Personality and Individual Differences 45 (2008) 584590

Contents lists available at ScienceDirect

Personality and Individual Differences

journal homepage: www.elsevier.com/locate/paid

Factor structure and behavioural correlates of the Psychopathy Checklist-Revised [PCL-R] in a Brazilian prisoner sample
Carmen E. Flores-Mendoza a,*, Marco Antnio Silva Alvarenga a, scar Herrero b, Francisco Jos Abad b
a

Laboratorio de Avaliao das Diferenas Individuais, Universidade Federal de Minas Gerais, Av. Antnio Carlos, 6627, FAFICH Depto. de Psicologia Gab. 4042, CEP: 31270-901, Brazil b Facultad de Psicologa, Universidad Autnoma de Madrid, 28049, Madrid, Spain

a r t i c l e

i n f o

a b s t r a c t
The present study examines the reliability, factor structure, and criterion-related validity of Hare (2003) Revised Psychopathy Checklist [PCL-R] in a Latin American forensic context. Brazilian male inmates (124) were administered the PCL-R, along with relevant subscales of an inventory of normal personality [Personality Factorial Inventory], an intelligence scale [Standard Progressive Matrices], and a semi-structured interview based in DSM IV. Criminal offence records were obtained from prison les. Several theoretical factorial structures for the PCL-R were tested and compared. A bifactor model with three factors, one general factor reecting the overlap across all items, and two independent subfactors reecting the unique covariation among particular groups of items, shows the better t. Overall scores on the PCL-R correlated with the number of criminal offenses committed by participants. No signicant correlations were found between PCL-R scores and personality or intelligence measures. It was concluded that the PCL-R is a reliable and valid instrument for the assessment of psychopathy in the Brazilian male forensic population. 2008 Elsevier Ltd. All rights reserved.

Article history: Received 5 January 2008 Received in revised form 17 April 2008 Accepted 16 June 2008 Available online 9 August 2008 Keywords: PCL-R Latin American Brazil Psychopathy

1. Introduction Psychopathy is a severe personality disorder characterized by a profound emotional decit accompanied by a lack of regard for societal norms. The best-validated instrument for the assessment of psychopathy is the Psychopathy Checklist-Revised (PCL-R; Hare, 2003). Psychopaths, as dened by the PCL-R, are callous, unemotional, remorseless, and manipulative individuals. Considerable research has been devoted to the study of a two factor model of the structure of the PCL-R. In this model, Factor 1 encompasses the interpersonal and emotional features of psychopathy. High scores on Factor 2 describe an impulsive and irresponsible person, without long-term goals. An alternative three-factor model was proposed by Cooke and Michie (2001), encompassing a select set of 13 PCL-R items. The original Factor 1 was divided into separate affective and interpersonal factors each comprising four items. Five items of the original Factor 2 were retained, reecting an impulsive and irresponsible behavioural style. Various studies have yielded support for this three-factor model in samples of criminal offenders (Patrick, Hicks, Krueger, & Lang, 2005; Weaver, Meyer, Van Nort, & Tristan, 2006). Hare and his colleagues have criticised the three-factor model and the statistical and methodological procedures from which it
* Corresponding author. Tel./fax: +55 31 3499 6277. E-mail addresses: carmenor@uol.com.br, carmencita@fach.ufmg.br (C.E. Flores-Mendoza). 0191-8869/$ - see front matter 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2008.06.016

derives. The author of the PCL-R has recently proposed that four factors are needed to describe the structure of psychopathy (Hare, 2003), representing interpersonal, affective, behavioural, and antisocial features of the disorder. Recent studies also support this four-factor solution (Hare & Neumann, 2006). All of the above described models assume that the factors of the PCL-R represent correlated facets or indicators of the psychopathy construct. In these oblique factor models (see Fig. 1, model 1) the correlation between two items loading in separate facets derives only from their loadings in their respective factors and from the correlation between the implied factors. One problem of the oblique factor models is that it does not indicate the facet or variable variance accounted for by the higher-order psychopathy construct. If there are more than two factors, a higher-order can be applied in which correlations among rst order factors are wholly accounted for by a general second-order factor and the strength of the relation between each lower-order factor and the higher-order factor can be determined (see Fig. 1, model 2). In this model the association between the second-order factor and the observed variables is indirect, derived from their loading in the lower-order factor(s) (Gignac, 2007). Recently, Patrick, Hicks, Nichol, and Krueger (2007) have proposed an alternative structural model of the PCL-R, the bifactor model, which accounts for the covariance among PCL-R items in terms of a general factor reecting the overlap across all items, in addition to separate uncorrelated subfactors reecting the covariation among particular subgroups of items (see Fig. 1, model 3). It

