.
Fill. .
.1... :3
x... ..

Luv“.

..
.=;.‘¢
r 3
:51

.759
Wm..- 52:. ..

(t:
.1 .A:
.5.» 11.413....:;
zr? . ‘ . .7! . t
5.3..

m???&

g

 

§

E

{Gnu e an.

(it?
I v1.3“:

‘ _
1.. :2... i r

.
Lsuarmmummdwuhh

.7...
:mmmw,

Juan:

:
. ,
Para..."
¥§.t‘?~rt

{4. I:
u ,l #1:]. a ‘1
fir?.fl.ul..¥1 .iam‘nm .
3 «13:11:91:

3..

.r .I.

'r4x.liiv

12%

.I
1.:
..«mmam»
.

.2)
'3‘!“
. AI. . v . 0

Egg . , iggéghfi ,. . 4%....»

1., .

 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

PSYCHOPATHY IN ADOLESCENCE:
CONCEPTUALIZATION, STABILITY, AND PREDICTION OF
AGGRESSION

presented by

JENNIFER SUZANNE DURST

has been accepted towards fulfillment
of the requirements for the

Doctoral degree in Psychology

 

 

' Maj r rofe'ssor’s Signature

May 9, 2005

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

PLACE IN REIURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 c:/CIRC/DatoDuo.Indd—n15

 

PSYCHOPATHY IN ADOLESCENCE:
CONCEPTUALIZATION, STABILITY, AND PREDICTION OF AGGRESSION

By

Jennifer Suzanne Durst

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Psychology

2005

ABSTRACT

PSYCHOPATHY IN ADOLESCENCE:
CONCEPTUALIZATION, STABILITY, AND PREDICTION OF AGGRESSION

By
Jennifer Suzanne Durst
In adults, research has consistently associated the personality disorder of
psychopathy with a particularly aggressive and violent form of criminal offending.
Preliminary studies have shown a similar link in adolescence. Yet, much more
knowledge is needed about the disorder in youth to determine whether it is a useful
construct for early identification of potentially severe offenders. The present study
investigated psychopathy in a sample of 149 adolescents involved with the juvenile
justice system. The research aims were to: (1) assess the fit of several proposed
conceptualizations of psychopathy; (2) test the stability of psychopathy and its
dimensions in adolescents over time; and (3) examine the predictive validity of
psychopathy for aggression in court-referred adolescents. Results indicated that models
of psychopathy that did not include social deviance criteria evidenced the best fit to the
data. As predicted, a three-factor model of psychopathy was the best-fitting model when
compared to a two-factor model. The affective dimension of psychopathy demonstrated
greater structural stability over a year’s time than the narcissistic or behavioral
dimensions, though other indicators of absolute and relative stability did not support the
stability of the affective factor. Lastly, results indicated that psychopathy significantly
predicts later aggression, after controlling for concurrent aggression and other variables
previously found to be related to aggression and violence (i.e., social anomie, association

with delinquent peers, and lack of parental supervision).

Dedicated to my parents for their unfaltering support in good times and in bad. Thank
you for always showing me how (and sometimes forcing me) to persevere. I would not

have made it through without you. I love you both very much.

iii

ACKNOWLEDGEMENTS

I appreciate the opportunity to also thank my committee members for their
continued assistance and support. Dr. Rick Deshon is a brilliant teacher and mentor who
repeatedly provided me with statistical guidance and friendship. Dr. Bertram Karon is an
outstanding supervisor, analyst, and teacher, who has served as a wonderful role model
throughout my years in graduate school. Dr. Robert Caldwell’s flexibility, feedback, and
superior knowledge of the APA Style Manual helped me tremendously.

Also, thank you to my guidance committee members, Dr. Joel Nigg and Dr. Kelly
Klump. Throughout my graduate school career, Joel has provided me with unequaled aid
and encouragement for which I am forever appreciative. Kelly has been extremely
generous with her time and guidance, and has pushed me to produce higher quality work.

To my friend and former labmate, Dr. Allison Schettini-Evans: if it weren’t for
your unending work, amazing organization, tireless leadership, and ceaseless support, this
dissertation would not exist. Also, a tremendous thank you to all the undergraduate
research assistants who spent hours of their time collecting and entering the data for this
project.

I would like extend a special thank you to my supervisors and colleagues at my
clinical internship at the Denver VA Medical Center: Dr. Sara LeCaptain, Dr. Allison
Krawiecki, Dr. Heather Nations, Dr. Ralph Wechsler, Dr. Larry Wahlberg, Dr. Catie
Johnston-Brooks, Dr. Stephanie Kleiner, Dr. Lisa Brenner, Dr. Michael Craine, and Dr.
Marty Smith. Their understanding and unwavering support as colleagues and friends

made it possible to complete this project.

iv

Most importantly, this dissertation and my eventual graduation would not have
been possible without the assistance and encouragement of two people: Dr. Susan Frank,
my advisor, and Dr. Edward Gibeau, one of my clinical supervisors. Both championed
me, sustained me, and cared for me before, but especially after, my accident. I would not
have continued in school if it weren’t for their kindness and support. I can never repay
all that they have provided for me.

Finally, I need to thank all of the doctors and physical therapists who have treated
me over the last four and a half years. Each had a part in helping me to be a whole

person again and I will be forever grateful.

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. vi
LIST OF FIGURES ................................................................................ viii
INTRODUCTION ................................................................................... 1
CHAPTER 1
CONCEPTUALIZATION OF PSYCHOPATHY ............................................... 4
Two-Factor Models ......................................................................... 4
Three-Factor Models ....................................................................... 12
Adult-Youth Differences ................................................................. 16
CHAPTER 2
ISSUES IN THE ASSESSMENT OF PSYCHOPATHY .................................... l9
Temporal Stability ......................................................................... 19
Gender ...................................................................................... 23
CHAPTER 3
PSYCHOPATHY AND ITS RELATIONSHIP TO VIOLENCE ............................ 28
Psychopathy and Conduct Disorder ..................................................... 28
Evidence of the Psychopathy-Violence Association in Adolescent Populations ..31
CHAPTER 4
CONCLUSION ...................................................................................... 42
Hypotheses .................................................................................. 42
CHAPTER 5
METHOD ............................................................................................ 44
Participants ................................................................................. 44
Measures .................................................................................... 45
Measures of Psychopathy ........................................................ 45
Measures of Additional Variables .............................................. 48
Procedure ................................................................................... 52
Time 1 .............................................................................. 52
Time 2 .............................................................................. 53
Missing Data ............................................................................... 53
Research Predictions ...................................................................... 54
CHAPTER 6
RESULTS ........................................................................................... 55
Attrition ..................................................................................... 55
Goodness of Fit of the Structural Models .............................................. 55

vi

Prediction 1 ........................................................................ 57

Prediction 2 ........................................................................ 59
Prediction 3 ........................................................................ 61
Structural Stability ....................................................... 61
Absolute Stability...................................‘ ..................... 63
Relative Stability ......................................................... 63
Prediction 4a ....................................................................... 64
Prediction 4b ....................................................................... 65
Predicting Changes in Aggression from Psychopathy ................................ 65
Prediction 5 ........................................................................ 65
CHAPTER 7

DISCUSSION ....................................................................................... 73
APPENDICES ...................................................................................... 87
Appendix A. Two—Factor Model of Hare's Psychopathy Checklist-Revised. 88
Appendix B. Two-F actor Model of APSD based on Frick et al. (1994) ........... 89
Appendix C. Two-F actor Model of LSRP based on Lynam et al. (1999) ......... 90

Appendix D. Three-Factor Model based on Cooke and Michie (2001)
and Frick et al. (1994) ..................................................... 91
Appendix E. Figure 1 ...................................................................... 92
Appendix F. Figure 2 ...................................................................... 93
Appendix G. Figure 3 ..................................................................... 94
Appendix H. Figure 4 ..................................................................... 95
Appendix I. Figure 5 ...................................................................... 96
REFERENCES ...................................................................................... 97

vii

LIST OF TABLES

Table 1. Cleckley criteria for psychopathy ........................................................ 5
Table 2. Models of psychopathy ................................................................... 6
Table 3. Sample demographics at Time 1 and Time 2 ......................................... 45
Table 4. Items from the Levenson Self Report of Psychopathy Scale (LSRP) ............ 47
Table 5. Items from the Antisocial Process Screening Device (APSD) ..................... 48
Table 6. Items from the Aggressive Behavior Scale ........................................... 51
Table 7. Fit indices for the deviance and non-deviance models of the total sample. 52
Table 8. Item loadings of cross—loaded items for the APSD three-factor model ........... 59

Table 9. Fit indices of the two- and three-factor models using both the LSRP and
the APSD .................................................................................. 60

Table 10. Fit indices for the LSRP factors assessed simultaneously at Time 1 and
Time 2 .................................................................................... 62

Table 11. Alpha coefficients for the LSRP three psychopathy factors at T1 and T2.... .. 62

Table 12. Mean differences between T1 and T2 for the LSRP three psychopathy

factors (n=87) ........................................................................... 64
Table 13. Test-retest correlations between times 1 and 2 for LSRP total scores and

factors (n = 87) .......................................................................... 64
Table 14. Fit indices for the LSRP factors assessed simultaneously across gender. 65
Table 15. Mean levels of psychopathy for both gender measured by the LSRP ........... 66
Table 16. Mean levels of psychopathy for both gender measured by the APSD .......... 66
Table 17. Intercorrelations among the variables in the SEM procedure .................... 67

Table 18. Standardized estimates for the prediction model of aggression with
youth report of psychopathy .......................................................... 69

viii

Table 19. Fit indices for the predictive model of aggression .................................. 69
Table 20. Standardized estimates for the trimmed prediction model of aggression. . . 70

Table 21. Fit indices for each psychopathy factor in the predictive model of
aggression ............................................................................... 71

ix

LIST OF FIGURES

Figure 1. Predictive model of the prediction of aggression
with youth report of psychopathy ................................................................. 92

Figure 2. Trimmed model of the prediction of aggression
with youth report of psychopathy ................................................................. 93

Figure 3. Model of the prediction of aggression with
Narcissistic factor of psychopathy ................................................................ 94

Figure 4. Model of the prediction of aggression with
Affective factor of psychopathy ................................................................... 95

Figure 5. Model of the prediction of aggression with
Behavioral factor of psychopathy ................................................................ 96

INTRODUCTION

Across many domains, there is interest in the development and maintenance of
antisocial behavior and aggression in adults, and more recently, children and adolescents.
This interest is spurred by the widespread economic, legal, and human costs incurred
because of antisocial, criminal, and violent behavior. Research has illustrated that there
is not just one “type” of antisocial individual and work has focused on denoting the
subtypes of offenders (Frick, 2000). With adults, the construct of psychopathy has
proven effective in classifying one subtype of antisocial offender, identifiable because of
a typically severe and violent behavior pattern (Edens, Skeem, Cruise, & Cauffman,
2001)

Psychopathy is a disorder characterized by affective, interpersonal, and behavioral
deficits. Adult criminal offenders who are also classified as psychopaths are more
aggressive and commit more than twice as many violent acts as nonpsychopathic
criminals (Hare, 1993). They also display higher rates of recidivism, commit more total
crimes, and spend more time in prison than their nonpsychopathic counterparts (Harpur
& Hare, 1994; Harris, Rice, & Corrnier, 1991; Selekin, Rogers, & Sewell, 1996). In
addition, the violent acts of psychopaths are more likely to be premeditated and
instrumental, used to satisfy an instrumental need or obtain something the psychopath
desires (F rick, 2000; Hare, 1993). Unlike nonpsychopathic criminals whose violence
generally occurs as a result of heated disputes, intense emotions, or precipitating distress,

the psychopathic criminal exhibits dispassionate, business-like violence and aggression

(Hare, 1993).

Similar relationships between psychopathy and violence have been found in
studies of adolescent offenders (Frick, 2000; Gretton, 1999). Yet, the investigation into
the nature of the psychopathy construct in adolescents and even younger children is still
in its infant stages. Work in this area is important. Interventions with psychopathic
adults have proven difficult (Hare, 1993; Ogloff, Wong, & Greenwood, 1990; Rice,
Harris, & Cormier, 1992; Serin, 1996), suggesting that by adulthood psychopaths are
relatively resistant to treatment interventions (McBride, 1998). Therefore, it becomes
important to intervene with this subgroup of offenders earlier. Identification and
understanding of the disorder in a younger cohort may lead to the development of more
effective and earlier interventions, leading to a reduction in future violence and antisocial
behavior.

One issue that has hindered the study of psychopathy in adults is disagreements
among researchers regarding its conceptualization and measurement (Cooke & Michie,
2001; Lilienfeld, 1994). Similar difficulties have also appeared in the adolescent and
child literature (Frick, 2000; Salekin, Rogers, & Machin, 2001). Yet, progress in the
research on psychopathy in younger cohorts depends a great deal upon better clarification
of the construct (Salekin et al., 2001). Adding to the conceptualization difficulty is the
fact that while some posit that psychopathic personality is established in childhood (see
Forth, Hart, & Hare, 1990; Frick, 2000); others wonder whether the concept of
psychopathy in adults can even be applied to adolescents and children (Edens et al., 2001;
McBride, 1998). This disagreement stems from the general temporal instability of
adolescent behavior and attitudes, and the rapid developmental changes that take place

during this stage of development.

A delineation of the appropriate conceptualization of psychopathy in adolescence
is only the beginning of the examination of the construct in this age group. It is also
necessary to determine whether psychopathy or its dimensions are useful in the prediction
of important outcomes like violence and aggression, especially over and above other
variables that have been significantly related to these outcomes in adolescents. Most of
the handful of studies that have examined the ability of psychopathy to predict future
violence (e.g. violent reoffense, violent or aggressive institutional infractions) have
reported a statistically significant relationship in the expected direction. However, only
one study (Brandt, Kennedy, Patrick, & Curtin, 1997) has examined the incremental
predictive validity of psychopathy to predict violent re-offense after other factors (i.e.,
education level, substance abuse, age at assessment, criminal history, number of Conduct
Disorder symptoms) were first considered and offered evidence that psychopathy
explains additional variance in future violence.

This study explored the nature and validity of the psychopathy construct in a
sample of court-referred adolescents. First, it assessed the “fit” of alternative models of
the disorder, previously identified in adults and children, in an adolescent sample using
two different measures of psychopathy; dimensions of the best fitting model were also
tested across gender. Second, it examined the absolute and relative stability of
psychopathy and its dimensions over time. Third, it explored whether and to what extent
psychopathy predicts future aggression both directly and over and above other factors,
namely association with delinquent peers, lack of parental supervision, and social

anomie, each of which has previously been associated with aggression.

CHAPTER 1

CONCEPTUALIZATION OF PSYCHOPATHY

Cleckley’s (1976) characterization of psychopathy is still considered the
framework for much of the current research on this disorder (Hare, 1993). According to
Cleckley, psychopathy includes affective and narcissistic [italics added] deficits reflected
in a lack of empathy, guilt or remorse, manipulation of others, superficial charm,
egocentricity, glibness, and untruthfulness. Cleckley proposes that psychopaths also
display several characterological deficient or immature behavioral patterns [italics
added], namely irresponsibility, unreliability, a failure to learn from experience, and a
lack of long-tenn goals (see Table 1). Cleckley maintained that because of these deficits
psychopaths are prone to antisocial behavior, most singly what he termed inadequately
motivated antisocial behavior, which can take place on a whim and lacks consistency. In
his view, an antisocial lifestyle is not a necessary consequence of the disorder. Cleckley
asserted that some psychopaths do not exhibit a history of antisocial behavior and may
even hold socially acceptable employment.

Currently there are no fewer than four alternative models outlined in the adult and
child literature (see Table 2), all reportedly based on Cleckley's characterization of
psychopathy. Differences in the measures and reporters employed, and age and gender
composition of the samples used, has muddied the search for the ideal construct. In
addition, rarely have any of these models been tested or compared in adolescent samples.
Two factor models

Hare and his colleagues (Hare, Harpur, Hakstian, & Forth, 1990; Harpur,

Hakstian, & Hare, 1988; Harpur, Hare, & Hakstian, 1989) proposed a two-factor model

of psychopathy defined as a higher order construct with two distinct factors or

dimensions (Hare et al., 1990). The first factor describes the affective and narcissistic

personality traits of the disorder and the second includes behavioral deficits such as

Table 1

C leckley criteria for psychopathy.

 

10.

ll.

12.

13.

. Considerable superficial charm and average or better intelligence

Absence of delusions or other signs of irrational thinking.
Absence of anxiety or other "neurotic" features.

Unreliability, disregard for obligations.

Untruthfulness and insincerity.

Lack of remorse or shame.

Inadequately motivated antisocial behavior.

Poor judgment and failure to learn from experience.
Pathological egocentricity, totally self-centered; incapacity for love.
General poverty of major emotional reactions.

Specific lack of insight, unable to see oneself the way others do.
Unresponsive to special considerations, kindness or trust.

Fantastic and uninviting behavior after drinking and sometimes with no drinking

(e.g., vulgarity, rudeness, quick mood shifts, pranks).

14. No genuine suicide attempts.

15.

16.

Sex life impersonal, trivial, and poorly integrated.

Failure to follow a life plan.

 

Table 2

Models of psychopathy.

 

Model

Background

Factors

 

2-factor model from
Hare et al. (1990)

2-factor model from
Frick et al. (1994)

2-factor model from
Lynam et al. (1999)

3-factor model from
Cooke and Michie
(2001) and Frick et al.
(2000)

Based on work with adult
criminal samples and the

Psychopathy Checklist (PCL)

Based on work with child

community and clinical samples
using the Antisocial Process

Screening Device (APSD),
which is based upon Hare's
PCL

Based on work with adult
(college age) community

samples and the Levenson Self-

Report Psychopathy Scale,
which is based upon Hare's
PCL

Based on work with both adult

criminal and child

community/clinical samples

using the PCL and APSD

1. Affective and
narcissistic personality
traits

2. Behavioral traits with
social deviance/antisocial
lifestyle criteria

1. Affective personality
traits

2. Narcissistic personality
traits and behavioral traits
with social deviance/
antisocial lifestyle criteria

1. Affective and
narcissistic personality
traits

2. Behavioral traits without
social deviance/antisocial
lifestyle criteria

1. Affective personality
traits

2. Narcissistic personality
traits

3. Behavioral traits without
social deviance/antisocial
lifestyle criteria

 

impulsivity, irresponsibility, proneness to boredom, parasitic lifestyle, and lack of long-

term goals (see Appendix A). Hare’s two-factor model of psychopathy was designed to

be based on Cleckley’s formulation and has been the dominant approach to the

assessment of psychopathy over the last decade (Edens et al., 2001). Operationalization

of Hare’s model has demonstrated a strong correlation with Cleckley’s criteria (r = .83;
Hare, 1991 as cited in Schmitt & Newman, 1999). Yet, Hare’s model arguably deviates
in some ways from Cleckley’s theory, representing or sharing in only seven of Cleckley’s
16 criteria (Salekin, Rogers, & Sewell, 1996). This deviation from the traditional view of
psychopathy stemmed from the fact that the goal of the model is to best discriminate
psychopathic from nonpsychopathic inmates. Because of this focus on criminal
populations, Hare’s model relies more heavily on criminal and social deviance
descriptors than outlined in Cleckley’s original criteria for the disorder (Edens et al.,
2001)

When developing the Psychopathy Checklist (PCL, Hare, 1980), his
Operationalization of his model, Hare selected items that tapped the personality variables
Cleckley proposed as the core of the disorder. However, because Hare was interested
primarily in criminal populations, he included items assessing involvement in criminal
activity as well. The 20-item revision of the PCL, the PCL-R, includes three social
deviance items that measure: (a) the presence or absence of juvenile delinquency, (b) the
revocation of conditional release, and (c) criminal versatility. The first two items load on
the PCL- R second factor; the third is only included in the total psychopathy score.
Trained raters generally score the PCL-R items after an interview and file review are
conducted.

