Screening for Psychopathy: Validation of the Psychopathic
Personality Inventory-Short Form with Reference Scores
Franca Tonnaer & Maaike Cima & Klaas Sijtsma &
Kasia Uzieblo & Scott O. Lilienfeld
Published online: 4 December 2012
# Springer Science+Business Media New York 2012
Abstract The current study reports validation results for the
for a newly developed PPI-Short Form (PPI-SF) in forensic
and non-forensic populations. We also provide criterion refer-
data from 1,065 participants and supplementary PCL-R data
from a subsample of 91 forensic offenders. Mokken scale
analysis was used to construct the PPI-SF. In Study 2, PPI-SF
yielded promising but preliminary support for the construct
validity of the PPI and the PPI-SF. The PPI-SF is of interest
for risk assessment because of its (a) strong relationship with
the PCL-R total score and (b) subscales known for their
predictive value for violence and criminal recidivism.
Psychopathy is a condition marked by a pattern of super-
ficial charm, callousness, dishonesty, egocentricity, and
unmotivated antisocial behavior (Hare 1991). The Psy-
chopathy Checklist-Revised ([PCL-R] Hare 2003) is the
most commonly used measure to assess psychopathy. Pro-
fessionals score the PCL-R based on an interview and
collateral file records. The instrument measures person-
ality traits and behaviors, including glibness, impulsivi-
ty, aggressive actions, and lack of empathy. The PCL-R
is a reliable and valid instrument for assessing psychop-
athy (Morrissey et al. 2005) but it has some pragmatic
limitations. First, the use of the PCL-R requires exten-
sive training and practice. Second, the PCL-R is time-
consuming, which is a potential limitation in many forensic
settings, in which the time allotted to obtain a diagnosis and
conduct a risk assessment is typically restricted. Third, collat-
eral files often are unavailable outside of prison or other
institutions, rendering PCL-R assessment impractical (Lilien-
feld and Andrews 1996).
In an effort to overcome these pragmatic limitations,
Lilienfeld and Andrews (1996) developed the Psychopathic
Personality Inventory (PPI) to assess the core features of
psychopathy in non-institutionalized settings. The PPI is a
self-report instrument developed and validated in under-
graduate samples, and its total score is moderately to highly
related to PCL-R total scores in prison samples (Poythress et
al. 1998). As the PPI was developed primarily as a research
tool, it lacks criterion reference tables for the interpretation
F. Tonnaer (*)
Department of Research, Forensic Psychiatric Centre De Rooyse
Wissel, P.O. Box 433, 5800 AK Venray, The Netherlands
Department of Clinical Psychological Science,
Maastricht University, Maastricht University, P.O. Box 433,
5800 AK Venray, The Netherlands
Department of Developmental, Clinical and Crosscultural
Psychology, Tilburg University, Tilburg, The Netherlands
Department of Research, Forensic Psychiatric Centre De Rooyse
Wissel, Venray, The Netherlands
Department of Methodology and Statistics, Tilburg University,
Tilburg, The Netherlands
Department of Applied Psychology, College University Lessius,
S. O. Lilienfeld
Department of Psychology and Interdisciplinary Sciences,
Emory University, Atlanta, USA
J Psychopathol Behav Assess (2013) 35:153–161
of individual scores. Later, Lilienfeld and Widows (2005)
developed a revised version of the PPI (the PPI-R) to elim-
inate psychometrically suboptimal items, lower its reading
level, and increase the measure’s cross-cultural applicability.
The goals of this study were to contribute to the valida-
tion of the PPI and to construct and validate a new PPI-Short
Form (PPI-SF) with reference scores. A short form of the
PPI has been developed previously (e.g., Lilienfeld and
Hess 2001; Vaughn, Howard and DeLisi 2008; Vaughn,
Newhill et al. 2008), but the potential disadvantage of this
short form is that it was developed by selecting subsets of 7
items each that loaded the most highly on each of 8 dimen-
sions identified in a factor analysis of the full PPI. This
strategy is defensible, but it may result in scales that are
overly narrow in content, because items with the highest
factor loadings will typically be those that display the high-
est inter-item correlations (stemming from high levels of
substantive overlap among items; Smith et al. 2000). An-
other disadvantage is that this approach may exclude items
that are rarely endorsed but highly discriminating at either
the low or high end of the psychopathy continuum. One
advantage of Mokken scale analysis (MSA), which was
used in the current study, is that it may identify items with
greater breadth better reflecting the ‘core’ of a construct, in
this case psychopathy. In this respect, this methodological
approach may produce a measure superior in some ways to
not only the PPI-SF but also the full PPI and PPI-R, as it
may target items with better discrimination power at both
the low and high ends of the psychopathy continuum.
The current study aims at developing a well-validated PPI-
discriminates among psychopathic, non-psychopathic, and
healthy comparison participants, in the first study we investi-
gated the psychometric qualities of the PPI. The second study
focused on the psychometric properties of the new PPI short
The first study had three aims. First, we examined the
reliability and the validity of the PPI. Second, we developed
criterion reference scores. Third, we designed the PPI-SF.