C.E. Flores-Mendoza et al. / Personality and Individual Differences 45 (2008) 584590

585

Fig. 1. Depiction of three-factor multifactor models in the PCL-R.

indicates that the items load directly in broad and specic factors that vary independently. Patrick et al. (2007) found that the bifactor model shows better t than the higher-order model. It suggests that the PCL-R contains subgroups of items that involve distinctive underlying constructs with separated etiologic processes, and which are differentially associated with several criteria and behavioural measures. The discussion regarding the factor structure of the PCL-R generally centres on data sets obtained from developed western coun-

tries. In the South American context, studies have been conducted in just two countries so far, Argentine and Brazil. In the rst, Folino and Hare (2005) analyzed 154 male convicts and they found correlations of 0.268 with the Barrats Motor Impulsiveness Scale and 0.758 with the Historical Clinical Risk-20 total score. In Brazil, Jozef and Silva (1999) investigated a sample of 29 murderers. Using a cut off score of 25 on the PCL-R, they found that 51% of individuals in this sample met criteria for psychopathy. A validity study, undertaken by Morana (2003), with 56 prisoners and 30 individuals

586

C.E. Flores-Mendoza et al. / Personality and Individual Differences 45 (2008) 584590

without criminal history, established a cut off score of 23 for PCL-R in Brazil. The Rorschach test was applied for this proposal. Considering that few Latin American studies have been conducted to date, no information yet exists regarding the factor structure of the PCLR for that context. The main goals of the present study are: 1. Investigate using conrmatory factor analysis (CFA) the t of several alternative factor models in PCL-R data collected from a Brazilian criminal offender sample. 2. Investigate associations between PCL-R psychopathy scores and criterion variables including criminal history, psychopathological, and personality measures in a Brazilian prisoner sample.

2. Method 2.1. Participants One-hundred and twenty four convicted male prisoners, selected from three prisons of the Belo Horizonte (Minas Gerais State), participated voluntarily in the study assessment, following an explanation concerning the projects objectives. There was no payment and participants received consent procedures in both written and oral form. The mean age was 30.19 (SD = 10.22; range = 1877). The majority of participants had an education equivalent to primary school (54%), self reported religion as catholic (56.5%), ethnicity/race as brown (46.8%), marital status as single (46%), and an income of between one to ve minimum wages. With respect to nationality, all but two participants were Brazilian (one participant was Paraguayan and another was Colombian). The criteria for inclusion of a prisoner in the study were: literate, without DSM-IV Axis I mental disorder, convicted (none of them were waiting for trial), and institutional criminal records available. 2.2. Measures and procedures The psychodiagnostic team, composed of three clinical psychologists, was trained in the use of the PCL-R by Christopher Patrick, Ph.D. from the Emotion Science Laboratory in the Department of Psychology at the University of Minnesota, as part of a collaboration between this laboratory and the Laboratory of Assessment of Individual Differences [UFMG]. The measures administered were: 2.2.1. Psychopathy The PCL-R consists of 20 items, which measure behaviour and personality traits of special relevance to psychopathy. Each item is rated from 0 to 2 (0 = does not apply, 1 = somewhat applies, 2 = denitely applies). The total score ranges from 0 to 40. Across 30 cases, intraclass correlations for measuring absolute agreement of each single measure were computed (McGraw & Wong, 1996). The correlations were 0.98, 0.95 and 0.98 for the total, Factor 1, and Factor 2 scores, respectively. With respect to internal consistency of the PCL-R, an alpha reliability index of .83 was found for the 20 items. The correlation between scores on Factor 1 and Factor 2 was 0.369. 2.2.2. DSM-IV psychopathology Number of symptoms of childhood Conduct Disorder and adult Antisocial Personality Disorder were assessed according to DSM-IV criteria on the basis of information collected from the Structured Interview for DSM-IV Personality Disorders (Pfohl, Blum, & Zimmerman, 1997). The scoring for statistical analyses was binary (presence = 1, absent = 0).