These types of concrete indicators of social deviance potentially improve the
reliability of an instrument such as the PCL-R for use in criminal populations, but
potentially sacrifice its construct validity. In fact, the second factor, termed social

deviance, along with the criminal versatility item, closely reflects the Diagnostic and

Statistical Manual, Fourth Edition (DSM-IV, American Psychiatric Association)
diagnosis of Antisocial Personality Disorder (ASPD). Because the inclusion of antisocial
behavior oversteps the boundaries of Cleckley’s original notion, classification is likely to
result in a heterogeneous group of individuals identified as psychopathic, all of whom
meet criteria for ASPD, but only some of who manifest the core personality traits of
psychopathy.

Although the PCL-R can be decomposed into two separate factors, identification
of “psychopathic” individuals relies on a total score as the measure of psychopathy.
Because total scores may be influenced by either or both factors, one or the other may be
weighted more heavily in the final classification decision (Selekin et al., 1996; Wootton,
F rick, Shelton, & Silverthom, 1997). Though the two dimensions are moderately
correlated, this level of association also suggests that some individuals may exhibit
characteristics of one dimension and not the other. Research indicates that the factors
have consistently shown differential correlates (Cooke & Michie, 2001; Harpur et al.,
1989; McBride, 1998). The personality-based factor scores correlate negatively with
measures of anxiety and empathy, and positively with scores on narcissism, instrumental
and predatory violence, and poor therapeutic change. Social deviance factor scores have
been negatively associated with IQ and socioeconomic status, and positively related to
sensation seeking and spontaneous, reactionary aggression.

In addition, investigations into the ability of each factor to predict violence have
revealed variable results. Lilienfeld’s (1994) review reported that measures that include
the core personality traits (e.g., PLC-R factor 1) of psychopathy are better predictors of

outcomes like recidivism and future violent offending than measures of ASPD criteria

alone. This report corresponded with Hart, Hare, and Forth’s (1994) contention that
generally factor 1 scores are as, if not more, predictive of violence than factor 2 scores.
Conversely, a review of the literature by Salekin, Rogers, and Sewell (1996) stated that
factor 2 scores were a better predictor of violent recidivism than factor 1. In either case,
confounding the two dimensions is potentially problematic because, as noted, not all of
the PCL items may actually tap into the traditional construct of psychopathy and high
scores on different combinations of items may in fact be measuring different constructs.

The inclusion of social deviance behavioral indicators in Hare’s two-factor model
makes it a behavior-based conceptualization of psychopathy. This conceptualization is
what Lilienfeld (1994) termed a “closed” concept because it is defined explicitly,
whereas the personality-based approach is an “open” concept. The open approach
generally provides a better fit to most diagnostic categories than a closed approach
because diagnostic categories are characterized by unclear boundaries, a potentially
infinite number of indicators for their criteria, and imperfect relationships between the
indicators and their criterion. Inclusion of closed criteria in diagnosis compromises the
construct validity because most psychiatric syndromes and their related diagnostic criteria
are generally open concepts.

In addition to social deviance indicators leading to possible misclassification, it
may be that the use of these closed behavioral indicators in factor 2 clouds the assessment
of its underlying personality traits. Harpur and Hare (1994) found that, afier age 40,
psychopaths’ factor 1 scores persisted whereas their factor 2 scores declined, showing a
significant difference in temporal stability. This decline in factor 2 scores mirrored

results typically found with Antisocial Personality Disorder. The addition of the

behavior-based indicators of criminal history and antisocial behavior may help to explain
why PCL factor 2 scores do not show the same stability over time as factor 1 scores.
Harpur and Hare stated that one possible explanation for the age-related differences in the
PCL factors is that the personality traits underlying factor 2 may be stable across the
lifespan, yet the behavioral indicators used to measure these traits show an age bias. This
explanation seems likely given the research indicating that personality traits generally
remain stable after age 30.

Most measures of psychopathy in children and adolescents are based on Hare’s
PCL-R and include items assessing social deviance. One of the most widely used
instruments with adolescents is the youth version of the PCL, the PCL: YV (Forth, Hare,
& Hart, 1990). This modified youth version employs the same two-factor structure as the
PCL-R, though it omits the items on parasitic lifestyle and marital history, items not as
applicable to adolescents. The Antisocial Process Screening Device (APSD, Frick &
Hare, in press) formally the Psychopathy Screening Device (PSD), has seen an increase
in use for the assessment of psychopathy in youth. The APSD is a 20—item parent- or
teacher-rated measure of psychopathy traits based on the PCL-R, but modified for
developmental appropriateness.

As would be expected being based on the PCL and representing downward
extensions of Hare’s adult model, both the PCL: YV and APSD include the assessment of
social deviance criteria to some degree. The PCL: YV still includes items measuring
juvenile delinquency and criminal versatility, although the scoring of these items is
modified. Instead of using a three-point scale, all violent crimes on the juvenile

delinquency item are rated a ‘2’ and all non-violent ones are rated ‘1.’ For criminal

10

versatility, four or more different types of offenses rates a ‘2’, three types scores a ‘ 1,’
and one or two types of crimes rates a ‘0.’ In contrast, only one of the 20 APSD items,
“engages in illegal activities,” assesses social deviance. As with adults, the arguments
against including social deviance criteria in the conceptualization and assessment of
psychopathy also apply to youth.

In adolescents, Hare’s two-factor model is generally employed, measured by the
PCL: YV, even though this model has received little empirical support. When tested via
confirmatory factor analysis procedures in an adolescent offender sample, the two-factor
model received only modest support with a reported comparative fit index of 0.83
(Brandt et al., 1997). In his unpublished dissertation study, Cruise (2001) reported a
slightly better fit of a two-factor oblique model with juvenile offenders, listing a
comparative fit index of 0.89 and x2 = 136.29, p = .01.

The original analysis of the APSD on a group of clinic-referred children aged 6 to
13 also yielded a two-factor model, but one quite different than that outlined by Hare in
adult populations (Frick, O'Brien, Wootton, & McBurnett, 1994). Principal component
analysis yielded one dimension that contained items not only assessing narcissistic traits,
but those tapping persistent characterological behavioral patterns and social deviance
criteria [italics added]. The second component contained items reflecting affective
deficits [italics added], which the authors termed Callous/Unemotional (see Appendix B).
This is in contrast to Hare’s adult two-factor model in which the narcissistic and affective
traits load on one factor and the behavioral and social deviance items load on another.

For both adults and youth, the inclusion of social deviance behavior-based criteria

in measures of psychopathy may limit the instrument’s construct and predictive validity

ll

when assessed in terms of certain important outcomes. Investigators have recently
questioned Hare and colleagues’ two-factor model of psychopathy along these and other
lines. In accordance with Cleckley’s (1976) position that psychopaths are found outside
the criminal domain, Cooke and Michie (2001) argue that social deviance criteria, for
example criminality and history of antisocial behavior, are not basic tendencies — core
personality features - of psychopathy, but possible characteristic adaptations of the
disorder. These characteristic adaptations result from the interplay between core traits
and environmental influences. Though his research focus is mainly criminal
psychopaths and his PCL criteria include antisocial behaviors, Hare (1993) also
acknowledged that not all psychopaths are criminals.

Following Cleckley’s original conceptualization and the arguments against the
inclusion of social deviance criteria in the conceptualization of psychopathy, recent
investigations have focused on the assessment of the construct in non-criminal,
community-based samples. Levenson, Kiehl, and Fitzpatrick (1995) developed the
Levenson Self-Report Psychopathy (LSRP) scale, a measure based on Hare’s PCL, for
use with noninstitutionalized populations. Like with versions of the PCL, analyses of the
LSRP demonstrated a two-factor model with affective and narcissistic traits loading on
one factor and behavioral traits loading on the second one (Levenson et al., 1995; Lynam,
Whiteside, & Jones, 1999). Yet, unlike the model based upon the PCL versions, the
Levenson model assumes the irrelevance of social deviance criteria. However, because
the LSRP does not include social deviance items, this assumption remains untested.
Three-factor models

Recently, Hare’s model of psychopathy has been questioned not only because of

12

its deviation from Cleckley’s original characterization and the inclusion of social
deviance criteria, but also because of its two-factor solution. Cooke and Michie (2001)
argued that previous investigators misinterpreted congruence coefficients as evidence for
the two-factor solution and their own attempts to replicate this model fell short in proving
its adequacy. These authors conducted both exploratory and confirmatory factor analyses
procedures on PCL-R data from a sample of 2,067 participants from North American
correctional and forensic institutions. Results from the exploratory analysis were
inconclusive, and the confirmatory procedure indicated that the two-factor solution did
not provide an acceptable fit to the data. Accordingly, the investigators concluded that
the two-factor solution was not “an adequate structural model for psychopathy” (p. 173).
Cooke and Michie (200]) subsequently reanalyzed a random half of their data
using newer statistical methods. An exploratory factor analysis produced five
interpretable factors, with most items loading on two factors. The social deviance
descriptors of juvenile delinquency, early behavioral problems, promiscuous sexual
behavior, many short term marriages, revocation of conditional release, and criminal
versatility formed couplets and defined the remaining three factors. The authors went on
to compare five different models following a model building process employing
confirmatory factor analysis procedures and informed by item response theory. The first
model, consisting of all five factors and including the social deviance behavioral
descriptors, resulted in the most unsatisfactory fit, with two other models showing an
adequate fit to the data. One of these adequate models consisted of 10 items described by
two correlated factors; the second included 13 items described by three correlated factors.

Both models demonstrated similar fit indices; the comparative fit indices of the two-

13

factor and three-factor models were 0.97 and 0.96, respectively. Cooke and Michie
argued that the three-factor model was superior because it included items traditionally
viewed as central to the psychopathy construct, namely glibness/superficial charm,
pathological lying, and manipulativeness. It should be noted that neither of the two best
fitting models included the social deviance descriptors. When cross-validated on the
other half of the data set, the new three-factor model exhibited a satisfactory fit
(comparative fit index = 0.97, RMSEA = 0.05). The authors also successfully replicated
the three-factor model using other psychopathy measures, including the short version of
the PCL (PCL-SV), the Psychopathy Criterion Set, and items from the ASPD and
Dissocial Personality Disorder criteria sets.

The three factors of Cooke and Michie’s (2001) new model are consistent with
the clinical tradition defining psychopathy in terms of three domains: narcissistic,
affective, and behavioral (see Appendix D). Cooke and Michie (2001) describe the first
factor as Arrogant and Deceitful Interpersonal Style, defined by items assessing glibness,
superficial charm, a grandiose sense of self, pathological lying, and
conning/manipulativeness. The second factor, termed Deficient Affective Experience, is
defined by items related to shallow affect, callousness and lack of empathy, lack of
remorse or guilt, and failure to accept responsibility for one’s actions. The last factor of
Impulsive and Irresponsible Behavioral Style includes items measuring need for
stimulation/proneness to boredom, impulsivity, irresponsibility, parasitic lifestyle, and
lack of realistic, long-term goals. Cooke and Michie’s conceptualization meshes more
closely with that of Cleckley than Hare’s model, though some differences still remain.

The authors still include impulsivity, which Cleckley did not posit to be a core trait of

14

psychopathy. In addition, because Cooke and Michie used PCL-R items as the basis for
their model, they did not include facets of psychopathy suggested by Cleckley but not
included in the PCL-R, such as absence of anxiety or above average intelligence.

In contrast to Hare and colleagues’ two-factor model, measures of criminal or
antisocial behavior throughout the lifespan do not contribute to any of the three
dimensions the Cooke and Michie model of psychopathy. This exclusion of social
deviance criteria coincides with Cleckley’s conception that antisocial and criminal
behavior is a common outcome of psychopathic personality, but is not a necessary
component of the disorder. The behavioral domain (defined by lack of long-term goals,
irresponsibility, and parasitic lifestyle) is still an important part of the psychopathy
construct in the three-factor model, but deviant behavioral indicators that may limit
construct validity are no longer included in its measurement (Lilienfeld, 1994).

A study by Frick and colleagues (Frick, Bodin, & Barry, 2000) suggested that a
three-factor model, similar to the one described by Cooke and Michie (2001), might also
provide a better fit in children (see Appendix D). Utilizing the APSD in a sample of
1,136 elementary school children, these authors found evidence for both the original two-
factor model (with narcissistic and impulsive traits combined on one factor and affective
traits on a separate factor) and the three-factor model (narcissistic, affective, and
behavioral traits separated) using exploratory factor analyses. Then, using confirmatory
factor analysis procedures, they tested the fit of both the two-factor and three-factor
models on the six- to 13-year-old clinic-referred sample described in Frick, O'Brien,
Wootton, and McBumett (I 994). The three-factor demonstrated only a slightly better fit

than the two-factor model (comparative fit indices of .92 and .91, respectively). But,

15

notably, the APSD social deviance item (i.e., engages in illegal activities) did not
contribute to the final three-factor model.

Only one investigation examined whether the three-factor solution outlined by
Cooke and Michie would replicate in an adolescent sample. Edens, Skeem, Cruise, and
Cauffman (2001) reexamined previous PCL-R data (Edens, Poythress, & Lilienfeld,
1999) for 50 youthful offenders utilizing the three-factor model suggested by Cooke and
Michie. Results of this reexamination were mixed. The Arrogant and Deceitful
Interpersonal Style factor related significantly to both verbal and physical aggression (r =
.29, p<.05). The correlation between the aggression outcome measures and the Deficient
Affective Experience factor was somewhat smaller (r = .21) and non-significant. The
Impulsive and Irresponsible Behavioral Style had the lowest association with aggression
(r = .14) and also had a relatively poor internal reliability afier the behavior-based items
were removed (or = .58, compared to the or = .75 of the original Social Deviance scale).
Though the reanalysis results were not completely supportive of the three-factor model,
the mixed findings and small sample employed indicate that investigation of a larger
sample is warranted.
Adult-youth differences

Even if it is found that the models of psychopathy found in adults generally
extend to adolescents, some investigators have posited that even minor differences in the
manifestation of the construct between adults and adolescents may still make this
downward extension problematic (Cruise, 2001; Salekin et al., 2001). For example,
adult-adolescent differences in item loadings on the dimensions may point to different

interpretations of the dimensions, with some items being more important to the

16

conceptualization of psychopathy in one age group versus the other (Cruise, 2001). In
addition, in some respects, adults, adolescents, and children may differ completely on the
symptoms that make up the disorder of psychopathy (Salekin et al., 2001).

In general, studies of psychopathy in adults utilizing the PCL and PCL-R have
shown that the items assessing egocentricity, grandiose sense of self, and superficial
charm have consistently related most strongly to factor 1 scores describing the core
personality characteristics of psychopaths (Cruise, 2001). This result suggests that
narcissism best captures the personality deficits of psychopathy. Yet, Cruise’s analysis of
PCLzYV data with adolescents demonstrated that affective deficits - callousness, lack of
empathy, and manipulation - were more highly related to factor 1 than narcissistic traits.
His conclusion was that adolescents might exhibit a different pattern of psychopathy than
adults.

An investigation attempting to clarify the diagnostic criteria for psychopathy in
youth (age range 6-18, M = 14.36 years) gathered data on “prototypical” characteristics
from 511 child psychologists who were members of the American Psychological
Association’s Section of Child Clinical Psychology (Salekin et al., 2001). Although the
authors found that many of the core features of psychopathy outlined for adults were also
assumed by professional experts to be prototypical for youth, many adult-to-child
differences were also outlined. Criterion such as “parasitic lifestyle,” “lack of long-term
planning,” “impulsivity,” and “failure to accept responsibility” were found not to apply to
youth. Still, the respondents endorsed other criteria considered to be core personality
features of psychopathy in adulthood, including “lies easily and skillfully,” “does not feel

bad or guilty,” “uses or cons others.” “is unconcerned about the feelings of others,”

17

“emotions are shallow,” and “acts charming in ways that seem insincere” as prototypical
of psychopathy in youth. Other behavioral indicators, some of which are included in

9, 6‘

adult assessments of psychopathy (i.e., “juvenile delinquency, early behavioral
problems”) and some that are not (i.e., “easily annoyed by others,” “has been physically
cruel to others”) were also considered prototypical features of psychopathy in youth.

In sum, in all age groups, the inclusion of behavioral-based, social deviance
criteria as core features of psychopathy is debated. In addition, there is still disagreement
among investigators about the appropriate structural model or factor structure to apply to
psychopathy. Again, this disagreement holds true for all age groups. Finally, although
overlap of core personality features seems to exist, the manifestation of psychopathy in
both children and adolescents may differ from the construct historically defined in adults.
Which personality features are more essential to the conceptualization of psychopathy
may differ by developmental age. While egocentricity and grandiosity appear most
fundamental to adult psychopathic personality traits, both the factor loadings of the
PCL:YV in adolescents and prototypical analysis of psychopathy in youth point to the
greater importance of affective deficits (e. g., lack of empathy) in defining psychopathy in
youth. In addition, some behavioral features may differ between adults and youth.
Distinguishing important adult-youth differences in the psychopathy construct will allow
researchers to better operationalize the construct in youth, leading to more accurate
associations of psychopathy with important outcomes in this age group, as well as more
targeted, successful interventions. At the same time, delineating the features common to

both adults and youth may make it easier to study psychopathy longitudinally across age

groups and shed light on the etiology of this disorder.

