Introduction
In the Netherlands, about 2000 adolescents exhibiting severe behavior problems are treated in compulsory residential youth care. Although officially these adolescents were not admitted because of criminal activities, 70% had contacts with the police (Nijhof et al.
2010). Generally, psychopathy is found to be more prevalent within adolescents than in childhood (Hare
2003) and we also would expect it to be present in a residential sample. However, the extent of psychopathic traits among adolescents in residential settings is unknown, as studies in these settings are currently lacking. In the present study, we examined whether a residential sample of adolescents can be classified into subgroups based on psychopathic traits and to what extent differences in internalizing behaviors, externalizing behaviors, and drug use characterize these subgroups.
A psychopathic personality involves an arrogant and deceitful interpersonal style, a defective emotional experience (e.g., shallow emotions and a lack of remorse, empathy, and responsibility for one’s own actions), and impulsive, irresponsible, and sensation-seeking behavior (Hare
1991). Studies examining psychopathy mostly identify interpersonal, affective and behavioral dimensions of the construct, although some studies also include a fourth antisocial dimension. Like all personality disorders, a psychopathic personality cannot be diagnosed before the age of 18 (American Psychiatric Association
1994); however, individuals can show psychopathic
traits before this age. The prevalence rates of psychopathic traits in a normal adolescent population were found to range between 5% and 6% (Andershed et al.
2002; Gustafson
2000). The prevalence of psychopathy is higher in boys than in girls (e.g., Bolt et al.
2004; Salekin et al.
1998; Sevecke et al.
2009b).
Psychopathic traits seem to be relatively stable over time (Forsman et al.
2008; Loney et al.
2007; Lynam et al.
2007). Genetic and environmental factors were found to be associated with psychopathic traits (Forsman et al.
2008). Larsson et al. (
2006) showed that common genetic influences explained between 43% and 56% of the variance in the three psychopathic dimensions, which is similar to other studies (Blonigen et al.
2003; Taylor et al.
2003). Larsson et al. (
2006) also revealed that non-shared environmental factors could explain 37% of variance in psychopathic traits, while shared environmental influences did not contribute.
The Youth Psychopathic traits Inventory (YPI; Andershed et al.
2001) and the Psychopathy Checklist: Youth version (PCL: YV; Forth et al.
2003) are widely used instruments to measure psychopathic traits in adolescence. In contrast to the PCL: YV, the YPI does not require rigorous training prior to conducting the assessment (Hillege et al.
2010). Moreover, it was found that self-report measures, such as the YPI, are useful in research on psychopathic traits (Andershed et al.
2002). Studies on the correlations between scores on the YPI and the PCL:YV report low to moderate correlations, suggesting that both instruments measure somewhat different constructs (Andershed et al.
2007; Cauffman et al.
2009; Dolan and Rennie
2006; Skeem and Cauffman
2003).
Although some studies include the antisocial dimension as a fourth dimension, the YPI assesses three dimensions: grandiose/manipulative (interpersonal), callous/unemotional (affective), and impulsive/irresponsible (behavioral) psychopathic personality dimensions (Hare
1991). Using the YPI, Andershed et al. (
2001) found that three subgroups of psychopathic adolescents could be identified within a normative sample. The first subgroup scored low on all three aspects, the second group showed average scores on the affective and interpersonal aspect and high scores on the behavioral aspect. The third group scored high on all three aspects. Using the PCL: YV, Andershed et al. (
2008) reported similar findings when distinguishing subgroups in a sample of male offenders.
Psychopathic traits, externalizing and internalizing behavior
Most research examining the relationship between psychopathy and externalizing problems includes conduct problems and attention deficit hyperactivity disorder (ADHD) as measures of externalizing problems. Both are found to be strongly related to psychopathic traits (e.g., Johansson et al.
2005; Forth et al.
2003; Sevecke et al.
2009b). More specifically, Abramowitz et al. (
2004) found that conduct problems, as well as ADHD, were important predictors of the behavioral dimension of psychopathy in adults. In contrast, for adolescents, Mathias et al. (
2007) found that only the behavioral factor, but not the other psychopathy factors, was related to symptoms of ADHD in partial correlation analyses. Colledge and Blair (
2001) also found that the impulsivity of ADHD among children was related to the behavioral factor of psychopathy. Sevecke et al. reported that conduct disorder (Sevecke et al.
2009a,
b) and ADHD (Sevecke et al.
2009a) contributed to the behavioral dimension of psychopathy in boys. Moreover, conduct disorder was also related to the interpersonal, affective and antisocial dimension (2009a) in boys, while for ADHD these links were not found. In girls, conduct disorder was linked with the affective, behavioral and antisocial factors of psychopathy, while ADHD contributed to all four dimensions of psychopathy (Sevecke et al.
