quite effective in predicting criminal behavior and particularly violence and to identify a subgroup of offenders at high risk for criminality, violence, (Douglas, Vincent, & Edens, 2006), and substance use problems (Taylor & Lang, 2006). However, it is less clear how useful the more personality-based model of psychopathy (Cooke & Michie, 2001) is in identifying a more problematic subgroup of offenders. The present study addressed this question in a German young offender population.

Most scholars in the field agree broadly that psychopathy is well described as a syndrome of extreme interpersonal, affective, and behavioral/lifestyle traits (Cooke and Michie, 2001, Hare, 2003a, Hare, 2003b; Johansson et al., 2002, Skeem et al., 2003). Exactly how many factors the construct consists of and the content of these factors is a subject of debate and research (see Cooke and Michie, 2001, Cooke et al., 2006, Cooke et al., 2004, Hare and Neumann, 2006, Neumann et al., 2005). Over the past 10–15 years the Psychopathy Checklist—Revised (PCL-R; Hare, 1991, Hare, 2003a) and its screening version (PCL:SV; Hart et al., 1995) have become somewhat of the gold standard for assessing psychopathy among adults in forensic, correctional and psychiatric settings. There is also a youth version of the PCL, adapted for use with adolescents (PCL:YV; Forth, Kosson, & Hare, 2003).

Factor analyses of the PCL-R and SV have identified two separate, but highly correlated factors: Factor I reflects the core affective and interpersonal features (e.g., callousness, grandiosity, superficial charm, shallow affect) of psychopathy, and Factor II represents the chronic unstable and antisocial lifestyle (e.g., irresponsibility, boredom proneness, parasitic lifestyle, impulsivity) (Hare, 1991, Hart et al., 1995). In the most recent version of the PCL-R (second edition; Hare, 2003a), the two-factor model was kept and it was proposed that psychopathy is underpinned by these two factors but that these two factors in turn are underpinned by two facets each. However, both the traditional two-factor model (Hare, 1991) and the newer two-factorial/four-faceted model (Hare, 2003a) have been questioned both on theoretical and empirical grounds (see e.g., Cooke and Michie, 2001, Cooke et al., 2004, Cooke et al., 2006, Lilienfeld, 1994, Lilienfeld, 1998).

Cooke and Michie (2001) have proposed an alternative hierarchical three-factor model of psychopathy. The long-standing clinical and theoretical notion that psychopathy manifests in people in three interrelated (interpersonal, affective, and behavioral) dimensions (e.g., Blackburn, 1998, Cleckley, 1976, Hare, 1991, Lilienfeld, 1994) was one reason for starting this work. Another starting point for developing this three-factor model has to do with the fact that the PCL initially was developed and validated primarily on forensic and correctional samples, and because of this, the PCL conceptualization of psychopathy is weighted quite heavily with items related to criminality and a socially deviant lifestyle (e.g., juvenile delinquency, revocation of conditional release, and criminal versatility). If one is interested in understanding psychopathy as a personality concept and its etiology and relation to criminal and other socially deviant behavior, one can argue that the conceptualization and measurement of psychopathy should not include criminal behavior (see e.g. Cooke et al., 2006). Psychopathy is likely to be linked to criminal behavior in many cases but perhaps not in all. Thus, actual criminal and socially deviant behaviors should perhaps not be part of the definition of the disorder but rather be treated as a potential co-variate of this disorder (Cooke and Michie, 2001, Cooke et al., 2004, Cooke et al., 2006). In fact, because antisocial behavior may be caused by a host of factors other than personality deviation, focusing on antisocial behavior may result in overdiagnosing of psychopathy (Blackburn, 1998, Hare et al., 1991, Harpur et al., 1989).

Cooke and Michie (2001) used a number of large independent forensic, correctional, and civil psychiatric samples in evaluating a new more personality-based three-factor model using PCL-R and PCL:SV data. The model posits that a reduced number of items of the PCL-R and PCL:SV assesses a superordinate factor of psychopathy which is underpinned by the three personality-based factors: (1) Arrogant and Deceitful Interpersonal Style (including items related to lying and deceitfulness), (2) Deficient Emotional Experience (including items related to a lack of empathy, remorse, guilt and inadequate depth of emotions), and (3) Impulsive and Irresponsible Behavioral Style (including items related to impulsivity, irresponsibility, and a lack of realistic long-term goals). This new hierarchical three-factor model was found to fit the data consistently and to fit significantly better than competing models. The three-factor model makes two crucial changes to the two-factor model. First, it divides the original Factor I into separate interpersonal (Arrogant and Deceitful Interpersonal Style) and affective (Deficient Affective Experience) factors. Also, almost half of the items from the PCL-R antisocial behavior factor are taken out based on Item Response Theory results showing that the items were poor indicators of psychopathy (Cooke & Michie, 2001). Since the original work of Cooke and Michie (2001), several studies on adult samples have now confirmed the three-factor model of the PCL (e.g., Cooke et al., 2001, Cooke et al., 2004, Johansson et al., 2002, Skeem et al., 2003).

