Youth with psychopathy features are not a ... - Wiley Online Library

7 downloads 61947 Views 625KB Size Report
1Sam Houston State University, USA; 2University of Southern Mississippi, USA; 3Simon ... structure, indicating that psychopathy features among youth are best.
Journal of Child Psychology and Psychiatry 48:7 (2007), pp 714–723

doi:10.1111/j.1469-7610.2007.01734.x

Youth with psychopathy features are not a discrete class: a taxometric analysis Daniel C. Murrie,1 David K. Marcus,2 Kevin S. Douglas,3 Zina Lee,4 Randall T. Salekin,4 and Gina Vincent5 1

Sam Houston State University, USA; 2University of Southern Mississippi, USA; 3Simon Fraser University, Canada; 4 University of Alabama, USA; 5University of Massachusetts Medical School, USA

Background: Recently, researchers have sought to measure psychopathy-like features among youth in hopes of identifying children who may be progressing toward a particularly destructive form of adult pathology. However, it remains unclear whether psychopathy-like personality features among youth are best conceptualized as dimensional (distributed along a continuum) or taxonic (such that youth with psychopathic personality characteristics are qualitatively distinct from non-psychopathic youth). Methods: This study applied taxometric analyses (MAMBAC, MAXEIG, and L-Mode) to scores from two primary measures of youth psychopathy features: the Psychopathy Checklist: Youth Version (N ¼ 757) and the self-report Antisocial Process Screening Device (N ¼ 489) among delinquent boys. Results: All analyses supported a dimensional structure, indicating that psychopathy features among youth are best understood as existing along a continuum. Conclusions: Although youth clearly vary in the degree to which they manifest psychopathy-like personality traits, there is no natural, discrete class of young ‘psychopaths.’ This finding has implications for developmental theory, treatment, assessment strategies, research, and clinical/forensic practice. Keywords: Psychopathy, juvenile psychopathy, callousunemotional, taxometric, dimensional, antisocial, conduct disorder. Abbreviations: PCL:YV: Psychopathy Checklist: Youth Version; APSD: Antisocial Processes Screening Device.

Psychopathy – a personality disorder characterized by shallow emotions, poor empathy, and an exploitive interpersonal style – is widely recognized as a predictor of violence and criminality among adults (Hare, 1996, 2003). Many researchers have expressed enthusiasm that extending the construct of adult psychopathy downward to youth could lead to a better understanding of youth who are progressing toward a particularly destructive form of adult psychopathology (e.g., Frick & Ellis, 1999; Lynam, 1996, 1998). Researchers have developed measures of psychopathic features that reliably identify child and adolescent personality features and behaviors that are phenotypically similar to the enduring traits characteristic of psychopathic adults (Lynam & Gudonis, 2005). For example, psychopathic features in youth have been related to prospective measures of violence and criminality (Frick et al., 2005; Gretton et al., 2004), institutional violence (Murrie et al., 2004; Stafford & Cornell, 2003), and self-reported aggression and conduct problems (Christian et al., 1997; Frick et al., 2003). Compared to low-psychopathy youth, youth scoring high on psychopathy measures tend to be less fearful (Barry et al., 2000) and less attuned to fear cues in others (Blair et al., 2005). Despite these findings, there remain important gaps in our understanding of psychopathy features among youth. One question is whether transient developmental characteristics could be mistaken for

indications of adult psychopathy. The Psychopathy Checklist: Youth Version (PCL:YV; Forth, Kosson, & Hare, 2003) includes criteria such as impulsivity, irresponsibility, and need for stimulation. Some have warned that such criteria are almost normative among adolescents, making it difficult to distinguish ‘psychopathic traits’ from common adolescent behavior (Seagrave & Grisso, 2002). Although this concern is primarily a question of measurement (i.e., can trained raters make reliable distinctions between age-typical versus atypical egocentricity or impulsivity?), it also evokes questions regarding underlying structure: Are youth who manifest psychopathy features qualitatively distinct from youth without psychopathy features, or is psychopathy distributed along a continuum? Stated differently, is ‘youth psychopathy’ taxonic or dimensional? Distinguishing whether youth psychopathy is taxonic or dimensional can have conceptual, methodological, and practical implications. Regarding etiology, taxa may result from causes ranging from a single gene to multiple factors; however, dimensional constructs cannot result from a single dichotomous causal factor (Meehl & Golden, 1982). Methodologically, a dimensional structure is better suited for analyses using classical test theory, whereas taxonic constructs often require categorical approaches such as Bayes’s Theorem (Ruscio & Ruscio, 2004). Regarding measurement, ‘forcing dimensional variation into spurious taxa can discard valuable information … whereas scaling taxonic constructs can increase measurement error’ (Ruscio & Ruscio,

Conflict of interest statement: No conflicts declared.  2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

Youth psychopathy features: a taxometric analysis

2004, p. 26). Practically, if a construct is dimensional, it does not lend itself to strict categorical classification. That is, we could not accurately state that a child or adolescent ‘is a psychopath’ (though Steinberg, 2002 cautioned that this language is used increasingly during legal proceedings).

