UNIVERSITY OF AMSTERDAM
The Multidimensional Personality
Questionnaire (MPQ) as a measure of
psychopathic traits
Date: 21-12-2012By Silva van Schagen, Bsc. Student Number: 5600731 Master: Brain and Cognitive Sciences
Track: Cognitive Neuroscience Supervisor: B.J. Verschuere, Ph.D Co-assessor: Prof. Dr. J.H. Kamphuis
Abstract
The Multidimensional Personality Questionnaire (MPQ) is a personality questionnaire that is based upon the assessment of several personality traits in order to capture behavioral
regulation, interpersonal and temperament style. Several studies have successfully tried to estimate psychopathic traits by using a specific constellation of MPQ items or scales, which suggests that the MPQ is also capable of measuring psychopathic traits. In study 1, using two large undergraduate samples, it was investigated to what extent the MPQ is able to map the 3-factor construct structure of psychopathy as assessed by the Youth Psychopathic Traits Inventory (YPI). The newly created MPQ psychopathy scale (MPQp) mapped the 3 facets of psychopathy, as confirmed by a battery of criterion-related questionnaires. Studies 2 and 3 examined whether the MPQp also captures the emotional and attentional deficits that characterize psychopathy. Using a fear-conditioning procedure, study 2 revealed an
unexpected increase in differentiation between the CS+ and the CS- in higher MPQp subjects for expectancy of the Unconditioned Stimulus (shock) as time passed during the fear
extinction phase. Indicating an inability of threat evaluation maintenance over time. In study 3 it was found that higher MPQp subjects showed deficient emotional responding compared to lower MPQp subjects, only when the attention was not focused on threat, confirming recent attentional views on psychopathy. Whereas, when the focus was on threat, reactions were very similar in both lower and higher MPQp subjects. Conceptually replicating the findings of study1, study3 also showed that the MPQp had high associations with a very widely used self-reported psychopathy instrument; the Psychopathic Personality Inventory Revised (PPI-R) and several additional criterion-related questionnaires. For the first time a 3-factor MPQ psychopathy instrument was created that is able to show relevant associations with criterion-related measures. Furthermore, since study 2 did not show the expected lack of fear
conditioning and in study 3 no fear deficit theory supporting results were found, it could be argued that the results of the present study provide more evidence for an attentional deficit in psychopathy, rather than an emotional deficit.
Psychopathy
Psychopathy is associated with several traits; superficial charm, fearlessness, an absence of anxiety, egocentricity, irresponsibility, inadequately motivated antisocial behavior, a lack of insight and learning by experience, the inability of attachment binding and an unreliable and untruthful personality (Cleckeley, 1941). Cleckeley (1941) has provided the foundation of the description of psychopathy, he wrote his ‘mask of sanity’ on basis of observations in patients that he treated in a psychiatric clinic. Cleckeley’s description is still the most widely used in order to understand and investigate psychopathy. Psychopathy is a widely studied
phenomenon, however, there remains much controversy about the psychopathy construct. For example, there are researchers who claim that psychopathy is nothing more than a mythical identity (Blackburn, 1988). Furthermore, none of the so far appeared Diagnostic and statistical manual’s (DSM; Lilienfeld, 1998) contain a diagnosis for psychopathy. Although this
controversy and lack of psychopathy diagnosis in the DSM, it should be emphasized that in the last few decades promising results, concerning the measures and methods that are used to investigate psychopathy, have been found (Lilienfeld, 1998).
Factor structure of psychopathy
The different traits, as described by Cleckeley (1941) and associated with psychopathy are considered to be related to different processes or deficits. Because of this, considerable research has been performed in order to better understand the construct of psychopathy. Superficial charm, fearlessness, an absence of anxiety, egocentricity and untruthfulness are thought to be related to the detachment of emotions in psychopathy (Lykken, 1995). Whereas irresponsibility, inadequately motivated antisocial behavior, unreliability, a lack of insight and learning by experience are thought to be related to disinhibited behavior in psychopathy (Lykken, 1995). The most dominant conception about the psychopathy construct and
differences in processes/deficits between the components of psychopathy is that there are two factors involved in psychopathy that are of importance; one describing the detachment of emotions, known as ‘primary’ psychopathy or factor 1 and another describing the
disinhibited behavior, known as ‘secondary’ psychopathy or factor 2 (Lykken, 1995). It has also been found that the disinhibited behavior seen in psychopathy is accompanied by a detachment of emotions, which emphasizes that the factor structure is needed to describe psychopathic traits (Patrick et al., 2009). Regarding the two factor structure in psychopathy it has been hypothesized that the disinhibited behavior component is related to a deficit in the prefrontal cortex of the brain and the emotional detachment component is related to a deficit in the amygdala (Patrick et al., 1993; Blair, 2004).
There is much controversy about this two factor structure of psychopathy and there are emerging studies that claim a three or even a four factor structure is needed in order to
describe psychopathic traits. For example, Cooke & Michie (2001) claim that the antisocial behavior seen in psychopathy cannot solely be attributed to pathologies of personality and therefore needs to be excluded from the psychopathy construct. Furthermore, they claim that the most dominant assumption of a two factor solution for psychopathy is not satisfactory and propose a three factor structure divided in a Deceitful Interpersonal style, Deficient Affective
experience and Behavioral Impulsive and Irresponsible lifestyle. In this model the factor 1 or
‘primary’ psychopathy factor is divided into the Interpersonal and Affective factor, creating three instead of two factors. Importantly, these authors argued that antisocial behavior is a consequence of the psychopathic personality, and therefore excluded explicit antisocial items from the Behavioral Impulsive factor.
Typically psychopathy is measured by the Psychopathic Checklist-Revised (PCL-R; Hare, 2003). The PCL-R is the best validated instrument in assessing psychopathy and tries to measure psychopathy as described in Cleckeley’s ‘Mask of Sanity’ (1941).
The PCL-R assesses psychopathy related personality items and behavioral items by means of intensive interview and file record investigation (e.g., criminal record; Hare et al., 1990). The investigated facets within the PCL-R are closely related to Cleckeley’s (1941) psychopathy description and can be divided into two factors; factor 1 measures callousness, egocentricity and remorselessness and factor 2 measures antisociality, instability and social deviance (Hare et al., 1990). Thus, the PCL-R can also be used to measure the three factor construct of psychopathy.
The PCL-R was mainly developed to assess psychopathy in incarcerated samples, this raises several issues when investigating psychopathy in the general population. First of all generally no file records are tracked in the “normal” population, secondly, administering the PCL-R is very time consuming and requires a trained interviewer. To overcome these disadvantages in using the PCL-R in the general population and thus in epidemiological studies to psychopathic traits, several self-reported questionnaires were developed to assess psychopathy in a more straightforward way. For example, a very reliable and profoundly validated self-reported psychopathy questionnaire is the Psychopathic Personality Inventory
Revised (PPI-R; Lilienfeld & Widows, 2005). The PPI-R was created in order to measure the
two factor construct of psychopathy and is one of the most used self-reported psychopathy instruments.
Emotional and attentional processes involved in psychopathic traits
Several studies suggest that subjects higher in psychopathic traits show deficits in emotional and attentional processes. For example, it has been found that psychopathic patients show deficient fear conditioning, they show reduced skin conductance response (SCR) in the acquisition phase and have problems with predicting harm from threat stimuli (Birbaumer et al., 2005; Rothemund et al., 2012). The reduced SCR found during the acquisition phase could indicate that subjects with higher psychopathic traits are less prone to experience anxiety and are less sensitive to punishment, which is thought to be an essential trait in psychopathy (Blair et al., 2004). This is suggested to be mediated by a lack of activation in the orbitofrontal cortex (OFC; Birbaumer et al., 2005). Furthermore, subjects with higher psychopathic traits show lower valence ratings for the conditioned stimulus, suggesting they have difficulty in predicting harm from threat (Birbaumer., et al 2005; Rothemund et al., 2012). It is suggested that this is due to deficient amygdala activation resulting in a lack of emotional processing and association between the Conditioned Stimulus (CS) and the Unconditioned Stimulus (UCS; Birbaumer et al., 2005; Rothemund et al., 2012). However, these differences for higher psychopathic subjects in fear conditioning paradigms are almost only present in the acquisition phase, suggesting normal unlearning of the CS during the extinction phase. Although, in the study by Rothemund et al. (2012) it was found that there was a smaller decrease of FPS during extinction, indicating more persistence in conditioned state. Furthermore, higher psychopathic subjects do seem to display normal estimations in expectancy ratings of the UCS (Flor et al., 2002; Rothemund et al., 2012). Suggesting that although psychopathic subjects are conscious of the CS/UCS relation, they do not fear the UCS.
