A Prospective Study on Self-Reported Psychopathy and Criminal Recidivism Among Incarcerated Male Juvenile Offenders

Abstract

The present study examines the utility of three self-report measures of psychopathic traits in predicting recidivism among a sample of incarcerated male juvenile offenders. Participants (N = 214, M = 16.40 years, SD = 1.29 years) from seven Portuguese juvenile detention centers were followed and prospectively classified as recidivists versus non-recidivists. Area under the curve (AUC) analysis revealed that the Antisocial Process Screening Device–Self-Report (APSD-SR) presented the best performance in terms of predicting general recidivism, with the Youth Psychopathic Traits Inventory (YPI) and the Childhood and Adolescent Taxon Scale–Self-Report (CATS-SR) presenting much poorer results. However, logistic regression models controlling for past frequency of crimes and age of first incarceration found that none of these self-report measures significantly predicted 1- or 3-year recidivism, whether general or violent. Findings suggest there are limitations in terms of the incremental utility of self-report measures of psychopathic traits in predicting recidivism among juveniles.

Keywords

crime juvenile offenders psychopathy recidivism self-report

Introduction

Psychopathy is usually considered a multidimensional construct composed of Affective (e.g., guiltlessness, callousness), Behavioral (e.g., impulsivity, low self-control), and Interpersonal (e.g., conning, manipulation) dimensions manifesting itself as a personality disorder that is associated with antisocial behaviors and lifestyle (see, for example, Blair & Mitchell, 2009; Cooke & Michie, 2001; Hare, 2003; Skeem, Polaschek, Patrick, & Lilienfeld, 2011). Studies have demonstrated that high psychopathy scores are usually correlated with higher rates of both violent and non-violent antisocial behaviors, such as early criminal onset, early contacts with the police, early juvenile court referrals, and aggression (e.g., Andershed, Kerr, Stattin, & Levander, 2002; DeLisi & Piquero, 2011; Farrington, 2005; Forth & Book, 2010; Vaughn, Howard, & DeLisi, 2008).

Although some authors have cautioned against the application of the psychopathy construct to children and adolescents (e.g., Edens, Skeem, Cruise, & Cauffman, 2001; Seagrave & Grisso, 2002), results from studies show that psychopathic traits tend to be at least moderately stable from childhood to adulthood (e.g., Colins, Andershed, Salekin, & Fanti, 2018; Lynam et al., 2009). Indeed, research has shown that delinquent youths have more stable and chronic criminal trajectories when the precursors of psychopathy are present (DeLisi, 2016; Forth & Burke, 1998; Lynam, 1996). It seems increasingly clear that early identification of children and youths with elevated psychopathic traits may prove to be very useful in identifying candidates for early intervention (Bayliss, Miller, & Henderson, 2010; Frick, 1998).

Several instruments have been developed during the last two decades to assess psychopathic traits in adolescents (see, for example, Ribeiro da Silva, Rijo, & Salekin, 2013, for a review). The Psychopathy Checklist: Youth Version (PCL:YV; Forth, Kosson, & Hare, 2003) is considered by many to be the gold standard (Forth & Mailloux, 2000; Murrie & Cornell, 2002). However, self-report measures of psychopathic traits among youth have gained traction among researchers because they are easier to use, require less time to administer, and are less expensive. Presently, the self-report version of the Antisocial Process Screening Device–Self-Report (APSD-SR; Caputo, Frick, & Brodsky, 1999) is the most used and investigated measure of psychopathic traits in adolescents that taps the Affective, Behavioral, and Interpersonal dimensions of psychopathy. Both the APSD (Frick & Hare, 2001) and the PCL:YV are essentially downward extensions of the Psychopathy Checklist–Revised (PCL-R; Hare, 2003) which was created to assess psychopathy among forensic samples of adults.

Another promising instrument is the Youth Psychopathic Traits Inventory (YPI; Andershed et al., 2002) which is based on Cooke and Michie’s (2001) three-dimension (i.e., Affective, Behavioral, and Interpersonal) conceptualization of psychopathy. The authors of the YPI intentionally avoided items that overlapped with behavioral outcomes (e.g., criminal behavior). This measure is commonly used for assessing non-forensic samples of youths (e.g., community samples, school samples) and for research purposes. Another interesting but much less investigated measure is the Childhood and Adolescent Taxon Scale–Self-Report (CATS-SR). This instrument was developed as a proxy measure of the persistent Antisocial Behavior dimension of psychopathy, based on childhood and adolescence risk/history antisocial behavior items (e.g., childhood behavior problem before age 15; arrested under the age of 16), that could also be obtained through reviews of institutional files (Harris, Rice, & Quinsey, 1994). It is important to mention that the CATS does not span the entire construct of psychopathy as it lacks the affective and interpersonal features often considered central to the construct. Some studies argue that the CATS is able to distinguish a taxon (i.e., a class) of individuals who can be designated as having psychopathy (e.g., Harris et al., 1994; Skilling, Quinsey, & Craig, 2001).

Some studies have compared the different measures (e.g., PCL:YV, YPI) and methods (e.g., self-report, interview) available to assess psychopathy among youth offenders based on how well they associate with criminal variables such as frequency of offenses, frequency of arrests, and age of crime onset (e.g., Boccaccini et al., 2007; Cauffman, Kimonis, Dmitrieva, & Monahan, 2009; Douglas, Epstein, & Poythress, 2008; Salekin, 2008; Spain, Douglas, Poythress, & Epstein, 2004). Results tended to show little overlap between the different measures and the different methods of assessment and that the predictive capability of some psychopathy measures usually disappeared or was substantially reduced after controlling for relevant variables such as past criminal offenses and conduct disorder diagnosis. Self-report measures of psychopathy in particular tended to show less utility because of reliability, convergent validity, and malingering problems (Munoz & Frick, 2007; Salekin, Leistico, Neumann, DiCicco, & Duros, 2004; Silva, Genoves, & Latorre, 2012). For example, Asscher et al. (2011) found that clinical ratings of psychopathy had stronger associations with criminal outcomes compared with self-report measures of psychopathy, possibly reflecting the impact of social desirability or other response biases on the latter.

Despite the fact that some limitations do exist in terms of assessment, research corroborates the notion that psychopathic traits in youths can be assessed with an adequate degree of validity and reliability, and that measures of psychopathic traits do have some potential utility in terms of legal decision making with justice-involved youth and clinical decision making (Salekin, 2015). If some psychopathic trait measures, for example, prove to be able to predict criminal behavior after controlling for previous offending among youths, they would be useful in terms of implementing specific therapeutic strategies that would prevent or diminish recidivism rate; if they prove to be useful for research purposes in settings where the number of youth far exceeds the number of assessments that can be completed in a timely way; or if they are related to young people scoring higher on the Needs (or criminogenic) factor of the Risk Need Responsivity model (Andrews & Bonta, 2017) and thus have an higher risk of recidivism they would require more intensive treatment.

Unfortunately, studies that have examined and compared the potential incremental utility of self-report measures of psychopathic traits in prospectively predicting future criminal behavior after controlling for previous criminal history are not abundant (e.g., Colins, Vermeiren, De Bolle, & Broekaert, 2012; Douglas et al., 2008). It is well-known that previous criminal activity is the best predictor of future criminal activity (see, for example, Cottle, Lee, & Heilbrun, 2001; Kennealy, Skeem, Walters, & Camp, 2010). Some authors argue that psychopathy measures should only be used if they add something above and beyond existing risk-assessment measures and variables (Edens, Campbell, & Weir, 2007) due to the serious impact of labeling youth as psychopathic (Blais & Forth, 2013). For example, Colins et al.’s (2012) study with detained male adolescents found that neither general Psychopathy nor the Affective dimension of psychopathy measured by the YPI was predictive of recidivism after controlling for criminal history. However, the Behavioral (i.e., impulsivity) and Interpersonal (i.e., narcissism) dimensions of the YPI significantly predicted substance-related recidivism. Likewise, Douglas et al. (2008) found that psychopathy as measured by the Childhood Psychopathy Scale (CPS; Lynam, 1996) was no longer predictive of recidivism once traditional risk factors (e.g., history of offending, conduct disorder) were taken into account. Thus, the few available studies suggest that the core psychopathic features may do little to enhance the prediction of recidivism among youth.