Taxometric research and rationale Researchers studying psychiatric disorders have increasingly adopted taxometric analysis to investigate whether the underlying structure of a particular disorder is best understood as categorical (i.e., a taxon) or dimensional (Cole, 2004). Most taxometric research has addressed disorders among adults rather than among youth. Nevertheless, Beauchaine (2003) emphasized that taxometric analyses may help a) identify children at risk for later pathology, b) identify subtypes of disorders, and c) locate sensitive periods in the development of discrete pathological traits. Notably, these proposed benefits of taxometric studies correspond closely with the rationale researchers have offered for studying juvenile psychopathy features. Specifically, researchers posit that studying juvenile psychopathy features may: a) identify youths likely to meet criteria for psychopathy as adults (Lynam, 1996, 1998); b) identify a psychopathy-like subtype of the broader, heterogeneous Conduct Disorder (Frick et al., 2003); and c) facilitate interventions with youth while their psychopathic features are malleable (Lynam, 1996).

Taxometric studies of adult psychopathy Two initial studies of adult psychopathy concluded that psychopathy was taxonic, a ‘discrete class’ likely to manifest at an early age. Harris et al. (1994) reported an adult psychopathy taxon based on PCLR (Hare, 1991) data from inmates in a maximumsecurity psychiatric institution. They also reported that childhood antisocial variables drawn from file data, which they labeled the Childhood and Adolescent Taxon Scale (CATS), yielded a taxon. However, Lilienfeld (1998) emphasized that the unique forensic psychiatric/offender sample (featuring many schizophrenic patients) may have increased the likelihood of erroneously identifying a taxon. Furthermore, Lilienfeld (1998) noted, Harris and colleagues did not find evidence of a taxon underlying PCL-R Factor 1, the callous/manipulative personality style considered most characteristic of psychopathy, but only for Factor 2, which is more saturated with behavioral and criminal justice indices. Skilling and colleagues (2002) re-examined data from the study by Harris and colleagues (1994) and reported the presence of a taxon underlying Antisocial Personality Disorder as well as psychopathy. However, more recent studies converge on a dimensional structure for adult psychopathy, which calls into question the possibility of a ‘youth

715

psychopathy taxon.’ Marcus and colleagues (2004) examined the latent structure of psychopathy among 309 prison inmates using the Psychopathic Personality Inventory (PPI; Lilienfeld & Andrews, 1996) and found a dimensional latent structure for psychopathic personality. Edens and colleagues (2006) used PCL-R data to examine the latent structure of psychopathy in a sample of 876 prison inmates. They found no evidence of a psychopathy taxon even when using the same indicators as Harris et al. (1994). Likewise, other researchers (Guay et al., in press) reported dimensional findings after examining PCL-R scores from a large forensic sample. Overall, of the five studies that examined the latent structure of adult psychopathy, only two (which used the same sample) reported a psychopathy taxon (see Edens et al., 2006 for a detailed critique of these two studies). Notably, taxometric studies that identified a taxon for antisocial behavior among adults (i.e., Harris et al., 1994; Skilling et al., 2001) run counter to studies that used other latent variable models, such as item response theory, which tend to find a non-taxonic latent structure (Krueger et al., 2005).

Taxometric studies of juvenile antisocial features Neither of the two studies that investigated the latent structure of psychopathy features among youth used the PCL:YV, probably the most comprehensive measure of psychopathy features (Forth et al., 2003). Skilling and colleagues (2001) used a community sample of boys from urban schools to explore whether persistent antisocial behavior was taxonic. Using self-report items in the study database, the authors ‘matched’ items similar to the eight CATS items, 14 of the DSM-IV Conduct Disorder (CD) criteria, and eight of 20 PCL:YV (Forth et al., 2003) items. They concluded that a taxon of ‘serious antisocial behavior’ could be identified among children, after reporting taxonic results for all three measures, with an average base rate of 9% for the identified taxon (Skilling et al., 2001). However, although 9% might be a plausible prevalence for CD in a community sample of boys, it is much higher than would be expected for psychopathy or ‘serious antisocial behavior’ (Vasey et al., 2005). Unfortunately, Skilling and colleagues (2001) did not report information essential for evaluating their taxometric analyses, such as the validity or skew of their indicators (skewed indicators can yield pseudo-taxonic graphs). Vasey and colleagues (2005) examined the latent structure of psychopathic features in two studies using the Antisocial Process Screening Device (APSD; Frick & Hare, 2001). In their first study, they used self-report, parent-report, and teacher-report versions of the APSD as the multiple construct indicators for the taxometric analyses. The data came from 326 youth pooled from two samples: one of 283 mixed-gender community youth, and another of 43

 2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health.

716

Daniel C. Murrie et al.

clinically referred males. Their analyses suggested a taxon with a base rate of around 23% across the entire sample. Noting the high base rates (70% within the clinical group and 17% within the nonclinical group), the authors doubted that the observed taxon was psychopathy per se, but more likely a ‘broader construct similar to that found by Skilling et al. (2001)’ (Vasey et al., 2005, p.420). Study 2 added 60 male adolescents from a juvenile justice sample to the aforementioned data to ‘increase the expected base rate of psychopathy in the sample’ (p. 415). The authors concluded that they identified a taxon with an average base rate of around 6%. They recommended that future research attempt to replicate this low base-rate taxon in samples in which psychopathy features are likely to be relatively common (i.e., juvenile justice samples). However, the Vasey et al. (2005) studies appear to have some methodological limitations, which suggest additional study remains necessary. Although taxometric procedures require multiple measures of the construct, the goal is to use multiple aspects of the construct rather than measures of the same aspects taken from multiple informants (i.e., parent-, teacher-, and self-report versions of the same measure). That is, ‘each indicator used in a given taxometric analysis should represent a phenotypically distinct dimension or symptom of the disorder’ (Cole, 2004, p. 5). A second limitation to the Vasey et al. study involves the hazards of combining disparate samples (i.e., mixed-gender community youth with clinicallyreferred or juvenile-justice males). Mixing substantially different (clinical and non-clinical) samples greatly increases the chances of identifying an institutional pseudo-taxon (Cole, 2004). A third concern is that it is difficult to understand what happened to the antisocial taxon that was identified in Study 1 when the data were analyzed in Study 2. As the larger category of antisocial behavior, it should have been present and overshadowed the smaller psychopathy taxon even in the Study 2 analyses.