The reduced responses to the CS+ during fear conditioning in higher psychopathic trait subjects could be related to a fear or amygdala deficit in psychopathy (Blair et al., 2005). This low fear theory is supported by Patrick et al. (1993), who found that subjects higher in
psychopathic traits showed reduced startle responses during presentation of unpleasant
pictures. The frontal cortex is considered to provide relevant information to the amygdala it is suggested that deficient responding of especially the orbitofrontal cortex (OFC) is important
regarding the fear deficit theory in psychopathy (Blair, 2008). Foremost, since the interaction between the amygdala and the OFC is very important in predicting consequences during learning (Schoenbaum et al., 1998). A well known orbitofrontal functioning experiment is the Wisconsin Card Sorting task (WCST). It has been found that subjects higher in psychopathic traits and especially higher in the impulsive-irresponsible psychopathic traits make more perseverative errors and finish less categories (Lapierre et al, 1995). In addition, Go/No-go paradigms revealed that psychopathic subjects make more errors of commission (responses on No-go trials), also suggesting an orbitofrontal deficit (Lapierre et al., 1995). An ERP
investigation revealed that psychopaths show larger positive ERP components for the No-go trials than for the Go trials which is opposite in normal subjects, however, they did not find any differences on reaction times or errors between higher and lower psychopathic trait subjects (Kiehl et al., 2000). The results from these studies suggest that there is an orbitofrontal deficit in psychopathy, which in turn could indicate a deficit in the amygdala/OFC interaction and thus supports the fear deficit theory in psychopathy.
Contrary to the fear deficit theory, Newman et al. (2010) propose that the lack of response during fear stimuli and on amygdala/orbitofrontal related tasks is not due to a deficit in fear reaction, but is caused by a deficit in processing of situational relevant information when their attention is focused elsewhere. Regarding this so called attentional deficit theory, Newman et al. (2010; Baskin-Sommers et al., 2012) directly tried to assess these two deficit theories by investigation of incarcerated men with PCL-R measured psychopathic traits in a fear conditioning experiment. They measured fear-potentiated startle (FPS) and created two distinct conditions; they focused the attention on fear information or focused the attention on alternative information. They found that subjects scoring higher on the PCL-R total and interpersonal/affective factor did not show any FPS differences compared with lower scoring subjects in the fear focused condition, but only showed diminished FPS for the alternative focused condition. Suggesting that attentional processes indeed are the underlying reason why psychopaths show smaller or less reactivity to emotional stimuli, rather than a fear deficit or amygdala dysfunction. Thus, the attentional deficit may explain the emotional deficit seen in psychopathy and for this reason it could be important to measure or manipulate attention in assessing psychopathy during emotional relevant tasks.
Psychopath or psychopathic?
An important feature of the psychopathy construct, regarding self-report in the “normal” population, is the dimensional structure of psychopathic traits instead of a more
straightforward categorical structure (Edens et al., 2006).
Edens et al. (2006) performed a taxometric analysis within the Psychopathic
Checklist-Revised (PCL-R; Hare, 2003) in male offenders. They did not find any evidence that
could indicate a taxonic difference distinguishing between psychopaths and non-psychopaths. Since the PCL-R can only be assessed in criminal or psychiatric patient samples, because of file record investigation, Walters et al. (2007) tried to investigate dimensionality by using the
Psychopathy Checklist: Screening Version (PCL: SV; Hart et al., 1995), in which no criminal
record is needed. They investigated a sample of incarcerated men and women and psychiatric patients. They also found support for a more quantitative difference instead of a qualitative difference regarding psychopathic traits. Last, to further support this dimensionality of psychopathy, Neumann & Hare. (2008) investigated psychopathy in a community sample of men and women by means of the PCL:SV. They found that although a community subject sample was investigated, the psychopathy latent model as seen in criminal samples remained.
All together these studies suggest that this dimensional structure in psychopathy makes it difficult to qualitatively separate psychopaths from the rest of the population and implicates that psychopathic traits maybe widespread among the “normal” population, with some people showing more and/or higher psychopathic traits than others.
Self-reported psychopathic traits
The most widely used and validated psychopathic trait self-reported inventory is the
Psychopathic Personality Inventory Revised (PPI-R; Lilienfeld & Widows, 2005) which
measures psychopathic traits divided into two distinct factors; Fearless Dominance (FD) and
Irresponsible Antisociality (IA). Fowles & Dindo (2009) state that a person high in FD factor
holds a high approach or reward motivation which results in strong reward-seeking behavior not retained by concern for others or fear. Furthermore, they emphasize the importance of egocentricity, achievement-oriented behavior and the need for physically dangerous and thrilling behavior in people that show high FD factor. The IA factor holds antisocial behavior characterized by impulsivity or disinhibited behavior, aggression/anger and has associations with internalizing and externalizing dimensions, such as alcohol and drug abuse, depression, attention deficit hyperactivity disorder (ADHD), conduct disorder (CD) and oppositional defiant disorder (ODD; Fowles & Dindo, 2009). Contrary to the PCL-R that measures psychopathy as one construct, the PPI-R was developed in order to assess all of the
psychopathy aspects as described by Cleckeley (1941) by using a personality based approach, in which the FD and IA factor are uncorrelated (Patrick et al, 2009). Although this
independence of the two factors of the PPI-R, when psychopathy is assessed by means of the PPI-R, both factors need to be taken into consideration (Patrick et al., 2009). Despite the differences between the PCL-R and the PPI-R, Benning et al. (2003) found that the factors of the PCL-R and the PPI had collateral relations with several questionnaires (e.g., MPQ), indicators (e.g., drug abuse) and behaviors (e.g., antisocial behavior).
On basis of the findings by Cooke & Michie (2001) which proposes a three factor structure of psychopathy, a self-reported psychopathy scale holding these three factors and is focused on investigating psychopathy in youth; the Youth Psychopathic Traits Inventory (YPI) was developed (Andershed et al., 2002). The YPI measures the Grandiose Manipulative (Interpersonal), Callous-Unemotional (Affective) and Impulsive-Irresponsible (Lifestyle) factors. The Interpersonal factor is associated with superficial charm, a grandiose sense of self-worth, untruthfulness and a manipulative nature. The Affective factor is associated with shallow affect, a lack of empathy, a lack of remorse and a lack of taking responsibilities for one’s own actions. The Lifestyle factor is associated with Impulsiveness, irresponsible
behavior, proneness to boredom, a lack of long-term goals and a parasitic lifestyle. Andershed et al. (2007) found that the three factors of the YPI showed significant parallel correlations with the three factors of the Psychopathic Checklist: Youth Version (PCL:YV; Forth et al., 2003). For an overview of the above mentioned psychopathy assessing instruments and their factors see table 1.