18

CHAPTER 2

ISSUES IN THE ASSESSMENT OF PSYCHOPATHY

Temporal stability

Whether a downward extension of the adult model or a modified depiction of
psychopathy is determined to be the best conceptualization of psychopathy in
adolescents, some argue that the construct of psychopathy cannot be appropriately
applied to adolescents given that many personal attributes and attitudes are in flux during
this period of development (Edens et al., 2001). To date, no published studies have
longitudinally assessed the stability of psychopathy in adolescents (Cruise, 2001). The
lack of evidence relating to the reliability of psychopathy over time in this age group
makes it impossible to know whether any model of psychopathy confirmed in adolescents
reflects a stable trait or merely coincides with normal adolescent-limited characteristics
that may change with age.

It is reasonable to question the stability of personality-based disorders in general,
as well as psychopathy, in particular, during adolescence. Investigators have argued
against the utility of the psychopathy construct during a developmental period marked by
rapidly changing attitudes and behavior because many of its core features increase
normatively during the adolescent years. Cognitive limitations in time and social
perspective-taking, and changes in identity that are normative during adolescence may in
large part account for high levels on several criteria of psychopathy, including a
proneness to boredom, impulsivity, a lack of long-term goals, and egocentricity, along

with social deviance (Amett, 1992; Edens et al., 2001; Elkind, 1967; Salekin et al., 2001;

19

Spear, 2000). Measurements of psychopathy, defined a priori by enduring features of
personality, may be confounded by these adolescent-limited characteristics.

Adolescence is a developmental period marked by a change in traits and
behaviors that work to help the adolescent negotiate a drive for independence and acquire
skills needed to function successfully in an adult world. This developmental change can
be seen not just in humans, but also across a number of species. For example, rats in
adolescence showed increases in social investigation and interaction, sensation seeking,
and risk-taking (Spear, 2000). Likewise, human adolescents exhibit a normative increase
in sensation seeking and risk taking. These increases may be related to changes to the
adolescent brain. Neuropsychological advances over the last decade have allowed us to
begin to outline the myriad changes to brain structure and connections that takes place
during adolescence. Some of these changes, especially alterations to the prefrontal cortex
and limbic regions, and modification of dopamine input to these regions, might be
associated with increases seen in certain motivated behavior (Spear). Increases in
sensation seeking and risk taking may impel adolescents to explore new places,
situations, and people that assist in the separation from parents and identity formation.

Though an increase in risk taking and sensation seeking is normative and may
promote important developmental outcomes, these increases can also lead to behavior
that closely mirrors some of the criteria of psychopathy. For example, risk taking might
manifest itself as involvement in illegal activities and antisocial behavior, and, because of
the experimental nature of this behavior, it might be quite varied (Amett, 1992; Spear,
2000). In addition, sensation seeking itself, as operationalized by Zuckerman, includes a

subscale tapping boredom susceptibility, which is a criteria included in most

20

conceptualizations of psychopathy (Amett). Further, high scores on the Disinhibition
component of sensation seeking have been found in adult psychopaths (Zuckerman,
1984). Sensation seeking and risk taking increases may also lead to increases in
impulsive behavior, as youth strive to experience more novel stimuli.

In terms of a lack of long-term goals, another psychopathy criterion, indecision
about the future or having several, perhaps conflicting, future plans is considered
normative for adolescents. In fact, in Loevinger’s (1976) system of ego development, it
is the last two stages - individual and autonomous — that are concerned with issues of
establishing a clear identity. Before this point, youth are in a state of identity diffusion,
marked by confusion about who they are and what they want to do in life. This
ambiguous state may last into the third decade of life for many. Adolescents in the midst
of identity diffusion will not have long-term plans. It is when they become adults and are
expected to enter the work force and develop more long-term social relationships that
aimlessness and a lack of clear goals will be judged as problematic.

Because these behavioral and social deviance criteria of psychopathy overlap with
normative adolescent changes in sensation seeking, risk taking and ego development,
psychopathy scores based on some items measuring these characteristics may be inflated.
Studies have found that scores on the behavioral/social deviance factor of measures based
upon the PCL-R were related to age in both adult and adolescent samples. In an
investigation of 14-18-year-old juvenile delinquents, the younger the adolescent the
lower the scores on the behavioral factor of the PCL: YV (Brandt et al., 1997). Harpur
and Hare (1994) found that factor 2 scores of the PCL-R decreased with age over the

lifespan of adult criminals. It should be noted that both of these studies were cross-

21

sectional in nature, so results should be interpreted cautiously.

In addition to certain behavioral criteria, some personality features of
psychopathy might also be related to normative developmental changes in adolescence.
Egocentrism is a failure to clearly understand where the self ends and the other begins
(Elkind, 1967). As egocentrism extends into adolescence, it is further marked by the
tendency to attribute one’s own thoughts to others. This tendency creates the belief in
adolescents that all others are just as preoccupied with them as they are with themselves.
They create, in Elkind’s terminology, an imaginary audience. The result of believing that
everyone is captivated by one’s life is a view of oneself as unique, special, and
exceptional. It is posited that such views of oneself during adolescence can also be tied
to newfound physical and cognitive abilities, and that such beliefs are integral to the
individuation process. Yet, such beliefs could also lead to self-centeredness and
grandiosity, inflating narcissism scores on psychopathy measures. There is a difference
between normative narcissism and the pathological narcissism of psychopathy, even in
adolescence (see Bleiberg, 1994), but it is possible that current measures of psychopathy,
especially self-reports, may not be sensitive enough to discern between the two.
Interestingly, many of the adolescent-limited characteristics outlined above that may
inflate psychopathy scores are not included in the prototype of the youthful psychopath
described by child psychologists (Salekin, et al., 2001).

Despite the potential overlap of psychopathic traits and normative adolescent-
limited characteristics, and no evidence of the temporal stability of psychopathic traits in
the same sample of adolescents (Cruise, 2001), cross-sectional studies provided some

evidence to dispute the claim that psychopathic personality traits are unstable in

22

adolescence. When investigators compared samples of younger and older adolescents,
they found similar levels of psychopathy across age groups (Forth, Hart, & Hare, 1990;
McBride, 1998). These traits may in fact be stable across this period. However, cross-
sectional data are not conclusive and longitudinal results are needed to address the
question of stability.

Gender

To date, most studies of psychopathy in both adult and youth populations have
utilized male-only samples. Because many of these studies involved criminal
populations, male samples were generally more prevalent and, hence, more accessible.
Yet, as with investigations of conduct disorder and antisocial behavior (e.g., F ergusson &
Horwood, 2002; Moffitt, Caspi, Rutter, & Silva, 2001; Silverthom, Frick, & Reynolds,
2001), researchers are beginning to examine psychopathy in females and uncovering
possible differences in the construct between the genders.

The term “gender differences” is one that has been used broadly. When
comparing males and females on variables of interest, those comparisons can be made on
a structural level, or on the basis of mean differences or differences in prevalence rates.
When examining gender differences in psychopathy, investigators have looked at all of
these types of differences.

Although studies of psychopathy in women are limited in number, in those
focusing on structural gender differences, preliminary evidence with adult females
suggests that differences in psychopathy between males and females stem largely from
differences in behavioral characteristics and social deviance items. Males and females do

not appear to differ greatly on the affective and narcissistic/interpersonal characteristics.

23

For example, Salekin, Rogers, and Sewell (1997), in a test of Hare’s two—factor model
with a sample of female offenders using the PCL-R, found that of the nine original factor
2 items, only three loaded cleanly on factor 2 for the females (“parasitic lifestyle,” “early
behavioral problems,” and “juvenile delinquency”). Four other original factor 2 items
cross-loaded across both factors in the female sample, whereas two others failed to load
above .40 on either factor. Additionally, two items — “promiscuous sexual behavior” and
“criminal versatility” - that are part of the total psychopathy score but not part of either
factor in the original model, load onto factor 2 for the females. Interestingly, factor 1
items remained the same for both males and females.

Another investigation by Frick, Barry, and Bodin (2000) compared elementary-
school age boys and girls on a three-factor model of psychopathy as measured by the
APSD. These authors found similar factor structure for both genders in this community
sample. Yet, two differences were noted. One, there were more cross-loadings between
items on the impulsive and narcissistic factors for girls than for boys. Two, the
dimension defined by the affective deficits of psychopathy explained the most variance in
the boys’ model, whereas this factor explained the least amount of variance in the girls’
model.

An investigation of children and adolescents found gender differences in the
conceptualization of psychopathy in this age group as rated by child clinical
psychologists (Salekin et al., 2000). Prototypical analyses differed for boys and girls age
6-18, driven mainly by differences in the behavioral criteria of psychopathy. Typical
features for boys included aggressive and violent behavior (e. g., rape, used a weapon,

bullies or intimidates others, physically cruel to animals, deliberately destroyed others'

24

property), whereas psychopathic girls were described as displaying less aggressive, but
still oppositional behavior. Psychopathic girls were also described as sexually
promiscuous, whereas this was not a criterion applied to boys. The affective and
narcissistic personality features proposed as prototypical of psychopathy, including “does

9’ ‘6'

uses or cons others, 1S not concerned about the feelings of

’9 66

not feel bad or guilty,
others,” and “lies skillfully and easily” were rated similarly for both genders. As with
adult samples, the prevalence rate of psychopathy for girls was less than for the boys.
Salekin et al. (2000) hypothesized that the lower prevalence rate of psychopathy for girls
found in their study and in previous ones might be explained by the behavioral
differences between boys and girls. Specifically, these authors suggested that boys and
girls might not differ on the personality facet of psychopathy, but that the prevalence for
girls declines when antisocial behaviors are added to the construct.

The gender differences in psychopathy demonstrated in both adult and youth
samples have also been explained as not due to actual differences in the construct of
psychopathy, but due instead to gender bias in the measurement of psychopathy. Grann
(2000) compared 36 male and female adult violent offenders matched for demographic
variables as well as index crime and number of previous violent convictions on each of
the 20 items of the PCL-R. For most items, no gender differences were discovered. Yet,
several items discriminated between male and female offenders. Men more often
endorsed the items lack of empathy/callousness and juvenile delinquency, while women
more often scored higher on sexual promiscuousness. Similar to findings in the studies
examining factor structures across gender, odds ratios demonstrated that most of the

items differing between men and women were behavioral/social deviance items (i.e.,

25

impulsivity, irresponsibility, criminal versatility). While Grann speculated that these
differences might reflect an actual difference in characteristics, he also hypothesized
possible gender bias on the part of the administrator of the PCL-R, which includes both
interview and file review procedures.

In terms of prevalence rates of psychopathy, Salekin et al. (1997) found lower
prevalence rates in their female sample than found previously in male samples. Whereas
prior investigations with male offender samples reported prevalence rates of
approximately 25%-30% using a PCL-R cut-score of 29; only 16% of the female sample
scored above this cut-score. It should be noted that the original two-factor model
proposed that the two dimensions are correlated while Salekin et al. (1997) employed an
orthogonal varimax rotation when testing this model in females.

An unpublished dissertation study (Murphy-Peaslee, 2000) provided evidence for
mean level gender differences in psychopathy, again stemming mostly from differences
in behavioral and social deviance criteria. This investigation compared the PCL-R scores
and Rorschach variables of psychopathic male and female adult offenders. Men and
women scored similarly on factor 1 of the PCL-R, but men scored higher on factor 2,
indicating higher levels antisocial/criminal behavior in the psychopathic men. The author
theorized that the behavioral expression of psychopathy might differ across genders.

Conversely, other studies have revealed similar mean levels of psychopathy for
men and women. In an investigation of violent offenders, mean PCL-R total scores were
slightly higher for men than women, but not significantly different (M = 19.42 and M =
17.78). This pattern also held for both factor scores (Grann, 2000). Subjects from a

methadone treatment program also only differed slightly on mean PCL-R total scores,

26

with men scoring slightly higher (Piotrowski, Tusel, Sees, Banys, & Hall, 1995). In
another examination of psychopathy among individuals in drug treatment, men scored
significantly higher than women on total, factor 1, and factor 2 (Rutherford, Alterman,
Cacciola, & McKay, 1998). Males and females in this study demonstrated similar
correlations between all PCL-R scores and antisocial behavior during youth (i.e., Conduct
Disorder) and adulthood (i.e., Antisocial Personality Disorder). Though, there was one
significant gender difference; a stronger association between total and factor 1 scores and
adult antisocial behavior was found for women as compared to men.

Similar discrepancies in the findings of gender differences in the prevalence of
psychopathy have been demonstrated in adolescent samples. In a small clinical sample
(N = 30), Myers, Burket, and Harris (1995) found a large, significant mean difference in
the total psychopathy scores of adolescent boys and girls as measured by the youth
version of the PCL-R. Oppositely, Loper, Hoffschmidt, and Ash (2001), using a
different measure of psychopathy, the Psychopathy Content Scale, found no mean
differences for male and female adolescents incarcerated in a state juvenile correctional
facility (M = 9.69 and 10.80, respectively).

Overall, studies have demonstrated variable prevalence rates of psychopathy
between males and females, for both adults and youth. Yet, many investigations
comparing the factor structures and item responses across genders have suggested that the
conceptualization of psychopathy also might differ between genders. This difference
appears to be mostly associated with the behavioral and social deviance criteria of
psychopathy, implying that men and women might manifest differing behavioral

expressions as part of the psychopathic disorder.

27

CHAPTER 3

PSYCHOPATHY AND ITS RELATIONSHIP TO VIOLENCE

Psychopathy and conduct disorder

Like adults diagnosed with ASPD, adolescents with Conduct Disorder (CD)
display significant levels of aggression, defiance of authority, and behavior that is
fi'ightening and disturbing to others. Many of these behaviors continue into adulthood
(Valentine, 2001). One difficulty facing those who treat CD adolescents and children is
the heterogeneity of those diagnosed with the disorder. As with antisocial adults, to
address this heterogeneity, investigators have attempted to delineate subgroups of
antisocial adolescents and children to better aide earlier intervention and treatment.
Previous attempts to classify these heterogeneous subgroups focused on the type, pattern,
developmental nature, or timing of conduct problems (Lynam, 1997). Longitudinal
studies have shown that different subgroups of delinquents have distinct developmental
correlates and trajectories (McBride, 1998). For example, children who are fearless and
impulsive develop empathy and guilt less easily than other children, which may lead to
higher rates of aggression (see Kochanska, 1991; Lykken, 1995).

Yet, it has only been recently that classification efforts have started to focus on
personality characteristics as a basis for groupings, a methodology that has shown
promise with antisocial adults. Lynam (1997) suggested that personality traits associated
with psychopathy might be used at the childhood level to describe a homogeneous
subgroup of CD children and adolescents. In adults, the psychopathic subgroup of

conduct disordered/antisocial persons appear homogeneous in terms of their pattern of

28

offending (more violent), resistance to treatment, and psychophysical and linguistic
measures (McBride, 1998; see Hare, 1998; Lykken, 1995). Antisocial youth who also
exhibit psychopathic personality traits might characterize a similar subgroup.

There have been past efforts to apply concepts similar to psychopathy to youth in
order to better understand subgroups of conduct disorder (Frick, 2000). Lynam (1996)
theorized that children high in both conduct disorder/oppositional defiant disorder and
Attention Deficit Hyperactivity Disorder (ADHD) exhibit antisocial behavior and
neuropsychological correlates similar to adults with psychopathy. Lynam (1998) showed
that 12-13 year old boys high in both ADHD and CD demonstrated several characteristics
associated with psychopathy; these characteristics were not associated with either ADHD
or CD alone. In a study of adult inmates, those retrospectively diagnosed with both
childhood CD and ADHD had significantly higher scores on PCL-R Total, factor 1, and
factor 2 scales than those with CD alone (Vitelli, 1998). The childhood CD/ADHD
group also was more likely to have a history of childhood violence. Similarly, in another
study, the undersocialized aggressive CD subtype demonstrated poorer treatment
prognosis, higher rates of failure on conditional release, and higher rates of recidivism
much like adult psychopaths (Quay, 1987). The undersocialized aggressive CD subgroup
was characterized by ADHD, aggression, poor peer relationships, and lack of empathy, a
description that closely resembles the depiction of psychopathy. Yet, the similarities
between the CD/ADHD and undersocialized aggressive subgroups, and adult
psychopaths appear to be mainly on a behavioral level (McBride, 1998).

Frick (2000) suggested that in delineating subgroups of CD it is important to

distinguish between children who do and do not demonstrate the affective and

29

interpersonal traits associated with psychopathy in adulthood. He and his colleagues
(Barry et al., 1999; Barry, Frick, DeShazo, McCoy, Ellis, & Loney, 2000; Christian,
Frick, Hill, Tyler, & Frazer, 1997; Frick, Cornell, Barry, Bodin, & Dane, 2001) have
found that children who display personality characteristics that correspond to
psychopathic traits in adults exhibit a greater severity of aggression than children who do
not have those characteristics. Barry et al. (2000) investigated 6- to 13-year old clinic-
referred children that presented with CD and ADHD or ADHD only. They also assessed
for the presence of callous-unemotional traits, which correspond to the affective and
interpersonal deficits associated with psychopathy, e.g., lack of guilt, absence of
empathy, and shallow and constricted emotions. They found that the children with
conduct problems and ADHD plus callous-unemotional traits (CU), as compared to a
conduct/ADHD group without CU traits and a control group, more often displayed
characteristics commonly associated with psychopathy in adulthood. These
characteristics included greater severity and variety of antisocial behavior, more early
contact with police, a reward-dominant response style/avoidance learning deficits, a
preference for thrill and adventure seeking activities, and fewer intellectual deficits.

The ADHD/conduct problem group also differed from the ADHD/conduct
problem plus CU trait group along other lines. Those with high levels of CU traits did
not have elevated levels of anxiety when compared to other groups, while those high on
CD/ADHD and low on CU traits had the highest level of anxiety of all the groups. Frick
et al. (2001) found similar results using a non-referred sample of elementary school
children. Those children assessed as high on both conduct problems and CU traits

exhibited extremely high levels of aggression after one year. In addition, these children

30

displayed high levels of both reactive and instrumental aggression. Children with
conduct problems but without CU traits exhibited lower levels of overall aggression and
were especially lower on instrumental aggression.