Over a period of three years, we collected PPI scores from
1,065 male participants (588 forensic patients and 477
healthy controls). The control sample included a student
sample (n0311) and a community sample (n0166). The
ethical commission of Maastricht University approved the
research protocol. Participants cooperated on a voluntary
basis. They received written and oral instructions emphasiz-
ing that participation was not related to prospects for parole
and that they were free to withdraw from the study at any
time. Participants ranged in age from 13 to 72 years (M0
31.3, SD012.2), with a mean age of 34.0 (SD011.6) for the
forensic patients, and 28.1 (SD012.1) for the healthy con-
trols. Forensic participants were older than the controls,
t(1053)08.07, p<.001, d0.50. Educational information
was available for most participants (72 %). Of this majority,
11 % attended elementary school and 38 % secondary school;
51 % had a bachelor-level education. For forensic individuals,
the corresponding percentages of these three levels of educa-
tion were 27 %, 70 %, and 3 % respectively, among healthy
controls, they were 0 %, 15 %, 84 %, respectively. These
findings indicated higher education levels for control partic-
ipants, t(762)034.5, p<.001, d02.5. For 57 % of the forensic
participants, offenses were known; 16 % were convicted for
(attempted) manslaughter or murder, 15 % for bodily harm,
14 % for property crimes, 13 % for property crimes with
violence, 11 % for drug and weapon related crimes, 7 % for
sexual offenses with minors, 6 % for sexual offences, 3 % for
deprivation of freedom, and 2 % for arson; 12 % were con-
167 male graduates and 165 prisoners were reported previ-
ously (Uzieblo et al. 2007; Verschuere et al. 2007). Supple-
mentary PCL-R data were available for a subsample (n091)
of forensic offenders, which were retrieved for diagnostic
purposes. Two trained forensic professionals scored and dis-
cussed all PCL-R interviews, resulting in a consensus score
based on discussion.
Psychopathic Personality Inventory ([PPI] Lilienfeld and
Andrews 1996) The PPI contains 187 self-report items, each
of which is answered on a 4-point Likert scale (10false and
40true). The PPI assesses a variety of psychopathic person-
ality characteristics that are grouped into eight factor-
analytically developed content subscales: (1) Machiavellian
Egocentricity (30 items) assesses ruthlessness and a willing-
ness to manipulate others, (2) Social Potency (24 items)
assesses charm and interpersonal dominance, (3) Coldheart-
edness (21 items) assesses callousness and an absence of
guilt, (4) Carefree Nonplanfulness (20 items) assesses a
failure to plan behavior and inhibit maladaptive impulses,
(5) Fearlessness (19 items) assesses a propensity for risk
taking behavior, (6) Blame Externalization (18 items)
assesses externalizing misbehavior, (7) Impulsive Nonconfor-
mity (17 items) assesses a lack of concern about social tradi-
tions, and (8) Stress Immunity (11 items) assesses the absence
of emotional reactions to potentially anxiety-provoking
154 J Psychopathol Behav Assess (2013) 35:153–161
events. Internal consistency of the PPI content scales is ade-
quate to high (alphas range from 0.70 to 0.93), and construct
validity is supported by moderate to high associations with
other measures of psychopathy and low associations with
desirability (Lilienfeld and Andrews 1996; Lilienfeld and
Psychopathy Checklist-Revised ([PCL-R] Hare 1991,
2003) The PCL-R is a 20-item measure of psychopathy
based on an interview and collateral file records. Each item
is evaluated on a 3-point scale (00does not apply and 20
does definitely apply). Hare (1991) proposed a two-factor
structure consisting of an Affective-Interpersonal factor and
a Behavioral Lifestyle factor. However, the most recent
version of the PCL-R manual presented a four-factor model
(Hare 2003). The first (Interpersonal) factor comprises
symptoms such as glibness and grandiose self-worth. The
second (Affective) factor comprises symptoms such as lack
of remorse or guilt, and irresponsibility. The third (Behav-
ioral Lifestyle) factor assesses features such as impulsivity
and lack of long-term goals. The fourth (Antisocial) factor
comprises poor behavioral control, juvenile delinquency,
and multifarious criminality. The internal consistency of
the PCL-R is adequate (alpha for the total score ranged from
.83 to .87, Hildebrand 2008). An extensive body of self-
report, interview, and psychophysiological research sup-
ports the PCL-R’s construct validity (Hare 2003).
The distributions of the PPI total and the subscale scores
were essentially normal1(see Table 1). Only 18 participants
had more than 3 missing item scores. Two-way missing-
value imputation (Van Ginkel and Van der Ark 2005) was
used to estimate the missing item scores. The internal con-
sistency of the PPI was estimated using coefficient alpha.
Group differences between forensic patients and healthy
controls were investigated using the independent-samples
t-test. Convergent validity was investigated using Pearson
product–moment correlations. Moreover, PPI criterion ref-
erence scores were collected in the total sample (N01065).