Inventory/Questionnaire Factors assessed assed by items/scales Psychopathic Checklist Revised (PCL-R) Emotional Detachment
or primary psychopathy Grandiose sense of self-worthGlibness/superficial charm Pathological lying Conning/manipulative
Lack of remorse or guilt Shallow affect Callous/lack of empathy Failure to accept responsibility
Impulsive-Antisocial Lifestyle or secondary psychopathy
Need for stimulation Parasitic lifestyle Poor behavioral controls Early behavior problems Lack of realistic goals
Impulsivity Irresponsibility Juvenile delinquency Revocation of conditional release Psychopathic Personality Inventory Revised
(PPI-R)
Fearless Dominance Social Influence Fearlessness Stress Immunity
Impulsive Antisociality Machiavellian Egocentricity Rebellious Nonconformity
Blame Externalization Carefree Nonplanfulness Youth Psychopathic Traits Inventory (YPI) Grandiose-Manipulative
(Interpersonal factor) Dishonest Charm Grandiosity Lying Manipulation Callous-Unemotional (Affective factor) Unemotionality Callousness Remorselessness Impulsive-Irresponsible (Lifestyle factor) Thrill-seeking Impulsiveness Irresponsibility
The Multidimensional Personality Questionnaire (MPQ) as a tool to measure psychopathic traits
The development of self-reported measures for psychopathic traits have made it possible to study psychopathy in the general population and has many benefits. However, the use of these self-reported measures are expensive, since most of them are copyrighted, besides they have to be available in the sample under investigation or need to be administered
deliberately/additionally. Since a lot of epidemiological studies use “normal” self-reported personality measures it would be of great benefit for studies into psychopathy if these self-reported instruments would hold a measure for psychopathic traits. This would save money, time and more importantly create many databases, already available and collected, in which psychopathic traits could be studied.
A very widely used and validated “normal” personality measure and used in all kinds of research is the Multidimensional Personality Questionnaire (MPQ; Tellegen, 1982). The MPQ was developed in order to assess personality traits divided in 11 primary scales, that measure emotional disposition, temperament, reversed impulsivity, conventionality, sensation seeking and the level of imagination in experiences (Patrick et al., 2002). The primary scales are thought to tab into three higher order factors; positive and negative emotionality (PEM and NEM respectively) and constraint (CON; Patrick et al., 2002). By measuring these primary scales and higher order factors, the MPQ can be used to capture behavioral regulation, interpersonal and temperament style which in turn could be important in more understanding of psychopathologies (Patrick et al., 2002).
Several studies suggest that the MPQ is capable of measuring psychopathic traits. Benning et al. (2005) used a regression equation to calculate MPQ estimated psychopathy factors by using the two factors of the PPI. They estimated the scores of the two PPI factors
by using the scores on the 11 primary scales of the MPQ, in which they relied on beta weights that were obtained from a regression equation taken from a “normal” population sample. After which they investigated three different samples to validate the MPQ estimated psychopathy factors; young adults from the community, an undergraduate sample and a sample of incarcerated men. They performed a correlational analysis of the MPQ estimated psychopathy factors with several relevant self-reported questionnaires for this validation. As expected, it was found that the estimated factor 1 was positively associated with narcissism, sensation seeking behavior, temperament and a sociable personality. Also they found negative associations of the estimated factor 1 with distress, empathy, anxiety and fear. For the
estimated factor 2 they found positive associations with impulsiveness, anger and anxiety, whereas there were negative associations with several sociability facets and empathy.
Blonigen et al. (2006) created a MPQ estimated psychopathy factor measure by correlational analyses of the factor scores on the PPI with the items of the MPQ in order to assess the development of psychopathic traits from late adolescence to early adulthood. They used the items in the Brief version of the MPQ and correlated every item with PPI-FD and IA factor. An item was selected and added in a factor whenever it correlated at least .20 with that factor and correlated less than half of this correlation with the other factor. Since the sample under investigation was assessed longitudinally they investigated the development of FD and IA factor as measured by the MPQ. They found the typically pattern that is seen in the
development of psychopathy; a decline of factor 2 or IA factor and stability of factor 1 or FD factor over the years. In further studies to the MPQ estimated PPI factors that Blonigen et al. (2006) created it was found that these MPQ based PPI measures had strong relations with the PCL-R and with other measures that are associated with the PPI factors, such as measures assessing affective content and narcissism (Witt & Donnellan, 2008; Witt et al., 2009). The above described studies suggest that the MPQ indeed could be used as an instrument to measure psychopathic traits.
Present study
Regarding the proposed 3-factor model by Cooke & Michie (2001), in the present study it was tried to create a 3-factor MPQ estimated psychopathy instrument (MPQp). This MPQp
instrument was first developed in a model-building sample and then cross-validated in a validation sample (Study1). In addition, fear conditioning data (Study2) and a large set of criterion-related questionnaires and tasks in the laboratory (Study3) were measured to further validate the MPQp and to exploratory assess its relation to emotional/attentional processes theorized to be involved in psychopathy.
In Study1, a MPQp instrument was created by selecting MPQ items that correlated with the factors of the YPI. One major advantage of using this psychopathy measure is that it measures the 3-factor model as proposed by Cooke & Michie (2001), but it can also be used as an instrument which measures the two factor model as described by the PCL-R. For the validation of the YPI based MPQp estimated measures, several questionnaires related to psychopathic traits were first correlated with the MPQp instrument in the model-building sample and then cross-validated in a validation sample.
In Study 2, in order to assess the fear deficit theory and to validate the MPQp, the MPQp measures were addressed by means of a fear-paradigm in a large sample. Skin
Conductance Response (SCR), FPS, UCS expectancy ratings were assessed and the MPQ was also measured. It was expected that subjects with higher MPQ estimated psychopathy traits would show less differentiation between the CS+ and the CS- for SCR and the FPS, however, they would show normal UCS expectancy ratings.
In Study 3, for explorative assessment of emotional/attentional processes and to further validate the MPQp, subjects completed the new created MPQp instrument and several criterion-related questionnaires. After which they participated in 3 different tasks, both
amygdala and orbitofrontal cortex related tasks were used (WCST, Emotional Go/No-go paradigm, Emotional Lexical decision task).
Study 1: MPQ Psychopathy measure creation and validation
Method
Subjects
Two samples of subjects were used; a model-building sample and a validation building sample. Data of first year undergraduate Psychology students from the University of
Amsterdam was analyzed in the study. Data of the model-building sample were obtained in October and November of 2011 (N=508) and data of the validation sample were obtained in October and November of 2010 (N=491).
For the model-building sample N=7 were excluded on basis of their short durations (<2,5 SD from the mean) on completing the questionnaires. Participants that had an age of >2.5 SD from the mean were also excluded from analyses (N=14). The final dataset for the model-building sample, used in creation of the MPQp instrument, consisted of N=487 (N=350 females and N=137 males), their mean age was 19.40 (SD=1.53, range=17-25).
In addition, following MPQp creation N=6 participants were removed on basis of their TRIN/VRIN scores (>3 SD from the mean) from the model-building sample1. The final
model-building sample used in the cross validation analysis consisted of N=481 (N=346 females and N=135 males), their mean age was 19.41 (SD=1.53, range=17-25).
For the validation sample N=2 were excluded on basis of their short durations (<2,5 SD from the mean) on completing the questionnaires. N=15 participants were removed on basis of their TRIN/VRIN scores of the MPQ (>3 SD from the mean). N=1 participant was removed because the MPQ was done twice with different outcomes. Participants that had an age of >2.5 SD from the mean were also excluded from analyses (N=23). The final validation sample consisted of N=451 (N=346 females and N=105 males), their mean age was 19.86 (SD=2.56, range=17-33).
Self-reported questionnaires
Youth Personality Traits Inventory (YPI; Andershed et al., 2002). The YPI is a self-report
measure consisting of 50 items and is scored on a four-point response scale. The YPI is thought to measure three factors of psychopathy; Grandiose-Manipulative (GM) or
Interpersonal, Callous-Unemotional (CU) or Affective and Impulsive-Irresponsible (I-I) or Lifestyle (Andershed et al., 2002). These factors are assessed by 10 subscales, with each scale related to one of the psychopathic factors. The Dishonest Charm (DC), Grandiosity (GR), Lying (LY), and Manipulation (MA) subscales belong to the GM factor. The Remorselessness (REM), Unemotionality (UNEMO) and Callousness (CAL) subscales belong to the CU factor. The Thrill-Seeking (THRILL), Impulsivity (IMPULS) and Irresponsibility (IRRESP)subscales belong to the I-I factor. In this study the Dutch version of the YPI developed by Das et al. (2002) was used. The YPI Total score (Cronbach’s α=.92), YPI-GM (Cronbach’s α=.89), YPI-CU (Cronbach’s α=.82) and YPI-I-I (Cronbach’s α=.83) were assessed.