2009a). Sevecke et al. (
2009b) also found that externalizing symptoms, measured using the Youth Self Report (YSR), were associated with the affective dimension of psychopathy in girls, whereas antisocial behavior was related to the behavioral and interpersonal dimension. Using a two-factor model, Abramowitz et al. (
2004) found that both ADHD and childhood conduct problems were stronger predictors for the antisocial lifestyle factor compared to the interpersonal affective factor. In sum, stronger relations were found between externalizing problems and the behavioral and antisocial factor scores than between externalizing problems and the affective and interpersonal factor scores, although some differences between boys and girls were found.
About 15% to 20% of the criminal population is diagnosed as having a psychopathic personality (Hart and Hare
1997). In adolescence, the psychopathy-like group shows the most severe pattern of antisocial behavior compared to other offenders (e.g., Andershed et al.
2002; Lynam and Gudonis
2005). Adolescents with psychopathic traits often engage in delinquent behaviors earlier, show more versatility in offending, commit more violent crimes, and show higher rates of recidivism than other offenders (Lynam and Gudonis
2005; Skeem and Cauffman
2003; Vincent et al.
2003; Walters
2003). Higher scores on psychopathy are related to a higher likelihood of delinquency in both boys and girls (Marsee et al.
2005). Regarding the relationships between the dimensions of psychopathy and delinquent behavior, Christian et al. (
1997) found that adolescents scoring high on the affective dimension, who also had childhood conduct disorder, showed a more serious criminal career than adolescents scoring lower on the affective dimension (see also Barry et al.
2000; Loney et al.
2003). More recent studies, however, revealed that recidivism is more strongly related to the behavioral dimension than to the interpersonal and affective dimension (Douglas et al.
2006; Edens et al.
2007). Salekin (
2008) investigated the association between psychopathy using the PCL:YV and both general as well as violent recidivism. Moderately positive correlations were found between the three dimensions of the factor model of psychopathy and general and violent offending, except for the interpersonal dimension and violent recidivism.
Additionally, drug use is positively related to psychopathic traits (Andershed et al.
2008, Murrie and Cornell
2000; Poythress et al.
2006). Hillege et al. (
2010) found that the interpersonal, affective as well as the behavioral dimension were related to drug use. However, for boys the impulsive/irresponsible dimension showed a stronger relationship with drug use than the other two dimensions, and for girls the affective dimension showed a stronger association with drug use. Poythress et al. (
2006) found that the behavioral dimension was mainly related to drug use in the past. However, Poythress et al. also found a significant association between drug use and the interpersonal dimension. In sum, psychopathic traits are positively associated with both delinquent behavior and drug use. In addition, the combination of psychopathic traits and drug abuse is associated with higher levels of recidivism (Taylor and Lang
2006). More specifically, prior studies suggest that the behavioral dimension is more strongly related to both drugs use and recidivism than the affective and interpersonal dimensions.
Concerning internalizing problems, Poythress et al. (
2006) found a positive direct relation between psychopathic traits and internalizing problems (self-esteem, withdrawal, anxiety, worrisome recurring thoughts, depression, fatigue or loss of interest). Salekin et al. (
2004) confirmed this finding. Moreover, several studies investigated anxiety and the link with psychopathy (e.g., Kosson et al.
2002; Kubak and Salekin
2009; Skeem and Cauffman
2003). In general it was found that anxiety was positively related to psychopathic traits in adolescence. In addition, gender differences emerged in the relationship between psychopathic traits and internalizing problems. Sevecke et al. (
2009b) found that a significant relationship between internalizing symptoms measured by the Youth Self Report (YSR) and the affective dimension of psychopathy existed only for boys. On the other hand, psychopathic traits are also associated with low levels of internalizing problems (Brandt et al.
1997; Frick et al.
1999; Lynam
1997). The relationship between internalizing behavior and psychopathy might be an indirect one because conduct problems are strongly associated with both psychopathic traits and internalizing problems. Moreover, internalizing problems are common in adolescents diagnosed with conduct problems (Lambert et al.
2001). For example, within clinical samples, between 60 to 75% of the children with a conduct disorder also showed anxiety disorders (Russo and Beidel
1993). In sum, psychopathic traits are linked to internalizing problems, probably due to direct or indirect associations with conduct problems.