2.2.3. Personality traits The Personality Factorial Inventory is a self-report personality test developed by Pasquali, Azevedo, and Ghest (1997) for the Brazilian population. It contains 155 items that involve 17 separate personality facets. The response format for items is a seven-point Likert scale. However for greater comprehension of the prisoners a ve-point Likert scale (1: strongly disagree to 5: strongly agree) was utilised. Also, to limit the time required for personality assessment, 36 items were administered from this inventory to index the following four facets: Aggression, Deference, Dominance, and Persistence. An Alpha reliability index of 0.73 (0.66), 0.64 (0.66), 0.70 (0.78) and 0.76 (0.79) was found for the four scales respectively in this study and for the Brazilian standardization (in parenthesis). The reasons for choosing IFP and its four scales were: a) the nonexistence of other personality measures appropriate for assessment of the Brazilian forensic population, and b) the four scales chosen contained items more appropriate to prisoners context than other scales available. An Oblimin rotated solution for these facets revealed that the Aggression and Dominance items collapsed into one factor. The Persistence and Deference items each loaded on their own factors. The raw scores for these two scales were used along with a composite of the Aggression and Dominance items. 2.2.4. General Intelligence The odd items were applied from Ravens Standard Progressive Matrices (Raven, Raven, & Court, 1993). The Alpha reliability index of 30 items was 0.87. This instrument was selected considering the low educational level of the sample (54% of the participants did not complete the basic educational level). The non-verbal nature of the items avoided the interference of low reading skills or limited general knowledge. 2.2.5. Socio-economic status (SES) SES was assessed through a semi-structured interview that included questions about age, education, ethnicity, religion, number of sons and daughters, occupation, and economic incomes. 2.2.6. Criminal records The criminal records (CR) set the number and type of crimes committed by each inmate. Following the psychological assessment procedures, the CR was veried to conrm the crimes verbalized in the interviews. Type and number of offences committed were quantied. First, each specic type of arrest was scored as binary (presence = 1, absent = 0). There were 25 different reasons for arrest, which were separated into 6 general categories: violent crimes (e.g. homicide, sexual assault), drug related crimes (e.g. drug trafcking, drug use), property crimes (e.g. theft, robbery), fraud (e.g. falsication of public and private documents, falsication of vehicle identication), kidnapping and nally arms carriage, indicating the number of different type of arrests within each category. The total number of crimes was then computed by the addition of all six general variables. The interviews were conducted in a room isolated from the rest of the prison activities. 2.3. Statistical analysis A key limitation of EFA is that no prior model is specied, and thus solutions can vary from dataset to dataset (Hare & Neumann, 2006). A Conrmatory Factor Analysis (CFA) approach is more logical because the question of PCL factorial structure has been intensively studied. In the current study, a conrmatory approach to model estimation was used. The t of the three oblique factor models (2-, 3-, and 4-factor models) was tested which have been

C.E. Flores-Mendoza et al. / Personality and Individual Differences 45 (2008) 584590