A large part of the interest in psychopathy construct as assessed by the PCL is due to its relationship to important outcomes such as the modest but consistent relationship to antisocial behavior, particularly future violence among released criminal offenders, both in youths and adults (see Douglas et al., 2006, Edens et al., 2007, Forth et al., 2003, Hare, 2003a, Walters, 2003 for reviews). The relationship between psychopathy and recidivism seems stronger among males than females (and even non-significant for females in some studies) and most studies have shown that the behavioral/lifestyle factor (Factor II) of the traditional two-factor model of the PCL (which includes criminal behavior) is a stronger predictor than the affective/interpersonal factor (Factor I) (see Douglas et al., 2006, Edens et al., 2007, Forth et al., 2003, Hare, 2003a, Walters, 2003 for reviews).

Furthermore, significant relations and a substantial overlap between the PCL and substance use problems have been observed in numerous studies, both in youths and adults (see Forth et al., 2003, Hare, 2003a, Taylor and Lang, 2006 for reviews). This combination has been found to be particularly problematic and related to higher recidivism rates (see Rutherford et al., 2000, Taylor and Lang, 2006), and is therefore important to acknowledge both in research and practice. As is the case with the relationship between criminality and violence and the PCL, the association between the PCL and substance use problems appears to be accounted for largely by the relation between substance use problems and the behavioral/lifestyle factor, that is Factor II of the two-factor model. An important question is whether the substantial overlap between psychopathy and substance use problems and the relation between psychopathy and criminality and violence, remains when applying the more personality-based three-factor model of psychopathy.

It is important to study the manifestations of psychopathy in youths in order to describe the manifestations of psychopathy in this age group, to identify the potential precursors and transitions to adult psychopathy, and to understand the development of severe antisocial behavior (see e.g., Farrington, 2005, Forth and Burke, 1998, Rutter, 2005). Also, it may be that young psychopathic offenders are more malleable than adults and may benefit more from treatment (Forth & Burke, 1998). In fact, recent research shows that the violence potential of adolescents with significant psychopathic features can be reduced through intensive treatment (Caldwell, Skeem, Salekin, & Van Rybroek, 2006). Thus, a proper identification and understanding of the psychopathic subgroup among young offenders seems important. Most studies using the youth version of the Psychopathy Checklist show that the construct operates largely in the same way in youths as in adults (see Edens et al., 2007, Forth et al., 2003, Salekin, 2006 for reviews).

There are at least two different approaches in studying the usefulness of the three-factor model. One is the variable-oriented approach in which the three factors are related (correlated) to outcomes of interest (see Skeem et al, 2003, although on adults). The other approach is a more person-oriented approach in which the three factors are used to derive groups and then group differences in various outcomes are studied (see Vincent, Vitacco, Grisso, & Corrado, 2003). Both approaches are needed in the research process but in the present study we chose to use a person-oriented approach because it is more directly relevant to clinical practice. One person-oriented approach is cluster analysis, where the three factors of the PCL are analyzed, identifying youth with similar patterns across the three factors. Subsequently, the derived groups or clusters can be compared on other variables.