Mokken scale analysis (MSA) using the program MSP
(Molenaar and Sijtsma 2000) investigated the scalability for
individual items and entire scale, estimated item popularity,
and item discrimination.
MSA is a frequently used scaling procedure but may re-
quire some introduction (see Emons et al. 2012). MSA is
based on a nonparametric item response theory (IRT) model
(Sijtsma and Molenaar 2002). Compared with factor analysis
(FA) and principal components analysis (PCA), MSA pos-
sesses several advantages. One advantage is that MSA is
well suited for discrete item scores, such as rating scale
scores, whereas FA and PCA assume item scores to be
continuous. MSA uses the scalability coefficient H for
dimensionality analysis and item selection. The researcher
chooses a minimum value c for H and an automatic item
selection procedure selects only those items in clusters that
produce a scale for which H≥c. We investigated the item
dimensionality by selecting different values for c starting
with c0.30 and raising c by .05 in each next selection
round until c0.50. Ultimately, studying the pattern of clus-
ter outcomes across the different c values indicates the true
dimensionality of items (Sijtsma and Molenaar 2002, chap.
6). Items H<.3 are unscalable (Sijtsma and Molenaar 2002,
p. 60), .3≤H<.4 indicates a weak scale, .4≤H<.5 a medi-
um scale, and H≥.5 a strong scale (Mokken, 1971, p. 185).
We used mean item scores as item locations, indicating the
popularity of the specific behavioral aspect with respect to
the personality scale. Item coefficient Hjindicates the ex-
tent to which item j is related to the other items in the
scale, with higher Hjvalues indicating better discriminating
power and greater contributions to a reliable person
Reliability of the PPI
In the total sample, total PPI score coefficient alpha was .92.
For the PPI subscales, alphas ranged from .70 (Stress Im-
munity) to .87 (Carefree Nonplanfulness). For the total PPI
score, coefficient alpha was .92 among forensic participants
and .91 among controls. For the forensic participants, alpha
for the PPI subscales ranged from .67 (Stress Immunity) to
.88 (Carefree Nonplanfulness). For the controls, alpha for
the PPI subscales ranged from .74 (Stress Immunity) to .86
(Carefree Nonplanfulness and Blame Externalization).
Validity of the PPI
Group Differences Independent-samples t-tests and effect
sizes (using Cohen’s d) were computed to examine whether
forensic participants and controls differed on the PPI sub-
scales. Table 1 shows that the forensic group had a signifi-
cantly higher mean total score than the controls. The forensic
group scored significantly higher on the PPI subscales of
Fearlessness, Blame Externalization and Stress Immunity.
Unexpectedly, controls scored significantly higher on Impul-
Machiavellian Egocentricity, Social Potency, Coldhearted-
ness, and Carefree Nonplanfulness.
1Because of acceptable skewness and kurtosis values, no extreme
scorers were excluded.
J Psychopathol Behav Assess (2013) 35:153–161 155
Convergent Validity In the forensic subsample, we found
significant correlations between the PPI total score and
PCL-R total score, r0.40, .21<r<.56, the PCL-R Lifestyle
factor, r0.48, .30<r<.63, and the PCL-R Antisocial factor,
PPI Criterion reference
Some previous researchers have used a mean or median total
PPI score as a cut-off for categorizing individuals as psy-
chopathic versus non-psychopathic (Vaughn, Howard and
Delisi 2008). We introduce the mean score of the healthy
participants as a criterion reference score for psychopathy.
The criterion reference score allows researches to compare
individual test scores with a standard criterion reference (to
generate an individual z-score indicating an individual’s
position with respect to the group) instead of a random study
Constructing the PPI-SF
Scalability We first conducted the dimensionality analysis
second random half to evaluate the influence of chance
capitalization. Both analyses yielded the same results; there-
fore only the results obtained in the first data set are
discussed. Dimensionality analysis on the 30 items for
Machiavellian Egocentricity suggested a medium 16-item
scale (H0.44). Dimensionality analysis for Social Potency
yielded anacceptable, weak 17-item scale (H0.35).For Cold-
heartedness we found a medium 9-item solution (H0.42).
Dimensionality analysis for Carefree Nonplanfulness yielded
a medium 13-item scale (H0.44). For Fearlessness a medium
12-item scale (H0.43) was accepted. For Blame Externaliza-
tion a strong 10-item scale (H0.50) was accepted. For Impul-
sive Nonconformity we found a weak 10-item scale (H0.33).
For Stress Immunity a medium 7-item scale (H0.49) was
accepted. Contrary to the results of Benning et al. (2003),
dimensionality analysis of the PPI two-factor structure did
the point of departure for further analyses pursued here.
Item Location Items exhibiting a mean score of approxi-
mately 2.5 reflect an average of the psychopathic trait re-
ferred to in the item text. For Machiavellian Egocentricity,
an average item was “I’m good at flattering important people
when it’s useful to do so.” (Item #66). An average item for
Social Potency was “If I really wanted to, I could convince
most people of just about anything.” (Item #49). An average
Coldheartedness item was “Seeing a poor or homeless person
walking the streets at night would really break my heart.”