587

proposed for representing the structure of PCL-R items. Items loading on each factor in these different models are shown in Table 1. With respect to the four-factor structure, the higher-order model was tested (the higher-order two-rst-order-factor structure is not identied and the t of the higher-order three-rst-order-factor structure is equivalent to the oblique three-factor solution). Finally, the t of the corresponding bifactor models applied at the same set of items (13, 17 and 18 items) was tested. For all the bifactor models, there was one broad factor underlying all variables, as well as other (two, three or four) uncorrelated unique factors underlying each specic group of variables. In this study, the alternative measurement models were estimated using the maximum likelihood method in the LISREL 8.54 program (Joreskog & Sorbom, 2003). Several t indices were considered. The standardised root-mean-square residual (SRMR) and the root-mean-square-error of approximation (RMSEA) are overall t indexes. The SRMR is an index of the average discrepancy between the model-estimated statistics and observed sample statistics, with the RMSEA providing an index of the discrepancy in model t per degree of freedom. In addition, two incremental t indexes were considered: the comparative t index (CFI) and the TuckerLewis index (TLI). Incremental t indices compare the model to a baseline model in which the covariances among all the variables are assumed to be zero. The TLI is moderately corrected for parsimony as it estimates the relative improvement in t per degree of freedom over the baseline model. Model comparisons with the same group of items can be evaluated from a practical signicance test, based on a difference of .01 or greater between TLI estimates, as recommended by Vandenberg and Lance (2000). Pearson correlations were computed to assess relations between PCL-R total and factor scores and criterion measures including criminality, personality and intellectual measures, and number of symptoms of APD and CD. Univariate mean comparisons (t-tests) were used to compare psychopathic (PCL-R total P 30) versus non-psychopathic groups (PCL-R total 6 30) on these different criminal, psychological, and psychopathological variables. Finally, a regression analysis was conducted using psychopathy group as the dependent variable. Variables that differentiated between psychopathic (PCL-R total P 30) and nonpsychopathic groups (PCL-R total 6 30) in the t-tests were included as predictors.

3. Results 3.1. Psychopathy scores The mean PCL-R total score for the sample was 21.32 (SD = 7.14, skewness 0.054, kurtosis 0.269). The mean PCL-R Factor 1 score was 8.12 (SD = 3.68), and for Factor 2 the mean score was 10.57 (SD = 4). The 13.7% of the sample scored above the cutoff point of 30 proposed by Hare (2003) for psychopathy. Table 1 shows mean, standard deviation, skewness and kurtosis for each PCL item in the Brazilian samples. 3.2. Factor structure Fit results for the CFA models are shown in Table 2. Among all the models, the bifactor model with two factors indicates the better t. In this model a small group of items only loads signicantly in the specic factor (items 13 and 16) or in the general factor (items 4, 5 and 19) whereas the remaining items load in both the general and the specic factor. Non-signicant loadings for the PCL items in the bifactor model were removed from the model. Results are shown in Table 3. 3.3. Psychopathy factors and criminal behaviour Offenses for which the participants were convicted included violent crimes, crimes against property, drug related crimes, fraud, kidnapping, and illegal arms carriage. Table 4 shows bivariate correlations between PCL-R scores and each of these criminal categories. Total PCL-R score was signicantly associated with property crimes, fraud, illegal arms carriage, and total number of crimes. PCL-R Factor 1 scores correlated positively with the number of violent crimes, fraud, property crimes, and total number of crimes. Factor 2 was only related to property crimes and to the total number of crimes committed. 3.4. Psychopathy factors and DSM-IV psychopathology PCL-R scores were also correlated with the number of APD and childhood Conduct Disorder symptoms. Results are presented also in Table 4.

Table 1 Hare Psychopathy Checklist-Revised (PCL-R) Two-factor model 1. Impression management 2. Grandiose sense of self-worth 3. Stimulation seeking 4. Pathological lying 5. Manipulation for personal gain 6. Lack of remorse 7. Shallow affect 8. Callous/lacking empathy 9. Parasitic orientation 10. Poor behaviour control 11. Impersonal sexual behaviour 12. Early behaviour problems 13. Lacks goals 14. Impulsivity 15. Irresponsibility 16. Failure to accept responsibility 17. Unstable interpersonal relation 18. Juvenile delinquency 19. Revocation of conditional release 20. Criminal versatility 1 1 2 1 1 1 1 1 2 2 * 2 2 2 2 1 * 2 2 * Three-factor model 1 1 3 1 1 2 2 2 3 * * * 3 3 3 2 * * * * Four-factor model 1 1 3 1 1 2 2 2 3 4 * 4 3 3 3 2 * 4 4 4 Mean 0.81 0.94 1.33 1.00 1.04 1.39 0.78 0.91 1.11 1.22 1.13 1.02 1.27 1.40 1.32 1.25 0.53 1.15 0.82 0.94 SD 0.75 0.72 0.68 0.79 0.78 0.61 0.70 0.71 0.61 0.75 0.73 0.69 0.72 0.60 0.66 0.75 0.73 0.71 0.87 0.80 Skew. 0.32 0.09 0.53 0.00 0.07 0.44 0.34 0.13 0.07 0.38 0.21 0.02 0.45 0.41 0.45 0.45 0.99 0.21 0.36 0.10 Kurtos. 1.15 1.07 0.77 1.38 1.35 0.64 0.94 0.99 0.36 1.13 1.10 0.85 0.98 0.67 0.71 1.10 0.42 0.97 1.61 1.42

In each model tested, the items are included and their descriptive statistics [Brazilian sample]. Note: * it means that item is not considered in the theoretical model.