Another important research question is whether researchers should focus on the general psychopathy construct or on one particular dimension. Recently, researchers investigated the utility of the Callous-Unemotional dimension, arguing that this dimension can be considered the most important and promising dimension in terms of identifying a subgroup of serious and violent youth offenders with persistent patterns of offending and whom have unique etiological, emotional, and cognitive factors associated with their offending (Frick, Ray, Thornton, & Kahn, 2014; Frick & White, 2008). The Callous-Unemotional dimension has gained relevance with the new Limited Prosocial Emotions specifier of the Conduct Disorder diagnosis of the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association [APA], 2013). However, few empirical studies using youth forensic samples have corroborated the importance of the Callous-Unemotional dimension in terms of differentiating severe antisocial youth and predicting future offending behavior (Brandt, Kennedy, Patrick, & Curtin, 1997; Caputo et al., 1999). Thus, there is limited support for favoring the Callous-Unemotional dimension over other dimensions such as the Impulsivity (or Behavioral) dimension (e.g., Cauffman et al., 2009) and the Antisocial dimension (e.g., Vitacco, Neumann, & Caldwell, 2010), with most studies using PCL family instruments. Because of this, new studies investigating the utility of the Callous-Unemotional dimension and comparing it with the utility of other dimensions of the psychopathy construct using different measures are needed, especially those using self-report measures.

Most juvenile recidivism studies reported in the literature were done in North America, with few studies done in Europe. Such studies are especially scarce in southern European countries such as Spain, Greece, or Italy (Loeber & Farrington, 2012; Zara & Farrington, 2016). In Portugal, no prospective juvenile recidivism studies are reported to exist, and we are aware of only one retrospective recidivism study using male juvenile offenders. Pechorro, Braga, Ray, Gonçalves, and Andershed (2019) found a relation between retrospective criminal recidivism and the Impulsivity dimension of the APSD-SR after controlling for age and socioeconomic status (SES). Results also showed that recidivism was associated with alcohol use, but not with drug use or crime seriousness.

Present Study

The question whether self-report measures of psychopathy have added value to predict the risk of reoffending among male juvenile offenders is highly relevant to professionals and researchers working in the judicial and forensic domain. This is the first study that prospectively examines the relationship between self-reported measures of psychopathic-like traits and future offending behavior among Portuguese youth. Because psychopathy seems to be influenced by culture and ethnicity (Edens et al., 2007; Sullivan & Kosson, 2006), our study adds to the literature examining whether self-report psychopathy measures are useful in terms of predicting recidivism among culturally distinct samples from countries were no such studies have been done before. Considering the literature reviewed above, we investigated the following research questions:

Research Question 1: Can self-reported psychopathic-like traits significantly predict recidivism?

Research Question 2: Which self-report measure (APSD, YPI, and CATS) and which dimension (Affective, Behavioral, and Interpersonal) of the psychopathy construct performs better in terms of predicting recidivism, after controlling for past criminal history?

Method

Participants

The sample was composed of 214 male youth (recidivists and non-recidivists; see the “Recidivism Outcomes” section below) recruited from seven nation-wide juvenile detention centers managed by the Portuguese Ministry of Justice. Custody in a juvenile detention center is the strictest measure the Portuguese courts can apply to youth offenders, with a maximum possible sentence of 3 years. In Portugal, the youth justice scheme applies to 12- to 15-year-olds. Offenders can be tried as adults beginning at age 16, but none of the participants were tried as adults. The seven detention centers are considered low to medium security and exclusively admit detainees who were tried for crimes according to Portuguese youth justice law (Lei tutelar-educativa).

The participants’ ages ranged from 12 to 19 years old (M = 16.4, SD = 1.3); some were older than 15 because they were still in custody for charges they incurred as youth. Most (94%) were from an urban background and were Portuguese nationals (84%). The majority were White (56%), followed by Black (19%), mixed race (18%), and other ethnic minorities (6%). Their SES was mostly low (96%), and they had completed an average of less than 7 years of education (M = 6.27 years, SD = 1.44 years). Approximately a quarter (28%) of the participants were taking prescription psychiatric medication. They got involved in crime at an early age (M = 11.3 years, SD = 2.2 years), had their first contact with the law on average before they were 13 years old (M = 12.7 years, SD = 1.9 years), were detained prior to the age of 16 (M = 15.3 years, SD = 1.2 years), and spent an average of 636.1 (SD = 258.1) days in detention until they were released and followed for the present study.

In terms of the seriousness of prior crimes, 78% were in the repeated serious delinquency category, 16% were in the serious delinquency category, and 6% were in the minor to moderate delinquency category (see General Delinquency Seriousness Classification below). Most of the participants (94%) used violence in committing the crimes (see the violent behaviors classification systems below).

With regard to the diversity of criminal history according to the classification system used in Portugal, the participants had committed an average of 3.2 (SD = 4.6, range = 0%-30%, 36% of total crimes) crimes against people, 4.8 (SD = 5.5, range = 0%-43%, 44% of total crimes) crimes against property, no crimes against cultural identity, 0.22 (SD = 0.61, range = 0%-5%, 8.26% of total crimes) crimes against living in society, 0.06 (SD = 0.29, range = 0%-3%, 2.36% of total crimes) crimes against the State, and 0.38 (SD = 1.26, range = 0%-15%, 9.67% of total crimes) diverse detached legislation crimes (see the crime classification system below for examples of these different crime categories).

Instruments

Sociodemographics

A questionnaire was constructed to describe the sociodemographic characteristics of the participants in a standardized way, including questions about participants’ age, ethnic group, urban versus rural origin, years of schooling completed, SES, nationality, taking of psychiatric medication, age of crime onset, first contact with the law, and cannabis use, drug use, and alcohol abuse during the 12 months before detention (these last three coded as 5-point ordinal variables; for example, “How often did you use cannabis in the 12 months before your detention?”). SES was measured by a combination of parental level of education and parental profession (Simões, 2000).

Psychopathy measures

The APSD-SR (Frick & Hare, 2001; Caputo et al., 1999) is a multidimensional 20-item measure designed to assess psychopathic traits in adolescents. Each item is scored on a 3-point ordinal scale ranging from 0 = never to 2 = often. The APSD-SR items comprise three dimensions: Callous-Unemotional dimension, Impulsivity dimension, and Narcissism dimension. The total score, as well the score of each dimension, is obtained by adding the respective items after reverse-scoring the appropriate items. Higher scores indicate higher psychopathic traits. The Portuguese version of the APSD-SR was used (Pechorro, Hidalgo, Nunes, & Jiménez, 2016). Internal consistency in the current sample, estimated by Cronbach’s alpha, was APSD Total = .79, Callous-Unemotional dimension = .65, Impulsivity dimension = .61, and Narcissism dimension = .72.

The YPI (Andershed et al., 2002) is a 50-item self-report measure designed to assess the core personality traits of the psychopathic personality constellation in youth aged 12 years old and above. Each item is scored on an ordinal 4-point Likert-type scale ranging from 0 = does not apply at all to 3 = applies very well. The YPI has 10 subscales with five items each representing Cooke and Michie’s (2001) three-dimensional conceptualization of psychopathy, namely, the Callous-Unemotional dimension (or Affective), the Impulsive-Irresponsible dimension (or Behavioral), and the Grandiose-Manipulative dimension (or Interpersonal). The total score, as well as each dimension score, is obtained by adding the respective items after reverse-scoring the appropriate items. Higher scores reflect an increased presence of the characteristics associated, namely, psychopathic traits. The Portuguese version of the YPI was used (Pechorro, Andershed, Ray, Maroco, & Gonçalves, 2015). Internal consistency in the current sample, estimated by Cronbach’s alpha, was YPI Total = .87, Affective dimension = .70, Behavioral dimension = .77, and Interpersonal dimension = .85.