Present study Despite the growing number of taxometric studies of psychopathy, there is no clear consensus about the latent structure of psychopathy features among youth, and no taxometric studies have used the PCL:YV. Those studies that have appeared to find a taxon have generally suffered from significant methodological limitations. Given the need for further research and the recommendation to study juvenile-justice samples in which a psychopathy taxon may be more common (Vasey et al., 2005), we conducted taxometric analyses using the two mostused and well-validated measures of youth psychopathy features in juvenile-justice samples: the PCL:YV and the self-report APSD. The PCL:YV is the adolescent version of Hare’s (2003) Psychopathy Checklist-Revised, which is the most widely used

measure of adult psychopathy (Conoley & Impara, 1995). The APSD is the most widely studied measure of psychopathic features in community children; the self-report APSD is most often used in studies of juvenile-justice samples. We examined large samples (Ns of 757 and 480) of adolescent males involved in the juvenile justice system, which should provide an adequate base rate of a psychopathy taxon, should one exist.

Method Data source Data from seven samples of male youths involved with the juvenile justice system were pooled to create two samples (one of PCL:YV scores, one of APSD scores). Sample 1 consisted of APSD and PCL:YV results from 113 incarcerated males admitted to the centralized intake center for the Virginia Department of Juvenile Justice (Murrie & Cornell, 2002). Sample 2 consisted of APSD and PCL:YV results from 85 incarcerated males in Florida (Spain et al., 2004). Sample 3 included 100 APSD (Lee et al., 2003) and 271 PCL:YV (Vincent, under review) results from boys incarcerated at secure and minimum security centers in British Columbia, Canada. Sample 4 included 86 APSD results from males involved in a court-diversion program in Florida (Poythress, Dembo, et al., 2006). Sample 5 included 110 APSD results from adolescent males from Florida who were prosecuted in adult court (Poythress, Lexcen et al., 2006). Sample 6 consisted of PCL:YV data for 82 nonincarcerated boys at the Court Evaluation Unit in Miami, Florida (Salekin et al., 2004). Sample 7 consisted of PCL:YV data for 206 incarcerated boys from a detention center in Alabama (Salekin et al., 2005). Mean scores for the PCL:YV varied somewhat across the samples [F (4,633) ¼ 3.58, p ¼ .01], though Tukey’s post-hoc comparison revealed that the only significant between-groups difference was between Sample 7 (M ¼ 19.50, SD ¼ 7.11) and Sample 2 (M ¼ 22.74, SD ¼ 5.98). Mean scores from all other samples fell between these two. Mean APSD scores differed more across samples [F (4,476) ¼ 29.38, p < .001] with the highest mean score (M ¼ 18.61, SD ¼ 5.49) for Sample 3 and the lowest (M ¼ 10.96, SD ¼ 5.23) for Sample 5. Tukey’s post-hoc revealed several significant differences among groups. Further details regarding sampling strategies, exclusion criteria, and descriptive statistics for the each sample are available in the original publications. Combining the samples yielded PCL:YV results for 757 males, who ranged in age from 13 to 18 years (M ¼ 16.06, SD ¼ 1.25), with a racial composition of 41% White, 45% Non-White, and 14% unreported. APSD results were available for 490 youth who ranged in age from 11 to 19 years (M ¼ 15.97, SD ¼ 1.55). Forty-eight percent of participants were White, with the remaining 52% Non-White.

Measures PCL:YV. The PCL:YV assesses psychopathic features using a semi-structured interview and data from the

 2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health.

Youth psychopathy features: a taxometric analysis

participant’s files. The PCL:YV manual (Forth et al., 2003) supports a two-factor model with four nested factors (called ‘facets’ in the manual) comparable to the model Hare (2003) proposed for adult samples. These include: Interpersonal and Affective facets (Factor 1) as well as Behavioral and Antisocial facets (Factor 2). The four facets served as indicators for the taxometric analyses in the present study. For the pooled sample in the present study, the Chronbach’s alpha for the total score was .83; facet level alpha values were .72 (Interpersonal), .68 (Affective), .53 (Behavioral), and .65 (Antisocial), consistent with the normative samples (Forth et al., 2003). Findings from adolescent samples, typically comprised of offenders, have demonstrated that PCL:YV scores have high interrater agreement (average total score ICC2 ¼ .96; Forth et al., 2003). All samples in the present study featured total-score interrater reliabilities ‡.89.

APSD. The APSD (Frick & Hare, 2001) represents an effort to adapt Hare’s (1991) PCL-R to measure psychopathy traits in adolescents by parent and teacher reports. Caputo et al. (1999) developed a self-report version of the measure that is now commonly applied to justice-involved adolescent samples. Self-report measures become more appropriate as a child enters adolescence and becomes a better informant regarding antisocial behaviors that may not be observable to adults (Frick et al., 2000). The APSD features a three-factor structure with Narcissism, Callous/Unemotional, and Impulsivity factors (Frick et al., 2000). We used a slightly modified version of this three-factor structure based on a recent factor-analytic study of the self-report APSD using a sample of justice-involved youth (Poythress, Dembo et al., 2006). Factor scores served as the indicators for the taxometric analyses. The alpha coefficient for APSD total score in the current sample was .78; factor-level alpha values were: .70 (Narcissism), .59 (Callous/Unemotional), and .62 (Impulsivity).

