Skip to main content
SearchLoginLogin or Signup

Psychopathic traits and offending trajectories from early adolescence to adulthood

Published onJan 01, 2014
Psychopathic traits and offending trajectories from early adolescence to adulthood


Purpose: Measures of adolescent psychopathy have yet to be examined in offending trajectory studies. This may explain why identifying etiological differences between individuals following high-rate and moderate-rate offending trajectories has remained elusive. The current study used the Psychopathy Checklist: Youth Version (PCL:YV) to examine psychopathic traits and offending trajectories within a sample of incarcerated offenders. Methods: Convictions were measured for Canadian male (n = 243) and female (n = 64) offenders at each year between ages 12 and 28. Semi-parametric group based modeling identified four unique trajectories: adolescence-limited (AL) (27.3% of sample), explosive-onset fast desister (EOFD) (30.6%), high-rate slow desister (HRSD) (14.6%), and high frequency chronic (HFC) (27.5%). Findings: Both a three and a four factor model of psychopathy were tested, and both factor structures were positively and significantly associated with the HRSD and HFC trajectories. Regarding individual factors of psychopathy, the ‘Antisocial’ factor of the PCL:YV was the only individual dimension significantly associated with membership in high-rate compared to moderate-rate offending trajectories. Conclusions: Psychopathic traits appear more commonly present amongst individuals who follow chronic versus moderate offending trajectories. Implications for early intervention and risk management of offenders are discussed.


Criminal trajectories; group-based trajectory modeling, juvenile delinquency, PCL:YV, psychopathy


Nagin and Land (1993) introduced semi-parametric group-based modeling (SPGM) as an analytic technique that could be used to examine whether offenders in chronic offending trajectories could be distinguished from offenders in moderate offending trajectories in terms of specific underlying risk factors. However, identifying these risk factors has remained elusive in criminal trajectory studies (see Piquero, 2008 for a review). This is despite a number of developmental life course studies arguing that etiological differences exist between these two groups (e.g., LeBlanc & Loeber, 1998; Moffitt, 1993; Patterson, Debaryshe, & Ramsey, 1989; Thornberry, 2004). This issue has likely persisted because of three unaddressed conceptual challenges associated with trajectory research. First, insufficient base rates of chronic offenders have made detecting significant differences between chronic and moderate offending trajectories difficult (van Domburgh, Vermeiren, Blokland, & Doreleijers, 2009). Second, detecting such differences has remained challenging because many studies have not included important neuropsychological deficits that have been hypothesized to differentiate chronic and moderate offenders (see van der Geest, Blokland, & Bijleveld, 2009). Third, many criminogenic factors measured in adolescence are only distally related to adult offending outcomes (Chung, Hill, Hawkins, Gilchrist, & Nagin, 2002) and thus risk factors that remain stable across the life course should be utilized in trajectory studies.

Retrospective longitudinal data from a Canadian sample of incarcerated offenders were used to examine male (n = 243) and female (n = 64) offending trajectories between the ages of 12 to 28. Although incarcerated samples do not allow for the types of generalizations that can be made using community samples, the use of such samples does ensure an adequate base rate of offenders in chronic offending trajectories. Symptoms of adolescent psychopathy, measured using the Psychopathy Checklist: Youth Version (PCL:YV), were introduced in the current study as relatively stable neuropsychological deficits that potentially differentiate individuals who follow chronic offending trajectories from individuals who follow moderate offending trajectories.