588

C.E. Flores-Mendoza et al. / Personality and Individual Differences 45 (2008) 584590

Table 2 Goodness of t statistics for alternative factor models of the PCL-R Model 17- Iitems Oblique factor model, 2 factors Bifactor, 2 specic factors 13-Items Oblique factor model, 3 factors Bifactor, 3 specic factors 18-Items 1 Higher-order factor, four lower-order factors Oblique factor model, four factors Bifactor, four specic factors

v2
186.1 134.4 114.1 86.0 278.9 220.0 216.3

df 118 103 62 52 132 128 117

RMSEA 0.069 0.051 0.083 0.073 0.095 0.076 0.083

SRMR 0.092 0.071 0.082 0.073 0.110 0.093 0.100

CFI 0.93 0.97 0.92 0.94 0.89 0.92 0.91

TLI 0.92 0.96 0.90 0.92 0.87 0.90 0.88

Note: N = 124, v2 = chi square goodness of t statistic; df = degrees of freedom; RMSEA = root-mean-square-error of approximation; SRMR = standardized root-mean-square residual; CFI = comparative t index; TLI = TuckerLewis t index.

Table 3 Loadings in the two factor bifactor model PCL items 1. Impression management 2. Grandiose sense of self-worth 3. Stimulation seeking 4. Pathological lying 5. Manipulation for personal gain 6. Lack of remorse 7. Shallow affect 8. Callous/lacking empathy 9. Parasitic orientation 10. Poor behaviour control 12. Early behaviour problems 13. Lacks goals 14. Impulsivity 15. Irresponsibility 16. Failure to accept responsibility 17. Juvenile delinquency 18. Revocation of conditional release Broad Factor 0.37 0.31 0.26 0.86 0.63 0.44 0.38 0.46 0.30 0.33 0.40 * 0.22 0.35 * 0.35 0.48 Specic Factor 1 0.35 0.62 * * 0.37 0.37 0.42 0.47 0.50 0.51 0.27 0.53 0.43 0.40 0.66 * 0.71 Specic Factor 2

late weakly with childhood conduct disorder. Both child and adult antisocial symptoms correlated with property crimes and total number of offenses. 3.5. Psychopathy factors, personality and intelligence No signicant correlation was found between psychopathy factors and the intellectual and personality traits considered by SPM and FPI. 3.6. Group differences between psychopathic and non-psychopathic inmates Recent taxometric analyses indicate the dimensional nature of the construct underlying PCL-R (Guay, Ruscio, Knight, & Hare, 2007). In spite of these results, a cut score of 30 is useful, in a research context, to compare individuals with high levels of psychopathy with persons with lower levels (Bolt, Hare, & Neumann, 2007). For this purpose, using the cutoff point of 30 (Hare, 2003), t-tests between psychopaths and non-psychopaths were conducted. Table 5 shows descriptive and t-scores. Some dependent variables had high levels of skewness and kurtosis. Groups were also compared, using the non-parametric WilcoxonMannWhitney two-sample rank-sum test. Group differences were statistically signicant (p < 0.05) for almost the same variables. Psychopathic inmates tended to commit a larger number of offenses especially against property, fraudulent offenses, and illegal arms carriage. They also displayed a higher number of APD and CD symptoms. No signicant differences were found for SPM and FPI scores.

Note: non-signicant loadings removed from the model.