717

between successive windows. In MAXEIG, for each window all of the indicators (except for the input indicator) are factor analyzed and the eigenvalue of the first principal factor is plotted on the y-axis, whereas in MAXCOV, the covariance between the two remaining indicators is plotted. If psychopathy is taxonic, the eigenvalue or covariance will be greatest in the subsample (i.e., window) that is most evenly divided between psychopaths and non-psychopaths; the graph will peak at this cut. In contrast, a MAXEIG or MAXCOV graph for a dimensional construct will appear concave, flat, or irregular. Research has clearly demonstrated that it is not possible to infer the underlying structure of a construct simply by graphing the distribution of raw scores of single measures of that construct (i.e., a dimensional construct may yield a bimodal distribution and a taxonic construct can yield a unimodal distribution, see Ruscio et al., 2006). But by factor analyzing multiple valid indicators of the construct and graphing the distribution of the first principle factor, it is possible to examine underlying structure. All of the indicators are factor analyzed in L-Mode and the distribution of scores on the first principal factor is graphed. A taxonic construct produces a bimodal graph, and a dimensional construct yields a unimodal graph. Ruscio’s (2005) taxometric programs were used to perform these analyses. Researchers using these programs can compare the graphs produced by actual data to those resulting from simulations (Ruscio, Ruscio, & Meron, in press). Ruscio’s programs calculate a curve fit index that ranges from 0 (consistent with a dimensional structure) to 1 (consistent with a taxonic structure). Ten sets of simulated dimensional data and 10 sets of simulated taxonic data were generated for each of the analyses. Finally, these taxometric analyses also yield base rate estimates, so one can compare the observed base rate from each taxometric procedure for consistency among each procedure and for consistency with what might be expected based on theory or previous research.

Procedure The data were analyzed using four of the taxometric procedures developed by Meehl and his colleagues: MAMBAC (Mean Above Minus Below A Cut; Meehl & Yonce, 1994), MAXCOV (MAXimum COVariance; Meehl & Yonce, 1996) or MAXEIG (MAXimum EIGenvalue; Waller & Meehl, 1998), and L-Mode (Latent mode factor analysis; Waller & Meehl, 1998). MAMBAC requires two indicators, an input indicator, which is graphed on the x-axis, and an output indicator y. Cuts (50 in this study) are made along the input indicator, and at each cut the difference between the mean score on the output indicator for those cases above the cut and the mean scores for those cases below the cut is graphed on the y-axis. A taxonic construct will have a single peak, indicating the base rate of the taxon. The MAMBAC graph of a dimensional construct will appear concave. In MAXEIG and MAXCOV the sample is divided into a series of overlapping windows along the input indicator. Each window is a segment of the data that has been sliced along the x-axis. In the present analyses, the sample was divided into 50 windows with .90 overlap

Results Selection of indicators Because the PCL:YV and APSD each yield facet or factor scores with non-overlapping items, the taxometric analyses were performed at the scale level. The four facet scores of the PCL-YV and the three factor scores of the APSD each had appropriate psychometric properties to serve as indicators for taxometric analyses (Meehl & Golden, 1982). For example, indicators should be correlated with one another in the overall sample, but there should be little correlation among the indicators for members of the taxon or complement (Meehl & Golden, 1982). The average correlation among the four PCL:YV indicators was .41 in the full sample. In contrast, the average correlation among these three scales for those in the lower quartile of PCL:YV total scores was .04; it was ).02 for those in the upper quartile. Results with the APSD were similar. Furthermore, the

 2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health.

718

Daniel C. Murrie et al.

estimated average indicator validities based on the subsequent MAMBAC, MAXEIG, and L-Mode analyses all exceeded the recommended minimum of 1.25 (Meehl & Yonce, 1996).

MAMBAC analyses PCL:YV. For each analysis, one scale served as the output indicator and the remaining three scales were combined to create the input indicator. These analyses were then repeated with each scale serving as the output indicator. None of these curves appeared taxonic (copies of all figures not pictured are available upon request). The average of these four curves had a concave appearance, producing a graph much more similar to the simulated dimensional data than to the simulated taxonic data, with a curve fit index of .14 (Figure 1). The base rate estimates were all considerably higher than would have been expected in this sample and they did not converge on a single value, but ranged from .39 to .69 (M ¼ .54, SD ¼ .13). Because facets 1 and 2 specifically focus on psychopathic personality (as opposed to antisocial behavior), we ran an additional MAMBAC analysis using only these two facets as indicators. Again, results clearly appeared dimensional. APSD. For the APSD data, one scale served as the output indicator and the other two scales were combined to create the input indicator, yielding three MAMBAC curves, all of which were concave. The average of the three curves was clearly U-shaped and

was much more similar to the graph produced by the simulated dimensional data than to the simulated taxonic data, with a curve fit index of .14. These curves also yielded high base-rate estimates (.42, .51, and .54; M ¼ .53, SD ¼ .06). Overall, the MAMBAC analyses provided strong evidence to suggest that youth psychopathy has a dimensional structure.