The Childhood and Adolescent Taxon Scale (CATS; Harris et al., 1994; Quinsey, Harris, Rice, & Cormier, 2006) is an actuarial rating scale developed from eight childhood and adolescent variables, most having to do with antisocial or aggressive behavior (e.g., “Childhood aggression problem,” “Arrested below the age of 16”). It has eight dichotomous items scored either 0 (no) or 1 (yes). Higher scores mean a greater likelihood of life course persistent antisocial behavior associated with psychopathy. Relevant information to score CATS items can be obtained through self-report (CATS-SR) or clinical rating (e.g., using clinical reviews, institutional files). A potential advantage of the CATS is that it can be objectively verified if necessary, which can be very useful in forensic assessments where a respondent might be less willing to admit to socially undesirable characteristics (Bolton, 2006). The Portuguese version of the CATS-SR was used (Pechorro, 2011). Internal consistency in the current sample, estimated by Cronbach’s alpha with items defined as categorical variables, was .71.

Antisocial and criminal history variables

A Conduct Disorder scale was also created using the 15 dichotomous criteria that assess conduct disorder (see, for example, Skilling et al., 2001). The criteria (0 = no, 1 = yes) were added to obtain a total continuous scale score, with higher scores indicating a higher number of symptoms. Internal consistency in the current sample was α = .77.

A version of the General Delinquency Seriousness Classification (GDSC; Loeber, Farrington, Stouthamer-Loeber, & van Kammen, 1998) was used to classify official crime reports: Level 0 consisted of no delinquency; Level 1 consisted of minor to moderate delinquency, including shoplifting, minor vandalism, stealing, minor drug dealing, minor fraud, pickpocketing, and carrying weapons; Level 2 consisted of serious delinquency, including assault, gang fighting, murder, rape, robbery, major drug dealing, carjacking, and arson; Level 3 consisted of repeated serious delinquency, including two or more serious Level 2 offenses. The first author scored the GDSC.

Violent behaviors were dichotomously classified (0 = absent, 1 = present). The following examples of violent behaviors were considered: gang fighting, strong-arming someone, armed robbery, and assault to hurt or kill (Loeber et al., 1998). The act of carrying a weapon (e.g., pocket knife) was not considered a violent behavior per se in the present classification because participants may carry it for protection, not for using it in proactively committing crimes.

The diversity of crimes was classified according to the six main categories specified by Portuguese legislation and used by the Portuguese Ministry of Justice and the police, namely, (a) crimes against people (e.g., assault, homicide), (b) crimes against property (e.g., stealing, shoplifting; it also includes robbery), (c) crimes against cultural identity (e.g., religious discrimination, racism), (d) crimes against living in society (e.g., inflicting harm to animals, forcing children to beg for money), (e) crimes against the State (e.g., corruption, money laundering), and (f) diverse detached legislation crimes (e.g., driving without a license, growing cannabis). The diversity of crimes variable can also be used as an ordinal variable in which individuals either had each category (=1) or did not (=0), and the values are then summed to compute a variety index (ordinal frequency of crime categories). Frequency of crimes was defined as the number of crimes committed, specified in the previously mentioned categories. Crime onset refers to the age in which the participants started to engage in crimes.

Recidivism outcomes

Official criminal data were supplied by the General Directorate of Rehabilitation and Prison Services (DGRSP) of the Portuguese Ministry of Justice. Participants with at least one new crime charge that lead to a new intervention by the DGRSP after they were assessed by us and released from the detention center were considered recidivists, whereas those who had no new charge during the 1- and 3-year follow-up periods were considered non-recidivists for those follow-up periods, respectively. Analyses reported below are for general and violent recidivism after 1 year and after 3 years. Violent recidivism was coded according to the violent behaviors classification described above. A 1-year follow-up was included because most studies tend to report 1-year recidivism rates, while a 3-year follow-up was also used because it is usually considered the minimum number of years acceptable in examining a longitudinal tendency (Virginia Department of Juvenile Justice, 2005).

Procedure

Authorization to assess youths was previously obtained from the ethics committee of the DGRSP. Male detainees from the Portuguese Juvenile Detention Centers were asked to voluntarily participate after being informed about the aim of the study. Informed consent was used. Only those who were detained for at least 2 months were approached due to the fact that youths can be temporarily committed for assessment purposes, with no subsequent charges. The participation rate was approximately 92%. Of the initial sample, 30 participants were later excluded, primarily because no official recidivism data were available (i.e., missing data), they could not be followed for the entire 3-year period (e.g., due to the length of their incarceration), or they were incarcerated in adult prisons (e.g., due to more recent changes). Information was obtained from multiple sources including official institutional files and self-reports (e.g., age of criminal activity onset). The youth were told that the data they provided were confidential, for research purposes only, and would not affect their treatment in any way. The questionnaires were administered individually. Only males were included in the present study due to the small number of incarcerated female youths in Portugal, and to limit the potential confounding effect of gender differences on associations with correlates. The participants were assessed in 2013 and 2014. Official recidivism data were supplied by the DGRSP taking into consideration the time interval (i.e., 1 year, 2 years, and 3 years) after each participant’s release from the detention centers (i.e., the follow-up upon release for each individual). Due to potential administrative delays in the processing of the recidivism data at a national level, an extra 6 months were used to guard against these delays.

Analytic Plan

The data were analyzed using SPSS v25 (IBM Corp., 2017). Areas under the curve (AUCs) were used to assess and compare the validity of psychopathy measures with recidivism outcomes. Binary logistic regression was used to test unique associations between the predictor variables and the dependent variable (recidivism status). The first block of each binary logistic regression model was used to control for retrospective crime frequency obtained from official records; the second block included the self-reported psychopathic traits measures. Four logistic regression models were used to predict four outcomes: general and violent recidivism after 1 year and after 3 years. We adjusted for the number of significance tests by using an alpha of .01. ANOVAs, exact Wilcoxon, McNemar, Mann–Whitney, and chi-square tests were used to compare groups according to the characteristics of the variables (scale, ordinal or nominal), and the characteristics of the subsamples (paired or independent). Pearson correlations (r) and partial correlations were used to analyze associations between scale variables, and effect sizes were also provided (Leech, Barrett, & Morgan, 2015).

Results

The analysis of the sociodemographic variables revealed no significant differences between non-recidivists (n = 116; 54.2%) and recidivists (n = 98; 45.8%) in terms of age, ethnic group, urban versus rural origin, years of schooling completed, SES, nationality, or taking of psychiatric medication. However, the recidivists reported having higher rates of alcohol use and cannabis use. In terms of the criminal variables, significant differences were found with regard to age of crime onset, age of first problem with the law, age of first entry into a juvenile detention center, retrospective frequency of crimes, and conduct disorder symptoms, with recidivists starting their criminal activity, having their first problem with the law, and being detained at an earlier age; they also present a higher frequency of crimes and conduct disorder symptoms. However, no differences were found in terms of average days in detention, crime seriousness and violent behaviors, or cocaine/heroin use (see Table 1).

Table 1.

Comparisons of Sociodemographic and Criminal Variables Between Overall Non-Recidivists and Recidivists.

	Non-recidivists (n = 116)	Recidivists (n = 98)	Test statistic (p value)	Effect size
Age	16.49 (1.28)	16.29 (1.29)	F = 1.365, p = .24	$η_{p}^{2}$ = .01
Socioeconomic status	95.14	98.08	U = 4,443, p = .70	r = .02
White ethnicity	53.8%	59.4%	χ² = 6.55, p = .49	Φ = .06
Psychiatric medications	22.2%	34.4%	χ² = 3.889, p = .06	Φ = .14
Urban origin	94.9%	92.7%	χ² = 0.451, p = .57	Φ = .05
Nationality	85.5%	82.3%	χ² = 3.375, p = .54	Φ = .13
Education (years)	6.33 (1.55)	6.20 (1.30)	F = 0.465, p = .50	$η_{p}^{2}$ = .002
Alcohol use	2.28 (1.34)	2.69 (1.24)	U = 4,543.5, p = .01	r = .16
Cannabis use	2.28 (1.65)	2.70 (1.61)	U = 4,670, p = .02	r = .16
Cocaine/heroin use	0.26 (0.77)	0.45 (1.01)	U = 5,206, p = .15	r = .09
Age of crime onset	11.52 (2.24)	10.94 (1.98)	F = 3.956, p = .04	$η_{p}^{2}$ = .02
Age of first law contact	12.97 (1.98)	12.43 (1.71)	F = 4.403, p = .03	$η_{p}^{2}$ = .02
Age of first detention	15.56 (1.22)	15.05 (1.23)	F = 8.88, p = .003	$η_{p}^{2}$ = .04
Crime frequency	6.18 (4.21)	10.90 (10.02)	F = 21.3124, p < .001	$η_{p}^{2}$ = .09
Days in detention	655.09 (278.56)	612.88 (229.96)	F = 1.414, p = .24	$η_{p}^{2}$ = .01
Crime seriousness	2.68 (0.60)	2.77 (0.52)	U = 5,263, p = .28	r = .07
Violent behaviors	92.3%	95.8%	χ² = 1.144, p = .39	Φ = .07
CD symptoms (out of 15)	6.57 (3.34)	8.11 (2.99)	F = 12.325, p < .001	$η_{p}^{2}$ = .06

Note. F = ANOVA statistic; U = Mann–Whitney statistic; χ² = chi-square statistic; CD symptoms = Conduct Disorder symptoms; $η_{p}^{2}$ = partial eta squared; Φ = phi.