Multidimensional Personality Questionnaire (MPQ; Tellegen, 1982). The MPQ is a
self-report measure consisting of 276 items and is thought to measure three higher order factors; 1 The VRIN/TRIN scales were not taken into account in creation of the MPQp measure since there were only N=6 that had to be removed on this basis and this made no difference for the MPQp item selection. However, for completeness in the cross validation analysis it was chosen to remove participants on basis of the TRIN/VRIN scales. Especially because of the relatively large amount of participants (N=15) that violated the TRIN/VRIN in the validation sample and had to be removed.
Positive Emotionality (PEM), Negative Emotionality (NEM) and Constraint (CON). These three higher order factors are assessed by 11 primary scales, with each scale related to one of the higher order factors. The Wellbeing (WB; Cronbach’s α=.79), Social Potency (SP; Cronbach’s α=.80), Achievement (AC; Cronbach’s α=.79) and Social Closeness (SC; Cronbach’s α=.67) scales belong to PEM. The Stress Reaction (SR; Cronbach’s α=.81), Aggression (AG; Cronbach’s α=.76) and Alienation (AL; Cronbach’s α=.77) scales belong to NEM. The Control (CO; Cronbach’s α=.81), Harm Avoidance (HA; Cronbach’s α=.67), and Traditionalism (TR; Cronbach’s α=.55) scales belong to CON. Only the Absorption (AB; Cronbach’s α=.72) scale cannot be classified into one of the higher order factors and stands alone. In this study the Dutch total version (276 items) and brief version(132 items) version of the MPQ (Eigenhuis et al., in press) were used.
Social Economical status (SES). In the present sample 10 items were used to calculate a total
SES scale. The items included: Education of both parents from the subject scored on a 5-point response scale (ranging from primary school to higher vocational education). The kind of paid job the parents of the subject were involved in, scored on a 4-point response scale (ranging from no work experience to highly qualified). The height of the disposable household income during the period before the eighteenth birthday of the subject, scored on a 5-point response scale (ranging from 500-1000 to 5000 or more euro per month). The highest level education completed by the subject scored on a 7-point response scale (ranging from primary school to science education). The number of years of schooling the subject received of all (part time) training of Basic Education to the present. The living circumstances applying to the subject, scored on a 4-point response scale (ranging from single or living independently to married). The current disposable income of the subject, scored on a 6-point response scale (ranging from <250 to >1250 euro per month). The amount of experience with paid work outside of the home of the subject, scored on a 7-point response scale (ranging from no work experience to >5 years).
Raven’s Progressive Matrices (RPM; Raven, 2000). The RPM or Raven is an intelligence test
in which one’s ability to distract meaning from perplexity is measured. It is thought that the RPM is a direct measure of general intelligence, which examines one of the two components of cognitive ability; educative ability as proposed by Spearman in 1923. In this version 35 items (figures) were presented that needed to be solved by selecting the correct part (several options) which was missing from the figure displayed. In the present study the score on Raven IQ was assessed.
Verbal Analogies VII (Guilford, 1956). The Verbal Analogies test tries to measure the degree
of “cognition of conceptual relations” , which is a component from the “structure of intellect” proposed by Guilford. Subjects have to decide which word (from a list) “belongs” to the presented word in the same way as the relationship between two words that are also presented indicate. For example apple : eat and Milk : ? subjects have to choose from; (white, drink, cow, sweat) with the correct answer being “drink”. In the present study 40 analogies had to be answered. The score on correct answers and the average score (correct minus error) was assessed.
The Vocabulary test (Elshout & Keizer, 1968). The Vocabulary test tries to measure the
degree of verbal skills or the cognition of semantic units and has associations with several components of the “Structure of intellect” proposed by Guilford (Guilford, 1956). In the present study subjects were presented with 40 words and had to indicate what the meaning of
the presented word was. The score on correct answers and the average score (correct minus error) was assessed.
Beck Depression Inventory (BDI; Beck et al., 1988). The BDI is self-reported questionnaire
developed to assess the presence of depression symptoms. The BDI consists of 21 items scored on a three point response scale (0 to 3; indicating intensity). In the present study the BDI total score was assessed.
The Alcohol Use Disorders Identification Test (AUDIT; Babor et al., 1992). The AUDIT is a
self-reported questionnaire developed to asses drinking behavior and related problems. For this study the total AUDIT score (Cronbach’s α= .80) was assessed.
The Rutgers Alcohol Problems Index (RAPI; White & Labouvi, 1989). The RAPI is a
self-reported questionnaire developed to assess problem drinking in adolescents. Subjects are asked to indicate how many times a certain problem related with their alcohol use has
occurred. In this study a brief version using 18 items of the original 53 item version was used and was scored on a four-point response scale (never, 1 to 2 times, sometimes, a lot). Total scores of the RAPI were assessed (Cronbach’s α= .89; only present in model-building sample)
The PANAS Trait Inventory (Watson et al., 1988). The PANAS Trait Inventory was developed
to assess positive and negative affect in general. The PANAS Trait Inventory consists of 20 items (words that describe feelings and emotions) that are scored on a five-point likert scale (ranging from very slightly or not at all to extremely). In the present study both negative (Cronbach’s α= .91) and positive affect (Cronbach’s α= .90) were assessed (only present in model-building sample).
The Anxiety Sensitivity Index (ASI; Reiss et al., 1986). The ASI is a self-reported questionnaire
which measures the degree of fear triggered by one’s anxiety feelings and symptoms. The ASI consists of 16 items scored on a five-point likert scale (ranging from hardly applicable to very applicable). The total ASI score (Cronbach’s α= .85) was assessed.
The Penn State Worry Questionnaire (PSWQ; Meyer et al., 1990). The PSWQ is a
self-reported questionnaire developed to assess clinical significant worry. The PSWQ consists of 16 items scored on a five-point likert scale (ranging from not characteristic to very
characteristic). The total PSWQ score (Cronbach’s α= .93) was assessed.
The State-Trait Anxiety Inventory (STAI; Spielberger et al., 1970). The STAI is a self-reported
questionnaire developed to assess anxiety as an emotional state (State Anxiety) and anxiety as a personality trait (Trait Anxiety). In the present study a brief form of the STAI was
usedwhich consists of 20 items scored on a four-point likert scale (ranging from not at all to very much so), only the Trait Anxiety scale was assessed.
The State-Trait Anger Expression Inventory (STAXI; Spielberger, 1983). The STAXI is a
self-reported questionnaire developed to assess anger as an emotional state (State Anger) and anger as a personality trait (Trait Anger). In the present study the STAXI consisted of 40 items scored on a four-point response scale (with either assessing the intensity of the feeling or the frequency of the feeling), only the Trait Anger scale was assessed.
The Behavioral Inhibition and Behavioral Activation Scales (BIS/BAS; Carver & White,
1994). The BIS/BAS is a self reported questionnaire in order to assess 4 factors; The BIS scale measures behavioral response inhibition and is associated with negative feelings (Cronbach’s α= .47), the three BAS scales are associated with positive feelings of which the
BAS fun seeking scale measures the tendency to impulsively search reward/pleasure
(Cronbach’s α= .53), the BAS Drive scale measures goal directed behavior (Cronbach’s α= . 78) and the BAS Reward Responsiveness scale measures the degree of positive response to reward (Cronbach’s α= .66). The BIS/BAS consists of 24 items that are scored on a four-point response scale (ranging from strongly disagree to strongly agree).
The Big Five Personality Questionnaire (5PFT; Elshout & Akkerman, 1975). The 5PFT is a
Dutch self-reported questionnaire created to test the big five personality factors; Extraversion, Neuroticism, Agreeableness, Conscientiousness and Openness to Experience. The 5PFT consists of 70 items (14 items per scale) scored on a seven-point likert scale. All five personality factors examined by the 5PFT were studied.
The Adjective Check List (ACL; Gough & Heilbrun, 1983). The ACL is a self-reported
personality questionnaire which measures 35 personality traits (scales). The ACL consists of 300 items (personality adjectives) that are scored on a five-point likert scale (ranging from not applicable to very applicable). The scales assessed in the present study were Dominance, Intraception, Aggression, Self Control and Confidence.