MAXEIG/MAXCOV analyses PCL-YV. None of the four MAXEIG curves had the clear inverted-U that would be found if psychopathy were taxonic; three of the curves appeared flat. The average curve produced by the research data did not have an inverted-U shape; it appeared more similar to the simulated dimensional data than to the taxonic data, with a curve fit index of .10 (Figure 2). The base-rate estimates for these curves varied considerably and were much higher than expected baserates, ranging from .46 to .78 (M ¼ .66, SD ¼ .15). APSD. One of the three MAXCOV curves appeared to have a peak as would be found if psychopathy were taxonic. However, the average of the three curves was more similar to the graph produced by the simulated dimensional data than to the simulated taxonic data, with a curve fit index of .18. These curves also yielded lower base-rate estimates than the MAMBAC analyses (.13, .15, and .25; M ¼ .17, SD ¼ .06). Overall, the MAXEIG analyses suggested that youth psychopathy has a dimensional structure.

Simulated Dimensional Data

0.07 0

100 200 300 400 500 Input (Cases)

600 700

0.06

0.06

0.07

0.08

Mean Difference 0.09 0.10 0.11

Mean Difference 0.08 0.09 0.10 0.11

0.12 0.13

0.12 0.13

Simulated Taxonic Data

0

100 200

300 400 500 600 700 Input (Cases)

Figure 1 Average MAMBAC curves for the PCL:YV factors Interpersonal, Affective, Behavioral, and Antisocial. The dark line represents the actual data; lighter lines represent one standard deviation above and below the average for each simulated data set. For each curve, the input (x-axis) was comprised of the three indicators that were not the output indicator. Data were sorted by the scores on this input indicator and 50 cuts were made along the input indicator. The y-axis represents the differences between the mean scores on the output indicator of individuals falling above and below each cut. These MAMBAC analyses were replicated 10 times by randomly shuffling the cases with equal scores before making the cut on the input indicator and recalculating the difference scores on the output indicator. The graphs for the simulated taxonic and dimensional data were produced by generating 10 data sets for each latent structure  2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health.

Youth psychopathy features: a taxometric analysis

Simulated Taxonic Data

719

1.0 Eigenvalue 0.4 0.6 0.8 0.2 0.0

0.0

0.2

Eigenvalue 0.4 0.6 0.8

1.0

Simulated Dimensional Data

-1.5 -1.0 -0.5 0.0 0.5 50 Windows

1.0

1.5

-1.5 -1.0 -0.5 0.0 0.5 50 Windows

1.0

1.5

Figure 2 MAXEIG curves for the four PCL:YV factors. The dark line represents the actual data and the lighter lines represent one standard deviation above and below the average for each simulated data set. Data were sorted along the x-axis by scores on the input indicator and then grouped into 50 subsamples with overlapping windows. Associations among the remaining indicators within each window were plotted on the y-axis. To stabilize the shape of each curve, these MAXEIG analyses were replicated 10 times by randomly shuffling the cases with equal scores before making the cut on the input indicator and recalculating the eigenvalues on the output indicator. Graphs for the simulated taxonic and dimensional data resulted from generating 10 data sets for each latent structure Research Data

Simulated Taxonic Data

0.4 0.1

Relative Frequency

Relative Frequency 0.2

0.0

0.1 0.0

0.1 0.0 -3 -2 -1 0 1 2 Factor Scores

3

0.2

0.3

0.4 0.3

0.4 0.3 0.2

Relative Frequency

Simulated Dimensional Data

-3 -2 -1 0 1 2 Factor Scores

3

-3 -2 -1 0 1 2 Factor Scores

3

Figure 3 Latent mode (L-Mode) factor analysis curves for the PCL:YV factors and for the simulated taxonic and dimensional data. Each graph represents the frequency distribution of scores on the first factor of a factor analysis of the indicator set. The graphs for the simulated taxonic and dimensional data resulted from generating 10 data sets for each latent structure (represented by the dotted lines); solid lines indicate the average of these data sets

L-Mode analysis L-Mode analyses of both the PCL-YV (Figure 3) and the APSD data yielded unimodal curves, consistent with a dimensional construct.

Discussion This study yielded consistent evidence that the psychopathy construct has a dimensional latent

structure among male adolescents. This finding held true whether the indicators were drawn from the clinician-scored PCL:YV or from the self-report APSD. The results of these analyses support the position that differences in psychopathic traits among young delinquents appear to be differences in degree, not kind. These results were also consistent with other recent studies of adult psychopathy and antisocial behavior that used taxometric or latent class analysis (e.g., Edens

 2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health.

720

Daniel C. Murrie et al.

et al., 2006; Krueger et al., 2005; Marcus et al., 2006). The absence of taxonic findings appears credible for at least four reasons. First, the current study utilized a large sample from a variety of juvenilejustice settings. This large sample provides a sufficient base rate of youth with psychopathic features to detect the putative taxon members (should they exist) utilizing MAMBAC, MAXEIG, or L-Mode. Second, the current study reduced the likelihood of identifying pseudo-taxons by analyzing data from a fairly homogenous sample of youth. Mixing very disparate samples increases the likelihood of falsely detecting a taxon. Third, although we studied only those who had come into contact with the courts, the sample represents youth from several states and one Canadian province. This feature of the study is important because it provides greater generalizability. Finally, this study used two primary youth psychopathy measures likely to be employed in research and clinical practice, thereby addressing some limitations in previous research. Results derived solely from informant rating-scale data (e.g., Vasey et al., 2005) are vulnerable to the possibility of ‘pseudotaxonicity’ if some raters tend to score participants in an all-or-none manner (Beauchaine & Waters, 2003). Results derived from convenience samples of items that are ‘matched’ to diagnostic criteria or other test items (e.g., Skilling et al., 2001) may not reflect the construct of psychopathy in the same manner that well-validated measures of psychopathy-features (i.e., the PCL:YV and the APSD) reflect the construct. The fact that two well-validated measures based on two different assessment modalities (i.e., self-report and expert-ratings by trained clinicians) both yielded such similar results further supports the credibility of the findings. Despite these strengths, it is important to consider two limitations to the study. First, although our samples were drawn from juvenile justice settings in which clinicians are most likely to employ psychopathy assessment, they feature only males. Although there is no particular reason to expect psychopathy would appear taxonic for girls but not boys, an obvious first step in testing the generalizability of these findings would be to replicate the study with girls. Secondly, although factor scores from wellvalidated instruments should be sensitive for detecting putative taxa (Beauchaine, 2003), the next steps in taxometric research are to conduct additional studies using indicators from other domains (e.g., physiological measures, behavioral tasks, emotional processing tasks, etc.).