The CD/ADHD plus CU traits group that demonstrated a continuity of aggression
and antisocial conduct over time best fits with Moffitt’s (1994) description of life-course
persistent antisocial behavior. Moffitt distinguished between life-persistent and
adolescent-limited antisocial behavior. The life-persistent course is characterized by
child-onset and continuity of antisocial behavior throughout adolescence and into
adulthood. Conversely, the adolescent-limited type is defined by adolescent onset,
temporal instability, and inconsistency across situations. Moffitt hypothesizes that
neuropsychological deficits lay the groundwork for the persistent course, but lasting
antisocial behavior only develops if certain environmental conditions are present to create
increasingly difficult behavior and a lack of prosocial skills. The theory of life-course
persistent antisocial behavior stresses the idea that the environmental conditions are
constantly reacting to a person’s underlying trait, leading to the pervasiveness of an
antisocial attitude throughout all domains of a person’s life. As Frick and his colleagues
work has demonstrated, traits associated with psychopathy appear to be related to
continuity of aggression over time in children, which is consistent with the theory of life-
course persistent antisocial behavior.

Evidence of the psychopathy-violence association in adolescent populations

Though Frick and colleagues have attempted to apply personality traits similar to

adult psychopathy to school-age children, other investigators (e.g., McBride, 1998;

Toupin et al., 1995) have suggested that more knowledge about the disorder in

31

adolescence is needed to better understand its relationship to conduct disorder, violence,
and adult psychopathy before applying it to younger children. Investigators have posited
that it is important to understand the underpinnings of antisocial behavior and violence in
adolescence because studies of crime reveal that criminals rarely begin their antisocial
activity as an adult. Instead, this activity is generally established by late adolescence
(Moffitt, 1994). To date, studies of adolescents utilizing the two-factor conceptualization
of psychopathy have found similar associations between psychopathy and violence and
recidivism to that found in adults. Using total psychopathy scores that combine both
factors, most investigations have demonstrated correlations between psychopathy and
diverse measures of aggression and violence (e.g., Conduct Disorder symptoms, verbal
and physical aggression, violent recidivism, previous violent offenses, violent
institutional infractions, time to violent re-offense) in the range of r = .20 to r = .60
(Brandt et al., 1997; Caputo, Frick, & Brodsky, 1999; Edens et al., 1999; Forth, Hart, &
Hare, 1990; Gretton, McBride, Hare, O’Shaughnessy, & Kumka, 2001; LOper,
Hoffschmidt, & Ash, 2001; Murdock Hicks, Rogers, & Cashel, 2000; Myers, Burket, &
Harris, 1995; Rogers, Johansen, Chang, & Salekin, 1997; Toupin, Mercier, Dery, Cote, &
Hodgins, 1995). Such similarity of associations between psychopathy and
violent/aggressive outcomes supplies some support for the downward extension of adult
conceptualization of psychopathy to adolescents.

Along these lines, most studies of psychopathy in adolescence have utilized the
PCL:YV. Using this instrument, Forth et al. (1990) showed that total psychopathy scores
of adolescent male offenders significantly correlated with the number of CD symptoms (r

= .64), the number of previous violent offenses (r = .27), the number of institutional

32

charges for violent or aggressive behavior (r = .46), and number of total charges or
convictions for violent offenses (r = .46). In an examination of 50 male, racially diverse
adolescents in a youthful offender prison, Edens et al. (1999) investigated the relationship
between both the modified youth version of the PCL-R and the Psychopathic Personality
Inventory (PPI, Lilienfeld & Andrews, 1996), and disciplinary infractions committed
while incarcerated. Both PCL-R and PPI total scores correlated significantly albeit not
that strongly with combined physically aggressive and verbally aggressive/defiant
infractions (r = .28 and r = .24, respectively).

Also, in an investigation of the relation between psychopathy and institutional
infractions involving 120 male adolescent offenders, researchers (Murdock-Hicks et al.,
2000) found that youth high in psychopathy (score 2 17 as measured by the 12-item
shortened version of the PCL-R, the PCL: SV) committed significantly more total,
nonviolent, and violent infractions than adolescents classified as non-psychopathic (score
< 17). Interestingly, these investigators also found that total PCL: SV scores were
significantly correlated with the number of violent and nonviolent infractions only for
African-American youth. Rogers et al. (1997) looked at PCL: YV psychopathy scores,
Oppositional Defiant Disorder (ODD) and Conduct Disorder (CD) symptoms, and
institutional infractions in a sample of 81 boys and girls in a residential treatment
program for dually diagnosed juvenile offenders. They found that neither the number of
ODD nor aggressive CD symptoms related significantly to verbally aggressive or
physically aggressive infractions, or treatment non-compliance. Conversely, PCL: YV
total scores were significantly correlated with physical aggression (r = .28) and treatment

non-compliance (r = .25).

33

Also utilizing the youth modified version of the PCL-R, Brandt et al. (1997)
examined 130 boys aged 14 to 18 released from a training facility for juvenile offenders
who had committed three or more felonies. Total psychopathy scores were significantly
related to a history of crime severity (r = .25), the number of prior commitments (r = .49),
number of conduct disorder symptoms measured during commitment at the training
facility (r = .48), Child Behavior Checklist Aggression scale (r = .31), and the number of
verbal (r = .31), physical (r = .28), and total (r = .24) major infractions committed during
the youths' stay at the facility. After release from the training facility, those adolescents
whose scores on the modified PCL-R (> 27) placed them in the high psychopathy group
had a significantly shorter length of time before re-arrest for a violent offense. In another
study of 220 male adolescent sex offenders who completed an outpatient sex offender
treatment program, those youth in the high psychopathy group (score 2 30) were
significantly more likely than the low group (score < 18) to commit general, violent, and
sexual offenses during an average follow-up period of 55 months (Gretton et al., 2001 ).
In addition, survival analysis showed that offenders in the high psychopathy group
committed general and violent offenses significantly sooner than those adolescents in the
low psychopathy group. Psychopathy, as measured by the PCL:YV, continued to
contribute to the prediction of both general and violent re-offense even after controlling
for offense history.

Examinations of psychopathy with adolescents in treatment facilities have yielded
similar results to those studies involving offender samples. In an investigation of 30
adolescent inpatients, those youth who reported committing several serious delinquent

behaviors, such as causing serious injury during a fight, stealing, vandalism, breaking and

34

entering, a history of arrests, and purposely injuring or killing animals, had significantly
higher youth-modified PCL-R total scores than those adolescents who reported not
participating in such behavior (Myers et al., 1995). Using the French version of the PCL—
RzYV with a sample of boys in various treatment programs, investigators found that total
psychopathy scores significantly predicted delinquency, aggressive behavior, and the
number of aggressive conduct—disorder symptoms after one year (Toupin et al., 1995).
Measures of psychopathy other than the PCL-R or its modifications have been
employed with adolescents and yielded similar results. In a sample of serious male
juvenile offenders, Kruh, F rick, and Clements (2001) found, after controlling for
historical risk indices, social desirability, and demographic variables like age and
education, that the presence of psychopathic personality traits measured by the APSD
significantly predicted severity and variety of violence history. Offenders whose
violence was characterized as indiscriminate had higher psychopathy scores than those
offenders whose violence was described as a provoked single incident. In addition,
adolescents with a history of violence against women, multiple violent incidents against
the same victim, unprovoked violence, or violence committed with a weapon had
significantly higher psychopathy scores than adolescence without such a history. Using
the Psychopathy Content Scale (Murrie & Cornell, 2000), a measure based on items fi'om
the Millon Adolescent Clinical Inventory and designed to have content similar to the
PCL-R, to measure psychopathy, investigators studied the instrumentality, empathy/ guilt,
and reactivity reported by adolescents for a particular violent incident. They found that
higher levels of instrumentality and lower levels of empathy/ guilt reported by juvenile

offenders were both significantly related to higher Psychopathy Content Scale scores

35

(Loper et al., 2001).

Unpublished dissertation studies investigating psychopathy and violence in
adolescents have reported similar results to those outlined in the published articles. In a
study of 233 male adolescent offenders from a sex offender treatment program assessed
with the PCL:YV, the number of nonviolent and violent crimes, and criminal versatility
were positively associated with psychopathy (McBride, 1998). Conversely, age at first
arrest was negatively related to total psychopathy and both PCL:YV factors. Ridenour
(1996) showed that adolescent psychopathic classification (score 2. 30) predicted future
incarceration after one year after administration of the PCL-R: W (80% of psychopaths
and 21% of non-psychopaths). In addition, youth-modified PCL-R scores accounted for
a significant percentage of variance in future violent (26% of variability) and non-violent
(40% of variability) offenses. Total psychopathy scores accounted for unique variance in
future offending after controlling for SES, age of first offense, and other variables related
to Moffrtt’s Conduct Disorder Taxonomy.

Also using the PCL: YV to assess psychopathy, Gretton (1999) found that
adjudicated adolescents who were retrospectively diagnosed with psychopathy exhibited
higher rates of violent offenses, committed crimes earlier following release, and had a
greater likelihood of escaping from custody over a 10-year follow-up period than non-
psychopaths. In addition, when the psychopathic group reached early adulthood, the
number of nonviolent crimes committed by this group did not differ from non-
psychopaths, whereas the difference between psychopaths and nonpsychopaths on
number of violent offenses increased as the participants aged. In a study of female

incarcerated adolescents, Bauer (2001) found that those classified as psychopathic by the

36

PCL: YV had a greater number of Conduct Disorder symptoms, greater alcohol
dependence, earlier onset of criminal behavior, and more institutional violence. Within
an inpatient hospital setting, adolescents with higher psychopathy scores demonstrated
more overall aggression and higher rates of aggression toward both peers and staff than
those scoring lower in psychopathy (Stafford, 1998). In addition, those youth higher in
psychopathy displayed both reactive and instrumental aggression while adolescents lower
in psychopathy demonstrated only reactive aggression.

Notably, much of the research assessing psychopathy in adolescence and its
relation to violence has utilized the two-factor model of psychopathy, but employed only
total psychopathy scores as the measure of psychopathy. Few published studies have
examined the two factors separately. Work by Brandt et al. (1997), Edens et al. (1999),
and Rogers et al. (1997) are exceptions, with all these groups of investigators finding
small but significant correlations between each of the factors and different types of
institutional infractions, as well as other measures of aggression and conduct symptoms.
The investigation by Edens et al. revealed that both factor 1 and 2 of the PCL: YV were
significantly correlated with combined physically aggressive and verbally
aggressive/defiant institutional infractions (r = .30 and r = .28), although only factor 2
scores related significantly to physically aggressive acts alone (r = .24). Conversely,
Rogers et al. found that both factor 1 and 2 related at the same significance level to
institutional infractions involving physically aggressive behavior (r = .30 for both). Like
the Rogers et al. investigation, Brandt et al. found that both factor 1 and 2 of the PCL:
YV related significantly to major verbal and physical aggressive infractions while

committed, as well as the number of prior commitments, the number of Conduct Disorder

37

symptoms, CBCL externalizing and aggression scales, and the number of Intensive
Supervision Program placements. Only factor 1 scores were significantly associated with
a history of crime severity, whereas only factor 2 scores were significantly correlated
with total number of major infractions committed and the ratio of negative to positive
reviews.

In an investigation of different types of juvenile offenders, namely sexual
offenders, other kinds of violent offenders, and non-contact offenders, researchers,
employing the APSD, found that the groups differed significantly on the scale assessing
Callous-Unemotional traits (analogous to affective deficits of PCL-R factor 1).
Specifically, 35% of sex offenders had elevated scores on the CU scale with only 6% and
7% of the violent offender and non-contact offender groups, respectively, showing
similar elevations. The groups did not differ on the lmpulsivity/Conduct Problems scale
(similar to PCL-R factor 2), with approximately 25% of each group demonstrating
elevations on the I/CP scale (Caputo et al., 1999).

In addition to these published reports, McBride (1998) reported in her dissertation
study that both PCL: YV factors had similar associations with outcome measures (i.e.,
number of violent/nonviolent crimes, criminal versatility). Similarly, Gretton’s (1999)
retrospective study revealed that both PCL:YV factors were similarly and significantly
associated with number of Conduct Disorder symptoms and number of violent offenses
committed per year (over a IO-year period). Yet, some studies do reveal differential
correlates for the two factors. A study of gang membership found that incarcerated
adolescents who reported gang affiliation exhibited higher levels of PCL factor 1 than

factor 2 scores (King, 1997). Ridenour (1996) found that PCL-R:YV factor 1 scores

38

were more significantly related to the number of violent offenses committed after release
from a treatment facility than were factor 2 scores (R2 = .23 and .19 respectively, though
both p < .01). Conversely, higher factor 2 scores were more associated with the number
of nonviolent offenses than were factor 1 scores (R2 = .36 and .29, respectively, both p <
.01). Cruise (2001) demonstrated that only PCL: YV factor 2 scores significantly
predicted total institutional infractions and new detentions for those juveniles released
from the detention facility. Conversely, in this study, factor 1 scores exhibited little
relationship to any external outcome measure (all assessed within three months of the
assessment of psychopathy).

Overall, these studies have shown that both the personality-based factor I and
social deviance factor 2 of the PCL-R exhibit similar levels of associations with
important outcomes like physically and verbally aggressive institutional infractions,
number of prior arrests, and measures of aggression and CD. Factor 1 scores appear to be
more related to crime severity and variety, and possibly more related to sexual types of
offenses. Factor 2 scores seem to relate more strongly to total number of institutional
infractions (both aggressive and non-aggressive).

Although several studies have examined the differential association of each of the
factors with violent and aggressive outcomes, only one published study to date (Brandt et
al., 1997) has assessed the incremental validity of psychopathy over other factors to
predict violence in adolescents. Yet, this type of analysis is necessary to determine
whether psychopathy and/or its dimensions are important over and above other violence-
related factors. When Brandt et a1. considered the incremental prediction of each of the

psychopathic factors, both added significantly to the prediction of the total number of

39

violent post—release offenses. After demographic, criminal history, and psychometric
(i.e., # of Conduct Disorder symptoms, MMPI and CBCL scales, IQ, and training facility
behavioral indicators) variables were entered into the hierarchical regression equation,
factor 2 scores accounted for an additional five percent of the variance while factor 1
scores accounted for three percent more of the variance over and above factor 2 scores.

Though Brandt et a1. (1997) assessed the predictive ability of other variables
before considering psychopathy, they did not consider several other factors found to be
related to violence in adolescent populations. Lipsey and Derzon (1998) used meta-
analytic procedures to determine the best predictors of violence for children age 6-11 and
adolescents age 12-14. For the latter age group, lack of social ties (social anomie) (.39),
involvement with antisocial peers (.37), and commission of general offenses (.26) were
the strongest predictors of violence. In addition, Hawkins et al. (2000) reviewed studies
assessing the longitudinal (retrospective or prospective) prediction of physical
interpersonal violence in juveniles. These authors listed delinquent peers and gang
membership, poor family management practices (no clear parental expectations of
children’s behavior, poor supervision, and severe and inconsistent discipline), and
physical abuse or neglect history as significant predictors of later violence. Early
aggressive behavior and involvement in other kinds of antisocial behavior were violence
predictors for males, while poor academic achievement was more strongly related to
violence for females than for males.

These findings of correlates to adolescent violence mesh with Moffitt’s (1994)
theory of life-course persistent antisocial behavior. As noted earlier, this theory posits

that underlying trait deficits continually interact with certain environmental conditions,

40

leading to continuity and pervasiveness of antisocial behavior. The type of
environmental conditions outlined by Moffitt include poor family attachment bonds and
inconsistent child-rearing practices, lack of school achievement, parent and sibling
deviance, and socioeconomic status, as well as other person variables. Additionally, the
theory of life-course persistent antisocial behavior hypothesizes that one issue that
maintains an antisocial lifestyle is a person’s dearth of knowledge of prosocial behavior.
The neuropsychological deficits lead to a difficult temperament, leading to both parent
and peer rejection. Because of such rejection, an individual is not able to establish a
repertoire of prosocial behavior, precluding him/her of creating bonds with prosocial
peers throughout development. This failure leads to more associations with other
antisocial peers. This theory and the empirical findings of important correlates to
violence and antisocial behavior make it important to consider other predictors of

violence in adolescents when investigating the predictive power of psychopathy.

41

CHAPTER 4

CONCLUSION

In sum, this study helps to elucidate the nature, structure and stability of
psychopathy and its relation to severe and violent offending during the adolescent years.
This study examines the fit of previously proposed models of psychopathy in an attempt
to find the best model for adolescents. In addition, it investigates the stability of
psychopathy and its dimensions over time during this stage of development. Finally, the
present investigation assesses the ability of psychopathy and its dimensions to predict
violent behavior after controlling for social anomie, association with delinquent peers,
and poor parental supervision that have been found to be related to violence in
adolescence.

Hypotheses

Hypotheses are as follows:

1. A non-deviance model of psychopathy will be superior to a model that includes criteria
assessing social deviance.

2. A theoretical model conceptualizing psychopathy as a multidimensional construct
defined by three factors - affective, narcissistic, and behavioral - will be superior to a
model that includes only two factors with the affective and narcissistic traits combined
onto the same factor.

3. The affective dimension will demonstrate greater structural stability and relative
stability over time than either the narcissistic or behavioral dimensions.

4a. The affective and narcissistic dimensions will replicate across gender better than the

42

structure of the behavioral factor.
4b. Males will display higher levels than women for each of the psychopathy dimensions.
5. Adolescent offenders with higher levels of psychopathy will exhibit higher levels of
aggression after a year’s duration than adolescents with lower levels of psychopathy,

independent of initial individual differences in family income, parental supervision, the

youth’s association with delinquent peers, and the youth’s lack of social ties.

43

CHAPTER 5

METHOD

Participants

The 149 youth who were part of this study were participating in a larger research
project looking at offender trajectory over a one-year period. These youth were between
the ages of 10 and 16 years of age (M = 14.7, SD = 1.35). Most (n =142) were recruited
over a two-year period following their appearance before a Hearing Officer in the Clinton
County, Michigan Juvenile Justice Court or were referred by a Court Probation Officer
during the same period (n = 7). Close to two-thirds of the sample was male. The
majority were Caucasian (see Table 3). The children’s grade level ranged from 4th to 11‘h
with a mean of 8.94 (SD = 1.4). A parent of each of the juveniles also participated. For a
majority of the youth, biological mothers completed the parent report (74.7%). Other
informants were the youths’ biological fathers (18.8%), grandmothers (3.2%), stepfathers
(1.9%), stepmothers (0.6%), or other relatives (0.6%).

At Time 2, 89 youth remained in the study (see Table 3). Most of the families
were lost at Time 2 because they could not be located after a year in spite of precautions
set up to avoid this source of attrition (i.e., participants gave a fi'iend’s number when they
signed up). A few families declined to participate in the second wave of data collection
and one youth died in the year between the first and second wave. At Time 2, the mean
age was 15.9 years (SD = 1.28) and ranged from 12 to 18. Similar to Time 1,
approximately two-thirds of the sample at Time 2 was male. The ethnic breakdown of

the youth remaining at Time 2 was also similar to Time I.