The present study
The present study will examine whether the three dimensions underlying the ten psychopathic traits of the YPI are present in a clinical adolescent sample referred to residential care. Most studies on psychopathic traits of adolescents included normative samples or samples consisting only of (male) offenders, while our study included males and females from residential settings. The first aim of the study was to test the validity of the YPI in a clinical sample, hypothesising the same factor structure as reported by Andershed et al. (
2001). The next aim was to test whether the adolescents could be classified into different subgroups based on psychopathic traits. Further, it was examined whether these subgroups show different levels of problem behavior, delinquent behavior, and substance use. It was hypothesized, based on the study of Andershed et al. (
2001), that three groups of adolescents with a psychopathic personality can be distinguished: 1) adolescents scoring low on the manipulative, unemotional, and irresponsible dimension (‘relatively normal group’), 2) adolescents scoring moderately on the manipulative and unemotional dimension and high on the irresponsible dimension (‘impulsive, non-psychopathic-like group’), and 3) adolescents scoring high on all three dimensions (‘psychopathy-like group’). Differences between the three subgroups were expected on externalizing problems, delinquent behavior, and substance use, with the psychopathy-like group showing more externalizing problems, delinquent behavior, and substance use compared to both other groups. It is also expected that higher psychopathic traits are related to more externalizing problems, delinquency and drugs use in boys as well as in girls. Concerning internalizing problems and psychopathic traits, the literature showed mixed findings, which makes it difficult to formulate hypotheses.
Discussion
The present study examined the presence of psychopathic traits within a residential sample of adolescents and their relationship with problem behavior. Confirmatory Factor Analyses (CFA) showed that the variance of the YPI subscales could be explained by the three latent constructs, the grandiose/manipulative dimension (interpersonal), the callous/unemotional dimension (affective), and the impulsive/irresponsible dimension (behavioral), replicating the findings of Andershed et al. (
2002). Poythress et al. (
2006) also partly confirmed this finding. These authors found a satisfactory fit when the subscale lying was excluded and when the error terms for callousness and thrill seeking with unemotionality were correlated.
As hypothesized, three meaningful subgroups within the psychopathic personality could be distinguished. One subgroup of adolescents scored low on all three dimensions (normal group), one subgroup had moderate scores on the first two dimensions and relatively higher scores on the impulsive/irresponsible dimension (impulsive, non-psychopathic-like group), and the third subgroup scored high on all three dimensions (psychopathy-like group).
In the next step, statistical differences between the three subgroups on problem behavior were tested, more specifically on internalizing behavior, externalizing behavior, delinquency, and drug use. Our findings showed, as expected, that the normal group showed significant lower levels of problem behavior compared to the impulsive, non-psychopathic-like group and the psychopathy-like group. No differences between the impulsive, non-psychopathic-like and psychopathy-like group were found for externalizing and internalizing problem behavior.
As said earlier, an indirect relationship might be expected because internalizing problems are associated with externalizing problems, which have been found to be related to psychopathic traits. This suggests a correlation between internalizing problems and the behavioral factor of psychopathy. For boys the results indeed revealed a significant correlation between the behavioral factor and internalizing problems and not between the affective and interpersonal factors of psychopathy and internalizing problems. For girls, there was a significant correlation between the behavioral as well as the interpersonal factor of psychopathy and internalizing problems and no significant correlation between the affective factor and internalizing behavior problems. For boys only, these findings suggest that the link between psychopathic traits and internalizing problems is mediated by externalizing problems. For girls this indirect link is partially confirmed. However, these findings are not consistent with the findings of Sevecke et al. (
2009b), who found an association between the affective factor and internalizing problems for boys. Possible explanations for this difference in results between the current study and the study of Sevecke et al. (
2009b) can be found in the different samples; Sevecke et al. (
2009b) included incarcerated adolescents, while the current study included adolescents admitted to residential care because they needed protection against themselves or their environment. It is also likely that the comorbidity rate between internalizing and externalizing problems in the current study (67%) is higher than that in the Sevecke study of offenders. In this study we did not have longitudinal data and therefore our data did not allow the test of mediation. Further research should use longitudinal data to test a mediation model.
Other explanations for the different findings between Sevecke et al. (
2009b) and the current study might include the different instruments used to measure psychopathic traits. Sevecke et al. (
2009b) used the PCL: YV to measure psychopathic traits, while in this study the YPI was used. Based on previous studies showing only low to moderate correlations between the YPI and the PCL:YV, it is obvious that these two instruments do not measure exactly the same concept. Also the mean ages of the samples differed; the mean age of the sample used by Sevecke et al. (
2009b) was 17.73 years of age for boys and 17.76 years of age for girls, while the mean age of the current study was 15.74 for boys and 15.78 for girls. Further research is needed to obtain a better insight into the associations between psychopathy and externalizing and internalizing problems, and possible moderation of these associations by age and gender.
Concerning delinquent behavior, our study found that the psychopathy-like group showed the highest levels of property and violent offences. Andershed et al. (
2001) also found the highest delinquency rates among male adolescents scoring high on psychopathic traits. Furthermore, in the present study, both the impulsive, non-psychopathic-like and psychopathy-like group showed higher scores on drugs use compared to the normal group. Overall, our study showed that the psychopathy-like group did not substantially differ from the impulsive, non-psychopathic-like group.