Table 2 displays the correlations between the main variables used in the present investigation, including the self-report psychopathy measures and criminal history variables. For the most part, the intercorrelations among the criminal variables were in the expected directions; however, age of crime onset was unrelated to frequency, serious, and diversity. The intercorrelations among the subscales of the YPI and APSD were generally significant and positive, with the conceptually overlapping subscales tending to show the strongest correlations. The APSD Callous-Unemotional dimension was an exception because it ought to have shown preferential associations with the affective factor of the YPI, but demonstrated similar correlation magnitudes with each factor. Also, it was the only dimension of psychopathy that was unrelated to the CATS-SR. Interestingly, psychopathic traits were most consistently associated with age of first crime and were less related to other variables (e.g., retrospective crime frequency, crime seriousness).

Table 2.

Pearson Correlation Matrix of the Main Variables (Recidivism Outcomes Not Included).

	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. Crime frequency	1
2. Crime diversity	.42**	1
3. Crime seriousness	.32**	.23*	1
4. Age of crime onset	−.03	−.04	−.02	1
5. Age of first incarceration	−.19**	.06	−.09	.18*	1
6. APSD-SR Total	.08	.09	.08	−.20*	−.15	1
7. APSD CU	.11	.03	.08	−.10	−.17	.61**	1
8. APSD Impulsivity	.08	.04	.06	−.19*	−.12	.77**	.28**	1
9. APSD Narcissism	.02	.10	.04	−.10	−.07	.82**	.19*	.53**	1
10. YPI Total	.10	.16*	−.01	−.22*	−.06	.63**	.28**	.45**	.56**	1
11. YPI Affective	.11	.20**	.03	−.15	−.04	.44**	.26**	.27**	.39**	.81**	1
12. YPI Behavioral	.15*	.09	.05	−.24**	−.08	.51**	.26**	.49**	.35**	.75**	.45**	1
13. YPI Interpersonal	.03	.14*	−.05	−.16	−.03	.58**	.21*	.37**	.59**	.90**	.65**	.46**	1
14. CATS-SR Total	.08	.09	.02	−.27**	−.03	.23*	.06	.26**	.18*	.31**	.18*	.42**	.19*	1

Note. APSD-SR = Antisocial Process Screening Device–Self-Report; CU = Callous-Unemotional; YPI = Youth Psychopathic Traits Inventory; CATS-SR = Childhood and Adolescent Taxon Scale–Self-Report.

p < .01. **p < .001.

In the total sample, general recidivism values were 37.9% for the first year, 18.2% for the second year, and 7.5% for the third year. Violent recidivism values were 20.3% for the first year, 8.4% for the second year, and 2% for the third year. It is important to mention that these recidivism rates are not cumulative (i.e., recidivism rates were calculated year by year). We compared general recidivism proportion per year using McNemar tests. In the total sample, general recidivism dropped from the first year to the second year (p < .001) and from the second year to the third year (p < .001), and significant results were also obtained when comparing the first year with the third year (p < .001). Violent recidivism dropped from the first year to the second year (p < .001) and from the second year to the third year (p < .001), and significant results were also obtained when comparing the first year with the third year (p = .001). Finally, we also compared prior and subsequent (i.e., previous and post-release) crime seriousness and violent criminal behaviors among the recidivists to determine if any changes occurred. Significant differences were found indicating lower crime seriousness (Wilcoxon Z = −11.607, p < .001) and violent criminal behaviors (McNemar χ² = 135.329, p < .001) in new offenses after the release from the juvenile detention centers.

Table 3 shows the AUCs pertaining to general recidivism and violent recidivism. In terms of benchmarks, AUC values ≥ .56 are considered small effect sizes, AUC values ≥ .64 are considered medium effect sizes, and AUC values ≥ .71 are considered large effect sizes (Rice & Harris, 2005). Our results show that none of the measures reached large effect sizes, but the APSD-SR Total reached a medium effect size when considering the 3-year general recidivism. The APSD-SR Total significantly predicted 1-year general recidivism, and the APSD-SR and the YPI Total significantly predicted 3-year general recidivism. The YPI Interpersonal dimension predicted 1-year and both the YPI Behavioral dimension and the APSD-SR Callous-Unemotional dimension predicted 3-year general recidivism. Interestingly, none of the measures and their respective dimensions were able to significantly predict violent recidivism at p < .01.

Table 3.

Predictive Validities of Psychopathy Measures With Recidivism Outcomes.

AUCs	General recidivism (95% CI)	Violent recidivism (95% CI)
1-year recidivism
APSD-SR Total	.62* [0.55, 0.70]	.61 [0.52, 0.70]
YPI Total	.59 [0.51, 0.67]	.57 [0.48, 0.68]
CATS-SR Total	.60 [0.52, 0.67]	.57 [0.47, 0.66]
APSD-SR CU	.60 [0.52, 0.68]	.60 [0.51, 0.69]
APSD-SR Impulsivity	.61 [0.53, 0.69]	.55 [0.45, 0.64]
APSD-SR Narcissism	.55 [0.47, 0.63]	.55 [0.46, 0.64]
YPI Affective	.55 [0.47, 0.63]	.55 [0.45, 0.65]
YPI Behavioral	.63* [0.55, 0.70]	.56 [0.46, 0.66]
YPI Interpersonal	.55 [0.47, 0.63]	.57 [0.48, 0.67]
3-year recidivism
APSD-SR Total	.64* [0.56, 0.71]	.61 [0.53, 0.70]
YPI Total	.62* [0.54, 0.70]	.59 [0.50, 0.68]
CATS-SR Total	.58 [0.50, 0.65]	.56 [0.47, 0.65]
APSD-SR CU	.62* [0.55, 0.70]	.59 [0.50, 0.67]
APSD-SR Impulsivity	.61 [0.54, 0.69]	.54 [0.45, 0.62]
APSD-SR Narcissism	.56 [0.49, 0.64]	.57 [0.48, 0.65]
YPI Affective	.55 [0.47, 0.63]	.54 [0.44, 0.64]
YPI Behavioral	.63* [0.55, 0.71]	.55 [0.46, 0.64]
YPI Interpersonal	.59 [0.51, 0.67]	.60 [0.51, 0.69]

Note. AUC = area under the curve; CI = confidence interval; APSD-SR = Antisocial Process Screening Device–Self-Report; CU = Callous-Unemotional dimension; YPI = Youth Psychopathic Traits Inventory; CATS = Childhood and Adolescent Taxon Scale–Self-Report.

p < .01.

Table 4 presents four hierarchical binary logistic regression models predicting the four recidivism outcomes (i.e., 1-year general recidivism, 3-year general recidivism, 1-year violent recidivism, and 3-year violent recidivism) controlling for past frequency of crimes and age of first incarceration in the first step, and then entering the APSD-SR, YPI, and CATS-SR as covariates in the second step. We did not conduct the corresponding analyses for second-year recidivism because the majority of new offenses occurred during the first year after release and the 3-year perspective is more interesting in terms of interpreting a trend (see the “Recidivism Outcomes” section above). We did not enter sociodemographic characteristics first in this analysis because there were no significant differences in our univariate analysis. None of the self-report measures significantly predicted recidivism at p < .01 after controlling for retrospective crime frequency and age of first incarceration.