The Dutch Personality Questionnaire (NPV; Luteijn et al., 2000). The NPV is a self-reported
questionnaire developed to assess seven personality traits; Inadequacy (Neuroticism), Social Inadequacy, Rigidity, Resentment, Egoism, Dominance and Self-esteem. The NPV consists of 132 items which are scored on a three-point response scale (True, Don’t know, False; Luteijn et al., 2000). In the present study only Neuroticism, Social Inadequacy, Resentment, Egoism, Dominance and Self-esteem were assessed.
Table 2: N of subjects, Mean (sd) of scores of the investigated questionnaires for the model-building sample (in black) and validation sample (in grey)
Results study 1
Table 2: N, means and SD’s for all questionnaires
in the building and validation datasets N of subjects Mean (SD) of scores Dataset: Building Validation Building Validation
SES 371 344 49,81(3,95) 50,10(4,82)
RAVEN IQ 472 445 113,11(9,55) 115,19(9,65)
Verbal Analogies VII (items correct) 476 447 21,52(4,94) 21,86(4,94)
Vocabulary test (items correct) 478 450 12,93(3,76) 12,74(3,89)
YPI Total score 481 - 91,74(16,56)
-YPI Grandiose Manipulative 481 - 33,84(8,72)
-YPI Callous Unemotional 481 - 23,67(6,07)
-YPI Impulsive Irresponsible 481 - 31,03(7,10)
-MPQ Wellbeing (WB) 481 451 9,03(2,77) 17,88(5,23) MPQ Social Potency (SP) 481 451 6,81(3,28) 13,05(5,83) MPQ Achievement (AC) 481 451 7,30(3,13) 11,09(4,62) MPQ Social Closeness (SC) 481 451 9,25(2,63) 15,69(4,03) MPQ Stress Reaction (SR) 481 451 6,08(3,32) 8,69(5,81) MPQ Aggression (AG) 481 451 3,49(2,71) 5,24(3,56) MPQ Alienation (AL) 481 451 2,28(2,41) 3,08(3,54) MPQ Control (C) 481 451 6,76(3,29) 13,64(4,77)
MPQ Harm Avoidance (HA) 481 451 6,42(2,65) 15,98(5,09)
MPQ Traditionalism (TR) 481 451 6,07(2,28) 14,80(3,93)
MPQ Absorption (AB) 481 451 7,53(2,80) 17,35(7,48)
MPQ Higher order PEM 481 451 57,54(11,05) 45,55(15,36)
MPQ Higher order NEM 481 451 18,82(9,03) 32,82(14,07)
MPQ Higher order CON 481 451 43,53(9,29) 45,48(14,24)
BDI 465 447 6,05(5,91) 6,06(5,46)
AUDIT total score 479 449 8,96(4,60) 8,21(4,78)
RAPI 479 - 23,26(6,53)
-PANAS Positive Affect 481 - 30,15(6,77)
-PANAS Negative Affect 481 - 16,25(6,00)
-ASI 465 448 13,90(8,08) 10,90(6,48)
PSWQ 469 439 48,43(12,62) 45,89(12,90)
STAI 476 451 39,10(9,97) 40,16(8,87)
STAXI 476 448 15,84(4,26) 16,75(4,39)
BISBAS BIS 471 450 19,75(2,74) 19,07(1,83)
BISBAS BAS Drive 471 450 12,62(2,02) 12,85(1,86)
BISBAS BAS Fun Seeking 471 450 12,06(1,76) 11,82(1,66)
BISBAS BAS Reward Responsiveness 471 450 15,46(1,61) 15,44(1,46)
5-PFT Extraversion 469 437 64,68(11,39) 63,84(11,54) 5-PFT Agreeableness 469 437 72,88(9,49) 72,23(8,71) 5-PFT Conscientiousness 469 437 63,26(11,03) 63,87(10,41) 5-PFT Neuroticism 469 437 47,17(12,80) 45,98(12,30) 5-PFT Openness 469 437 64,01(10,20) 65,15(9,71) ACL Dominance 479 435 34,11(5,12) 34,02(5,24) ACL Intraception 479 435 35,29(5,29) 36,91(4,50) ACL Aggression 479 435 22,65(5,65) 24,22(5,54)
ACL Self Control 479 435 -.5,08(9,23) -.7,11(9,10)
ACL Confidence 479 435 34,49(13,57) 37,88(13,28) NPV Neuroticism 480 450 15,20(7,98) 14,29(7,22) NPV Social Inadequacy 480 450 8,24(6,37) 8,76(6,56) NPV Resentment 480 450 16,91(6,73) 16,69(6,36) NPV Egoism 480 450 9,58(4,59) 10,32(4,52) NPV Dominance 480 450 18,72(6,04) 18,61(5,89) NPV Self-esteem 480 450 27,87(5,76) 27,91(5,41)
Item selection data analysis
In the model-building sample, it was tried to develop a MPQ-based psychopathy scale by correlating individual MPQ items with Youth Psychopathic Traits Inventory (YPI; Andershed et al, 2002) factor scores. We used two-tailed Spearmans rho bivariate correlation because MPQ items were rated on a two-point scale causing a nonparametric distribution. This was done for the 3-factor model of the YPI (GM or Interpersonal, CU or Affective and I-I or Lifestyle)2. In order to select items we used the following sequential criteria3:
1. MPQ items needed to show convergent validity and were required to have a minimum correlation of .20 with one of the YPI factors.
2. MPQ items needed to show discriminant validity in the sense that they were required to show lower correlation with the other two YPI factors (e.g., Correlation .30 with the relevant factor and a smaller correlation with both of the other factors).
The newly created MPQp scale consisted of 50 items in total, for item content see appendix 1. - MPQp Interpersonal factor: 19 items (e.g., I enjoy being in the spotlight / I am very
good at influencing people).
- MPQp Affective factor: 15 items (e.g., If I have a humiliating experience I get over it very quickly / I like to watch a good, vicious fight ).
- MPQp Lifestyle factor: 16 items (e.g., I often act without thinking / I am more likely to be fast and careless than to be slow and plodding).
Internal consistency
Internal Consistency of the MPQ estimated psychopathy factors and total score were assessed. In the model-building sample the cronbach’s α of all factor and total scores were > .65 (see table 3). In the validation sample, cronbach’s α decreased, however, they remained > .58 (see table 3). The mean factor scores and standard deviation on the different MPQp factor and total scores within the model-building and validation sample were very close together (see table 3).
Table 3 N of subjects Mean score (SD) Cronbach's α
Sample Building Validation Building Validation Building Validation
MPQp Interpersonal 481 451 8,2 (3,8) 8,3 (3.4) .74 .67
MPQp Affective 481 451 5,8 (2.9) 6.0 (2.6) .66 .59
MPQp Lifestyle 481 451 7,2 (3.9) 6,8 (3.6) .81 .77
MPQp Total 481 451 21,2 (8.3) 28,8 (9.9) .86 .80
Table 3: N of subjects, mean (sd) of scores and reliability for the MPQ estimated factors and total for the model-building sample (in black) and the validation sample (in grey).
Inter correlations between MPQp factors and total score
2 It was also tried to create a 2-factor MPQp measure. However, the 3-factor MPQp scale showed better discriminant results
in the cross validation analysis. When comparing the 2-factor and 3-factor solution within the MPQ it seems that there are several reasons to describe psychopathy with a 3-factor model. The most important reason was the difference in correlations of the MPQp Interpersonal and MPQp Affective factors within the 3-factor solution with relevant questionnaires measuring anxiety, anger and worry. These correlations completely disappeared in the 2-factor solution.
3The criteria Blonigen et al. (2006) used were adjusted, since the low amount of items, reliability and correlations with
relevant YPI factors it gave in the described sample. Also, several different more strict criteria requirements were tried. However, the criteria described in the results gave the best solution, regarding item amount, reliability and correlations with the YPI.
Bivariate Pearson Inter-correlations between the MPQp factors and total scores are significant p<.01 in both the model-building and validation sample; see table 4 and 5. Inter correlations within the model-building sample are somewhat higher than in the validation sample.