(Item #69). An average Carefree Nonplanfulness item was “I
usually strive to be the best at whatever I do.” (Item #24). An
average item for Fearlessness was “When my life becomes
boringI liketotakesomechances tomakethingsinteresting.”
(Item #45). An average item for Blame Externalization was
“I’ve been the victim of a lot of bad luck in my life.” (Item
#44). Within the Impulsive Nonconformity scale, an average
item was “I would enjoy hitch-hiking my way across the
United States with no prearranged plans.” (Item #28). An
average item for Stress Immunity was “I am easily ‘rattled’
at critical moments.” (Item #63).
Table 1 Group Differences for PPI Total Scale and Subscales Between Forensic and Control Groups; Skewness and Kurtosis, Independent-
Samples t-Tests, and Effect Size (d)
M SDM SD
2Results can be obtained on request from the first author.
156J Psychopathol Behav Assess (2013) 35:153–161
Item Discrimination For Impulsive Nonconformity, item
scalability was weak (Hjin the range .3-.4); for Machi-
avellian Egocentricity, Social Potency, and Stress Immu-
nity, weak to medium (Hj: .3-.5); for Fearlessness, weak to
strong(Hj:.3-.56);and for Coldheartedness, Carefree Non-
planfulness, and Blame Externalization, medium to strong
Convergent Validity PPI-SF total score correlated significant
with PCL-R total score, r0.42 [.24<r<.58], the PCL-R Life-
style factor, r0.48 [.30<r<.63], and the PCL-R Antisocial
factor, r0.40 [ .21<r<.56].
PPI-SF Criterion Reference Table 2 shows criterion refer-
ence scores for the PPI-SF (n060). We used the mean score
of the healthy comparison participants as a criterion refer-
ence score for psychopathy, making it possible to produce
an individual z-score indicating the individual’s position
with respect to the group.
The aim of this first study was threefold: first, to examine
the reliability and validity of the PPI; second, to develop
criterion reference scores; and third, to design a novel PPI
Short Form (PPI-SF) using more sophisticated psychometric
methods than were used to develop the PPI or PPI-R.
Results were consistent with previous research on the orig-
inal PPI (Ross et al. 2009) in showing satisfactory reliability
and broadly satisfactory validity for the PPI. The results
revealed a significant difference between forensic and healthy
participants, with higher total score for the former group.
Unexpectedly, however, we did not find a significant dif-
ference between the forensic and healthy controls for PPI
Machiavellian Egocentricity, Social Potency, Carefree Non-
planfulness, or Coldheartedness. Originally, the PPI was
developed for measuring psychopathic personality charac-
teristics within healthy control populations and does not
explicitly assess criminal behavior. Given this purpose,
one might expect healthy controls to score high on certain
subscales, especially those measuring somewhat adaptive
functioning (e.g., Social Potency). However, PPI scores of
healthy controls in the current study did not deviate from
those in other healthy control samples (Carlson and Thái
2010; Fecteau et al. 2008). Most important, PPI total scores
did not show significant differences between the healthy
controls from the current study and the comparison healthy
control samples used in other studies (Carlson and Thái
2010; Fecteau et al. 2008). We therefore recommend inter-
preting individual subscale scores only in light of the PPI
total score and the criterion reference score.
Moreover, the current study replicated earlier findings
(Poythress et al. 1998) concerning a moderate to high cor-
relation between PPI and PCL-R total scores. Further, PPI
data were used to construct a less time-consuming screening
instrument (PPI-SF) and corresponding criterion reference
scores. MSA results for item scalability, item popularity, and
item discrimination showed a differentiation in scales with
weak (Impulsive Nonconformity and Social Potency) and
strong scalability (Coldheartedness, Blame Externalization,
and Carefree Nonplanfulness). In addition to the PPI results,
we found significant relationships between PPI-SF and
PCL-R total scores and between the Behavioral Lifestyle
Table 2 Group Differences for PPI-SF Total Scale and Subscales Between Forensic and Control Groups; Skewness and Kurtosis, Independent-
Samples t-Tests, and Effect Size (d)
Forensic psychopathic Forensic Non-psychopathicControls Skewness Kurtosistd
M (SD)M (SD)M (SD)
J Psychopathol Behav Assess (2013) 35:153–161157
and the Antisocial factors of the PCL-R, which are
known for their predictive value for violence, antisocial
conduct and criminal recidivism (Leistico et al. 2008;
Walters et al. 2008).