It is sometimes suggested that individuals with psychopathic personality traits consist of two groups that can be differentiated based on genetic and environmental influences (Skeem et al.
2007). The ‘primary psychopaths’ have a genetic basis for their psychopathy while the ‘secondary psychopaths’ have an environmental basis. This distinction is based on the extent of anxiety, in which the primary psychopaths experience lower anxiety compared to the secondary psychopath. These authors also found that secondary psychopaths appeared to show behavior that is more withdrawn and emotionally more unstable compared to a non-psychopathic, violent control group. Based on this theory about subgroups of psychopaths, Wareham et al. (
2009) examined subgroups of youth with psychopathic traits including indicators of anxiety. They found four subtypes, of which two types differed in the extent of anxiety, low versus high. The first subgroup is called the impulsive, non-psychopathic like group and the second group is called the impulsive-anxious group. The other two groups (the non-psychopathic group and the psychopath-like group) showed low levels of anxiety. Wareham et al. (
2009) concluded that the psychopathy-like group reflects the primary psychopaths. Concerning the secondary psychopaths, they stated that it is more difficult to conclude which group they can be compared to, but individuals classified into the impulsive classes seem to have some characteristics of the secondary psychopaths. Looking at the indirect link between internalizing problems and psychopathic traits and the differences found between the study of Sevecke et al. (
2009b) and our study, it might be possible that our sample reflect the secondary psychopaths showing especially reactive aggression and the study of Sevecke et al. reflect the primary psychopaths showing proactive aggression.
The strong association between psychopathic traits and externalizing behavior provides a solid foundation for future research and contributes to the existing literature (Forsman et al.
2007; Hart and Hare
1997; Lynam and Gudonis
2005). Our findings suggest that youth with psychopathic traits and youths showing impulsive behavior do not differ on externalizing problems. Consistent with the literature (e.g., Douglas et al.
2006; Forth et al.
2003; Hillege et al.
2010; Poythress et al.
2006), especially the impulsive/irresponsible dimension of psychopathy showed the strongest correlations with all measured problem behavior (internalizing, externalizing, substance use and delinquent behavior), which might explain the few differences found between both subgroups. This is also consistent with prior studies (Abramowitz et al.
2004; Colledge and Blair
2001; Mathias et al.
2007; Sevecke et al.
2009b), who found that externalizing problems are strongly related to the antisocial and behavioral dimensions of psychopathy. This suggests that the differences between the two groups are based on the interpersonal and affective dimensions of psychopathy. The support for the association between internalizing behavior and psychopathic traits is less straightforward. Prior studies (e.g. Brandt et al.
1997; Frick et al.
1999; Lambert et al.
2001; Lynam
1997; Poythress et al.
2006; Salekin et al.
2004) reported contradictive findings. Frick et al. (
1999) stated that utilizing different measures or different concepts of a psychopathic personality (a single dimension versus separate dimensions) to test the relationships could explain the mixed findings in the literature.
Limitations
One of the shortcomings of the current study is that exclusively self-reports were used to examine psychopathic traits. Although self-reports can be seen as valid and reliable measures (Loeber et al.
1989), and some even say that they give a better insight into the subjective aspects of psychopathic traits (Andershed et al.
2002), still the risk of underreporting as a consequence of vulnerability to social desirability is present. Another shortcoming is that only 65% of the adolescents entering one of the participating institutions participated in the current study. The number and nature of the identified latent classes are specific to the sample used. Still, there were no indications that the drop out was not at random because the populations of the participating institutions did not differ from each other according to the later data assessments.
Implications
To our knowledge, most studies using the YPI to examine psychopathic traits are conducted on samples of male offenders, or normative samples. Our study extends these studies and overcomes this eminent shortcoming in this area of research by including males and females admitted to compulsory residential treatment. The findings of this study are consistent with previous findings, supporting the existence of the same three subgroups among different populations. Being able to distinguish subgroups within a residential sample showing severe problem behavior is important for prevention as well as intervention. Some scholars state that adolescents may benefit more from treatment because of their young age (Forth and Burke
1998). Our findings imply that the main treatment goals should be different based on the psychopathic traits of adolescents. For this reason, it is pivotal that further research includes follow-up data. One of the main questions then has to be whether the psychopathy-like group shows more negative outcomes in other areas, for example, living situation, school/work, contacts with the family, and behavior problems. Further research should also link core elements of this treatment to follow-up data as well as psychopathic traits. This would give more insight into the role of psychopathic traits as a moderator of treatment.