Table 4
MPQp Interpersonal MPQp Affective MPQp Lifestyle
MPQp Interpersonal 1 -
-MPQp Affective .49** 1
-MPQp Lifestyle .35** .45** 1
MPQp Total .79** .78** .79**
Table 4 Inter-correlations of the model-building sample *p<.05 and **p<.01
Table 5
MPQp Interpersonal MPQp Affective MPQp Lifestyle
MPQp Interpersonal 1 -
-MPQp Affective .39** 1
-MPQp Lifestyle .21** .32** 1
MPQp Total .74** .73** .73**
Table 5 Inter-correlations of the validation sample *p<.05 and **p<.01
Pearson and partial correlational analysis of MPQp instrument with YPI factors for the model-building sample
Because of the high inter-correlations between factors, partial correlations were also calculated between MPQp factor and total scores, controlling for the remaining MPQp factors. Bivariate and partial correlations of the MPQp factors and total score with the YPI factors and total score were high, see table 6. When comparing the bivariate and partial correlations, the partial correlations were a lot less strong than the bivariate correlations.
Table 6: Pearson bivariate and partial correlations
of MPQp and YPI
MPQp Interpersonal MPQp Affective MPQp Lifestyle MPQp Total
Correlations Bivariate/Partial Bivariate/Partial Bivariate/Partial Bivariate YPI Interpersonal (GM) .67**/.57** .46**/.15** .36**/.17** .63** YPI Affective (CU) .44**/.25** .55**/.40** .28**/.03 .52** YPI Lifestyle (I-I) .43**/.31** .35**/-.02 .60**/.49** .60** YPI Total score .62**/.49** .49**/.14** .50**/.34** .69**
Table 6: Correlations of the different MPQ psychopathy measures with the YPI total and factor scores in the model-building sample. *p<.05 and **p<.01, In bold factor correlations of all three factors significantly differ from each other according to the Steiger’s Z test and
in grey one factor correlation of the three factor solution differs significantly from correlation of the other two factors.
Correlational analysis of MPQp instrument with the MPQ scales and higher order factors
The SP and AG scale had a strong correlation with MPQp Interpersonal r >.60, the SR and HA scale had a strong negative correlations with MPQp Affective r < -.50 and the AC and C scale had a strong negative correlation with MPQp Lifestyle r < -.35 (see table 7 and 8). For the higher order factors of the MPQ, MPQp Interpersonal had strong positive correlations with both PEM r > .32 and NEM r > .24 (see table 7 and 8). MPQp Affective had a negative correlation with NEM r < -.12 and a strong negative correlation with Constraint r < -.46 (see
table 7 and 8). Last MPQp Lifestyle had a very strong negative correlation with Constraint r < -.72 (see table 7 and 8).
Table 7:
MPQp Interpersonal MPQp Affective MPQp Lifestyle MPQp Total
Correlations Bivariate Partial Bivariate Partial Bivariate Partial Bivariate MPQ Wellbeing (WB) .23** .17** .18** .03 .15** .03 .24** Social Potency (SP) .76** .74** .37** -.00 .25** -.05 .59** Achievement (AC) .04 .34** -.25 -.22** -.40 -.44** -.26 Social Closeness (SC) -.06 -.02 -.14** -.21** .13** .25** -.02 Stress Reaction (SR) -.12** .16** -.50** -.50** -.22** .01 -.33** Aggression (AG) .66** .62** .31** -.04 .22** .04 .51** Alienation (AL) .18** .24** -.06 -.13* -.03 -.04 .05 Control (C) -.34** -.04 -.42** .09 -.97** -.97** -.76**
Harm Avoidance (HA) -.38** -.05 -.69** -.59** -.37** -.06 -.58** Traditionalism (TR) -.04 .10 -.22** -.13* -.22** -.19** -.20**
Absorption (AB) .19** .20** .08 .01 .04 -.05 .13**
Higher order PEM .42** .46** .12** -.11* .08 -.08 .27**
Higher order NEM .24** .39** -.20** -.35** -.06 -.00 .01
Higher order CON -.43** -.08 -.68** -.54** -.80** -.76** -.81** *p<.05 and **p<.01; In bold factor correlations of all two or three factors significantly differ from each other according to the Steiger’s Z test and in grey one factor correlation of the three factor solution differs significantly from correlation of the other two factors.
Table 8:
MPQp Interpersonal MPQp Affective MPQp Lifestyle MPQp Total
Correlations Bivariate Partial Bivariate Partial Bivariate Partial Bivariate MPQ Wellbeing (WB) .09 .05 .11* .05 .08 .03 .13** Social Potency (SP) .68** .67** .22** -.09 .16** -.01 .49** Achievement (AC) -.00 .17** -.25** -.20** -.42** -.36** -.31** Social Closeness (SC) .04 .11 -.15** -.23** .12** .18** .02 Stress Reaction (SR) -.08 .14* -.53** -.51** -.20** -.04 -.33** Aggression (AG) .67** .61** .38** .17** .18** .01 .55** Alienation (AL) .19** .26** -.04 -.10 -.00 .01 .07 Control (C) -.14** .11 -.27** .06 -.92** -.91** -.64**
Harm Avoidance (HA) -.39** -.17** -.69** -.59** -.31** -.16** -.60** Traditionalism (TR) .02 .16** -.23** -.17** -.25** -.22** -.20**
Absorption (AB) .21** .15** .13** .08 .05 -.03 .17
Higher order PEM .32** .41** -.05 -.20** -.04 -.08 .12*
Higher order NEM .41** .52** -.12** -.29** -.03 -.02 .13**
Higher order CON -.22** .11* -.56** -.46** -.73** -.72** -.69** *p<.05 and **p<.01. In bold factor correlations of all three factors significantly differ from each other according to the Steiger’s Z test and in grey one factor correlation of the three factor solution differs significantly from correlation of the other two factors.
Only the bivariate Pearson correlations are mentioned in the results, for partial correlation results refer to table 9 and 10. Because the correlational analysis showed much resemblance between the three MPQ estimated psychopathy factors, a Steiger’s Z-test was performed to compare the correlations between the three factors. Significant differences are only reported if they significantly differ from both other factors. Results from the model-building and the validation sample were compared.
MPQp Interpersonal factor
Both the model-building and validation sample showed significant positive correlations for MPQp Interpersonal factor with the AUDIT, STAXI , BAS Drive, BAS Fun Seeking, 5-PFT Extraversion, 5-PFT Openness, ACL Dominance, ACL Aggression, ACL Confidence, NPV Resentment, NPV Egoism, NPV Dominance and NPV Self esteem. The MPQp Interpersonal factor showed significant negative correlations with PSWQ, 5-PFT Agreeableness, ACL Self Control and NPV Social inadequacy (see table 9 and 10). The RAPI and PANAS positive affect questionnaires correlated positively with the model-building sample. However these questionnaires were not present in the validation sample.
The MPQp Interpersonal factor had significantly higher positive correlations compared to the other MPQp factors in both the model-building and validation sample for: STAXI; Z=2.70, p<.01 and Z=2.55, p<.05 in the model-building sample and Z=6.15, p<.01 and Z=2.44, p<.05 in the validation sample. BAS Drive; Z=2.01, p<.05 and Z=4.45, p<.01 in the model-building sample and Z=3.75 en Z=5.70, p<.01 in the validation sample. NPV resentment; Z=2.98 and Z=3.59, p<.01 in the model-building sample and Z=4.21 and Z=3.76, p<.01 in the validation sample.NPV Dominance; Z=5.64 and Z=8.29, p<.01 in the model-building sample and Z=5.78 and Z=6.93, p<.01 in the validation sample. The MPQp Interpersonal factor had significantly higher negative correlations compared to the other MPQp factors for none of the questionnaires.