44

The juveniles came before the court for a variety of offenses ranging from assault
and battery or arson to minor in possession of tobacco or curfew violations. Although

Table 3

Sample demographics at Time 1 and Time 2.

 

 

Tim; Ling;
Demographic Percent n Percent n
Gender
Male 61.7 92 65.2 58
Female 38.3 57 34.8 31
Ethnicity
Caucasian 73.8 1 10 78.7 70
Mixed 14.8 22 10.1 9
Hispanic 4.7 7 5 .6 5
Native American 4.7 7 4.5 4
African American 1.3 2 0.0 0
Asian 0.7 I 1 .1 1

 

most (86.8%) were facing charges on only one offense, this number ranged from one to
four. Fifty-one percent of the sample reported no prior arrests and approximately 30%
reported one prior arrest. Approximately 8% of the sample reported two previous arrests
and 11% said they had been arrested three times before participating in the study.
Measures

Measures of psychopathy. The study used two measures of psychopathy. The
first was the Levenson Self-Report Psychopathy Scale (LSRP; Levenson, 1995). This 24-

item adolescent self-report measure is designed to assess the two dimensions of

45

psychopathy measured by the PCL-R in a non-criminal population (see Table 4). Items
are assessed on a 1 to 4 scale (strongly disagree to strongly agree). For this study,
approximately half of the items were reworded to more closely match the education and
reading level of the subjects. For example, the item “success is based on survival of the
fittest; I am not concerned about the losers” became “success depends on winning and
how strong a person is. I don’t worry about the losers.” Three items (“I let others worry
about higher values; my main concern is with the bottom line,” “I don’t plan anything
very far in advance,” “Love is overrated”) from the original LSRP were dropped because
they were judged to be age inappropriate. The scale was imbedded in a longer (139 item)
self-report instrument, the Youth on the Fringe Survey (YOF S, Frank & Schettini, 2001).

The YOFS is actually a compilation of items from several different already
published instruments for adolescents. All 149 youth in the study had LSRP data.
Because the LSRP did not include social deviance items, three items stemming from
other measures, sexual promiscuity, criminal versatility, and juvenile delinquency, were
added to the LSRP when it was used in a social deviance model. The sexual promiscuity
item came directly from the adolescent version of the Functional Impairment Scale
(FISCA) described below (i.e., “During the past 3 months, how often did you act sexually
promiscuous or loose, engage in high risk sexual behaviors, or have unprotected sex”).
Juvenile delinquency was also based on the mean score of the Delinquency scale from the
FISCA. The criminal versatility item was the sum of the number of crimes charged in the
current Hearing and the number of times arrested reported in the past, which also came
from the FISCA.

The second psychopathy measure employed in the study was the Antisocial

46

Table 4

Items fiom the Levenson Self Report of Psychopathy Scale (LSRP).

 

5‘8"!”

.V‘

8.
9.

10.
ll.
12.
13.
14.
15.
I6.
17.
18.
I9.
20.
21.
22.
23.

I am often bored

I’m able to go after one goal for a long time*

I feel bad if my words or actions make someone else feel bad”

Success depends on winning and how strong a person is. I don't worry about the
losers

People who are stupid enough to get ripped off usually deserve it

I tell other people what they want to hear so that they will do what I want them to do
(I butter them up)

I have been in a lot of shouting matches with other people

For me, what's right is whatever I can get away with

I like playing with other people's feelings

I quickly lose interest in the tasks I start.

I try not to hurt other people when I go after what I want”

Making sure I look out for myself is the most important thing to me

I often look up to people who can rip people off (cheat others) and get away with it
Even if I were trying to sell something, I wouldn't lie about it*

In today's world, I feel okay about doing anything I can get away with to succeed
I would be upset if I succeeded because of someone else's loss“

Cheating is not fair because it is unfair to others*

Making a lot of money is my most important goal

Most of my problems are due to the fact that other people just don't understand me
Before I do anything, I think about all the things that can go right or wrong“

When I get frustrated, I often "let off steam" by blowing up

I say no when someone wants me to do things I know are wrong or dangerous*

My main purpose in life is getting as many goodies as I can

*indicates reversed-scored item

 

47

Process Screening Device (APSD; F rick & Hare, in press). The APSD is a 20-item
parent-report scale derived from Hare’s PC L-R to assess the parents’ view of
psychopathic traits in their children or adolescents (see Table 5). The measure has been
used with youth age 6 to 18. Items are scored on a three-point scale from 1 (the item
does not apply at all to your child) to 3 (applies very much to your child). The APSD
was introduced into the larger study of which this investigation is a part after the study
was underway. Therefore, data for the APSD is not available for all of the 149
participants, but for 79 participants at Time 1. The APSD includes a single social
deviance item (item 20).

Measures of additional variables. An Aggressive Behavior Scale was deveIOped
for this study using items from the Functional Impairment Scale for Children and
Adolescents (FISCA; Frank, Paul, Marks & Van Egeren, 2000; Frank, Van Egeren,

F ortier, & Chase, 2000). The FISCA is available in a parent and youth self-report version
(both were used in this study), identical in content, and each consisting of 184 items used
to identify mild, moderate, or severe impairment in eight domains of child and adolescent
functioning. The present study sampled items from the Delinquency and Control of
Aggression domains for each respondent, yielding identical parent and youth versions of
the Aggressive Behavior Scale. Each item was scored on a 3-point Likert scale (0 to 2),
indicating frequency of occurrence (never, occasionally, often).

An exploratory factor analysis (EF A) performed on 38 items from the
Delinquency and Aggression domains used data at time 1 to test the prediction that items
which assess violent behavior and aggression would hang together, forming an aggressive

behavior scale. The EFA produced two scales, one of which was made up of all 21 items

48

Table 5

Items from the Antisocial Process Screening Device (APSD).

 

l.

2.

10.

ll.

12.

13.

I4.

15.

16.

17.

18.

I9.

20.

Blames others for mistakes

Engages in illegal activities*

ls concerned about how well does at school/work (r)
Acts without thinking of the consequences

Emotions seem shallow not genuine

Lies easily and skillfully

Good at keeping promises (r)

Brags excessively about abilities, accomplishments
Gets bored easily

Uses or cons others to get what want

Teases or makes fun of others

Feels bad or guilty when he does something wrong (r)
Engages in risky or dangerous activities

Can be charming at times - insincere or superficial
Becomes angry when corrected or punished

Seems to think he or she is more important than others
Does not plan ahead

Is concerned about the feelings of others (r)

Does not show feelings or emotions

Keeps the same friends (r)

*social deviance item

 

49

from the Control of Aggression scale plus five items from the Delinquency scale, all
involving aggression towards property or person (see Table 6). Internal consistency,
measured by alpha coefficients, was .91 for the parent report version of the Aggressive
Behavior Scale and .85 for the youth-report version. Reliability using time 2 data was
comparable (or = .91 for the parent scale and or = .81 for the youth scale). The second
factor consisting of 12 items and assessing non-aggressive behavior was not used in this
study.

The three additional measures that were used in the predictive analyses, the social
anomie, association with delinquent peers, and lack of parental supervision scales, were
each tested with a CPA procedure to assess that all items loaded onto the construct as
predicted; all items loaded as predicted on the models, at or above .10 (standardized
estimate). The @ociation with delinquent peers fig]; included eight negative peer items
from the YOFS. These items assessed how often in the past year the youth’s friends had
participated in various types of delinquent behavior (i.e., take things that do not belong to
them, bang up or destroy things of some value on purpose, fight with someone with a
weapon, smoke cigarettes, speak foul language, have sexual intercourse, read
pornography). The items were measured on a five-point scale (never to very often). The
alpha coefficients at time 1 for the association with delinquent peers scale was .89 and
.88 at time 2.

The social anomie scale, measuring the juveniles’ lack of social ties, was made up
of six items from the YOF S: “I have trouble getting along with adults,” “I feel like I am
part of this school” (reversed), “I feel close to people at this school “(r), “I can’t seem to

keep people as friends for very long,” “I am not really accepted and liked,” and “I often

50

Table 6

Items from the Aggressive Behavior Scale.

 

1. Argue or fight with other kids
2. Bully, threaten, or shove other kids
3. Tease, ridicule, or pick on other kids

4. Lie, con, manipulate or take
advantage of others

5. “Blow up” or get annoyed at other
kids over little things

6. Act too young or immature around
kids the same age, or prefer to play with
younger kids

7. Have trouble getting along with other
kids the same age, or find it hard to
make or keep friends

8. Have trouble getting along with adults

9. Annoy others on purpose, or damage
their belongs on purpose

10. Say mean or cruel things, or verbally
abuse others

11. Act very cruel to animals
12. Do things impulsively (without
thinking) that were dangerous and could

injure others

13. Have temper tantrums or outbursts of
anger

14. Destroy or damage property in the
home when he or she was angry

* Items from the Delinquency domain

 

51

15. Get very upset if he or she could not
do or have something right away

16. Act inappropriately seductive, or do
unusual or inappropriate things of a
sexual nature in public

17. Act sexually promiscuous or loose,
engage in high risk sexual behaviors, or
have unprotected sex

18. Participate in gang activities that
included harassing or bullying others

19. Bite or throw things at others

20. Physically attack or really try to hurt
another child, a parent, or some other
adult living in the child’s/your home

21. Physically attack or really try to hurt
some other child or adult not living in
the child’s/your home

22. Vandalize or deface property*

23. Act so out of control that someone
filed a complaint"

24. Go joy riding in a car without
permission*

25. Play with fire so that damage to
property or people was likely“

26. Threaten someone with a weapon“

 

go to church or some place of worship” (r). These items were measured on a four-point
Likert scale (disagree a lot to agree a lot). The alpha coefficients at time I and time 2
were .61 and .59, respectively.

The four YOF S items comprising the lack of parental supervision scale measured
the amount of adult supervision youth received during the work week and weekend, as
well as parents’ knowledge of the youths’ whereabouts and activities. Items were
measured on a 4-point scale, yielding alpha coefficients of .67 at time 1 and .77 at time 2.
A single item from Social History Questionnaire completed by the parent informant
provided information on famin income. Income was measured on an eight-point scale,
each point of the scale representing a successive category of income (from $8,000 —
11,999 to over $100,000).

Procedure

Time 1. After a determination was made by the Hearing Officer as to the
dispensation of the case (i.e., warning and dismissal, consent probation, formal probation,
or referral for a formal hearing), the Hearing Officer introduced the adolescent and
his/her parent to an undergraduate research assistant involved with the study. The
adolescent and parent were informed that their decision to participate in the study did not
affect any outcomes involved with the court system. In addition, research assistants told
potential participants that they would each be reimbursed $20.00 for their initial
participation, that their information would be kept confidential, and that they would be
contacted later for additional testing sessions. If they agreed to participate, subjects were
given the option to complete the measures immediately at the courthouse or to set up a

testing appointment for a later date either at the research office located in the court or at

52

their home. If potential participants were unsure of their upcoming availability, they
were asked to provide a phone number at which one of the research staff could reach
them at a later time. All participants choosing to be tested at a later time were called and
reminded the night before their appointment. At the time of testing, before measures
were completed, research assistants obtained written consent from the parent and written
assent from the adolescent. Completion of all measures required between an hour and an
hour and a half for the majority of participants.

Time 2. Approximately one year after completing the Time 1 assessment, families
were recontacted and asked to participate in the follow-up phase of the study by
responding to the same measures completed at Time 1. Testing took place at the court
offices, at the participants’ home, or at another agreed upon location (e.g., local library,
university office). At this time, second written consent and assent forms were obtained,
and both the adolescents and parents were again each reimbursed $20.00 for their follow-
up participation.

Missing Data

All missing data were replaced in the data sets so that means and intercepts would
not have to be estimated in confirmatory factor analysis and structural equation modeling
procedures, allowing for the calculation of goodness of fit and other statistics. Only ten
of the 149 youth at time 1 (6.7%) had missing data for one of the variables employed in
the CFA analyses. Even fewer participants were missing data in each of the other data
sets (i.e., parents at both time points and adolescents at time 2). The series mean

procedure of SPSS was used to replace missing values.

53

Research predictions

Predictions are as follows:
I. Confirmatory factor analysis procedures will show that non-deviance models of
psychopathy will provide a better fit to the observed data than models that includes
criteria assessing social deviance.
2. Confirmatory factor analysis procedures will demonstrate that a three-factor model of
psychopathy provides a better fit to the observed data than a two-factor model
3. The affective dimension of psychopathy will demonstrate greater structural and
relative stability over time (i.e., the factor structures will replicate better and test-retest
correlations will be higher) than either the narcissistic or behavioral dimensions.
4a. Confirmatory factor analytic procedures will demonstrate that the affective and
narcissistic dimensions of psychopathy will replicate across gender better than the
behavioral dimension.
4b. Males will display higher mean levels of psychopathy than women for each of the
psychopathy dimensions.
5. Structural equation modeling will show that individual differences in levels of
adolescent psychopathy at time I can predict changes in adolescent aggression over a one
year period after controlling for initial difference in adolescent aggression, family
income, lack of parental supervision, association with delinquent peers, and lack of social

ties.

54

CHAPTER 6

RESULTS

Attrition

As noted previously, at time I, 149 participants had data for the LSRP and 79 had
information for the APSD. At time 1, the total sample of 149 participants (all of whom
had LSRP data) was comparable in age, grade and ethnic distribution to the subsample of
79 participants who also had APSD data. However, the proportion of males in the APSD
subgroup was somewhat smaller than in sample as a whole (46 of 79 or 58.2% versus 92
of 149 or 65.2%).

At time 2, 89 subjects had data for LSRP and only 32 had data for the APSD. Of
the 89 with LSRP, 58 were male whereas 31 were female. Participants with LSRP data at
time 2 did not differ greatly from the entire sample in percent of each gender (males =
64.4%), mean age (M = 14.8, SD = 1.2), grade (M = 9.0, SD = 1.3), or proportion of
Caucasian versus non-Caucasian youth. The 32 youth with APSD data also were
demographically comparable to the sample at time 1 (22 were male, M age = 14. 4, SD =
1.3, M grade = 8.7, SD = 1.2, percent Caucasian = 75%).

As noted earlier, fewer participants had APSD data because this measure was
added to the larger study several months after it began. Because of the small sample size,
several predictions are tested using only the adolescent report LSRP data.

Goodness of fit of the structural models
A major goal of this study was to compare the fit of several different models of

psychopathy in adolescents. Several two and three factor models suggested in the adult

55

and child literature were initially examined (see Table 2 and Appendices A-D).
Alternative two-factor models included one with social deviance criteria and one without
social deviance criteria; these models were tested using both the LSRP and the APSD.
Comparisons of the two- and three-factor models were also tested using both the LSRP
and APSD data sets.

Model comparisons were made within but not between data sets allowing the use
of statistics for nested models. Models compared using these nested designs shared all
the same free parameters except for certain parameters that were constrained to zero in
one but not in the other model (Maruyama, 1998). In the present study, parameters were
removed from the models instead of constrained; From (2003) has postulated that
removing variables is the same as constraining them to zero (R.P. Deshon, personal
communication, July 2003). To compare the nested models, the chi-square difference
test was utilized and other appropriate fit indices also were compared.

In regards to fit indices, the literature suggests that the comparative fit index (CF I)
and root mean square residual (RMR) are more appropriate for this study than others
made available by the AMOS statistical package. Maruyama (1998) does not
recommend the normed fit index (N F1) in spite of its popularity as a relative fit index
because it does not do well with small sample sizes and is affected by changes in sample
size. Relative Type 2 indices, such as the Tucker-Lewis Index (TLI), are much more
consistent across sample size than absolute or relative Type I indices. Hu and Bentler
(I 998) also recommend the use of the RMR in addition to other indices such as the TLI,
the CF 1, and/or the root mean square error of approximation (RMSEA) to evaluate model

fit. However, because the TLI and RMSEA are more affected by small sample sizes than

56

the RMR or CF I (Hu and Bentler, 1998), the present study focused on the RMR and CF].
Hu and Bentler (1999) suggest that a cutoff value close to 0.08 for the RMR and a value
close to 0.95 for the CF] are needed to conclude good model fit whereas Maruyama
suggests values over 0.9 for the CF] are sufficient to indicate good fit. When the fit of
these models to the data was examined, the following results were found:

Prediction 1 proposes that two-factor models of psychopathy that do not include
social deviance criteria will demonstrate better fit to the data. These alternative two
factor models were tested employing confirmatory factor analysis (CFA) procedures; all
of the CF As used data collected at Time 1, so that Time 2 data could later be utilized to
test the stability of the best model over time (see Prediction 3). If any items loaded
poorly (< .10), they were removed and the models rerun.

The first two-factor model tested was that derived from Hare using the youth-
report LSRP. Although the LSRP is modeled after Hare’s PCL, it does not include social
deviance items. Therefore, social deviance items from the youth version of the FISCA
that most mirrored those from the PCL: YV (i.e., juvenile delinquency, sexual
promiscuity, criminal versatility) were included with the LSRP when assessing the
deviance model. For both the deviance and non-deviance models employing the LSRP,
item 20 — “before I do anything, I think about all the things that can go right or wrong” —
was removed because the factor loadings were under .10 (deviance model -— .08; non-
deviance model — .07). When the models were rerun without item 20, goodness of fit
statistics were essentially the same for the non-deviance and deviance models (see Table
7). Hence, Prediction 1 was not supported by the LSRP data.

In contrast, the second two-factor model tested was that outlined by Frick et al.

57

(1994) and based on findings using the parent-report APSD. CF A analysis provided
some tentative support for the prediction that the non-deviance APSD model would
demonstrate a better fit to the data than the model including the deviance item (i.e.
engages in illegal activity). That is results of the )8 difference test indicated that the

Table 7

Fit indices for the deviance and non-deviance models of the total sample.

 

Model x2 CFI RMR xzdifl“

 

1. LSRP 2-factor with
social deviance criteria 423.2 .763 .087
(N = 149)

2. LSRP 2-factor non-deviance 340.8 .778 .081
(N = 149)

Difference between Models 1 & 2 82.4
3. APSD 2-factor with
social deviance criteria 179.0 .736 .041

(N= 78)

4. APSD 2-factor non-deviance 146.0 .776 .040
(N = 78)

Difference between Models 3 & 4 33.03

3'p < .05

 

models differed significantly (p = .003) even though the fit indices for the two models
were comparable (see Table 7).