There is one previous study that has done just this in a sample of 259 males, ranging from 12 to 19 years of age (M = 16.97) who were primarily from maximum custody settings (Vincent et al., 2003). The three PCL:YV factors were clustered using common (as opposed to model-based) cluster analysis (Vincent et al., 2003). Based on previous research with children obtaining four clusters, the authors were interested in replicating a four-cluster solution based on the two-factor model of psychopathy (Christian, Frick, Hill, Tyler, & Frazer, 1997; Frick, O'Brien, Wootton, & McBurnett, 1994). Thus, four clusters were specified and the following were gained; (i) a Low traits cluster (n = 74): Low on all three factors, (ii) a Callous-deceitful cluster (n = 63): High on Arrogant and Deceitful Interpersonal Style and Deficient Affective Experience, and relatively low on the Impulsive and Irresponsible Behavioral Style, (iii) an Impulsive cluster (n = 75): Low on the Arrogant and Deceitful Interpersonal Style, relatively low on the Deficient Affective Experience, and high on the Impulsive and Irresponsible Behavioral Style, and (iv) a Psychopathic traits cluster (n = 47): high on all three factors. The four clusters were then compared to external dimensions of interest. In general, past antisocial behavior patterns did not differ between young offenders in the impulsive and psychopathic clusters. Relative to the two cluster groups without impulsive symptoms, both clusters with behavioral symptoms had serious criminal histories characterized by a higher density and higher severity of offenses (more violent offenses). Also, both clusters tended to start problem behavior and drug use at quite a young age, but the psychopathic traits cluster initiated criminality and conduct problems significantly earlier than any other cluster. The psychopathic traits and impulsive clusters were the most serious young offenders in this study. They had engaged in more frequent and severe offending patterns up to the current incarceration. The psychopathic traits cluster was more likely to re-offend sooner and to re-offend violently than the impulsive cluster. Although the impulsive cluster had a high level of any recidivism, this group did not engage in a high level of violent recidivism. The authors concluded that the impulsive cluster seemed to show a trend toward desistance whereas the psychopathic traits cluster was not (Vincent et al., 2003). Limitations of this study include the forced four-cluster solution, the use of common cluster analysis, and the fact that competing cluster solutions were not tested and compared, as is done in model-based cluster analysis (Fraley & Raftery, 2000).

The main question of the present study was if and to what extent the three-factor model of psychopathy can identify a problematic subgroup of young offenders. A sample of German male incarcerated young offenders was assessed and as a first step, we used model-based cluster analysis in clustering the three factors to determine the best fitting types and number of clusters. In the second step, the derived clusters were compared on external dimensions of interest. We expected to find a psychopathic personality cluster with elevated levels on all three factors and that this cluster would be the most problematic in terms of the external (antisocial) dimensions studied (including past criminality, incarcerations, conduct disorder, and substance use problems), as compared to the other clusters derived.

Section snippets

Participants

All males admitted to a juvenile justice institution in a northern county in Germany during the years 2001 and 2003 comprised the target sample. A total of 148 males, newly and consecutively incarcerated delinquents were recruited for the present study (only 12 youths admitted to the institution refused participation). Age ranged between 15 and 25 years (mean age = 19.07, SD = 2.11). Number of registered crimes ranged between 0 and 20 (mean = 2.24, SD = 3.15). Thus, criminal behavior by the youth, but

Results

The focus of the present study is on the potential usefulness of the three-factor model (Cooke & Michie, 2001) of the PCL:SV. Therefore, we subjected the nine PCL:SV items included in the three-factor model to exploratory factor analysis. The items: Poor behavioral controls, Adolescent antisocial behavior, and Adult antisocial behavior are not included in this model. The present sample was too small (n < 200) for conducting confirmatory factor analysis in a reliable way. In the exploratory factor

Discussion

It is not clear as of yet how useful the more personality-based three-factor model of psychopathy (Cooke & Michie, 2001) is in identifying a more problematic group of offenders. The present study addressed this very question. The first most basic and important step of the present study was to investigate which groups or clusters of young offenders could be identified through model-based clustering of the three personality-based factors of psychopathy. Three clusters were identified; (1) an

References (46)

  • CookeD.J. et al.

    Reconstructing psychopathy: Clarifying the significance of antisocial and socially deviant behavior in the diagnosis of psychopathic personality disorder

    Journal of Personality Disorders

    (2004)
  • DouglasK.S. et al.

    Risk for criminal recidivism: The role of psychopathy

  • EdensJ.F. et al.

    Youth psychopathy and criminal recidivism: A meta-analysis of the psychopathy checklist measures

    Law and Human Behavior

    (2007)
  • EdensJ.F. et al.

    Assessment of “juvenile psychopathy” and its association with violence: A critical review

    Behavioral Sciences and the Law

    (2001)
  • FarringtonD.P.

    The importance of child and adolescent psychopathy

    Journal of Abnormal Child Psychology

    (2005)
  • ForthA.E. et al.

    Psychopathy in adolescence: Assessment, violence, and developmental precursors

  • ForthA.E. et al.