In the validation sample the MPQp Interpersonal factor had significantly higher positive correlations compared to the other MPQp factors for: 5-PFT Extraversion ; Z=5.34 and Z=2,68, P<.01, ACL Dominance; Z=2.87, p<.01 and Z=1.98, p<.05, ACL Confidence; Z=2.11, p<.05 and Z=2.84, p<.01 (see table 10). Furthermore, in the validation sample the MPQp Interpersonal factor had significantly higher positive correlations compared to the other MPQp factors for 5-PFT Agreeableness; Z=-2.68 and Z=-4.15, p<.01 (see table 10).
MPQp Affective factor
Both the model-building and validation sample showed significant positive correlations for MPQp Affective factor with the Vocabulary test (items correct), AUDIT, BAS Fun Seeking, 5-PFT Openness, ACL Confidence, NPV Egoism and NPV Dominance. The MPQp Affective factor showed significant negative correlations with BDI Total score, ASI, PSWQ, STAI, BIS, BAS Reward Responsiveness, 5-PFTConscientiousness and Neuroticism, ACL Aggression, NPV Neuroticism and Social inadequacy (see table 9 and 10). The RAPI questionnaires correlated positively and the PANAS negative affect negatively with the model-building sample. However these questionnaires were not present in the validation sample.
The MPQp Affective factor had no significantly higher positive correlations, for both the model-building and validation sample, compared to the other MPQp factors. The MPQp Affective factor had significantly higher negative correlations compared to the other MPQp factors for: 5-PFT Neuroticism; Z=-3.74 and Z=-3.59 p<.01, in the model-building sample and 5.40 and 4.22, p<.01 in the validation building sample. NPV Neuroticism; Z=-3.62 and Z=-2.85, p<.01 in the model-building sample and Z=-4.58 and Z=-3.98, p<.01 in the validation sample (see table 9 and 10).
In the model-building sample the MPQp Affective factor had significantly higher negative correlations compared to the other MPQp factors for: PANAS Negative affect; Z=-3.42, p<.01 and Z=-2.03, p<.05 and BIS; Z=-4.68 and Z=-4.36, p<.01 (see table 9). Whereas in the validation sample the MPQp Affective factor had significantly higher negative
correlations compared to the other MPQp factors for: BDI; Z=-2.25 and Z=-1.96, p<.05, PSWQ; Z=-6.55 and Z=-3.17, p<.01, STAI; Z=-4.62 and Z=-3.58, p<.01, , p<.01, BAS Reward responsiveness; Z=-2.99, p<.01 and Z=-2.39, p<.05, ACL Aggression; Z=-5.75 and Z=-5.16, p<.01(see table 10). Furthermore, in the validation sample the MPQp Affective factor had significantly higher positive correlations compared to the other MPQp factors for NPV Egoism; Z=2.45 and Z=4.09, p<.01 (see table 10).
MPQp Lifestyle factor
Both the model-building and validation sample showed significant positive correlations for MPQp Lifestyle factor with the AUDIT, BAS Fun Seeking , 5-PFT Extraversion, ACL Dominance and NPV Dominance. The MPQp Lifestyle factor showed significant negative correlations with ASI, PSWQ, STAI, BAS Drive, 5-PFT Conscientiousness and Neuroticism, ACL Intraception, ACL Self Control and NPV Social inadequacy (see table 9 and 10). The RAPI questionnaires correlated positively and the PANAS positive affect negatively with the model-building sample. However these questionnaires were not present in the validation sample.
The MPQp Lifestyle factor had no significantly higher positive correlations, for both the model-building and validation sample, compared to the other MPQp factors. The MPQp Lifestyle factor had significantly higher negative correlations compared to the other MPQp factors for: BAS Drive; Z=-4.45, P<.01 and Z=-2.45, p<.05 in the model-building sample and Z=-5.70, p<.01 and Z=-1.97 in the validation sample. 5-PFT Conscientiousness Z=-7.16 and Z=-4.19, p<.01 in the model-building sample and Z=-8.60 and Z=-6.55, p<.01 in the
validation sample. ACL Intraception; 4.24, P<.01 in the model-building sample and Z=-5.75 and Z=-4.57, p<.01 in the validation building sample (see table 9 and 10).
In the model-building sample the MPQp Lifestyle factor had significantly higher negative correlations compared to the other MPQp factors for: PANAS positive affect; Z=-3.09, p<.01 and Z=-2.32, P<.05 (see table 9). Whereas, in the validation sample the MPQp Lifestyle factor had significantly higher positive correlations compared to the other MPQp factors for AUDIT; Z=4.17 and Z=3.87, p<.01 (see table 10).
Table 9:
3-factor MPQ psychopathy solution; Model-building sample
Correlations Bivariate Partial Bivariate Partial Bivariate Partial Bivariate
SES .26** .18** .20** .06 .12** -.01 .25**
IQ
RAVEN -.00 -.08 .09 .11* .09 .06 .07
Verbal Analogies VII (items correct) -.01 -.09 .05 .16** .01 -.07 .01 Verbal Analogies VII (Total) -.03 -.11* .06 .17** -.01 -.07 .00 Vocabulary test (items correct) .11* -.01 .11* .14** -.04 -.11* .07
Vocabulary test (Total) -.00 -.08 .10* .17** -.02 -.10 .03
BDI total score -.01 .13* -.16** -.18** -.06 -.02 -.09
AUDIT Total score .29** .18** .20** -.01 .34** .27** .36**
RAPI .29** .21** .20** -.08 .32** .32** .35** PANAS Positive Affect .10* .16** .05 .04 -.10* -.20 .02 Negative Affect .03 .14** -.19** -.25** -.06 -.00 -.08 ASI -.01 .11* -.16** -.21** -.13** -.09 -.12* PSWQ -.22** .13* -.46** -.44** -.38** -.19** -.43** STAI -.11* .08 -.22** -.27** -.12** -.03 -.18** STAXI .29** .34** .12* -.19** .13** .08 .23** BISBAS BIS -.29** -.00 -.55** -.46** -.31** -.06 -.47** BAS Drive .19** .24** .06 .03 -.10* -.20** .06
BAS Fun Seeking .28** .20** .20** -.00 .33** .25** .35**
BAS Reward Responsiveness -.08 -.03 -.14** -.11* -.04 .03 -.10* 5-PFT Extraversion .33** .31** .11* -.13* .25** .19** .31** Agreeableness -.28** -.20** -.19** -.09 -.06 .06 -.22** Conscientiousness -.20** .10 -.34** -.08 -.60** -.56** -.49** Neuroticism -.16** .06 -.39** -.36** -.17** .01 -.29** Openness .20** .20** .23** .26** -.09 -.31** .13** ACL Dominance .23** .19** .16* .03 .14** .03 .22** Intraception -.01 .02 -.01 .16** -.28** -.35** -.13** Aggression .20** .28** -.09* -.25** .08 .05 .09* Self Control -.32** -.28** -.17** .10 -.35** -.27** -.37** Confidence .13** .06 .15** .10 .07 -.03 .15** NPV Neuroticism -.03 .14** -.26** -.26** -.08 .02 -.14* Social Inadequacy -.28** -.19** -.23** -.05 -.27** -.16** -.33** Resentment .23** .27** .04** -.05 -.00 -.05 .12* Egoism .23** .15** .26** .23** .01 -.15** .20** Dominance .67** .62** .40** .12* .24** -.06 .55** Self-esteem .15** .09 .20** .18** -.03 -.13* .12*
*p<.05 and **p<.01,. In bold factor correlations of all two or three factors significantly differ from each other according to the Steiger’s Z test and in grey one factor correlation differs significantly from correlation of the other two factors.