Table 4.

Logistic Regression Coefficients of the APSD-SR, YPI, and CATS-SR Predicting Recidivism.

	B (SE)	Wald	p value	Exp(B)	95% CI Exp(B)
1-year general recidivism
First step
Crime frequency	0.053 (0.022)	5.842	.016	1.055	[1.010, 1.101]
Age of first incarceration	−0.333 (0.124)	7.277	.007	0.717	[0.563, 0.913]
Constant	4.144 (1.907)	4.721	.030	63.041
Second step
Crime frequency	0.049 (0.022)	5.074	.024	1.051	[1.006, 1.097]
Age of first incarceration	−0.322 (0.127)	6.369	.012	0.725	[0.565, 0.931]
APSD-SR	0.050 (0.039)	1.628	.202	1.051	[0.974, 1.135]
YPI	0.000 (0.010)	0.001	.973	1.000	[0.980, 1.021]
CATS-SR	0.263 (0.128)	4.261	.039	1.301	[1.013, 1.671]
Constant	1.552 (2.128)	0.532	.466	4.720
3-year general recidivism
First step
Crime frequency	0.110 (0.028)	15.220	<.001	1.117	[1.056, 1.180]
Age of first incarceration	−0.284 (0.124)	5.218	.022	0.753	[0.590, 0.961]
Constant	3.306 (1.922)	2.960	.085	27.283
Second step
Crime frequency	0.104 (0.028)	13.835	<.001	1.109	[1.050, 1.172]
Age of first incarceration	−0.265 (0.128)	4.296	.038	0.767	[0.597, 0.986]
APSD-SR	0.049 (0.039)	1.601	.206	1.050	[0.974, 1.133]
YPI	0.007 (0.010)	0.461	.497	1.007	[0.987, 1.027]
CATS-SR	0.125 (0.120)	1.084	.298	1.133	[0.896, 1.433]
Constant	1.031 (2.132)	0.234	.629	2.803
1-year violent recidivism
First step
Crime frequency	0.024 (0.021)	1.342	.247	1.025	[0.983, 1.068]
Age of first incarceration	−0.474 (0.152)	9.772	.002	0.622	[0.462, 0.838]
Constant	5.581 (2.310)	5.836	.016	265.387
Second step
Crime frequency	0.023 (0.021)	1.144	.285	1.023	[0.981, 1.067]
Age of first incarceration	−0.465 (0.155)	8.958	.003	0.628	[0.463, 0.852]
APSD-SR	0.017 (0.048)	0.134	.714	1.018	[0.927, 1.117]
YPI	0.004 (0.013)	0.106	.745	1.004	[0.979, 1.030]
CATS-SR	0.212 (0.156)	1.834	.176	1.236	[0.910, 1.679]
Constant	3.569 (2.614)	1.863	.172	35.468
3-year violent recidivism
First step
Crime frequency	0.017 (0.020)	0.675	.411	1.017	[0.977, 1.058]
Age of first incarceration	−0.339 (0.140)	5.890	.015	0.713	[0.542, 0.937]
Constant	3.808 (2.149)	3.139	.076	45.053
Second step
Crime frequency	0.014 (0.020)	0.502	.479	1.015	[0.975, 1.056]
Age of first incarceration	−0.324 (0.143)	5.155	.023	0.724	[0.547, 0.957]
APSD-SR	0.023 (0.045)	0.256	.613	1.023	[0.937, 1.116]
YPI	0.006 (0.012)	0.281	.596	1.006	[0.983, 1.030]
CATS-SR	0.148 (0.142)	1.082	.298	1.159	[0.878, 1.531]
Constant	1.878 (2.425)	0.600	.439	6.543

Note. APSD-SR = Antisocial Process Screening Device–Self-Report; YPI = Youth Psychopathic Traits Inventory; CATS-SR = Childhood and Adolescent Taxon Scale–Self-Report; CI = confidence interval; crime frequency = retrospective number of official crimes.

Table 5 displays four hierarchical binary logistic regression models predicting the four recidivism outcomes (i.e., 1-year general recidivism, 3-year general recidivism, 1-year violent recidivism, and 3-year violent recidivism) controlling for past frequency of crimes and age of first incarceration in the first step, and then entering the APSD-SR Callous-Unemotional dimension, the APSD-SR Impulsivity dimension, and the APSD-SR Narcissism dimension in the second step. None of the dimensions of the APSD-SR were statistically significant.

Table 5.

Logistic Regression Coefficients of Dimensions of the APSD-SR Predicting Recidivism.

	B (SE)	Wald	p value	Exp(B)	95% CI Exp(B)
1-year general recidivism
First step
Crime frequency	0.053 (0.022)	5.842	.016	1.055	[1.010, 1.101]
Age of first incarceration	−0.333 (0.124)	7.277	.007	0.717	[0.563, 0.913]
Constant	4.144 (1.907)	4.721	.030	63.041
Second step
Crime frequency	0.051 (0.022)	5.059	.024	1.052	[1.007, 1.099]
Age of first incarceration	−0.293 (0.128)	5.275	.022	0.746	[0.581, 0.958]
Callous-Unemotional	0.133 (0.080)	2.774	.096	1.143	[0.977, 1.337]
Impulsivity	0.202 (0.112)	3.291	.070	1.224	[0.984, 1.524]
Narcissism	−0.056 (0.071)	0.616	.433	0.946	[0.823, 1.087]
Constant	2.181 (2.090)	1.089	.297	8.854
3-year general recidivism
First step
Crime frequency	0.110 (0.028)	15.220	<.001	1.117	[1.056, 1.180]
Age of first incarceration	−0.284 (0.124)	5.218	.022	0.753	[0.590, 0.961]
Constant	3.306 (1.922)	2.960	.085	27.283
Second step
Crime frequency	0.109 (0.029)	14.302	<.001	1.115	[1.054, 1.180]
Age of first incarceration	−0.241 (0.129)	3.523	.061	0.786	[0.611, 1.011]
Callous-Unemotional	0.153 (0.081)	3.575	.059	1.165	[0.994, 1.365]
Impulsivity	0.182 (0.112)	2.635	.105	1.200	[0.963, 1.495]
Narcissism	−0.036 (0.071)	0.255	.614	0.965	[0.839, 1.109]
Constant	1.215 (2.105)	0.333	.564	3.370
1-year violent recidivism
First step
Crime frequency	0.024 (0.021)	1.342	.247	1.025	[0.983, 1.068]
Age of first incarceration	−0.474 (0.152)	9.772	.002	0.622	[0.462, 0.838]
Constant	5.581 (2.310)	5.836	.016	265.387
Second step
Crime frequency	0.024 (0.022)	1.191	.275	1.024	[0.981, 1.068]
Age of first incarceration	−0.445 (0.155)	8.280	.004	0.641	[0.473, 0.868]
Callous-Unemotional	0.079 (0.096)	0.676	.411	1.083	[0.896, 1.308]
Impulsivity	0.028 (0.135)	0.043	.836	1.029	[0.789, 1.341]
Narcissism	0.020 (0.086)	0.056	.813	1.020	[0.862, 1.208]
Constant	4.509 (2.537)	3.159	.075	90.822
3-year violent recidivism
First step
Crime frequency	0.017 (0.020)	0.675	.411	1.017	[0.977, 1.058]
Age of first incarceration	−0.339 (0.140)	5.890	.015	0.713	[0.542, 0.937]
Constant	3.808 (2.149)	3.139	.076	45.053
Second step
Crime frequency	0.016 (0.020)	0.608	.435	1.016	[0.976, 1.058]
Age of first incarceration	−0.309 (0.142)	4.740	.029	0.734	[0.555, .970]
Callous-Unemotional	0.071 (0.091)	0.615	.455	1.074	[0.899, 1.284]
Impulsivity	−0.003 (0.128)	0.001	.980	0.997	[0.776, 1.281]
Narcissism	0.058 (0.080)	0.528	.467	1.060	[0.906, 1.240]
Constant	2.756 (2.361)	1.362	.243	15.734

Note. APSD-SR = Antisocial Process Screening Device–Self-Report; CI = confidence interval; crime frequency = retrospective number of official crimes.