    The Hare PCL:YV

    (2003)
  • FraleyC. et al.

    Model-based clustering, discriminant analysis, and density estimation

  • FraleyC. et al.

    MCLUST: Software for model-based clustering, density estimation and discriminant analysis

  • FreeseR.

    Hare PCL:SV. German manual supplement. Deutschsprachige Handbuchbeilage

    (1999)
  • FrickP.J. et al.

    Psychopathy and conduct problems in children

    Journal of Abnormal Psychology

    (1994)
  • FydrichT. et al.

    Strukturiertes Klinisches Interview für DSM-IV. Achse II: Persönlichkeitsstörungen. Interviewheft. Göttingen, Hogrefe [Gibbon M, Spitzer RL, Williams JBW, Smith L, Benjamin, First MB (1997) Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II)

    (1997)
  • HareR.D.

    The Hare Psychopathy Checklist Revised

    (1991)
  • Cited by (40)

    • Validation of the French Youth Psychopathic Traits Inventory-Short Version in a general population sample of emerging adults in France

      2024, Psychologie Francaise
      Citation Excerpt :

      Nonetheless, we could reply that a recent study using person-oriented analysis shows that it is possible to identify, both in men and women, a group of young adults (aged 20 to 24 years old) that were high on all three psychopathy dimensions as measured by this scale (Colins, Fanti, Larsson, & Andershed, 2017). Furthermore, prior person-oriented studies also identified this psychopathic personality group in justice-involved adolescents and adults (Andershed et al., 2008; Vincent et al., 2003), and in community adolescents (Colins, Noom et al., 2012), using various instruments, suggesting that this group can be robustly identified. In view of Cronbach's α obtained, the YPI-S scores demonstrated marginal to acceptable good internal consistency, except for the Behavioral score (.59) in the total sample and in women (.58).

    • Subtypes of homicide offenders based on psychopathic traits

      2017, International Journal of Law and Psychiatry
      Citation Excerpt :

      Relative to primary psychopaths, the secondary psychopaths manifested higher levels of internalizing problems including anxiety and depression, more pronounced borderline personality traits, and inferior interpersonal functioning (Drislane et al., 2014; Gill & Stickle, 2016; Skeem et al., 2007). However, the results of cluster-analytic studies that employed heterogeneous samples were less congruent – most studies suggested a four-group solution (Falkenbach, Barese, Balash, Reinhard, & Hughs, 2015; Falkenbach, Poythress, & Creevy, 2008; Vassileva et al., 2005), yet three-, five-, or even six-cluster solutions also appeared (Andershed, Köhler, Eno Louden, & Hinrichs, 2008; Coid et al., 2012; Falkenbach et al., 2014). Despite important differences between these studies, including the number of derived clusters, many authors concluded that two clusters interpreted as primary and secondary psychopathy had emerged (Falkenbach et al., 2008; Falkenbach et al., 2015; Gill & Stickle, 2016; Vassileva et al., 2005).

    • Gender differences in psychiatric disorders and clusters of self-esteem among detained adolescents

      2014, Psychiatry Research
      Citation Excerpt :

      Fifth, some potentially interesting variables when studying high self-esteem in detained youths were not considered. For example, a small but substantial subgroup of detained adolescents shows the interpersonal, affective and behavioral features of the psychopathy construct (Andershed et al., 2008; Colins and Andershed, 2014). These adolescents feel superior to others and are characterized by a grandiose sense of self-esteem (Vincent et al., 2003).

    • Psychopathic severity profiles: A latent profile analysis in youth samples with implications for the diagnosis of conduct disorder

      2019, Journal of Criminal Justice
      Citation Excerpt :

      Finally, we compared the psychopathic profiles of the forensic sample on key outcome variables (CD diagnosis, comorbidity, recidivism risk, and aggression). We expected to find similar psychopathic profiles in the forensic and community samples, with at least one group with low scores on all three psychopathic traits and another group with high scores on all three psychopathic traits (Andershed et al., 2008; Lee et al., 2010; Nijhof et al., 2011). We also expected to find at least another intermediary group, though it is difficult to formulate hypothesis regarding the trait composition of this group, because of the mixed findings of previous research (Andershed et al., 2008; Lee et al., 2010; Neumann, Schmitt, Carter, Embley, & Hare, 2012; Nijhof et al., 2011).

    View all citing articles on Scopus
    View full text