Prediction 2: The three-factor models will exhibit a somewhat better fit than the
two-factor models. The three-factor model outlined by Cooke and Michie (2001) and that

delineated by Frick et al. (2000) employing the APSD differed slightly in item loadings,

58

so in a preliminary analysis four items were allowed to cross-load (see Appendix D).
This analysis demonstrated that loadings were more consistent with the Frick et al.
model. The loadings of three of the items (1, 3r, and 5) supported Frick et al.’s model;
only one reversed item — “Is good at keeping promises” — loaded according to the Cooke
and Michie model. In subsequent analyses, these items were loaded onto the factors on
which they loaded more strongly in this preliminary analysis. See Table 8 for factor
loadings of cross-loaded items.

Table 8

Item loadings of cross-loaded items for the APSD three-factor model.

 

 

Item Cooke and Michie Frick et al.
factor loading factor loading

1. Blames other for his/her mistakes Affective (.133) Behavioral (.539)
3. Is concerned about how well s/he

does at school/work” Behavioral (- .050) Affective (.468)
5. His/her emotions seem shallow

and not genuine Affective (.031) Narcissistic (.559)
7. Is good at keeping promisesa Behavioral (.503) Affective (.230)

areversed scored items

 

An initial test of the three-factor model employing the LSRP yielded modification
indices suggesting that the reversed item 22 — “I say no when someone wants me to do
things I know are wrong or dangerous” — would load better on the affective factor than
the behavioral one. In addition, as in the comparison of the deviance and non-deviance

two-factor models above, the LSRP item “before I do anything, I think about all the

59

things that can go right or wrong” was omitted because it failed to contribute
substantially to any factor (factor loading of .07).

The final LSRP three-factor model without these two items was compared to the
LSRP two-factor non-deviance model. The )8 difference test between these two models
indicated that the goodness of fit of the two models differed significantly (x2 = 38.4, df=
2, p = .00). Also, consistent with Prediction 2, the fit indices indicated that the 3-factor
model evidenced a better fit to the data (see Table 9). Similarly, the 3-factor model using
the APSD data was compared to the 2-factor non-deviance model using the APSD. In
Table 9

Fit indices of the two— and three-factor models using both the LSRP and the APSD.

 

 

Model N x2 CFI RMR
1. 2-factor non-deviance using LSRP 149 340.8 .778 .081
2. 3-factor model using LSRP 149 302.4 .839 .074

Difference between models 1 and 2 = 38.4a

3. 2-factor non-deviance using APSD 78 146.0 .776 .040
4. 3-factor model using APSD 78 153.3 .875 .035
Difference between models 3 and 4 = 7.3

ap = .00

 

general, this analysis provided very weak support for Prediction 2, that is the )8 difference
test indicated that the two models did not differ significantly and the RMR values were
comparable. However, the CPI was higher for the 3-factor model.

Prediction 3 proposes that the aflective dimension of psychopathy will

demonstrate greater structural and relative stability over time than either the narcissistic

60

or behavioral dimensions. Whereas scores on the factor assessing the affective features
of psychopathy are expected to be relatively stable over a one-year period, scores on
factors describing narcissistic and the behavioral aspects of psychopathy are expected to
show greater fluctuation. Determining temporal stability is a complex task and, ideally,
data should be examined at four time points to accurately separate error from stability in
scores. Yet, certain procedures are available to estimate stability using two time points.
First, using only the LSRP three-factor model, structural stability was evaluated by
assessing model fit across the one year period between the first and second assessments.

Structural stability

It was not expected that the total three factor model would demonstrate structural
stability because it is hypothesized that not all the factors of psychopathy are equally
stable over time in adolescents. In addition, the limited sample size at time 2 made an
adequate test of the stability of the total model unfeasible. Instead, the structural stability
of each of the three dimensions was assessed separately over both time points using CF A
procedures, with the expectation that the affective factor would show greater structural
consistency across time points than either the narcissistic or behavioral factor.

This prediction was supported by the CFA procedures. As predicted, results
showed that the LSRP affective factor structure was stable across the two time points.
The x2 was insignificant (p = .4), indicating that the data from both time points run
simultaneously did not differ significantly from the model; the CF I fell above the gold
standard value of 0.95, the RMR fell below 0.08, and the RMSEA fell below 0.06. As
expected, the fit indices for the LSRP behavioral factor did not evidence as adequate

stability across the two time points. The x2 was significant (p = .044) and CF I fell below

61

0.9 (0.897). Likewise, consistent with expectations, the narcissistic factor showed a
similar poor fit across the two time points with the CFA yielding a significant x2 (p = .00)
and a CFI values below 0.9 (see Table 10).

Another method used to demonstrate structural stability in the factors over time
was to show that the amount of error in measurement was consistent across each time
point. Alpha coefficients for each dimension were calculated and compared at time 1 and
time 2. Contrary to what was predicted, coefficients for the LSRP scales were more
similar for the behavioral and narcissistic factors than for the affective factor (see Table
Table 10

Fit indices for the LSRP factors assessed simultaneously at Time 1 and Time 2.

 

 

 

Model x2 CFI RMR GFI RMSEA
LSRP Affective factor 18.83 .991 .061 .965 .016
LSRP Behavioral factor 29.41al .897 .074 .943 .061
LSRP Narcissistic factor 123.98 .866 .077 .873 .067
8p < .05

Table 11

Alpha coeflicients for the LSRP three psychopathy factors at T1 and T2.

 

 

Model Time 1 Time 2
Affective factor .58 .69
Behavioral factor .65 .64
Narcissistic factor .81 .81

 

62

11). Yet, the higher alpha coefficients simply mean that the items measure a more
unitary construct, not necessarily the same construct over time.

Absolute stability

Mean differences between time 1 and time 2 for all three factors were also
calculated. Much like the results of the alpha coefficients, the affective factor
demonstrated the largest mean difference between the two time points (see Table 12).
The narcissistic factor also evidenced a significant mean difference across the two time
points. The affective factor mean level increased from time 1 to time 2 whereas means
for each of the other factors decreased over time.

Relative stability

Test-retest correlations between time 1 and time 2 were calculated for each factor
to examine whether individual adolescents maintained their relative positions (high or
low) in level of psychopathy on each of the three psychopathy dimensions. More
specifically, it was predicted that the correlation for the affective dimension would be
higher than for either the narcissistic or behavioral factors. However, as can be seen in
Table 13, the data did not support this prediction.

In sum, support for the prediction that the affective factor would show greater
structural, absolute and relative stability was mixed. CFA procedures showed greater
structural stability over time for the affective as compared to the narcissistic or behavioral
factor, but the alpha coefficients for the affective factor proved to be more inconsistent
across time points than those for the other two factors. In addition, the affective factor

had the lowest test-retest correlation of the three psychopathy factors.

63

Table 12

Mean differences between T1 and T 2 for the LSRP three psychopathy factors (n=87).

 

 

 

Model T1 T2 M difference p
Affective factor 2.11 2.28 - .170 .035
Behavioral factor 2.45 2.37 .085 .180
Narcissistic factor 2.09 1.96 .125 .043
Table 13

Test-retest correlations between times 1 and 2 for LSRP total scores and factors (n = 87).

 

 

Model Correlation
LSRP Total .6228
LSRP Affective .251b
LSRP Behavioral .5653l
LSRP Narcissistic .5968

3p < .001

bp < .05

 

Prediction 4a proposes that the structure of the aflective and narcissistic
psychopathy factors will replicate better across gender than the behavioral factor. Only
the LSRP three-factor model was used to test this prediction due to the small sample size
for the APSD. LSRP data were available for 92 males and 57 females. The CFA did not
completely support the prediction. Both the behavioral and affective factors

demonstrated good fit, though the fit of the narcissistic factor was only moderate (see

64

Table 14). The statistics for the affective factor were generally good with only a lower
CFl value than the behavioral factor. The narcissistic factor displayed a significant 12,
but lower CPI and GF I values, and a higher RMSEA value than those for the other
factors.

Table 14

F it indices for the LSRP factors assessed simultaneously across gender.

 

 

Factor x2 CFI RMR GFI RMSEA
LSRP Affective factor 24.3 .883 .074 .952 .049
LSRP Narcissistic factor 120.821 .847 .078 .874 .070
LSRP Behavioral factor 20.8 .971 .073 .956 .033

8p = .00

 

Prediction 4b: Males will display higher mean levels of psychopathy than females
for each of the psychopathy dimensions. Mean levels of the three psychopathy factors
were calculated for each gender using both the youth-report LSRP and the parent report
APSD at time 1. For the LSRP, the prediction was only partially supported. On the
LSRP, males demonstrated higher mean levels for the affective and narcissistic factors,
but were lower than the females on the behavioral dimension. Only the narcissistic
dimension demonstrated a significant difference (see Table 15). For the APSD, support
for the prediction was in the expected direction. Boys displayed higher mean levels than
girls on all factors, yet none of the differences were significant (see Table 16).
Predicting changes in aggression from psychopathy

Prediction 5 proposes that adolescents ’ psychopathy scores at time 1 will predict

aggressive behavior one year later even after controlling for other variables associated

65

Table 15

Mean levels of psychopathy for both gender measured by the LSRP.

 

mow-:92) MOP”)
Affective factor 2.14 (.57) 2.03 (.60)
Narcissistic factora 2.16 (.60) 1.91 (.55)
Behavioral factor 2.35 (.66) 2.54 (.62)

almean difference significant at p = .013

 

Table 16

Mean levels of psychopathy for both gender measured by the APSD.

 

Male 01:46) W 01:33)
Affective factor 0.83 (.37) 0.71 (.41)
Narcissistic factor 0.75 (.45) 0.64 (.39)
Behavioral factor 1.12 (.40) 1.01 (.46)

 

with violence, namely, household income, lack of social ties, association with delinquent
peers, poor and parental supervision. Structural equation modeling provided a test of
this hypothesis using youth self-report of psychopathy on the LSRP. Measures of social
ties, involvement with delinquent peers, and parental supervision used items from the
YOF S and hence, also depended on youth report. In the initial SEM analysis, aggressive
behavior at time 1 and time 2 were measured using the Aggressive Behavior scale from

the self-report FISCA. In addition, the total psychopathy score was used.

66

To assess the predictive power of psychopathy over and above these other
violence-related variables, structural equation modeling procedures were used and a
predictive model was tested. In the model, psychopathy, lack of social ties, association
with delinquent peers, and income were measured at time 1. Aggressive Behavior at
Time I was also controlled for. Aggressive Behavior measured at Time 2 served as the
outcome variable. Correlations for all variables are shown in Table 17.

Table 17

Intercorrelations among the variables in the SEM procedure.

 

 

Subscale 1 2 3 4 5 6 7
(n = 89)

1. T1 LSRP psychopathy -- .509a .459a .166 .482a .4362' .207
2. T1 youth-report aggression -- .526a .341a .347a .487a .139
3. T2 youth-report aggression -- .132 .290a .379a .246b
4. Income -- .158 .238b .021
5. Social anomie -- .373a .253b
6. Assn. with delinquent peers -- .028

7. Lack of supervision

2'p _<_ .01
bp 5 .05

 

Notably, all of the time 1 variables except for income (i.e., social anomie,
association with delinquent peers, lack of supervision, in addition to psychopathy) were

significantly related to aggression at time 2. Additionally, psychopathy was significantly

67

associated to lack of social ties, association with delinquent peers, and lack of
supervision.

When the predictive structural equation model was run (see Table 18 and Figure
1), the fit indices indicated a good fit to the data (see Table 19). The x2 was not
significant ()8 = 10.2, p = .071), indicating that the data did not differ significantly from
the model. According to the standardized estimates, psychopathy was still significantly
associated to both lack of social ties (b = .47, p = .01) and association with delinquent
peers (b = .41, p = .01), and aggression at time 1 (b = .33, p = .01). Psychopathy’s
association to lack of supervision bordered on significance (b = .21, p = .07). In line with
Moffitt’s theory, psychopathy significantly predicted aggression indirectly through these
other variables (p = .01). Psychopathy’s direct relationship to aggressive behavior at time
2 bordered on significance (b = .20, p = .07). None of the other variables in the model
except for aggression at time 1 (p = .01) were significantly associated with aggression at
time 2 (lack of supervision; lack of social ties; association with delinquent peers). All of
the significant associations between these other variables and aggression dropped out
when psychopathy was included in the model. Association with delinquent peers lost its
significant relation to aggressive behavior (b = .12), as did lack of social ties (b = -.01)
and lack of parental supervision (b = .16).

Because the relationship of many of the variables to time 2 aggression was far
from significant, a trimmed model was run. In this trimmed model, only those variables
with pathways to time 2 aggression significant at the .10 level or better were allowed to
predict time 2 aggression, that is only psychopathy and aggression at time 1 served as

predictors for time 2 aggression in the trimmed model. Because the original model

68

Table 18

Standardized estimates for the prediction model of aggression with youth report of
psychopathy.

 

 

 

 

 

Predictor social ties lack of sup. delinq. peers T1 agg. T2 agg.
(n = 89)

T1 LSRP psychopathy .47a .21b .41a .33a .20b

Income .08 .01 .17 .22 --

T1 lack of social ties -- -- -- .04 -.01

T1 lack of supervision -- -- -- .05 .16

T1 assn. w/ delinq. peers -- -- -- .28a .12

T1 youth-report agg -- -- -- -- .34a

ap s .01

bp = .0

Table 19

Fit indices for the predictive model of aggression.

Model n x2 CFI RMR

Youth- report predictive model 89 10.22' .958 .044

Trimmed model 89 14.2b .949 .044

8p = .071

bp = .077

 

proved to be a good fit to the data, all other paths — significant or not — were allowed to
remain in the model (see Table 20 and Figure 2). Fit indices for the trimmed model were

similar to those found for the original model (see Table 19). In the trimmed model,

69

psychopathy at time 1 significantly and directly predicted aggression at time 2 (b = .26, p
= .021). The significance value for all other remaining relationships in the model went
unchanged. Findings supported the prediction that psychopathy significantly predicted
aggression over time both directly and indirectly through other variables previously
found to be associated to aggression and violence.

Table 20

Standardized estimates for the trimmed prediction model of aggression.

 

 

Predictor social ties lack of sup. delinq. peers Tl agg. T2 agg.
(n = 89)

T1 LSRP psychopathy .47a .21b .41a .33a 26"
Income .08 -.01 .17 .22 --

T1 lack of social ties -- -- -— .04 ~-

Tl lack of supervision -- -- -- .05 —-

T1 assn. w/ delinq. peers -- -- -- .28a --

T1 youth-report agg. -- -- -- -- .39a

 

As a follow-up analysis, each psychopathy factor replaced total psychopathy in
the original model to examine each dimension’s ability to predict time 2 aggression. All
three models demonstrated fit indices indicating that they were good fits to the data (see
Table 21). Like the model with the total psychopathy scale, all three factors

demonstrated a significant indirect relationship to time 2 aggression. Also, the significant

70

associations between the other violence-related variables and time 2 aggression became
non-significant when the factors were included in the models; the only exception to this
was supervision, which still evidenced a significant (p = .05) relationship to time 2
aggression in the affective factor model. Yet, the models differed in some ways in terms
of the standardized estimates. Only the narcissistic factor demonstrated a significant
direct relationship to time 2 aggression. Additionally, neither narcissism nor the affective
characteristics were significantly associated to time 1 aggression, though the path from
Table 21

Fit indices for each psychopathy factor in the predictive model of aggression.

 

 

Model N x2 CFI RMR
Narcissistic 89 12.9a .926 .041
Behavioral 89 15.7b .909 .055
Affective 89 126° .922 .042
ap = .024

bp = .008

°p = .028

 

narcissism to time 1 aggression showed a statistical trend (p = .06). See Figures 3
through 5 for the SEM models for the three factors.

The relationship between total psychOpathy and time 2 aggression was also tested
using parent-reported psychopathy (APSD). Because of the smaller number of
participants with APSD data, hierarchical regression was used and the only variable
controlled for was aggression at time 1. The results indicated that time 1 and time 2

aggression were significantly related (b = .60, p = .001), but, in contrast to the youth

71

model, psychopathy did not explain significant additional variance in aggression at time 2

after controlling for time 1 aggression (AR2 = .867, p = .359).

72

CHAPTER 7

DISCUSSION

The first goal of the present study was to examine the structure of psychopathy in
adolescence. The results indicated that the best fitting structural model for the adolescent
data was a three-factor model when psychopathy was defined by youth report. This
finding corresponds to previous results in both the child (Frick et al., 2000) and adult
(Cooke & Michie, 2001) literature. In the previous work with children and adults, the
measures used were either Hare’s PCL-R (adults) or a measure based on the PCL-R,
revised to be appropriate for use with children. The youth measure employed in the
present investigation was also based upon Hare’s PCL-R, but was modified for use with
noncriminal populations. The PCL-R and its derivative measures originally had two-
factor solutions (see Frick et al., 1994; Hare et al., 1990; Lynam et al., 1999), but
subsequently demonstrated an equal or better fit to a three-factor model solution when
compared to a two-factor one. The fact that a three-factor solution has been found to be
present in samples of children (age 6-13), adults, and now adolescents suggests that this
conceptualization of psychopathy might be one that starts early and persists through
development, possibly only manifesting in different ways. Of course, the only
methodology that can truly answer that question is a longitudinal design that follows the
same group of individuals from childhood into adulthood.

The second goal of the current study was to demonstrate the superior stability of
the affective dimension of psychopathy compared to the behavioral and narcissistic

dimensions. When using the youth-report of psychopathy, evidence of greater stability of

73

the affective factor was mixed. When examining structural stability, referred to as
statistical factorial invariance over time, the affective factor showed greatest stability
over a one—year period. Yet, when alpha coefficients provided the definition of stability,
stability was found for the narcissistic and behavioral factors, but not for the affective
factor. An examination of the alpha coefficients indicated that there was more error in
measurement at time 1 than at time 2 for the affective factor. An examination of the
mean levels of each factor at both time points revealed that the affective factor varied
most whereas the behavioral dimension was the only one without a significant mean
difference between times 1 and 2. The reason that these indicators of structural stability
did not mesh is unclear. Relative stability, as measured by test-rest correlations for all
three factors, was similar to the alpha coefficients. Test-retest correlation coefficients for
the narcissistic and behavioral factors each exceeded r = .5 (p < .001) whereas the
stability coefficient for the affective factor, though statistically significant (p < .05), was
relatively low (r = .251).

For the affective factor, the thinking behind the original stability hypothesis was
that the narcissistic and behavioral aspects of psychopathy are confounded by processes
generally associated with normal adolescent development. As adolescents develop, these
types of characteristics should fluctuate in quality and quantity and then dissipate as
youths get older. Conversely, affective characteristics such as lack of empathy and
remorse that are not as commonly associated to adolescence, and found to be present in
children (see Frick et al., 1994 and 2000), would remain more stable.