Table 10:
3-factor MPQ psychopathy solution; Validation sample
Correlations Bivariate Partial Bivariate Partial Bivariate Partial Bivariate
SES .08 .05 .03 .01 .05 .02 .08
IQ
RAVEN .04 -.05 .13** .07 .11* .08 .12**
Verbal Analogies VII (items correct) .07 .06 .07 .01 .10* .06 .11*
Verbal Analogies VII (Total) .02 .08 .13** .04 .07 .03 .09
Vocabulary test (items correct) .05 .03 .14** .09 -.01 -.06 .08
Vocabulary test (Total) .01 -.06 .14** .15** -.06 -.11 .03
BDI total score -.03 .02 -.18** -.15** -.05 .03 -.11*
AUDIT Total score .16** .04 .18** -.01 .42** .40** .36**
ASI -.07 .02 -.27** -.20** -.17** -.11 -.22** PSWQ -.13** .04 -.53** -.46** -.35** -.22** -.44** STAI -.03 .06 -.33** -.31** -.10* -.00 -.19** STAXI .25** .34** -.16** -.29** .09 .11* .11* BISBAS BIS .02 .07 -.10* -.15** -.05 .03 -.05 BAS Drive .18** .28** -.07 -.10 -.20** -.22** -.04
BAS Fun Seeking .22** .21** .14** .06 .14** .06 .23**
BAS Reward Responsiveness .09 .11 -.11* -.12* .05 .11 .03
5-PFT Extraversion .37** .42** .02 -.23** .20** .16** .28** Agreeableness -.21** -.22** -.03 .03 .07 .11 -.08 Conscientiousness -.07 .09 -.21** -.11 -.59** -.58** -.42** Neuroticism -.05 .05 -.40** -.36** -.13** -.01 -.24** Openness .21** .17** .17** .11* -.07 -.16** .13** ACL Dominance .28** .33** .09 -.08 .15** .12* .25** Intraception .19** .20** .11* .03 -.20** -.25** .03 Aggression .19** .27** -.20** -.35** .15** .23** .09 Self Control -.37** -.41** -.03 .25** -.34** -.35** -.36** Confidence .24** .30** .10* -.04 .05 .00 .17** NPV Neuroticism .00 .06 -.30** -.30** -.04 .04 -.13** Social Inadequacy -.32** -.32** -.21** .00 -.22** -.14* -.34** Resentment .20 .25 -.08 -.15 -.05 -.06 .08 Egoism .10* .05 .26** .23** -.01 -.11 .14** Dominance .56** .52** .29** .04 .15** -.00 .46** Self-esteem .10* .11* .12* .08 -.05 -.08 .07
p<.05 and **p<.01. In bold factor correlations of all two or three factors significantly differ from each other according to the Steiger’s Z test and in grey one factor correlation differs significantly from correlation of the other two factors.
Discussion study 1
The results indicate that indeed the MPQ can be used for measuring psychopathic traits. The reliability analysis shows acceptable reliability of the scales for almost all factors (>.65) except for the MPQp affective factor in the validation sample. This might be due to the relative low number of items (N=15) of the factor. However, since the reliability is .66 in the
model-building sample, it might also be that this lower reliability is caused by the difference between MPQ versions taken from both samples. In the model-building sample the brief version of the MPQ was administered, whereas, in the validation sample the full MPQ was administrated.
Correlations of the MPQp factors and total score and the YPI psychopathy scales were high, which also indicates the ability of the MPQ to measure the by the YPI described
psychopathic traits. Furthermore there were significant correlations, of the MPQp model, with many of the criterion-related questionnaires. There were positive relations with alcohol use, anger, extraversion, dominance, social adequacy, egoism and the tendency to impulsively seek reward or pleasure and negative relations with neuroticism, anxiety, worry, self control and conscientiousness. Also, as expected, several of these correlations correspond
significantly with only one of the factors or had a significantly higher positive/negative correlation with one of the factors compared with the other factor(s).
The MPQp Interpersonal factor was predominantly positively associated with extraversion, confidence, dominance, anger, resentment and goal directed behavior. The MPQp Affective factor was predominantly negatively associated with depression, anxiety, negative affect, neuroticism, worry, aggression and response inhibition and positively
associated with egoism. The MPQp Lifestyle factor was predominantly negatively associated with goal directed behavior, conscientiousness and intraception; best described as the
tendency to think in a social relevant way. Furthermore, the MPQp Lifestyle factor was positively associated with alcohol use.
These results under scribe the normally associated personality traits with the
psychopathic factors. However, there were some discrepancies in the results, generally it is thought that the Lifestyle factor of psychopathy is related to aggression and anger. However in the present study aggression and anger was predominantly associated with the MPQp Interpersonal factor. This could be caused by the aggression items that were left in the item selection of the MPQp Interpersonal factor. It was decided not to exclude these items on basis of face validity because reliability of the scale and total score dropped dramatically after exclusion of these items. Furthermore, since Cooke & Michie (2001) suggested that antisocial behavior should not be taken into account in psychopathy, the lifestyle factor within the YPI does not contain antisocial items. The lifestyle factor of the YPI only contains items related to impulsivity, responsibility and thrill seeking behavior, this lack of antisocial items could also have caused this shift in aggression items and association to the Interpersonal facet of the MPQp.
Study 2: Fear conditioning in psychopathy as measured by the MPQp
Method
Subjects
13 studies already collected within the clinical research department of the University of Amsterdam were analyzed. Undergraduate psychology students (N=297) participated.
Stimuli
The Conditioned Stimulus (CS) consisted either of fear relevant pictures (spiders, snakes), neutral stimuli (square and circle) or pictures of men faces depending on the study. The CS was presented on a computer screen. During the acquisition phase one CS was most of the time followed by the UCS (CS+) and the other CS was never followed by the UCS (CS-; Gazendam & Kindt, 2012). CS+ and CS- assignment was counterbalanced across subjects. The Unconditioned Stimulus (UCS) consisted of a self-calibrated electric shock delivered to the left hand by means of two electrodes. The shock was calibrated per subject up to the point
that it was not painful, however, difficult to tolerate (Gazendam & Kindt, 2012). During the extinction phase the CS+ was not followed by a shock anymore
Psychophysiological assessment
In this study focus was on the acquisition and extinction phase for fear-potentiated startle response (FPS) as measured by electromyogram (EMG), skin conductance response (SCR) and electric shock expectancy (UCS expectancy).
Fear-potentiated startle (FPS) was measured by means of eye blink response to an acoustic startle with orbicularis oculi electromyogram (EMG) collected from two electrodes positioned around the eye. The acoustic startle was presented through a headphone and consisted of white noise (±104dB).
Skin Conductance Response (SCR) was measured by means of two electrodes that were connected to two fingers (second and fourth) of the non-dominant hand. The SCR to the Conditioned Stimulus was measured.
UCS expectancy was measured by means of online rating during the presentation of the Conditioned Stimulus (CS). Ratings were measured on a scale from “certainly a shock
expected” to “certainly no shock expected” and these ratings were registered on a scale of
200; from -100 (certainly no shock) to 100 (certainly a shock).
Procedure
During habituation phase subjects received several acoustic startle probes, this in order to reduce initial startle reactivity and to be able to assess emotional effects during the testing phase (Gazendam & Kindt, 2012). During the acquisition phase the CS was presented on a computer screen for 8 seconds. Startle probe was delivered at 7 seconds after stimulus onset and the UCS was delivered following the CS+ at 7.5 seconds after stimulus onset. UCS expectancy was rated during or following stimulus onset. ITI was jittered and around 22 seconds in which a startle probe was administrated to measure Noise Alone (NA) startle response. The extinction phase was implemented on a different day and yielded the same procedure as described for the acquisition phase, except in this phase the CS+ was not followed by the UCS anymore.
Fear extinction paradigm
Most studies included several conditions. In the present study experimental conditions (administration of drugs or providing specific instructions to reduce fear memory) were excluded, only the control conditions were included as these followed a highly similar
procedure. As most manipulations were given after acquisition (before extinction), extinction phase for these studies were not included in the analysis. Number of trials differed between studies and because of this only the first 7 trials of acquisition and first 9 trials of extinction were analyzed, which were measured in all studies. The acquisition phase data was divided in 3 trial blocks (block 1: trial 1&2, block 2:trial 3,4&5, block 3: trial 6&7) and this was also done for the extinction phase data (block1:trial 1&2, block2: trial 4,5&6, block3: trial 8&9).
Data reduction
N=4 subjects were excluded on basis of their score on the TRIN and VRIN scales within the MPQ (>3sd from the mean). For all acquisition datasets 1 experimental condition (N=13) was excluded. For all extinction datasets 10 experimental conditions (N=111) were excluded. Later on it was also decided to exclude the placebo condition of one of the studies for the