Table 6 presents four hierarchical binary logistic regression models predicting the four recidivism outcomes (i.e., 1-year general recidivism, 3-year general recidivism, 1-year violent recidivism, and 3-year violent recidivism) controlling for past frequency of crimes and age of first incarceration in the first step, and then entering the YPI Affective (or Callous-Unemotional), the Behavioral (or Impulsive-Irresponsible), and the Interpersonal (or Grandiose-Manipulative) dimension in the second step. None of the dimensions of the YPI significantly predicted recidivism at p < .01 after controlling for retrospective crime frequency and age of first incarceration. It is also worth mentioning that crime frequency tended to be a predictor of general recidivism but not of violent recidivism.

Table 6.

Logistic Regression Coefficients of Dimensions of the YPI Predicting Recidivism.

	B (SE)	Wald	p value	Exp(B)	95% CI Exp(B)
1-year general recidivism
First step
Crime frequency	0.053 (0.022)	5.842	.016	1.055	[1.010, 1.101]
Age of first incarceration	−0.333 (0.124)	7.277	.007	0.717	[0.563, 0.913]
Constant	4.144 (1.907)	4.721	.030	63.041
Second step
Crime frequency	0.048 (0.022)	4.547	.033	1.049	[1.004, 1.096]
Age of first incarceration	−0.327 (0.125)	6.831	.009	0.721	[0.564, .921]
Affective	0.003 (0.034)	0.010	.921	1.003	[0.939, 1.072]
Behavioral	0.051 (0.025)	4.196	.041	1.053	[1.002, 1.106]
Interpersonal	−0.002 (0.019)	.007	.931	0.998	[0.962, 1.036]
Constant	2.610 (2.039)	1.639	.200	13.594
3-year general recidivism
First step
Crime frequency	0.110 (0.028)	15.220	<.001	1.117	[1.056, 1.180]
Age of first incarceration	−0.284 (0.124)	5.218	.022	0.753	[0.590, 0.961]
Constant	3.306 (1.922)	2.960	.085	27.283
Second step
Crime frequency	0.108 (0.029)	14.133	<.001	1.115	[1.053, 1.179]
Age of first incarceration	−0.266 (0.126)	4.451	.035	0.766	[0.598, 0.981]
Affective	−0.038 (0.035)	1.166	.280	0.963	[0.900, 1.031]
Behavioral	0.055 (0.026)	4.560	.033	1.057	[1.005, 1.111]
Interpersonal	0.025 (0.019)	1.647	.199	1.025	[0.987, 1.064]
Constant	1.714 (2.086)	0.675	.411	5.552
1-year violent recidivism
First step
Crime frequency	0.024 (0.021)	1.342	.247	1.025	[0.983, 1.068]
Age of first incarceration	−0.474 (0.152)	9.772	.002	0.622	[0.462, 0.838]
Constant	5.581 (2.310)	5.836	.016	265.387
Second step
Crime frequency	0.026 (0.022)	1.406	.236	1.026	[0.983, 1.071]
Age of first incarceration	−0.470 (0.152)	9.611	.002	0.625	[0.464, .841]
Affective	−0.008 (0.041)	0.042	.837	0.992	[0.915, 1.075]
Behavioral	−0.003 (0.030)	0.009	.926	0.997	[0.940, 1.058]
Interpersonal	0.028 (0.023)	1.466	.226	1.028	[0.983, 1.076]
Constant	5.115 (2.455)	4.341	.037	166.493
3-year violent recidivism
First step
Crime frequency	0.017 (0.020)	0.675	.411	1.017	[0.977, 1.058]
Age of first incarceration	−0.339 (0.140)	5.890	.015	0.713	[0.542, 0.937]
Constant	3.808 (2.149)	3.139	.076	45.053
Second step
Crime frequency	0.020 (0.021)	0.886	.347	1.020	[0.979, 1.062]
Age of first incarceration	−0.336 (0.140)	5.780	.016	0.714	[0.543, 0.940]
Affective	−0.034 (0.039)	0.754	.385	0.967	[0.896, 1.043]
Behavioral	−0.006 (0.029)	0.046	.830	0.994	[0.940, 1.051]
Interpersonal	0.046 (0.022)	4.217	.040	1.047	[1.002, 1.093]
Constant	3.557 (2.311)	2.369	.124	35.043

Note. YPI = Youth Psychopathic Traits Inventory; CI = confidence interval; crime frequency = retrospective number of official crimes.

Discussion

Addressing our first research question, the APSD-SR Total score consistently predicted 1-year and 3-year general recidivism in the receiver operating characteristic (ROC) analyses, while the YPI Total score only predicted 3-year general recidivism. Interestingly, the CATS-SR Total score did not predict general or violent recidivism which is somewhat surprising given that it includes only items about prior offending. In terms of scale dimensions, the Behavioral dimension of the YPI consistently predicted 1-year and 3-year general recidivism while the Callous-Unemotional dimension of the APSD-SR only predicted 3-year general recidivism. None of the measures or their dimensions were able to significantly predict violent recidivism. Based on these findings, the APSD-SR seems to be the most useful measure in terms of predicting general recidivism among detained adolescent youth on its own.

However, addressing our second research question, none of these self-report measures significantly predicted recidivism among delinquent Portuguese youth after taking age at first incarceration and criminal history into account in regression analyses. Not even the Behavioral/Lifestyle dimension of psychopathy was significant, though it has been shown to be the strongest predictor dimension in some studies (e.g., Colins et al., 2012; Vincent, Odgers, McCormick, & Corrado, 2008). Results do not support the notion that the Callous-Unemotional dimension of the APSD-SR is especially relevant in identifying serious youth offenders and in predicting recidivism (Frick & White, 2008). Other studies have also put in to question the utility of the Callous-Unemotional traits after controlling for criminal history (e.g., Lahey, 2014; Pechorro, Nunes, Jiménez, & Hidalgo, 2015). This may also prove to be true from a clinical point of view because the new Limited Prosocial Emotions specifier of the Conduct Disorder diagnosis (APA, 2013) was partly derived from the items of the Callous-Unemotional dimension of the APSD. Interestingly, the crime frequency covariate tended to be a predictor of general recidivism but not of violent recidivism. This can possibly be attributed to the fact that being incarcerated at an earlier age is predictive of a more violent pattern of offending due to the nature of the early behavior whereas frequency does not capture the level of severity.

Our study has some strengths and limitation that must be mentioned. Strengths include the prospective design with a 3-year follow-up, the use of the retrospective frequency of crimes variable as a co-variable in the regression models, the simultaneous use of three self-report measures that tap the psychopathy construct for comparison purposes, and the official crime data provided by the Portuguese Ministry of Justice. Also, this was the first study of its kind among Portuguese youth in detention, so this study helps to fill a gap in the literature by examining the relation between recidivism and psychopathic traits among youths from a southern-European country (Zara & Farrington, 2016).

In terms of the limitations of our study, the absence of females in our sample did not allow us to examine gender effects, the sample size did not allow us to control for more variables that might influence the association between psychopathy and recidivism, the low internal consistency of some of the dimensions of the APSD-SR constrained their predictive potential, we did not have a social desirability measure to control for impression management, and we did not have a self-report delinquency measure that could complement the official crime data. The absence of the self-report delinquency measure is particularly relevant because it is known that minor offenses are usually underreported in official data and that serious offending is often underreported in self-report studies (Babinski, Hartsough, & Lambert, 2001; Hoeve et al., 2008). Another limitation is that recidivism was not recorded cumulatively, but year by year instead. Time at risk was not available for analysis, so we could not control for differences in time at risk and our analyses did not account for censoring, as when someone who reoffended in Year 1 might receive a custodial sentence and thus not be at risk in the second or even third year of follow-up. A final limitation is that using the CATS-SR from a dimensional perspective can be questionable because this instrument was originally conceived from a taxonomic approach.