Implications for research Because adolescent psychopathy features appear to have a dimensional distribution, researchers may opt to rely primarily on correlational designs, rather than high–low comparison groups, which tend to

hide nonlinear relationships between degree-ofsymptoms and other variables of interest (Ruscio & Ruscio, 2004). Given that dimensional constructs are more likely to result from multifactorial causes, cross-disciplinary research strategies from multiple perspectives (e.g., physiological, genetic, developmental, systems-theory, personality theory) will likely be necessary to shed light on the development of psychopathy features (Richters & Cicchetti, 1993).

Implications for practice and policy Beauchaine (2003) suggested that taxometric analyses might be useful to locate ‘bifurcation points in the development of discrete traits’ (p. 518), and he suggested psychopathy, in particular, as warranting investigation. Specifically, he suggested taxometrics might identify a period when psychopathic traits are continuous, before they crystallize into taxonic traits (Harris et al., 1994) that are presumably less malleable. Intervention, he argued, should be delivered before the stage at which psychopathy traits appear taxonic. Our results suggest psychopathy features are not taxonic in adolescence. Recent rigorous taxometric studies also suggest that psychopathy is not taxonic among adults (Edens et al., 2006; Marcus et al., 2004). Thus, increasing evidence suggests that psychopathy traits are likely distributed along a continuum regardless of age. Of course, early intervention is preferable for most forms of pathology, and there may be many reasons to treat as early as possible the callous-unemotional traits that are hallmarks of psychopathy. However, taxometric research alone offers no justification for intervening at a particular phase or declining to intervene in later phases of development. These non-taxonic, or dimensional, results also suggest it is misleading to pinpoint a particular PCL:YV score as precisely delineating ‘psychopaths’ from ‘non-psychopaths.’ Instead, a score is better understood as quantifying ‘how psychopathic’ an individual is. More generally, our results suggest that it is misleading to speak of juvenile ‘psychopaths’ (see Steinberg, 2002), a terminology that implies bright line distinctions separating qualitatively different juvenile offenders. Although semantically less convenient, it would probably be more accurate to describe a youth scoring high on the PCL:YV as ‘relatively high in psychopathy features.’ This enhanced precision is particularly important for clinicians practicing in forensic contexts, where the legal system may make significant decisions related to diagnoses. On a related note, these dimensional results offer no support for institutional or legal policies that require definitive distinctions between psychopathic and non-psychopathic youth. Though we are unaware of any such policies in the juvenile justice system at present, the adult criminal justice system has long featured such legislation (e.g., ‘sexual

 2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health.

Youth psychopathy features: a taxometric analysis

psychopath laws’) and case law (see Edens & Petrila, 2006; Edens et al., 2006). This is not to say that there are no meaningful differences between youth scoring relatively high versus relatively low in psychopathy, but only that any precise categorical distinctions should be acknowledged as arbitrary. Similarly, while it may prove feasible to identify scores on psychopathy measures that best identify youth at risk for community violence or institutional misbehavior, these scores should be identified based on research, rather than assumptions about a score’s correspondence to an assumed diagnostic threshold, such as the commonly used cutoff score of 30 on the adult and adolescent PCL measures (see Edens et al., 2006 for a similar discussion of adult psychopathy scores). Of course, the lack of any definitive marker between ‘psychopathic’ and ‘not-psychopathic’ youth creates challenges for clinicians. As Ruscio and Ruscio (2004) explained, although taxa can in effect simply be read off of nature, the decision about where to draw the line between normality and pathology for dimensional constructs is a value judgment open to debate. If scholars (e.g., Seagrave & Grisso, 2002) are correct in predicting that youth psychopathy assessment will be regularly employed in clinical and legal settings, then it will be necessary to determine how varying degrees of psychopathy features correspond with outcomes or conditions of interest, and whether some specific symptoms or criteria are more closely related to these outcomes than others. It will be important to consider carefully the implications of ‘setting the bar too low’ (i.e., assuming that all research on juvenile psychopathy applies to youth who manifest only a few, or transient, psychopathy criteria) as well as the implications of ‘setting the bar too high’ (failing to apply our extant knowledge base to youth who genuinely appear high in psychopathy features).

Conclusion Although the assessment of psychopathy among youth is controversial, there is substantial evidence to suggest that the features characteristic of psychopathic adults can also be identified in some youth (although their expression may understandably differ). However, this study provides evidence that psychopathic features occur along a continuum; they are not distributed in a manner that creates a unique or qualitatively distinct class of youth.

Acknowledgements We thank Dr. Norman Poythress and Dr. Raymond Corrado for contributing data to this study. We thank the Social Sciences and Humanities Research Council (Grant R-410-88-1246 awarded to Raymond

721

Corrado) for supporting collection of the Canada data, and the University of Virginia’s Center for Children, Family, and the Law for supporting collection of the Virginia data.