The number of adult symptoms of APD was strongly correlated with the manifestation of early behavioural problems. Both sets of symptoms correlate with PCL-R scores. Factor 2 scores correlate with CD and APD, whereas Factor 1 seems to be more strongly related to APD, and to a lesser extent with CD. Factor 1 scores corre-

Table 4 Association between PCL-R factors, criminal behaviour, APD, and CD PCL-R F1 PCL-R total PCL-R F1 PCL-R F2 Violent crimes Drug related crimes Property crimes Fraud Kidnapping Arms carriage Total number of crimes APD 0.818** PCL-R F2 0.807** 0.369** Violent crimes 0.128 0.229* 0.068 Drug related crimes 0.052 0.025 0.026 0.007 Property crimes 0.44** 0.25** 0.442** 0.091 0.068 Fraud 0.204* 0.242** 0.097 0.033 0.122 0.229* Kidnapping 0.029 0.018 0.061 0.15 0.003 0.024 0.057 Arms carriage 0.219* 0.171 0.145 0.076 0.129 0.167 0.01 0.003 Total number of crimes 0.45** 0.364** 0.312** 0.319** 0.452** 0.655** 0.523** 0.156 0.422** APD 0.776** 0.602** 0.672** 0.122 0.012 0.400** 0.163 0.037 0.109 0.377** CD 0.522** 0.28** 0.553** 0.013 0.041 0.481** 0.132 0.032 0.014 0.331** 0.702**
* **

Note: APD = Antisocial Personality Disorder; CD = Conduct Disorder. p < 0.05; p < 0.01.

C.E. Flores-Mendoza et al. / Personality and Individual Differences 45 (2008) 584590 Table 5 Mean comparisons in psychopathy and non-psychopath groups Total sample Skewness Violent crimes Drug crimes Property crimes Fraud Kidnapping Arms carriage Total number of crimes APD CD SPM Aggressiveness Perseverance Agreeableness 1.595 0.934 0.662 6.477 4.261 1.864 1.521 0.081 0.513 0.036 0.485 0.405 0.343 Kurtosis 1.656 0.237 0.032 52.330 16.420 1.500 2.605 0.095 0.358 0.808 0.233 0.078 0.373 NPS [n = 107] M 0.27 0.50 0.98 0.1 0.05 0.12 2 5.3 3.9 15.1 46.7 35.7 35.7 SD 0.54 0.64 0.83 0.53 0.23 0.32 1.4 1.5 2.8 5.36 10.31 5.1 4.5 Ps [n = 17] M 0.53 0.53 1.82 0.35 0 0.41 3.6 7.3 6.1 15.8 35.7 34.8 34.9 SD 0.51 0.71 1.1 0.6 0 0.5 1.4 1 2.7 6.9 5.16 5.8 5 1.838 0.144 3.628* 1.77* 0.997 3.11* 4.235* 6.872* 2.974* 0.414 0.239 0.566 0.64 0.016* 964 0.003* 0.003* 0.319 0.003* 0.000* 0.000* 0.004* 0.573 0.573 0.462 0.652 t

589

U-MannWhitney p

Note: *p < 0.05. Persons with scores above 30 were termed as psychopaths, and participants scoring under 30 as non-psychopaths.

Table 6 Logistic regression analysis with the psychopath group (PCL-R P 30 or PCL-R < 30) Wald Property crimes Fraud Total number of crimes APD CD Illegal arms carriage 0.795 0.088 0.063 10.7 0.117 2.261 Signicance 0.373 0.767 0.802 0.001 0.732 0.133 Exp(B) 1.563 1.178 1.1 2.687 0.951 4.186

Note: APD = Antisocial Personality Disorder; CD = Conduct Disorder.