This study suffered from three limitations. First, the
healthy comparison participants showed a more pronounced
tendency to answer in a more socially desirable direction
than the forensic participants (see Table 1). We cannot
readily explain this result because we had only limited
access to background information in the former group;
nevertheless, the results may simply reflect higher psycho-
logical health among individuals in this group. Second,
PCL-R scores were available only for a subset of the full
sample. Because the PCL-R still is the most commonly used
measure to assess psychopathy, the PPI-SF validation study
(Study 2) included a PCL-R assessment for all participants
as a standard procedure. Third, during the data collection for
the current study, Lilienfeld and Widows (2005) published a
shortened (154 item) revised version of the PPI, the PPI-
Revised (PPI-R). The PPI-R excluded items that were prob-
lematic psychometrically as well as items that were difficult
to read or culturally specific. In contrast, in the present
study, a sophisticated psychometric analysis was used to
exclude items from the PPI-SF, resulting in the shortest
questionnaire justifiable. Given the need for a less time
consuming screening device, the PPI-SF may be the best
alternative. Moreover, as of this writing the original PPI
continues to be used extensively in research and clinical
settings. Therefore, we based the analytic approach on the
original PPI. Clearly, in further research the current ap-
proach should be extended to the PPI-R as well.
The second study had two major goals. First, we exam-
ined the reliability and the validity of the newly developed
PPI-SF. Second, we developed criterion reference scores for
PPI-SF and PCL-R data were collected from 60 male
participants (15 forensic psychopathic, 25 forensic non-
psychopathic, and 20 healthy controls). The Ethical Commit-
tee of Maastricht University and the Ministry of Justice ap-
proved the research protocol. A psychologist recruited
forensic participants at Penitentiary Institution ‘De Geerhorst’
Sittard. Healthy comparison participants were recruited at
Maastricht University. All participants cooperated voluntarily
and received written and oral instructions. Forensic partici-
pants were informed that participation was not related to
treatment or criteria for parole. All participants were free to
withdraw from the study at any time. Participants were in
majority Caucasian (87 % of forensic psychopathic, 92 % of
forensic non-psychopathic, and 100 % of healthy controls)
and ranged in age from 19 to 59 years (M032.4, SD010.3).
Forensic participants had a mean age of 32.4 years (SD0
10.3), and healthy controls had a mean age of 24.7 years,
SD03.4, t(53)04.3, p0.00, d01.23. Twenty-five percent of
the forensic participants were convicted for (attempted) man-
slaughter or murder, 30 % for bodily harm, 23 % for property
crime with violence, 13 % for property crime, 8 % for sexual
offences, and 2 % were convicted for other crimes. PCL-R
data were collected from all participants. Supplementary in-
stitutional files and collateral information were available for
the forensic participants.
Psychopathic Personality Inventory-Short Form ([PPI-SF]
Lilienfeld and Andrews 1996) The 100-item PPI-SF from
Study 1 (see above) was used.
Psychopathy Checklist-Revised ([PCL-R] Hare 1991,
2003) The PCL-R (see Study 1) was used. All PCL-R inter-
views were videotaped and scored by two supervised, trained
masters’students. PCL-R assessments in a subsample (n015)
of the current forensic sample were double-scored with a
consensus rating arrived at by discussion. Inter-rater reliabil-
ity, expressed by the intraclass-correlation coefficient (ICC),
equaled .98 for the total PCL-R score (Landis & Koch, 1977).
ICC for the PCL-R factors ranged from .90 (Interpersonal
factor) to .99 (Antisocial factor).
Reactive-Proactive Aggression Questionnaire ([RPQ] Raine
et al. 2006) The RPQ consists of 23 items rated on a 3-point
scale (00never and 20often). The RPQ contains two sub-
scales: a Proactive subscale containing 12 items assessing
predatory aggression and a Reactive subscale consisting of
11 items assessing impulsive aggression. Earlier studies
showed good reliability, good convergent and discriminant
validity for RPQ scores with coefficient alpha for the Reac-
tive subscale ranging from .81 to .86, and for the Proactive
subscale scores ranging from .84 to .86 (Raine et al. 2006).
The Aggression Questionnaire ([AQ] Buss and Perry
3Age-corrected results were similar to the reported results (with one
exception: no significant group difference was found in the age-
corrected PPI-SF analysis in Carefree Nonplanfullness). Both results
can be obtained on request from the first author.
158 J Psychopathol Behav Assess (2013) 35:153–161
29 items that are answered on a 5-point scale (00definitely
disagree and 40definitely agree). The AQ has a four-factor
structure comprising physical aggression, verbal aggression,
hostility, and anger. The test-retest reliability of the AQ ranges
between .72 and .80 across various intervals (Hornsveld,
Muris, Kraaimaat, & Meesters, 2009). The AQ also displays
strong convergent validity, as demonstrated by high correla-
The Behavioral Inhibition System and Behavioral Activation
System scales ([BIS-BAS] Carver and White 1994) The
BIS-BAS is a self-report measure consisting of 20 items,
rated on a 4-point scale (10strongly agree and 40strongly
disagree). The BIS-BAS contains four subscales: the BIS
scale, assessing sensitivity to punishment cues; the BAS-
Reward Responsiveness scale, assessingsensitivity to rewards
and reward cues; the BAS-Fun Seeking scale, assessing
motivation to achieve goals; and the BAS-Drive scale,
assessing the willingness to rewarding stimuli without re-
flection. The BIS-BAS scales show moderate to good inter-
nal consistency reliability, with alphas ranging from .59
to.79 (Fayers and Machin 2007).