Correspondence to Daniel C. Murrie, Department of Psychology, Sam Houston State University, PO Box 2447, Huntsville, Texas 77341, USA; Tel: 936.294.4161; Fax: 936. 294.3798; Email: [email protected]

References Barry, C.T., Frick, P.J., Grooms, T., McCoy, M.G., Ellis, M.L., & Loney, B.R. (2000). The importance of callous-unemotional traits for extending the concept of psychopathy to children. Journal of Abnormal Psychology, 109, 335–340. Beauchaine, T.P. (2003). Taxometrics and developmental psychopathology. Development and Psychopathology, 15, 501–527. Beauchaine, T.P., & Waters, E. (2003). Pseudotaxonicity in MAMBAC and MAXCOV analyses of rating scale data: Turning continua into classes by manipulating observer’s expectations. Psychological Methods, 8, 3–15. Blair, R.J., Budhani, S., Colledge, E., & Scott, S. (2005). Deafness to fear in boys with psychopathic tendencies. Journal of Child Psychology and Psychiatry, 46, 372–336. Caputo, A.A., Frick, P.J., & Brodsky, S.L. (1999). Family violence and juvenile sex offending: Potential mediating roles of psychopathic traits and negative attitudes toward women. Criminal Justice and Behavior, 26, 338–356. Christian, R.E., Frick, P.J., Hill, N.L., Tyler, L., & Frazer, D.R. (1997). Psychopathy and conduct problems in children II: Implications for subtyping children with conduct problems. Journal of the American Academy of Child and Adolescent Psychiatry, 36, 233–241. Cole, D.A. (2004). Taxometrics in psychopathology research: An introduction to some of the procedures and related methodological issues. Journal of Abnormal Psychology, 113, 3–9. Conoley, J., & Impara, J. (1995). 12th mental measurements yearbook. Lincoln, NE: Burros Institute. Edens, J., Marcus, D., Lilienfeld, S., & Poythress, N. (2006). Psychopathic, not psychopath: Taxometric evidence for the dimensional structure of psychopathy. Journal of Abnormal Psychology, 115, 131– 144. Edens, J., & Petrila, J. (2006). Legal and ethical issues in the assessment and treatment of psychopathy. In C. Patrick (Ed.), Handbook of psychopathy (pp. 573– 588). New York: Guilford Press. Forth, A.E., Kosson, D.S., & Hare, R.D. (2003). The Hare Psychopathy Checklist: Youth Version manual. Toronto: Multi-Health Systems. Frick, P.J., Bodin, S.D., & Barry, C.T. (2000). Psychopathic traits and conduct problems in community and clinic-referred samples of children: Further

 2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health.

722

Daniel C. Murrie et al.

development of the Psychopathy Screening Device. Psychological Assessment, 12, 382–393. Frick, P.J., Cornell, A.H., Barry, C.T., Bodin, S.D., & Dane, H.A. (2003). Callous-unemotional traits and conduct problems in the prediction of conduct problem severity, aggression, and self-report of delinquency. Journal of Abnormal Child Psychology, 31, 457–470. Frick, P.J., & Ellis, M. (1999). Callous–unemotional traits and subtypes of conduct disorder. Clinical Child and Family Psychology Review, 2, 149–168. Frick, P.J., & Hare, R.D. (2001). The Antisocial Process Screening Device. Toronto: Multi-Health Systems. Frick, P.J., Stickle, T., Dandreaux, D., Farrell, J., & Kimonis, E. (2005). Callous-unemotional traits in predicting the severity and stability of conduct problems and delinquency. Journal of Abnormal Child Psychology, 33, 471–487. Gretton, H.M., Hare, R.D., & Catchpole, R. (2004). Psychopathy and offending from adolescence to adulthood: A 10-year follow up. Journal of Consulting and Clinical Psychology, 72, 636–645. Guay, J., Ruscio, J., Hare, R., & Knight, R.A. (in press). A taxometric study of the latent structure of psychopathy: Evidence for dimensionality. Journal of Abnormal Psychology. Hare, R.D. (1991). The Hare Psychopathy ChecklistRevised. Toronto, Canada: Multi-Health Systems. Hare, R.D. (1996). Psychopathy a clinical construct whose time has come. Criminal Justice and Behavior, 23, 25–54. Hare, R.D. (2003). Hare Psychopathy Checklist-Revised (PCL-R): 2nd Edition. Technical Manual. North Tonawanda, NY: Multi-Health Systems. Harris, G.T., Rice, M.E., & Quinsey, V. (1994). Psychopathy as a taxon: Evidence that psychopaths are a discrete class. Journal of Consulting and Clinical Psychology, 62, 387–397. Krueger, R.F., Markon, K.E., Patrick, C.J., & Iacono, W.G. (2005). Externalizing psychopathology in adulthood: A dimensional-spectrum conceptualization and its implications for DSM-V. Journal of Abnormal Psychology, 114, 537–550. Lee, Z., Vincent, G.M., Hart, S.D., & Corrado, R.R. (2003). The validity of the Antisocial Process Screening Device as a self-report measure of psychopathy in adolescent offenders. Behavioral Sciences and the Law, 21, 771–786. Lilienfeld, S.O. (1998). Methodological advances and developments in the assessment of psychopathy. Behaviour Research and Therapy, 36, 99–125. Lilienfeld, S.O., & Andrews, B.P. (1996). Development and preliminary validation of a self-report measure of psychopathic personality traits in noncriminal populations. Journal of Personality Assessment, 66, 488– 524. Lynam, D.R. (1996). Early identification of chronic offenders: Who is the fledgling psychopath? Psychological Bulletin, 120, 209–234. Lynam, D.R. (1998). Early identification of the fledgling psychopath: Locating the psychopathic child in current nomenclature. Journal of Abnormal Psychology, 107, 566–575. Lynam, D.R., & Gudonis, L. (2005). The development of psychopathy. Annual Review of Clinical Psychology, 1, 381–407.