We found that self-report psychopathy measures did not incrementally explain criminal recidivism among our sample of detained male youths. This is consistent with some previous studies using youth and adult samples (e.g., Colins, van Damme, Andershed, Fanti, & DeLisi, 2017; Colins, Vermeiren, De Bolle, & Broekaert, 2012; Rock, Sellbom, Ben-Porath, & Salekin, 2013). Self-report measures may have some relevance if used with caution as preliminary screening devices (e.g., Silva et al., 2012), but they did not predict recidivism after taking aspects of criminal history (crime frequency and age at first incarceration) into account. Given that criminal justice systems routinely have access to criminal history information, self-reported psychopathy may not provide additional value in terms of predicting recidivism in these settings (Asscher et al., 2011; Lilienfeld & Fowler, 2006).

Footnotes

Acknowledgements

The authors wish to thank the following Portuguese juvenile detention centers: Bela Vista, Mondego, Navarro de Paiva, Olivais, Padre António Oliveira, Santo António, and Santa Clara.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Pedro Pechorro

References

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Washington, DC: American Psychiatric Publishing.

Andershed

Kerr

Stattin

Levander

(2002). Psychopathic traits in non-referred youths: Initial test of a new assessment tool. In Blaauw

Sheridan

(Eds.), Psychopaths: Current international perspectives (pp. 131-158). Haag, The Netherlands: Elsevier.

Andrews

Bonta

(2017). The psychology of criminal conduct (6th ed.). London, England: Routledge.

Asscher

van Vugt

Stams

Deković

Eichelsheim

Yousfi

(2011). The relationship between juvenile psychopathic traits, delinquency and (violent) recidivism: A meta-analysis. Journal of Child Psychology and Psychiatry, 52, 1134-1143. doi:10.1111/j.1469-7610.2011.02412.x

Babinski

Hartsough

Lambert

(2001). A comparison of self-report of criminal involvement and official arrest records. Aggressive Behavior, 27, 44-54. doi:10.1002/1098-2337(20010101/31)27:1<44::AID-AB4>3.0.CO;2-O

Bayliss

Miller

Henderson

(2010). Psychopathy development and implications for early interventions. Journal of Cognitive Psychotherapy: An International Quarterly, 24, 71-80. doi:10.1891/0889-8391.24.2.71

Blair

Mitchell

(2009). Psychopathy, attentions and emotion. Psychological Medicine, 39, 543-555. doi:10.1017/S0033291708003991

Blais

Forth

(2013). Potential labeling effects: Influence of psychopathy diagnosis, defendant age, and defendant gender on mock jurors’ decisions. Psychology, Crime & Law, 20, 116-134. doi:10.1080/1068316X.2012.749473

Boccaccini

Epstein

Poythress

Douglas

Campbell

Gardner

Falkenbach

(2007). Self-report measures of child and adolescent psychopathy as predictors of offending in four samples of justice-involved youth. Assessment, 14, 361-374. doi:10.1177/1073191107303569

10.

Bolton

(2006). Assessing risk for violent recidivism: A comparison of the Psychopathy Checklist-Revised and the Child and Adolescent Taxon Scale when calculating Violence Risk Appraisal Guide scores (Doctoral dissertation, Indiana State University). Dissertation Abstracts International, 67(5-B), 2823.

11.

Brandt

Kennedy

Patrick

Curtin

(1997). Assessment of psychopathy in a population of incarcerated adolescent offenders. Psychological Assessment, 9, 429-435. doi:10.1037/1040-3590.9.4.429

12.

Caputo

Frick

Brodsky

(1999). Family violence and juvenile sex offending: The potential mediating role of psychopathic traits and negative attitudes toward women. Criminal Justice and Behavior, 26, 338-356. doi:10.1177/0093854899026003004

13.

Cauffman

Kimonis

Dmitrieva

Monahan

(2009). A multimethod assessment of juvenile psychopathy: Comparing the predictive utility of the PCL:YV, YPI, and NEO PRI. Psychological Assessment, 21, 528-542. doi:10.1037/a0017367

14.

Colins

Andershed

Salekin

Fanti

(2018). Comparing different approaches for subtyping children with conduct problems: Callous-unemotional traits only versus the multidimensional psychopathy construct. Journal of Psychopathology and Behavioral Assessment, 40(1), 6-15. doi:10.1007/s10862-018-9653-y

15.

Colins

van Damme

Andershed

Fanti

DeLisi

(2017). Self-reported psychopathic traits and antisocial outcomes in detained girls: A prospective study. Youth Violence and Juvenile Justice, 15, 138-153. doi:10.1177/1541204015619659

16.

Colins

Vermeiren

De Bolle

Broekaert

(2012). Self-reported psychopathic-like traits as predictors of recidivism in detained male adolescents. Criminal Justice and Behavior, 39, 1421-1435. doi:10.1177/0093854812456526

17.

Cooke

Michie

(2001). Refining the construct of psychopathy: Toward a hierarchical model. Psychological Assessment, 13, 171-188. doi:10.1037/1040-3590.13.2.171

18.

Cottle

Lee

Heilbrun

(2001). The prediction of criminal recidivism in juveniles: A meta-analysis. Criminal Justice and Behavior, 28, 367-394. doi:10.1177/0093854801028003005

19.

DeLisi

(2016). Psychopathy as a unified theory of crime. New York, NY: Palgrave MacMillan.

20.

DeLisi

Piquero

(2011). New frontiers in criminal careers research, 2000–2011: A state-of-the-art review. Journal of Criminal Justice, 39, 289-301. doi:10.1016/j.jcrimjus.2011.05.001

21.

Douglas

Epstein

Poythress

(2008). Criminal recidivism among juvenile offenders: Testing the incremental and predictive validity of three measures of psychopathic features. Law and Human Behavior, 32, 423-438. doi:10.1007/s10979-007-9114-8

22.

Edens

Campbell

Weir

(2007). Youth psychopathy and criminal recidivism: A meta-analysis of the psychopathy checklist measures. Law and Human Behavior, 31, 53-75. doi:10.1007/s10979-006-9019-y

23.

Edens

Skeem

Cruise

Cauffman

(2001). Assessment of “juvenile psychopathy” and its association with violence: A critical review. Behavioral Sciences & the Law, 19, 53-80. doi:10.1002/bsl.425

24.

Farrington

(2005). The importance of child and adolescent psychopathy. Journal of Abnormal Child Psychology, 33, 489-497. doi:10.1007/s10802-005-5729-8

25.

Forth

Book

(2010). Psychopathic traits in children and adolescents: The relationship with antisocial behaviors and aggression. In Salekin

Lynam

(Eds.), Handbook of child and adolescent psychopathy (pp. 251-283). New York, NY: Guilford Press.

26.

Forth

Burke

(1998). Psychopathy in adolescence: Assessment, violence, and developmental precursors. In Cooke

Forth

Hare

(Eds.), Psychopathy: Theory, research and implications for society (pp. 205-299). Boston, MA: Kluwer Academic.

27.

Forth

Kosson

Hare

(2003). The Psychopathy Checklist: Youth Version (Technical manual). Toronto, Ontario, Canada: Multi-Health System.

28.

Forth

Mailloux

(2000). Psychopathy in youth: What do we know? In Gacono

(Ed.), The clinical and forensic assessment of psychopathy: A practioner’s guide (pp. 25-54). Mahwah, NJ: Lawrence Erlbaum.

29.

Frick

(1998). Conduct disorder and severe antisocial behavior. New York, NY: Plenum.

30.

Frick

Hare

(2001). The Antisocial Process Screening Device: Technical manual. Toronto, Ontario, Canada: Multi-Health System.

31.

Frick

Ray

Thornton

Kahn

(2014). Can callous-unemotional traits enhance the understanding, diagnosis, and treatment of serious conduct problems in children and adolescents? A comprehensive review. Psychological Bulletin, 140, 1-57. doi:1037/a0033076

32.

Frick

White

(2008). Research review: The importance of callous-unemotional traits for developmental models of aggressive and antisocial behavior. Journal of Child Psychology and Psychiatry, 49, 359-375. doi:10.1111/j.1469-7610.2007.01862.x

33.

Hare

(2003). Hare Psychopathy Checklist—Revised. Toronto, Ontario, Canada: Multi-Health Systems.

34.

Harris

Rice

Quinsey

(1994). Psychopathy as a taxon: Evidence that psychopaths are a discrete class. Journal of Consulting and Clinical Psychology, 62, 387-397. doi:10.1037/0022-006X.62.2.387

35.