The distribution of the total PPI-SF scores was approximately
normal (Table 2). Internal consistency of the PPI-SF was
estimated by means of coefficient alpha. One-way multivari-
ate analysis of variance (MANOVA) was used to evaluate
group differences among forensic psychopathic, forensic
non-psychopathic, and healthy participants (based on PCL-R
assessment with a cut-off score of 264). Convergent validity
was investigated using correlations between PPI-SF and ag-
gression measures, as various studies demonstrated increased
levels of aggression and insensitivity to punishment cues in
psychopathy (Uzieblo et al. 2007), as well as with the BIS
scale of the BIS-BAS, which measures sensitivity to punish-
mentcues. The validityofthe PPI-SF was further investigated
using logistic regression of the target variable PCL-R cut-off
score4on the PPI-SF total score.
Reliability of the PPI-SF
In the total sample, coefficient alpha for the complete PPI-
SF equalled .86. For the PPI-SF subscales, alphas ranged
from .64 (Stress Immunity) to .87 (Blame Externalization).
Alpha for the complete PPI-SF was .85 for forensic partici-
pants and .89 for the controls. For forensic participants, PPI-
SF subscales exhibited alphas ranging from.57 (Stress Immu-
nity) to .82 (Machiavellian Egocentricityand Social Potency).
For controls, alphas ranged from .65 (Carefree Nonplanful-
ness) to .85 (Blame Externalization).
Validity of the PPI-SF
Group Differences The results showed predicted and signif-
icant group differences3with respect to PPI-SF total scores
on the one hand, and the PPI-SF-subscales of Machiavellian
Egocentricity, Coldheartedness, Carefree Nonplanfulness,
Fearlessness, Blame Externalization, and Impulsive Non-
conformity, on the other (Table 2). No group differences
were found for the PPI-SF-subscales of Social Potency and
Bonferroni-corrected post-hoc tests3revealed a statistically
significant group difference for the PPI-SF-total score, for the
forensic psychopathic versus forensic non-psychopathic par-
ticipants, t(56)04.01, p<.01, d0.1.07, and for forensic psy-
chopathic versus healthy participants, t(56)02.49, p<.05,
d0.67. Moreover, forensic psychopathic participants scored
significantly higher than forensic non-psychopathic partici-
pants on Machiavellian Egocentricity, t(56)03.18, p<.01,
d0.85, and Fearlessness, t(56)02.68, p<.05, d0.72. Both
forensic groupsscored significantly higherthan healthy partic-
ipants on Blame Externalization (forensic psychopathic partic-
ipants, t(56)06.06, p<.001, d01.62, and forensic non-
psychopathic participants, t(56)06.30, p<.001, d01.68).
Healthy controls did notdiffer significantly from psychopathic
participants on Carefree Nonplanfulness and Coldheartedness.
Convergent Validity PPI-SF total score correlated signifi-
cantly with the PCL-R total score, r0.39, .15<r<.593and
all PCL-R factors (Factor 1, r0.32 [.07<r<.53], Factor 2,
r0.40 [.16<r<.59], Factor 3, r0.32 [.07<r<.53], and Fac-
tor 4, r0.30 [.05<r<.51]).
Correlations between the PPI-SF total score and the AQ
were significant for the AQ total score (r0.47 [.25<r<.65])
and three AQ subscale scores: Physical Aggression (r0.53
[.32<r<.69]); Verbal Aggression (r0.50 [.28<r<.67]); and
Anger (r0.40 [.16<r<.59]), but for Hostility the correlation
was not significant.
Correlations between the total PPI-SF score and the BAS
subscales of the BIS-BAS were not significant. Correlations
between the PPI-SF total score and the RPQ yielded signif-
icant results for the RPQ total scale, r0.63 [.45<r<.76], the
Proactive aggression scale, r0.61 [.42<r<.75], and the Re-
active aggression scale, r0.53 [.32<r<.69].
PPI-SF and the BIS scale of the BIS-BAS correlated sig-
nificantly, r0−.39 [−.59<r<−.15]3.
4The PCL-R cut-off score of 26 was used as a dichotomous target
versus non-psychopathic (PCL-R total score<26). Recently, a lower cut-
off score of 26 was suggested for psychopathy in Europe (Cooke 1995;
Grann et al. 1998).
J Psychopathol Behav Assess (2013) 35:153–161 159
Results of the logistic regression of PCL-R cut-off
score4on the PPI-SF total score showed that the PPI-
SF displays criterion-related validity for predicting psycho-
pathic personality traits, B0.08, SE(B)0.03, β08.83, p<.01,
The aim of the second study was to investigate the reliability
and the validity of the PPI-SF and develop criterion refer-
ence scores. Results were broadly consistent with previous
research on the PPI (Ross et al. 2009) and with those of the
first study, revealing adequate reliability and validity for the
PPI-SF. First, analyses examining group differences repli-
cated Study 1 by showing a significantly higher PPI-SF total
score for the forensic psychopathic participants compared
with the other groups. Moreover, forensic participants
obtained significantly higher scores on the PPI-SF subscales
of Machiavellian Egocentricity, Fearlessness and Blame
Externalization, but unexpectedly, healthy participants
scored significantly higher than forensic non-psychopathic
participants on Coldheartedness and Carefree Nonplanful-
ness. Item-loading difficulties (cross loadings) with these
content scales and possible elimination of the Coldhearted-
ness scale have been discussed before (Neumann et al.