One implication of the finding that the narcissistic and behavioral traits might

remain stable over time is that the narcissistic and behavioral traits of psychopathy differ

74

from adolescent-limited characteristics. For example, the egocentricity generally
associated with adolescence, which is thought to vary and diminish as youth age, might
diverge from the narcissism that remains stable in psychopaths. Egocentricity consists of
a self-focus and difficulty separating me from not-me (Elkind, 1967), but might not also
entail the more pathological facets, such as incapacity for love of other or guilt, that is
included in the narcissism of psychopathy.

As mentioned earlier, determining the stability of characteristics across only two
time points is not the ideal and is a limitation of the present investigation; preferably, four
time points should be utilized. One difficulty in using only two time points is that levels
of the target variable might fluctuate between the two end-point measurements.
Therefore, a characteristic will appear stable through the measurement of the two time
points when it is in fact instable. Without the additional measurements, potential
fluctuations are missed. These potential fluctuations might have been present in the
narcissistic and behavioral psychopathy factors, but went undetected in the relative
stability measurements due to the lack of measurements.

In addition, it is possible that the one-year gap between the measured time points
of the present study is an insufficient amount of time to pick up on major developmental
changes that would result in the recession of adolescent-limited narcissistic and acting-
out behavior. The mean age of the youth at time I was approximately 15 years and,
therefore, 16 years at time 2. At these ages, youth are generally in high school, moving
from ninth to tenth grade. Therefore, this time period misses a critical move from junior
high to high school, as well as a move into later high school years when the next

developmental stage for the youth (i.e., college, full-time employment) is on the horizon,

75

possibly spurring developmental growth.

But what about the absolute and relative stability of the affective dimension?
Although it was hypothesized that the affective deficits of psychopathy are not as
confounded by ‘normal’ adolescent development as those that are part of the narcissistic
and behavioral dimensions, this might not be the case. In retrospect, adolescents have
been characterized as callous and unempathic, with these traits theorized as necessary for
normal adolescent development to take place. For example, in Blos’ theory of adolescent
development, he contends that adolescents must undergo a second individuation process
during which there is a major restructuring of ego functions and drive organizations
(Muuss, 1988). During adolescence, there is a weakening of the self in comparison to
more primitive drives as parents are rejected as extensions to the youths’ ego, leading to
emotional turmoil and rage. Additionally, this process leads to increases in anxiety.
Defenses against such anxiety might involve regression to more primitive responses such
as sadism, leading to callousness. Blos described such regressions as usual and necessary
for an adolescent to establish autonomy and organize his/her own psychic structure.

This study also began with the hypothesis that males and females would
demonstrate similar model fit for the affective and narcissistic dimensions of
psychopathy, but not for the behavioral factor. In the few studies that have examined
gender differences in psychopathy, the differences have generally stemmed from the
behavioral and/or social deviance characteristics of psychopathy, whereas the affective
and interpersonal traits have been found to be similar for males and females (Salekin et
al., 1997, Salekin et al., 2000). In the present investigation, this prediction was not

supported. Contrary to past findings, males and females demonstrated the most similarity

76

on the behavioral factor, indicating that the genders did not differ on the characterological
behavior patterns of psychopathy.

If males and females do not necessarily differ on the behavioral personality
characteristics of psychopathy, it is possible that the behavioral expression, or
characteristic adaptation (Cooke & Michie, 2001), of psychopathy differs across gender.
This difference would occur because males high in psychopathy might tend toward more
overt aggressive and violent behavior than psychopathic females. For example,
Hamburger, Lilienfeld, and Hogben (1996) found that the relationship between
psychopathy and Antisocial Personality Disorder (ASPD) and Histrionic Personality
Disorder (HPD) was moderated by biological gender. Males high in psychopathy were
more likely to express behaviors consistent with ASPD (e.g., history of antisocial,
criminal behaviors) whereas females high in psychopathy were more likely to express
histrionic behaviors (i.e., self-dramatization, attention seeking, dishonesty, sexual
provocation).

The possibility that there might be a gender difference in the behavioral
expression of psychopathy was examined post hoc in a number of ways. First, to test the
follow-up prediction that males and females might not differ on the characterological
behavior patterns associated with psychopathy, but more so on the behavioral
manifestations of the disorder, males and females were compared again on the behavioral
factor, but with social deviance items (i.e., criminal versatility, juvenile delinquency, and
sexual promiscuity) included. When this model was run, the behavioral factor plus social
deviance did not evidence as good a fit across gender as it had without the social

deviance items. As compared to the behavioral factor without the social deviance items,

77

the x2 increased from 20.8 to 76.6 and became significant (p = .023). The CF I and GFI
values dropped to 0.822 and 0.898, respectively (from 0.971 and 0.956). The RMR and
RMSEA values increased to 0.122 and 0.053 (from 0.073 and 0.033).

Second, to examine specifically whether psychopathic males would present with
more aggressive behavior than psychopathic females, correlations between psychopathy
and both time 1 and time 2 aggression for each gender was tested post hoc. Additionally,
both youth and parent report of aggression was examined. Overall, the relationship
between psychopathy and aggression was greater for males. Specifically, in terms of
youth report, males and females demonstrated comparable significant relationships
between psychopathy and time I aggression. However, males displayed a significant
association between psychopathy and time 2 aggression whereas females did not.
Additionally, when a gender interaction variable was added to the SEM trimmed model
and used to predict time 2 youth-reported aggression, the estimate bordered on significant
(b = -.188, p = .068). The negative sign of the estimate indicates that males had a
stronger relationship between psychopathy and time 2 aggression than females. In terms
of parent report of psychopathy, males evidenced a significant relationship between
psychopathy and both time 1 and time 2 aggression. Yet, for females, there was virtually
no association between psychopathy and aggression at either time point (time 1: r = .098
and time 2: r = - .006).

In sum, although males do not appear to differ from females in the model fit of
the behavioral factor as hypothesized, they do appear to differ on this dimension when
social deviance criteria are added; perhaps it is not in the characterological behavior

patterns in which they differ, but in the behavioral manifestations (i.e., juvenile

78

delinquency, sexual promiscuity, criminal versatility) of psychopathy. Additionally,
males showed a stronger overall relationship between psychopathy and aggression than
did females.

The final objective of the present investigation was to examine the ability of
psychopathy to predict aggressive behavior after a year’s time, and over and above other
variables previously found to be related to aggression and violence. The total youth
sample was examined using structural equation modeling procedures to investigate the
model for the development of stable aggressive/violent behavior (Moffitt, 1994). In this
model, a temperamental/personality predisposition (e. g., psychopathy) lays the
groundwork for violence, but that predisposition more likely results in violent or
aggressive behavior when other variables, such as a lack of interaction with prosocial
peers (e.g., association with delinquent peers, lack of social ties) and lack of parental
involvement (e.g., lack of supervision), are present. Moffitt also posited that low income
plays a role in the development of aggression, mostly because it allows children and
adolescents fewer opportunities and, again, hinders prosocial development.

The SEM analyses provided support for Moffitt’s (1994) theory. Moffitt argues
for the strong impact of personality or temperament characteristics on the development of
violence/aggression and the current study demonstrated that psychopathy significantly
predicted later aggression both directly and indirectly. The indirect prediction occurred
via other variables (i.e., social anomie, lack of parental supervision, and association with
delinquent peers) that have previously been found to be highly associated with violence
and aggression (e.g., Hawkins et al., 2000; Lipsey & Derzon, 1998; Moffitt). These

variables lost their significant association to later aggression when psychopathy was

79

included in the model. This finding indicates that psychopathy is accounting for
aggression’s relationship to these variables and that previous studies finding strong
associations between such variables and aggression/violence might have missed
examining the personality characteristics driving the associations.

In addition to the indirect association to later violence, psychopathy also directly
accounted for variance in aggression over and above these other variables. Interestingly,
the follow-up analyses separately examining each of the psychopathy factors in the
predictive model found that this direct relationship between psychopathy and later
aggression was driven largely by the narcissistic characteristics. The narcissistic factor
contains items that relate to a disregard of others stemming from a focus on being better
than others and the importance of self-preservation. For example, items like “success
depends on winning and how strong a person is. I don’t worry about the losers” and
“people who are stupid enough to get ripped off usually deserve it” suggest that when a
person thinks they are stronger and smarter, the weaker and dumber are losers who
deserve whatever they get. Items like “making sure I look out for myself is the most
important thing to me” and “in today’s world, I feel okay about doing anything I can get
away with to succeed” point to the importance of self-preservation in the narcissistic
dimension and the lengths taken to maintain it. Lack of empathy and remorse that are the
keys to the affective factor might create a lack of concern in psychopathic youth about
what happens to others, but it might not be the driving force leading them to aggress
against others. The affective deficits cause psychopathic youth not to care when others
are hurt, but isn’t necessarily behind them causing the actual hurt. Instead, aggression

seems to stem from the superior and “taking care of number one” attitudes of the

80

narcissistic dimension.

An important finding in this study is that data from both the youth and parent
measure supported the three-factor model. However, an equally important caveat is the
presence of multiple potential confounds, namely in the areas of measurement, reporter
and sample size. Two different measures of psychopathy were used for youth and
parents. The parent report, the APSD, is a validated measure for use with parents of
youth; on the other hand, the youth measure was a revision of an adult measure for use
with adolescents and one that had not been used with this population before. Ideally, the
same measure would have been used for both reporters so it could have been compared
directly. There currently is an adolescent self-report version of the APSD, but this
version was not available at the time of data collection for the present study. Because the
measures of psychopathy cannot be compared directly, we do not know whether
differences in model fit occurred because of measurement differences, reporter
differences, or actual differences between the proposed models.

A related caution is that the youth measure of psychopathy is a measure that has
been validated for use with adults, but not for use with adolescents; items were revised
for use with youth in the current study. Youth- and parent-measured psychopathy
differed in their relation to aggression, suggesting that they might not be measuring the
same construct of psychopathy. Because the LSRP is not a validated measure of
psychopathy in this age group, it becomes suspect on these grounds. Yet, parent report is
also suspect due to the fact that parents are more than likely less aware of their children’s
internal thoughts and feelings. This lack of awareness is especially true during the

adolescent years as youth gain more autonomy and begin to break away from parents’

81

ideas. This growing autonomy might also make parents less aware of their children’s
behavior, leading to less knowledge of their level of aggressiveness. Again, it is difficult
to know whether differences are driven by problems with measures, reporter problems, or
actual differences.

Sample size is another limitation. Because for each reporter the sample size was
different, differences in results between the two measures/reporters could have stemmed
from differences in sample size. When determining model fit, it is best to have large
samples. As sample size decreases, the amount of confidence one may have in the model
fit statistics diminishes as well. The decay in participants between time 1 and time 2 led
to a decrease in sample size from the original N of 149. The most affected were the
analyses involving the parent report because the parent measure was added to the study
after the youth-report measure, leading to a lower initial sample size for analyses
involving parent report. Therefore, fit indices for models using parent-reported
psychopathy were potentially affected by the smaller sample size. The differences in
results between analyses involving youth and parent report might have stemmed from
these sample size issues.

Although there was a decrease in sample size from time 1 to time 2, there was no
revealing information available regarding the subjects who dropped out of the study
between time points, though an examination of demographic variables exhibited very
little change in mean age or grade, or frequencies of ethnicities. As noted earlier, the
attrition resulted because several participants moved without notice and it was impossible
to locate them. It can be speculated that these families are more disorganized or more

impoverished than those who remained in the study, forced to move frequently and

82

without friends or family who know of their new whereabouts (families were asked to
give a name of such a person when they sign up). It can only be postulated what
difference the presence of these more disorganized youth might have had on results.

Most likely, the presence of these youth would add to the numbers of youth lacking social
ties. It is also likely that these youth were lower income than those that remained in the
study, which might have influenced the effect of income in the predictive modeling.

In addition, there was no information collected at time 2 regarding any
intervention or treatment that youth might have received in the year between time 1 and
2. There was generally no formal treatment required by the juvenile court because most
youths’ cases were dispensed with only a fine. Still, it is reasonable to assume that some
parents might have pursued treatment for their children after involvement with the court
system. Yet, it was not possible to make any comparison between those who potentially
received treatment and the rest of the subjects.

Another issue to consider is that the present study utilized a court-referred sample.
Because this was not a community sample, it is difficult to generalize the present results
to the population as a whole. Additionally, the lack of gender differences in levels of
psychopathy found in the present study might result from the use of an offender sample.
Females in this type of sample might possess higher levels of psychopathic characteristics
than found in a community population. In past studies, the variability in gender
differences on mean levels of psychopathy related to whether the sample was drawn from
a clinical, community or offender population (e.g., Myers et al., 1995; Loper et al., 2001).
Clinical and community samples tended to demonstrated significant mean gender

differences whereas offender samples did not. Females in the current investigation might

83

also be more aggressive than their community counterparts. Whereas males generally
demonstrate higher levels of overt aggression than females, there was no significant
gender difference for either youth- or parent-reported aggression found in the present
study. In fact, although the difference was not significant, the mean level of youth-
reported aggression was higher for females than for males.

Another issue is the possible effect of self-report. A problem related to the use of
self-report measures for the assessment of psychopathy is the issue that lying and
dishonesty can be features of the disorder; leading to questions about the accuracy of the
self-report data. Additionally, psychopaths’ self-reports might be questioned on different
grounds. Psychopaths might be limited in their ability to introspect and understand their
own experience, leading to false reports (Lilienfeld, 1994). Alternatively, in some
investigations, psychopathy scores on self-report measures have been shown to be
somewhat negatively correlated with self-report measures of lying (Lilienfeld, 1994),
indicating that psychopaths might be willing to reveal negative traits. Levenson et al.
(1995) posited that self-report instruments might be feasible for use with psychopaths,
especially in noninstitutionalized samples, because of psychopaths “meta attitude” that it
is acceptable and possibly favorable to possess psychopathic characteristics. Other
studies have shown that the relationship to lying may be variably related to the different
dimensions associated with psychopathy. Zagon and Jackson (1 994) found that a
measure of the core personality traits of psychopathy correlated positively and
significantly with the MMPI Lie Scale and the Crowne-Marlowe Social Desirability
Scale. Conversely, a measure of behavioral traits related to psychopathy was

significantly negatively associated to scores on the Lie Scale.

84

Future research should attempt to address some of the limitations of the current
investigation. In terms of model fit, the present finding of the superior fit of a three-
factor solution for adolescents needs to be replicated. The mixed picture regarding the
stability of psychopathy indicates this is an area to be explored further; data collection at
a greater number of time points and a larger sample size would aid and improve such
exploration. In terms of the predictive power of psychopathy of important outcomes like
future aggression, future investigations should again attempt to follow the youth for a
longer time period. During adolescence, there is a peak of antisocial behavior. Youth get
into more behavioral trouble and flirt with illegal activities. Yet, most youth do not
remain on this path; they are what Moffitt (1994) termed adolescent-limited offenders.
According to Moffitt, adolescent-limited offenders are more likely to be involved in
behavior that attempts to display autonomy, adult privilege, or “status” (e.g., running
away, substance abuse, vandalism, public order offenses). Those on the life-persistent
course are more likely to participate in a variety of offenses, those outlined for the
adolescent-limited, as well as more victim-oriented offenses (e. g., violence, aggression,
fraud) and the types of crimes committed by lone offenders. As outlined earlier,
psychopathic characteristics are more likely related to life—course persistent behavior.

Moffitt pointed out that during the height of participation in delinquency of these
adolescent-limited youth, they will disguise the more persistent adolescents and create
‘noise’ in investigations of the more persistent group. Through the use of the Aggressive
Behavior scale as the outcome measure in the present study, an attempt was made to
‘weed out’ the effect of the adolescent-limited offender because of the inclusion of more

violent behavior. Yet, the best way to limit the ‘noise’ produced by the adolescent-limited

85

offender is to follow the youth for a longer time period.

In addition to delinquent behavior, there can also be an increase in psychopathic-
like characteristics during adolescence. Adolescents can be mean and spiteful, appearing
uncaring and unempathic. The present study hypothesized that the affective traits of
psychopathy would present differently in adolescence than the narcissistic or behavioral
ones, but it is possible that characteristics such as lack of empathy and remorse also
increase during adolescence as part of this adolescent-limited trend. Again, following
adolescents over a longer period of time would allow for a reduction in this ‘noise’
caused by these adolescent-limited youth.

Hare (1993) posits that investigations of psychopathy across the life course are
needed to determine what factors are involved when psychopaths become violent
offenders rather than con artists or even productive members of society. A greater
understanding of psychopathic traits in youth can lead to a measurement technology for
identifying a developmental subgroup of antisocial individuals typically marked by
patterns of severe aggression and violence (F rick, 2000). This technology would then
allow for intervention at an earlier stage of development when the traits or their

behavioral consequences may be more responsive to treatment.

86

APPENDICES

87

APPENDIX A

Two-factor model of Hare’s Psychopathy Checklist-Revised (PCL-R, Hare et al., 1990)

Factor 1 - personality items.

1. Glibness/Superficial Charm
Grandiose Sense of Self
Pathological Lying
Coming-Manipulative
Lacks Remorse or Guilt
Shallow Affect
Callousness/Lack of Empathy

@SQWPPN

Failure to Accept Responsibility

Factor 2 - behavioral/social deviance items.
9. Need for Stimulation/ Proneness to Boredom
10. Parasitic Lifestyle (not included on youth version)
1 1. Poor Behavioral Controls
12. Early Behavioral Problems
13. Lacks Realistic Goals
14. Impulsivity
15. Irresponsibility
*16. Juvenile Delinquency (scored differently on youth version)

*17. Revocation of Conditional Release
Items that do not load on either factor, but add to the total score
*18. Promiscuous Sexual Behavior
19. Many Short-term Marital Relationships (not included on youth version)

*20. Criminal Versatility (scored differently on youth version)

* indicates social deviance items

88

APPENDIX B

Two-factor model of APSD based on Frick et al. (1994)

Factor I - impulsivity/conduct problems.

1. Blames other for his/her mistakes

2. Engages in illegal activities‘

3. Acts without thinking of the consequences

4. Becomes angry when corrected or punished

5. Brags excessively about his/her abilities, accomplishments, or possessions
6. Seems to think that he or she is better or more important than other people
7. Teases or makes fun of other people

8. Engages in risky or dangerous activities

9. Keeps the same friends (r)

10. Gets bored easily

Factor 2 - callous-unemotional traits.

11. Feels bad or guilty when he/she does something wrong (r)

12. Is concerned about how well he/she does at school/work (r)

13. Does not show feelings or emotions

14. His/her emotions seem shallow and not genuine

15. Can be charming at times, but in ways that seem insincere or superficial

16. Is concerned about the feelings of others (r)

Items from the APSD not included in the model due to multiple loadings (a) or theoretical
reasons (b).