Finally, a regression analysis was conducted with the psychopathy group (psychopathic and non-psychopathic, i.e PCL-R P 30 or PCL-R < 30) as the dependent variable. The objective of this analysis was to determinate the capacity of the measures that emerged as signicantly different in the univariate analysis to differentiate between psychopathic and non-psychopathic inmates. The individual contribution of every variable to the regression equation is indicative of the importance of that measure to the discrimination between high and low PCL-R scorers. Results are shown in Table 6. The overall predictive model was signicant (X2 = 34.809, gl 6, p < 0.05), and overall correct classication based on the model was 88.7%. 4. Discussion Several studies support the validity of the PCL-R psychopathy construct across cultures and ethnic groups (Sullivan & Kosson, 2006). In the present study, some ndings from North American samples were replicated, whereas other results point to possible cultural differences in the Brazilian prison population. In the present study, the mean score was close to the mean reported by Hare (2003) for North American prison inmates (M = 22.1). Using the cutoff point of 30, prevalence of psychopathy in this sample was 13.7%. In a recent review by Sullivan and Kosson (2006), the prevalence of psychopathy among prison inmates ranged from 15% to 38% in studies from different countries. Further research and support is needed for the current Brazilian cut-off point. Conversely, initially no signicant correlations were found between psychopathy and normal personality. Lynam and Derenko (2006) conducted a meta analysis of studies concerning the relation of psychopathy with several structural models of personality. The most signicant effect sizes emerged for the Five-Factor Model. Future research in Brazilian inmates should consider the rela-

tion between this model and PCL-R scores, because regrettably the former (NEO-PI-R) was not available for this current study. In the current sample psychopaths tended to show more prolic criminal careers than their non-psychopathic peers. However, when specic types of crime were considered, PCL-R total scores were not associated with the total number of violent offences. Douglas, Vincent, and Edens (2006) list several factors that mediate the relation between psychopathy and violent recidivism, including age, gender, institutional setting, race and country. It is thus advisable to simultaneously use PCL-R and an appropriate risk assessment scheme such as the Historical Clinical Risk-20 or Sexual Violence Risk-20 in forensic settings. However, signicant relations were found for property and fraudulent crimes. When psychopathy factors were considered, Factor 2 showed higher correlations with property offences, whereas Factor 1 was the only psychopathy facet linked with fraud and violent crimes. Thus, consistent with previous studies, violent behaviour among this sample of psychopaths seemed to be more instrumental than expressive (Woodworth & Porter, 2002). It also appeared that the deceitful and manipulative tendencies of psychopaths were a powerful inuence in the commission of fraudulent crimes. A similar pattern of correlations was reported by Molt, Poy, and Torrubia (2000) for a Spanish offender sample. Levels of APD and CD symptoms were associated with both PCLR Factor 1 and 2. Within our sample, it appears that emotionalinterpersonal features of psychopathy (F1) were related to both adult and early antisocial behaviour, even though prior studies have shown APD symptoms to be more associated with F2 (Patrick et al., 2005). APD is thought to exclude the callous-emotional traits of psychopaths, with more exclusive focus on an impulsive and irresponsible antisocial lifestyle. Nevertheless, Widiger (2006) noted that some personality features present in PCL-R F1 are included in the adult APD criteria set, including deceitfulness and lack of remorse. Factor 2 also includes personality traits like sensation seeking, impulsivity, and irresponsibility. Among Brazilian criminal psychopaths, personality traits included in F1 are clearly correlated with different criminal behaviours and with chronic antisocial lifestyle dened by APD. We conclude that psychopathy is a valid construct in the Brazilian population, and PCL-R a reliable instrument to assess it. Some of our results mirror those of the previous literature. References
Bolt, D. M., Hare, R. D., & Neumann, C. S. (2007). Score metric equivalence of the Psychopathy Checklist-Revised (PCL-R) across criminal offenders in North America and the United Kingdom: A critique of Cooke, Michie, Hart, and Clark (2005) and new analyses. Assessment, 14(1), 4456.