Marcus, D.K., Lilienfeld, S.O., Edens, J.F., & Poythress, N.G. (2006). Is antisocial personality disorder continuous or categorical? A taxometric analysis. Psychological Medicine, 36, 1571–1582. Marcus, D.K., John, S., & Edens, J.F., (2004). A taxometric analysis of psychopathic personality. Journal of Abnormal Psychology, 113, 626–635. Meehl, P.E., & Golden, R. (1982). Taxometric methods. In P. Kendall & J. Butcher (Eds.), Handbook of research methods in clinical psychology (pp. 127– 181). New York: Wiley. Meehl, P.E., & Yonce, L.J. (1994). Taxometric analysis: I. Detecting taxonicity with two quantitative indicators using means above and below a sliding cut (MAMBAC procedure). Psychological Reports, 74, 1059–1274. Meehl, P.E., & Yonce, L.J. (1996). Taxometric analysis: II. Detecting taxonicity using covariance of two quantitative indicators in successive intervals of a third indicator (MAXCOV procedure). Psychological Reports, 78, 1091–1227. Murrie, D.C., & Cornell, D.G. (2002). Psychopathy screening of incarcerated juveniles: A comparison of measures. Psychological Assessment, 14, 390–396. Murrie, D.C., Cornell, D.G., Kaplan, S., McConville, D., & Levy, A.G. (2004). Psychopathy scores and violence among juvenile offenders: A multi-measure study. Behavioral Sciences and the Law, 49, 22–67. Poythress, N., Dembo, R., Wareham, J., & Greenbaum, P. (2006). Construct validity of the Youth Psychopathic Traits Inventory and the Antisocial Process Screening Device with justice-involved adolescents. Criminal Justice and Behavior, 33, 26–55. Poythress, N., Lexcen, F.J., Grisso, T., & Steinberg, L. (2006). The competence-related abilities of adolescent defendants in criminal court. Law and Human Behavior, 30, 75–92. Richters, J.E., & Cicchetti, D. (1993). Towards a developmental perspective on conduct disorder. Development and Psychopathology, 5, 1–4. Ruscio, J. (2005). Taxometric program documentation, R language [Computer software and manual]. Available at http://www.taxometricmethod.com. Ruscio, J., Haslam, N., & Ruscio, A.M. (2006). Introduction to the taxometric method: A practical guide. Mahwah, NJ: Lawrence Erlbaum Associates. Ruscio, J., & Ruscio, A.M. (2004). Clarifying boundary issues in psychopathology: The role of taxometrics in a comprehensive program of structural research. Journal of Abnormal Psychology, 113, 24–38. Ruscio, A.M., Ruscio, J., & Meron, M. (in press). Applying the bootstrap to taxometric analysis: Generating empirical sampling distributions to help interpret results. Multivariate Behavioral Research. Salekin, R.T., Leistico, A.R., Trobst, K.K., Schrum, C.L., & Lochman, J.E. (2005). Adolescent psychopathy and personality theory – the interpersonal circumplex: Expanding evidence of a nomological net. Journal of Abnormal Child Psychology, 33, 445–460. Salekin, R.T., Neumann, C.S., Leistico, A.R., DiCicco, T.M., & Duros, R.L. (2004). Psychopathy and comorbidity in a young offender sample: Taking a closer look at psychopathy’s potential importance over disruptive behavior disorders. Journal of Abnormal Psychology, 113, 416–427.

 2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health.

Youth psychopathy features: a taxometric analysis

Seagrave, D., & Grisso, T. (2002). Adolescent development and the measurement of juvenile psychopathy. Law and Human Behavior, 26, 219–230. Skilling, T.A., Harris, G.T., Rice, M.E., & Quinsey, V.L. (2002). Identifying persistently antisocial offenders using the Hare Psychopathy Checklist and DSM Antisocial Personality Disorder criteria. Psychological Assessment, 14, 27–38. Skilling, T.A., Quinsey, V.L., & Craig, W.M. (2001). Evidence of a taxon underlying serious antisocial behavior in boys. Criminal Justice and Behavior, 28, 450–470. Spain, S.E., Douglas, K.S., Poythress, N.G., & Epstein, M. (2004). The relationship between psychopathic features, violence and treatment outcome: The comparison of three youth measures of psychopathic features. Behavioral Sciences and the Law, 22, 85–102.

723

Stafford, E., & Cornell, D.G. (2003). Psychopathy scores predict adolescent inpatient aggression. Assessment, 10, 102–112. Steinberg, L. (2002). The juvenile psychopath: Fads, fictions, and facts. National Institutes of Justice Perspectives on Crime and Justice: 2001 Lecture Series 5, 35–64. Vasey, M.W., Kotov, R., Frick, P.J., & Loney, B.R. (2005). The latent structure of psychopathy in youth: A taxometric investigation. Journal of Abnormal Child Psychology, 33, 411–429. Waller, N.G., & Meehl, P.E. (1998). Multivariate taxometric procedures: Distinguishing types from continua. Newbury Park, CA: Sage. Manuscript accepted 5 December 2006

 2007 The Authors Journal compilation  2007 Association for Child and Adolescent Mental Health.