2008). We advise interpreting individual subscale scores
only in light of the PPI total and PPI criterion reference
score results. Second, results replicated earlier findings
(Poythress et al. 1998) and the results of Study 1, showing
a significant correlation between PPI-SF and PCL-R scores.
Moreover, the PPI-SF was associated with total PCL-R
scores. Furthermore, consistent with earlier findings (Hare
2003; Raine et al. 2006), we found a strong relation between
psychopathy and aggression and between psychopathy and
self-reported insensitivity to cues of punishment (Uzieblo et
al. 2007). However, we did not find a significant relation
with hostility, which contrasts with earlier research (e.g.,
Newman et al. 2005).
The reasons for the unanticipated group differences be-
tween forensic non-psychopathic participants and other par-
ticipants on Coldheartedness and Carefree Nonplanfulness
are unclear. These findings could raise questions regarding
the validity of these scales; alternatively, they might be
related to individual prospects for release and inmates’
personal situation, as all forensic individuals were held in
custody in a Penitentiary Institution in anticipation of a
possible conviction. Further research is needed to replicate
these unanticipated findings in other samples.
As noted earlier, a 56-item short form of the PPI (PPI-SF)
has been constructed. (Lilienfeld and Hess 2001). Neverthe-
less, formal criterion reference scores for this measure have
not been developed. Moreover, the original short form of the
PPI was developed by selecting items with high factor
loadings, which may result in narrower scales than the more
methodologically sophisticated approach used here (Smith
et al. 2000). Therefore, we emphasize the importance of a
well-validated PPI-SF as presented in the current article.
Summary and Concluding Discussion
We found that the PPI is a valid instrument useful in forensic
and non-forensic settings for measuring psychopathic person-
ality characteristics with respect to criterion reference scores.
Forpracticalsettings,inwhichthe PPI istimeconsuming,this
study presents a valid screening device with corresponding
criterion reference scores. The PPI-SF may be helpful in
forensic and non-forensic populations for detecting psycho-
pathic personality traits. The PPI-SF may also be of interest
for risk assessment, as the PPI-SF showed a significant rela-
tionship with the PCL-R and aggression. A general screening
for psychopathy may be important in many forensic settings
given the well-documented implications of this construct for
criminal recidivism (Hare 2003).
collection, Dorus van de Bosch for the support in the Penitentiary Insti-
tution, Bruno Verschuere for providing PPI Data, and Nienke Laan and
We would like to thank Anne Douven for data
Benning, S. D., Patrick, C. J., Hicks, B. M., Blonigen, D. M., &
Krueger, R. F. (2003). Factor structure of the psychopathic per-
sonality inventory: validity and implications for clinical assess-
ment. Psychological Assessment, 15, 340–350.
Buss, A. H., & Perry, M. (1992). The aggression questionnaire. Jour-
nal of Personality and Social Psychology, 63, 452–459.
Carlson, S. R., & Thái, S. (2010). ERPs on a continuous performance
task and self-reported psychopathic traits: P3 and CNV augmen-
tation are associated with fearless dominance. Biological Psychol-
ogy, 85, 318–330.
Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral
activation, and affective responses to impending reward and pun-
ishment: the BIS/BAS scales. Journal of Personality and Social
Psychology, 67, 319–333.
Cooke, D. J. (1995). Psychopathic disturbance in the Scottish prison
population: cross-cultural generalizability of the hare psychopathy
checklist. Psychology, Crime and Law, 2, 101–118.
Emons, W. H. M., Sijtsma, K., & Pedersen, S. S. (2012). Dimension-
ality of the Hospital Anxiety and Depression Scale (HADS) in
cardiac patients: comparison of mokken scale analysis and factor
analysis. Assessment, 19, 337–353.
Fayers, P., & Machin, D. (2007). Quality of life: the assessment,
analysis and interpretation of patient-reported outcomes (2nd
ed.). West-Sussex: Wiley Franken, I. H., & Muris, P. (2006).
BIS/BAS personality characteristics and college students’ sub-
stance use. Personality and Individual Differences, 40, 1497–
160 J Psychopathol Behav Assess (2013) 35:153–161
Fecteau, S., Pascual-Leone, A., & Théoret, H. (2008). Psychopathy Download full-text
and the mirror neuron system preliminary findings from a non
psychiatric sample. Psychiatry Research, 160, 137–144.
Grann, M., Långström, N., Tengström, A., & Stålenheim, E. G. (1998).
Reliability of file-based retrospective ratings of psychopathy with
the PCL-R. Journal of Personality Assessment, 70, 416–426.
Hare, R. D. (1991). Manual of the Psychopathic Checklist-Revised
(PCL-R). North Tonawanda: Multi-Health Systems.