17. Lies easily and skillfullya
18. Is good at keeping promises (r)‘1
19. Uses or "cons" other people to get what he/she wantsa

20. Does not plan ahead, or leaves things to the “last minute”b

’ . . . . .
1nd1cates soc1al dev1ance rtem

89

APPENDIX C

Two-factor model of LSRP based on Lynam et al. (1999)

Factor 1 - personality traits.

l.
2.

I feel bad if my words or actions make someone else feel bad (r)

Success depends on winning and how strong a person is. I don't worry about the
losers.

People who are stupid enough to get ripped off usually deserve it.

I tell other people what they want to hear so that they will do what I want them to do
(I butter them up).

For me, what's right is whatever I can get away with.

I like playing with other people's feelings.

I try not to hurt other people when I go after what I want. (r)

Making sure I look out for myself is the most important thing to me.

I often look up to people who can rip people off (cheat others) and get away with it.

. Even if I were trying to sell something, I wouldn't lie about it. (r)

. In today's world, I feel okay about doing anything I can get away with to succeed.
. I would be upset if I succeeded because of someone else's loss. (r)

. Cheating is not fair because it is unfair to others. (r)

. Making a lot of money is my most important goal.

15.

My main purpose in life is getting as many goodies as I can.

Factor 2 - behavioral traits.

16.
l7.
l8.
I9.
20.
21.
22.
23.

I am often bored.

I'm able to go after one goal for a long time. (r)

I have been in a lot of shouting matches with other people.

I quickly lose interest in the tasks I start.

Most of my problems are due to the fact that other people just don't understand me.
Before I do anything, I think about all the things that can go right or wrong. (r)
When I get frustrated, I ofien "let off steam" by blowing up.

I say no when someone wants me to do things I know are wrong or dangerous. (r)

90

APPENDIX D

Three-factor model based on Cooke and Michie (2001) and F rick et al. (1994)

Items

Becomes angry when corrected or punished

Brags excessively about his/her abilities,
accomplishments, or possessions

Uses or "cons" other people to get what
he/she wants

Seems to think that he or she is better or more
important than other people

Teases or makes fun of other people

*His/her emotions seem shallow and not
genuine

Keeps the same friends (r)

Feels bad or guilty when he/she does
something wrong (r)

*Is concerned about how well he/she does at
school/work (r)

Does not show feelings or emotions
Is concerned about the feelings of others (r)

”‘13 good at keeping promises (r)

Gets bored easily

*Blames other for his/her mistakes

Acts without thinking of the consequences
Engages in risky or dangerous activities

Does not plan ahead, or leaves things to the
"last minute"

Cooke and Michie
NAR AFF BEH

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

Frick etal.
NAR AFF BEH
x

X

X

X

X

X

X

X
X
X
X
X
X

NAR = narcissism (Frick et al.)/Deceitful Interpersonal Style (Cooke and Michie)
AF F = Callous-Unemotional Traits/Deficient Affective Experience
BEH = Impulsivity/Impulsive and Irresponsible Behavioral Style

*indicates cross-loaded items

91

APPENDIX E

Figure 1.

Predictive model of the prediction of aggression with youth report of psychopathy

et2agg

 

t2 youth-report
aggression

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.12
.2
.16
esuper
t1 youth-report
aggression 28
.04
t1 lack of TI '31"?k _0f ndglszgggh
social ties supervusnon
-4 .08 .21 lo/17
t1 lsrp .
psychopathy Income
total score

 

 

 

 

 

 

92

APPENDIX F

Figure 2.

Trimmed model of the prediction of aggression with youth report of psychopathy

et2agg

 

t2 youth-report
aggression

.39 .26

 

 

 

 

 

 

 

 

 

 

   
    

 

 

 

 

 

 

 

 

 

 

7 esupervsr
04 t1 youth-report 28
' aggression '
.33 '22 .05
t1 lack of t1 lack of t1 assn with 3&3:ng
social ties supervision del peers
‘ .08
.21 T -41 .17
.47 -.01
t1 Isrp
psychopathy income
total score

 

 

 

 

 

 

93

APPENDIX G

Figure 3.

Model of the prediction of aggression with narcissistic factor of psychopathy

et2agg

 

 
  
   
    

t2 youth-report

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

aggression .09
L
t1 youth-report @
aggression .3
.19
.23 -03
t1 lack of t1 assn with et1assn
Stgéglllig; supervision del peers delpee
.13
'31 .12 .01 -35 .21
t1hlsrp h
psyc opat y .
narciss. factor Income
score

 

 

 

 

 

94

APPENDIX H

Figure 4.

Model of the prediction of aggression with affective factor of psychopathy

et2agg

 

    

.09 t2youth-report .19

 

 

 

 

 

 

 

 

 

   

 

 

 

 

 

 

 

 

 

 

aggression
-.16 .21
.39
.16 t1 youth-report
aggression
:09 I
-.03 .23
- et1assn
t1 lack of t1 lack of N dafsn With delpeers
social ties supervision e peers
.1
37 .27 -.01 .25 .21
t1 Isrp
psychopathy .
affective factor Income
score

 

 

 

 

95

APPENDIX I

Figure 5.

Model of the prediction of aggression with behavioral factor of psychopathy

et2agg

 

    

t2 youth-report 16

 

 

aggression
.38 .05 -'8
esupervsr
t1 youth-report
aggression
.31

 

 

 

 

 

  
    

  

 

 

 

 

 

 

 

 

 

45 19 08
- t1assn
t1 lack Of t1 assn Wlth g I
élc'g‘i'iig'; supervision del peers
.08
-41 .11 .oo -28 '19

 

 

t1 Isrp
psychopathy
behavior factor
score

income

 

 

 

 

 

 

96

REFERENCES

97

REFERENCES

Amett, J. (1992). Reckless behavior in adolescence: A developmental perspective.
Developmental Review, 12, 339-373.

Barry, C. T., Frick, P. J ., DeShazo, T. M., McCoy, M., Ellis, M. & Loney, B. R. (2000).
The importance of callous-unemotional traits for extending the concept of
psychopathy to children. Journal of Abnormal Psychology, 109(2), 335-340.

Brandt, J. R., Kennedy, W. A., Patrick, C. J ., & Curtin, J. J. (1997). Assessment of
psychopathy in a population of incarcerated adolescent offenders. Psychological
Assessment, 9, 909-923.

Brown, B. (1990). Peer groups. In S. F eldman and G. Elliott, (Eds). At the threshold:
The developing adolescent (pp. 171-196). Cambridge, MA: Harvard University
Press.

Caputo, A. A., F rick, P. J ., & Brodsky, S. L. (1999). Family violence and juvenile sex
offending: The potential role of psychopathic traits and negative attitudes toward
women. Criminal Justice and Behavior, 26, 33 8-3 56.

Christian, R. E., Frick, P. J ., I-Iill, 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.

Cleckley, H. (1976). The mask of sanity. Mosby: Saint Louis, MO.

Cooke, D. J. & Michie, C. (2001). Refining the construct of psychopathy: Towards a
hierarchical model. Psychological Assessment, 13(2), 171-188.

Cruise, KR. (2001). Measurement of adolescent psychopathy: Construct and predictive
validity in two samples of juvenile offenders. Dissertation Abstracts
International, Section B, 61(10-B), 5556.

Dowson, J. H., Sussams, P., Grounds, A. T., Taylor, J. C. (2001). Associations of past
conduct disorder with personality disorders in ‘non-psychotic’ psychiatric
patients. European Psychiatry, 16, 49-56.

Edens, J. F., Poythress, N. G., & Lilienfeld, S. O. (1999). Identifying inmates at risk for
disciplinary infractions: A comparison of two measures of psychopathy.
Behavioral Sciences and the Law, 1 7, 435-443.

Edens, J. F., Skeem, J. L., Cruise, K. R., & Cauffman, E. (2001). Assessment of “juvenile

98

psychopathy” and its association with violence: A critical review. Behavioral
Sciences and the Law, 19, 53-80.

Elkind, D. (1967). Egocentrism in adolescence. Child Development, 38, 1025-1034.

Forth, A. E., Hart, S. D., & Hare, R. D. (1990). Assessment of psychopathy in male
young offenders. Psychological Assessment, 2, 342-344.

Frank, S.J., Paul, J. 8., Marks, M., & Van Egeren, LA. (2000). Initial validation of the
Functional Impairment Scale for Children and Adolescents. Journal of the
American Academy of Child and Adolescent Psychiatry, 3 9(1 0), 1300-1308.

Frank, S.J., Van Egeren, L.A., F ortier, J .L., & Chase, P. (2000a). Structural, relative, and
absolute agreement between parents' and adolescent inpatients' reports of

adolescent functional impairment. Journal of Abnormal Child Psychology, 28,
395-402.

Fergusson, D. M. & Horwood, L. J. (2002). Male and female offending trajectories.
Development and Psychopathology, 14, 159-177.

Frick, P. J. (2000, October). Using the construct of psychopathy to understand antisocial
and violent youth. Paper presented at Festschrift, University of British Columbia,
Vancouver, CA.

Frick, P. J ., & Hare, R. D. (In press). The Psychopathy Screening Device. Multi-Health
Systems: Toronto.

Frick, P. J ., Bodin, S. D., & Barry, C. T. (2000). Psychopathic traits and conduct
problems in community and clinic-referred samples of children: Further

development of the Psychopathy Screening Device. Psychological Assessment,
12, 382-393.

Frick, P, J, O’Brien, B. S., Wootton, J. M., & McBumett, K. (1994). Psychopathy and
conduct problems in children. Journal of A bnormal Psychology, 103, 700-707.

Grann, M. (2000). The PCL-R and gender. European Journal of Psychological
Assessment, 16(3), 147-149.

Gretton, HM. (1999). Psychopathy and recidivism in adolescence: A ten-year
retrospective follow-up. Dissertation Abstracts International: Section B, 59, 6488.

Gretton, H. M., McBride, M., Hare, R. D., O’Shaughnessy, R., & Kumka, G. (2001).
Psychopathy and recidivism in adolescent sex offenders. Criminal Justice and
Behavior, 28(4), 427-449.

Hamburger, M. E., Lilienfeld, S. 0., & Hogben, M. (1996). Psychopathy, gender, and

99

gender roles: Implications for antisocial and histrionic personality disorders.
Journal of Personality Disorders, 10(1), 41-55.

Hare, R. D. (1993). Without conscience: The disturbing world of the psychopaths among
us. New York: Guilford Press.

Hare, R. D., Harpur T. J ., Hakstian, A. R., & Forth, A. E. (1990). The revised
Psychopathy Checklist: Reliability and factor structure. Psychological
Assessment, 2(3), 338-341.

Harpur, T. J ., Hakstian, A. R., & Hare, R. D. (1988). Factor structure of the Psychopathy
Checklist. Journal of Consulting and Clinical Psychology, 56, 741-747.

Harpur, T. J. & Hare, R. D. (1994). Assessment of psychopathy as a function of age.
Journal of A bnormal Psychology, 103(4), 604-609.

Harpur, T. J ., Hare, R. D., & Hakstian, A. R. (1989). Two-factor conceptualization of
psychopathy: Construct validity and assessment implications. Psychological
Assessment, 1(1), 6-17.

Harris, G. T., Rice, M. E. & Cormier, C. A. (1991). Psychopathy and violent recidivism.
Law and Human Behavior, 15(6), 625-637.

Hawkins, J. D, Herrenkohl, T. 1., Farrington, D., Devon Brewer, D., Catalano, R. F.,
Harachi, T. W., & Cothem, L. (2000). Predictors of Youth Violence. Juvenile
Justice Bulletin (OJJDP Publication No. 071 8-A-09). Washington, DC: US.
Dept. of Justice, Office of Justice Programs.

Hu, L. & Bentler, P. M. (1998). Fit indices in covariance structure modeling: Sensitivity
to underparameterized model misspecification. Psychological Methods, 3(4), 424-
453.

Hu, L. & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural Equation
Modeling, 6(1), 1-55.

Kruh, I. P., Frick, P. J ., & Clements, C. B. (2001). Historical and Personality Correlates
to the Violence Patterns of Juveniles Tried as Adults. Manuscript submitted for
publication.

Levenson, M. R., Kiehl, K. A. & Fitzpatrick, C. M. (1995). Assessing psychopathic
attributes in a non-institutionalized population. Journal of Personality and Social
Psychology, 68(1), 151-158.

Lilienfeld, S. O. (1994). Conceptual problems in the assessment of psychopathy. Clinical
Psychology Review, 14(1), 17-38.

100

Lipsey, M.W., and Derzon, J .H. 1998. Predictors of violent or serious delinquency in
adolescence and early adulthood. In Serious and Violent Juvenile Offenders: Risk
Factors and Successful Interventions, edited by R. Loeber and DP. F arrington.
Thousand Oaks, CA: Sage Publications, Inc., pp. 86-105.

Loeber, R. (1990). Development and risk factors of juvenile antisocial behavior and
delinquency. Clinical Psychology Review, 10(1), 1-41.

Loevinger, J. (1976). Ego development: Conceptions and theories. San Francisco :
Jossey-Bass Publishers.

Lynam, D. R. (1997). Pursuing the psychopath: Capturing the fledgling psychopath in a
nomological net. Journal of A bnormal Psychology, 106(3), 425-43 8.

Lynam, D. R., Whiteside, S., Jones, S. (1999). Self-reported psychopathy: A validation
study. Journal of Personality Assessment, 73(1), 110-132.

McBride, ML. (1998). Individual and familial risk factors for adolescent psychopathy.
Dissertation Abstracts International: Section B, 59 (12-B), 6507.

Moffitt, T. E. (1994). Adolescence-limited and life-course persistent antisocial behavior:
A developmental taxonomy. Psychology Review, 100, 674-701.

Moffitt, T. E., Caspi, A., Rutter, M., & Silva, P. A. (2001). Sex diflerences in antisocial
behavior: Conduct disorder, delinquency, and violence in the Dunedin
longitudinal study. Cambridge: Cambridge University Press.

Murdock Hicks, M., Rogers, R., & Cashel, M. (2000). Predictions of violent and total
infractions among institutionalized male juvenile offenders._Journal of the
American Academy of Psychiatry and the Law, 28, 183-190.

Murphy-Peaslee, M. (1992). An investigation of incarcerated females: Rorschach indices
and Psychopathy Checklist scores. Dissertation Abstracts International: Section

B, 59, 6488

Murrie, D. C. & Cornell, D. G. (2000). The Millon Adolescent Clinical Inventory and
psychopathy. Journal of Personality Assessment, 75(1), 110-125.

Muuss, R. E. (1988). Theories of Adolescence (5“ ed). Random House: New York.
Myers, W. C., Burket, R. C., & Harris, H. E. (1995). Adolescent psychopathy in relation
to delinquent behaviors, conduct disorder, and personality disorders. Journal of

Forensic Sciences, 40(3), 436-440.

Ogloff, J ., Wong, 8., & Greenwood, A. (1990). Treating criminal psychopaths in a

101

 

therapeutic community program. Behavioral Sciences and the Law, 8, 81-90.

Piotrowski, N. A., Tusel, D. J ., Sees, K. L., Banys, P., & Hall, S. M. (1995). Psychopathy
and antisocial personality in men and women with primary opioid dependence.
Issues in Criminological and Legal Psychology, 24, 123-126.

Rice, M. E., Harris, G. T., & Cormier, C. A. (1992). An evaluation of a maximum
security therapeutic community for psychopaths and other mentally disordered
offenders. Law and Human Behavior, 16, 399-412.

Ridenour, TA. (1996). Utility analyses of the Psychopathy Checklist, Revised and
Moffit’s taxonomy for a rehabilitation program for juvenile delinquents.
Dissertation Abstracts International: Section B, 5 7, 4086.

Rogers, R., Johansen, J ., Chang, J. J., & Salekin, R. T. (1997). Predictors of adolescent

psychopathy: Oppositional and conduct-disordered symptoms. Journal of the
American Academy of Psychiatry and the Law, 25, 261-271 .

Rutherford, M. J ., Alterman, A. I., Cacciola, J. S., & McKay, J. R. (1998). Gender
differences in the relationships of antisocial personality disorder criteria to
Psychopathy Checklist-Revised scores. Journal of Personality Disorders, 12(1),
69-76.

Salekin, R. T., Rogers, R., & Machin, D. (2001). Psychopathy in youth: Pursuing
diagnostic clarity. Journal of Youth and Adolescence, 30(2), 173-195.

Salekin, R. T., Rogers, R., & Sewell, K. W. (1996). A review and meta-analysis of the
Psychopathy Checklist and Psychopathy Checklist-Revised: Predictive validity of
dangerousness. Clinical Psychology: Science and Practice, 3, 203-215.

Salekin, R. T., Rogers, R. & Sewell, K. W. (1997). Construct validity of psychopathy in a
female offender sample. Journal of A bnormal Psychology, 106, 576-585.

Schmitt, W. A., & Newman, J. P. (1999). Are all psychopathic individuals low-anxious?
Journal of A bnormal Psychology, 108, 353-358.

Serin, R. C. (1996). Violent recidivism in criminal psychopaths. Law and Human
Behavior, 20(2), 207-217.

Silverthom, P., Frick, P. J ., & Reynolds, R. (2001). Timing of onset and correlates of
severe conduct problems in adjudicated girls and boys. Journal of

Psychopatholgy and Behavioral Assessment, 23(3), 171-181.

Spear, L. P. (2000). Neurobehavioral changes in adolescence. Current Directions in
Psychological Science, 9(4), 111-114.

102

 

Stafford, J .E. (1998). Psychopathy as a predictor of adolescents at risk for inpatient
violence. Dissertation Abstracts International: Section B, 58, 3934.

Toupin, J ., Mercier, 11., Dery, M., Cote, G., & Hodgins, S. (1995). Validity of the PCL-R
for adolescents. Issues in Criminological and Legal Psychology, 24, 143-145.

Vitelli, R. (1998). Childhood disruptive behavior disorders and adult psychopathy.
American Journal of Forensic Psychology, I 6(4), 29-3 7.

Wootton, J. M., Frick, P. J ., Shelton, K. K., & Silverthom, P. (1997). Ineffective
parenting and childhood conduct problems: The moderating role of callous-
unemotional traits. Journal of Consulting and Clinical Psychology, 65(2), 301-
308.

Zimmerman, M. (1984). Sensation seeking: A comparative approach to a human trait.
Behavioral and Brain Sciences, 7, 413-471.

103

 

IIIII

llislllllllzlljlllfllfllllflll11111111111