Hoeve

Blokland

Dubas

Loeber

Gerris

van der Laan

(2008). Trajectories of delinquency and parenting styles. Journal of Abnormal Child Psychology, 36, 223-235. doi:10.1007/s10802-007-9172-x

36.

IBM Corp. (2017). IBM SPSS Statistics for Windows (Version 25). Armonk, NY: Author.

37.

Kennealy

Skeem

Walters

Camp

(2010). Do core interpersonal and affective traits of PCL-R psychopathy interact with antisocial behavior and disinhibition to predict violence? Psychological Assessment, 22, 569-580. doi:10.1037/a0019618

38.

Lahey

(2014). What we need to know about callous-unemotional traits: Comment on Frick, Ray, Thornton, and Kahn (2014). Psychological Bulletin, 140, 58-63. doi:10.1037/a0033387

39.

Leech

Barrett

Morgan

(2015). IBM SPSS for intermediate statistics: Use and interpretation (5th ed.). New York, NY: Lawrence Erlbaum Associates.

40.

Lilienfeld

Fowler

(2006). The self-report assessment of psychopathy: Problems, pitfalls, and promises. In Patrick

(Ed.), Handbook of psychopathy (pp. 107-132). New York, NY: Guilford Press.

41.

Loeber

Farrington

Stouthamer-Loeber

van Kammen

(1998). Antisocial behavior and mental health problems: Explanatory factors in childhood and adolescence. Mahwah, NJ: Lawrence Erlbaum.

42.

Loeber

Farrington

D. P

. (Eds.). (2012). From juvenile delinquency to adult crime: Criminal careers, justice policy, and prevention. Oxford, UK: Oxford University Press.

43.

Lynam

(1996). Early identification of chronic offenders: Who is the fledgling psychopath? Psychological Bulletin, 120, 209-234. doi:10.1037//0033-2909.120.2.209

44.

Lynam

Charnigo

Moffitt

Raine

Loeber

Stouthamer-Loeber

(2009). The stability of psychopathy across adolescence. Development and Psychopathology, 21, 1133-1153. doi:10.1017/S0954579409990083

45.

Munoz

Frick

(2007). The reliability, stability, and predictive utility of the self-report version of the Antisocial Process Screening Device. Scandinavian Journal of Psychology, 48, 299-312. doi:10.1111/j.1467-9450.2007.00560.x

46.

Murrie

Cornell

(2002). Psychopathy screening of incarcerated juveniles: A comparison of measures. Psychological Assessment, 14, 390-396. doi:10.1037/1040-3590.14.4.390

47.

Pechorro

(2011). Juvenile delinquency: Study of some psychological and relational variables with an emphasis on psychopathic traits (Unpublished doctoral thesis). Faculdade de Medicina da Universidade de Lisboa, Lisbon, Portugal.

48.

Pechorro

Andershed

Ray

Maroco

Gonçalves

(2015). Validation of the Youth Psychopathic Traits Inventory and Youth Psychopathic Traits Inventory–Short among incarcerated juvenile delinquents. Journal of Psychopathology and Behavioral Assessment, 37, 576-586. doi:10.1007/s10862-015-9490-1

49.

Pechorro

Braga

Ray

Gonçalves

Andershed

. (2019). Do incarcerated male juvenile recidivists differ from first-time offenders on self-reported psychopathic traits? A retrospective study. European Journal of Criminology, 16, 413-431. doi:10.1177/1477370818772774

50.

Pechorro

Hidalgo

Nunes

Jiménez

(2016). Confirmatory factor analysis of the Antisocial Process Screening Device: Self-Report among incarcerated male juvenile offenders. International Journal of Offender Therapy and Comparative Criminology, 60, 1856-1872. doi:10.1177/0306624X15588903

51.

Pechorro

Nunes

Jiménez

Hidalgo

(2015). Incarcerated youths with high or low callous–unemotional traits: A comparison controlling for age of crime onset. Journal of Forensic Psychiatry and Psychology, 26, 78-93. doi:10.1080/14789949.2014.971853

52.

Quinsey

Harris

Rice

Cormier

(2006). Violent offenders: Appraising and managing risk (2nd ed.). Washington, DC: American Psychological Association.

53.

Ribeiro

Silva

Rijo

Salekin

(2013). Child and adolescent psychopathy: Assessment issues and treatment needs. Aggression and Violent Behavior, 18, 71-78. doi:10.1016/j.avb.2012.10.003

54.

Rice

Harris

(2005). Comparing effect sizes in follow-up studies: ROC Area, Cohen’s d, and r. Law and Human Behavior, 29, 615-620. doi:10.1007/s10979-005-6832-7

55.

Rock

R. C.

Sellbom

Ben-Porath

Y. S.

Salekin

R. T.

(2013). Concurrent and predictive validity of psychopathy in a batterers’ intervention sample. Law and Human Behavior, 37, 145-154. doi:10.1037/lhb0000006

56.

Salekin

(2008). Psychopathy and recidivism from mid-adolescence to young adulthood: Cumulating legal problems and limiting life opportunities. Journal of Abnormal Psychology, 117, 386-395. doi:10.1037/0021-843X.117.2.386

57.

Salekin

(2015). Forensic evaluation and treatment of juveniles: Innovation and best practice. Washington, DC: American Psychological Association.

58.

Salekin

Leistico

Neumann

DiCicco

Duros

(2004). Psychopathy and comorbidity in a young offender sample: Taking a closer look at psychopathy’s potential importance over disruptive behavior disorders. Journal of Abnormal Psychology, 113, 416-427. doi:10.1037/0021-843X.113.3.416

59.

Seagrave

Grisso

(2002). Adolescent development and the measurement of juvenile psychopathy. Law and Human Behavior, 26, 219-239. doi:10.1023/A:1014696110850

60.

Silva

Genoves

Latorre

(2012). The use of a screening device to assess psychopathy in young offenders. The Spanish Journal of Psychology, 15, 724-735. doi:10.5209/rev_SJOP.2012.v15.n2.38884

61.

Simões

(2000). Investigações no âmbito da Aferição Nacional do Teste das Matrizes Progressivas Coloridas de Raven [Investigations in the context of the national standardization of the Raven Coloured Progressive Matrices test]. Lisbon, Portugal: Fundação Calouste Gulbenkian.

62.

Skeem

Polaschek

Patrick

Lilienfeld

(2011). Psychopathic personality: Bridging the gap between scientific evidence and public policy. Psychological Science in the Public Interest, 12, 95-162. doi:10.1177/1529100611426706

63.

Skilling

T. A.

Quinsey

V. L.

Craig

W. M

. (2001). Evidence of a taxon underlying serious antisocial behavior in boys. Criminal Justice and Behavior, 28, 450-470. doi:10.1177/009385480102800404

64.

Spain

Douglas

Poythress

Epstein

(2004). The relationship between psychopathic features, violence and treatment outcome: The comparison of three youth measures of psychopathic features. Behavioral Sciences & the Law, 22, 85-102. doi:10.1002/bsl.576

65.

Sullivan

Kosson

(2006). Ethnic and cultural variations in psychopathy. In Patrick

C. J.

(Ed.), Handbook of psychopathy (pp. 437-458). New York, NY: Guilford Press.

66.

Vaughn

Howard

DeLisi

(2008). Psychopathic personality traits and delinquent careers: An empirical examination. International Journal of Law and Psychiatry, 31, 407-416. doi:10.1016/j.ijlp.2008.08.001

67.

Vincent

Odgers

McCormick

Corrado

(2008). The PCL: YV and recidivism in male and female juveniles: A follow-up into young adulthood. International Journal of Law and Psychiatry, 31, 287-296. doi:10.1016/j.ijlp.2008.04.012

68.

Virginia Department of Juvenile Justice. (2005). Juvenile recidivism in Virginia. DJJ Research Quarterly, 3, 1-12.

69.

Vitacco

Neumann

Caldwell

(2010). Predicting antisocial behavior in high-risk male adolescents: Contributions of psychopathy and instrumental violence. Criminal Justice and Behavior, 37, 833-846. doi:10.1177/0093854810371358

70.

Zara

Farrington

D. P.

(2016). Criminal recidivism: Explanation, prediction and prevention. New York, NY: Routledge.