590

C.E. Flores-Mendoza et al. / Personality and Individual Differences 45 (2008) 584590 Morana, H. C. P. (2003). Identicao do ponto de corte para a escala PCL-R em Populao Brasileira Forense: Caracterizao de dois Subtipos da Personalidade; Transtorno Global e Parcial [Identifying the cutoff point for the scale PCL-R in Brazilian Population Forensic: Characterization of two types of personality; disorder Global and Partial] (178 p). So Paulo: Tese de Doutorado, Faculdade de Medicina, Universidade de So Paulo. Pasquali, L., Azevedo, M. M., & Ghest, J. (1997). Inventrio Fatorial de Personalidade [Personality Factorial Inventory]. So Paulo: Casa do Psiclogo. Patrick, C. J., Hicks, B., Krueger, R. F., & Lang, A. R. (2005). Relations between psychopathy facets and externalizing in a criminal offender sample. Journal of Personality Disorders, 19, 339356. Patrick, C. J., Hicks, B. M., Nichol, P. E., & Krueger, R. F. (2007). A bifactor approach to modeling the structure of the Psychopathy Checklist-Revised. Journal of Personality Disorders, 21(2), 118141. Pfohl, B. M. D., Blum, N., & Zimmerman, M. (1997). Structured interview for DSM-IV personality disorders. American Psychiatric Publishing, Inc. Raven, J., Raven, J. C., & Court, H. (1993). Raven manual: Section 1, general overview. Oxford: Oxford Psychologist Press. Sullivan, E. A., & Kosson, D. S. (2006). Ethnic and cultural variations in psychopathy. In C. J. Patrick (Ed.), Handbook of psychopathy (pp. 437458). NY: Guilford. Vandenberg, R. J., & Lance, C. E. (2000). A review and synthesis of the measurement invariance literature: Suggestions, practices, and recommendations for organizational research. Organizational Research Methods, 3, 470. Weaver, C. M., Meyer, R. G., Van Nort, J. J., & Tristan, L. (2006). Two-, three-, and four-factor PCL-R models in applied sex offender risk assessments. Assessment, 13(2), 208216. Widiger, T. A. (2006). Psychopathy and DSM-IV psychopathology. In C. J. Patrick (Ed.), Handbook of psychopathy (pp. 156171). NY: Guilford. Woodworth, M., & Porter, S. (2002). In cold blood: Characteristics of criminal homicides as a function of psychopathy. Journal of Abnormal Psychology, 111(3), 436445.

Cooke, D. J., & Michie, C. (2001). Rening the construct of psychopathy: Towards a hierarchical model. Psychological Assessment, 13, 171188. Douglas, K. S., Vincent, G. M., & Edens, J. F. (2006). Risk for criminal recidivism: The role of psychopathy. In C. J. Patrick (Ed.), Handbook of Psychopathy (pp. 533554). NY: Guilford. Folino, J. O., & Hare, R. D. (2005). Listado revisado para vericacin de la psicopata: Su estandarizacin y validacin en la Argentina [List revised to assess the psychopathy: Standardisation and validation for the Argentina context]. Acta Psiquitrica y Psicolgica de Amrica, 51(2), 94104. Gignac, G. (2007). Multi-factor modeling in individual differences research: Some recommendations and suggestions. Personality and Individual Differences, 42, 3748. Guay, J. P., Ruscio, J., Knight, R. A., & Hare, R. D. (2007). A taxometric analysis of the latent structure of psychopathy: Evidence for dimensionality. Journal of Abnormal Psychology, 116(4), 701716. Hare, R. D. (2003). The Hare Psychopathy Checklist-Revised (2nd ed.). Toronto, ON: Multi-Health Systems. Hare, R. D., & Neumann, C. S. (2006). The PCL-R assessment of psychopathy: Development, structural properties, and new directions. In C. J. Patrick (Ed.). Handbook of Psychopathy (pp. 5888). NY: Guilford. Joreskog, K. G., & Sorbom, D. (2003). LISREL 8.54. Lincolnwood, IL: Scientic Software International. Jozef, F., & Silva, J. A. R. (1999). Homicidio y Psicopata en Brasil. Estudio clnico y neurpsicolgico de un grupo de homicidas en Ro de Janeiro [Homicide and Psychopathy in Brazil. Clinical and neuropsychological study of a group of murderers from Rio de Janeiro]. Revista Espaola de Psiquiatra Forense, Psicologa Forense y Criminologa, 9, 1942. Lynam, D. R., & Derenko, K. J. (2006). Psychopathy and Personality. In C. J. Patrick (Ed.), Handbook of Psychopathy (pp. 133155). NY: Guilford. McGraw, K. O., & Wong, S. P. (1996). Forming inferences about some intraclass correlation coefcients. Psychological Methods, 1, 3046. Molt, J., Poy, R., & Torrubia, R. (2000). Standardization of the Hare Psychopathy Checklist-Revised in a Spanish prison sample. Journal of Personality Disorders, 14, 8496.