Hare, R. D. (2003). Manual for the hare psychopathy checklist-revises
(2nd ed.). Toronto: Multi-Health Systems.
Hildebrand, M. (2008). Psychopathy checklist-revised en psychopathy
checklist: youth version. In T. Giesbrecht, C. de Ruiter, & M.
Jelicic (Eds.), Forensisch psychodiagnostisch gereedschap.
Hornsveld, R. H. J., Muris, P., Kraaimaat, F. W., & Meesters, C.
(2009). The Aggression Questionnaire in Dutch violent forensic
psychiatric patients and secondary vocational students. Assess-
ment, 16, 181-192.
Landis, J. R., & Koch, G.G. (1977). The measurement of observer
agreement for categorical data. Biometrics, 33, 159–174.
Leistico, A.-M. R., Salekin, R. T., DeCoster, J., & Rogers, R. (2008). A
large-scale meta-analysis relating the hare measures of psychop-
athy to antisocial conduct. Law and Human Behavior, 32, 28–45.
Lilienfeld, S. O., & Andrews, B. P. (1996). Development and prelim-
inary validation of a self-report measure of psychopathic person-
ality traits in noncriminal populations. Journal of Personality
Assessment, 66, 488–524.
Lilienfeld, S. O., & Hess, T. H. (2001). Psychopathic personality traits
and somatization: Sex differences and the mediating role of neg-
ative emotionality. Journal of Psychopathology and Behavioral
Assessment, 23, 11–24.
Lilienfeld, S. O., & Widows, M. R. (2005). The psychopathic person-
ality inventory-revised. Lutz, Florida: Psychological Assessment
Mokken, R. J. (1971). A Theory and Procedure of Scale Analysis.
Berlin: De Gruyter.
Molenaar, I. W., & Sijtsma, K. (2000). MSP5 for windows. Version 5.0.
Groningen: Iec ProGAMMA.
Morrissey, C., Hogue, T., Mooney, P., Lindsay, W. R., Steptoe, L.,
Taylor, J., & Johnston, S. (2005). Applicability, reliability, and
validity of the psychopathy checklist – revised in offenders with
intellectual disabilities: some initial findings. International Jour-
nal of Forensic Mental Health, 4, 207–220.
Neumann, C. S., Malterer, M. B., & Newman, J. P. (2008). Factor
structure of the psychopathic personality inventory (PPI): findings
from a large incarcerated sample. Psychological Assessment, 20,
Newman, J. P., MacCoon, D. G., Vaughn, L. J., & Sadeh, N. (2005).
Validating a distinction between primary and secondary psychop-
athy with measures of Gray’s BIS and BAS constructs. Journal of
Abnormal Psychology, 114, 319–323.
Poythress, N. G., Edens, J. F., & Lilienfeld, S. O. (1998). Criterion-
related validity of the psychopathic personality inventory in a
prison sample. Psychological Assessment, 10, 426–430.
C., & Liu, J. (2006). The reactive-proactive agression questionaire:
differential correlates of reactive and proactive aggression in adoles-
cent boys. Aggressive Behavior, 32, 159–171.
Ross, S. R., Benning, S. D., Patrick, C. J., Thompson, A., & Thurston,
A. (2009). Factors of the psychopathic personality inventory.
Criterion-related validity and relationship to the factors of the
psychopathic personality inventory. Assessment, 16, 71–87.
Sijtsma, K., & Molenaar, I. W. (2002). Introduction to nonparametric
item response theory. Thousand Oaks: Sage.
Smith, G.T., McCarthy, D. M.,& Anderson, K. G. (2000).Onthe sins of
short-form development. Psychological Assessment, 12, 102–111.
Uzieblo, K., Verschuere, B., & Crombez, G. (2007). The psychopathic
personality inventory: construct validation of the two-factor struc-
ture. Personality and Individual Differences, 43, 657–667.
Van Ginkel, J. R., & Van der Ark, L. A. (2005). SPSS syntax for
missing value imputation in test and questionnaire data. Applied
Psychological Measurement, 29, 152–153.
Vaughn, M. G., Howard, M. O., & Delisi, M. (2008). Psychopathic
personality traits and delinquent careers: an empirical examina-
tion. International Journal of Law and Psychiatry, 31, 407–416.
Vaughn, M. G., Newhill, C. E., DeLisi, M., Beaver, K. M., & Howard,
M. O. (2008). An investigation of psychopathic features among
delinquent girls: violence, theft, and drug abuse. Youth Violence
and Juvenile Justice, 6, 240–255.
Verschuere, B., Crombez, G., Koster, E. H. W., & Clercq, D. (2007).
Antisociality, underarousal, and the validity of the concealed
information polygraph test. Biological Psychology, 74, 309–318.
Walters, G. D., Knight, R. A., Grann, M., & Dahle, K. P. (2008).
Incremental validity of the psychopathy checklist facet scores:
predicting release outcome in six samples. Journal of Abnormal
Psychology, 117, 396–405.
J Psychopathol Behav Assess (2013) 35:153–161161