Trajectory Research and the Developmental Life Course Perspective

Trajectories refer to the patterns and sequences of an outcome over age or time and can be used to explain the evolution of crime across the life course (Nagin, 2005; Nagin & Tremblay, 2005). The trajectory methodology is consistent with the developmental life course’s (DLC) emphasis on person-oriented methodological approaches (Magnusson & Bergman, 1988). The person-oriented approach focuses on persons rather than variables to facilitate the simultaneous examination of within-individual and between-individual differences in offending over time (Lussier & Davies, 2011; Magnusson & Bergman, 1988; Moffitt, 1993). The DLC perspective aims to explain the evolution of crime and deviance at the individual level from childhood to adulthood by considering how life conditions and other risk factors can influence the onset, persistence, and desistance of offending (Farrington, 2005; Loeber & LeBlanc, 1990; Nagin & Paternoster, 2000). Trajectory research can help provide a framework for addressing core DLC questions related to offending onset, persistence, and desistence. Although some have critiqued the SPGM method the basis of its ability to test taxonomic and other theories (e.g., Skardhamar, 2009; 2010), others have noted that this has never been the purpose of the SPGM approach (Brame, Paternoster, & Piquero, 2012). The meaningfulness of groups is not determined by the statistical method, but rather the connection between the groups identified and the theory examined. If the groups are as predicted by a theory, than that is support for the theory (Brame et al., 2012). Another critique of trajectory studies more generally has been related to the need for research to measure early neuropsychological deficits in offender samples and then track the development of the criminal trajectories of these offenders into adulthood (see Blokland, Nagin, & Nieuwbeerta, 2005; van der Geest et al., 2009; van Domburgh et al., 2009; Fergusson, Horwood, & Nagin, 2000).

In prior trajectory studies, use of the person-oriented approach to examine within-individual differences in offending over time has revealed that the number of unique offending trajectories representing incarcerated samples ranges from four to six. Of these trajectories, there is almost always at least one chronic trajectory and one adolescent-limited trajectory (Jennings & Reingle, 2012; Piquero, 2008). Although etiological differences between chronic and adolescent-limited offenders have long been hypothesized (e.g., LeBlanc & Frechette, 1989; Moffitt, 1993), studies that have compared chronic offending trajectories to adolescent-limited and other offending trajectories have had difficulty identifying developmental risk factors that distinguish these groups (e.g., Day et al., 2012; Fergusson et al., 2000; Landsheer & van Dijkum, 2005; Nagin, Farrington, & Moffitt, 2005; Odgers et al., 2008; Piquero, 2008).

Conceptual Challenges in Trajectory Studies

In the current study, focus was given to addressing three conceptual challenges associated with examining whether offenders in chronic offending trajectories have characteristics that differ from individuals who follow low or moderate offending trajectories. These conceptual issues are related to (1) sample selection (2) inclusion of appropriate risk factors, and (3) the temporal relationship between risk factor and offense trajectory. First, regarding sample selection, van Domburgh et al. (2009) explained that comparisons between high-rate and other offending trajectories have often yielded insignificant results because of an insufficient base rate of offenders following this high-rate and persistent offending trajectory. In other words, sufficient base rates are needed to perform the types of statistical analyses used to examine between-group differences in trajectory types (see Copas & Tarling, 1986; MacLennan, 1988). In effect, the theoretical relevance of studies that found no differences between chronic and moderate offending trajectories, but relied on low-risk, population-based samples, is limited (see van der Geest et al., 2009). Sampling directly from offender populations is needed to obtain adequate base rates (Blokland & Nieuwbeerta, 2005; Blokland et al., 2005; Piquero; 2008; van der Geest et al., 2009; van Domburgh et al., 2009).

Regarding the second conceptual issue, most prior studies have not included the types of neuropsychological measures that have been hypothesized to differentiate offenders following chronic versus moderate offending trajectories. Instead, predictors of offending trajectories have included parental divorce, religious involvement, school performance and IQ, impulsivity, poor concentration, early onset of antisocial behavior, criminal record of parents/siblings, and parenting style (e.g., Blokland et al., 2005; Day et al., 2012; Fergusson et al., 2000; Landsheer & van Dijkum, 2005; Nagin et al., 2005; Odgers et al., 2008; van der Geest et al., 2009; van Domburgh et al., 2009; Ward et al., 2010). Even measures of psychiatric disorder such as ADHD, conduct disorder (CD), and other behavioral or attention disorders should not be expected to help differentiate between trajectory types because most adjudicated1 adolescent offenders have these types of disorders (Forth, 1995) and these disorders are therefore rather unhelpful in predicting future offending (e.g., Gretton, Hare, & Catchpole, 2004). This limitation can be rectified by including the types of psychopathological disorders that are predominant within chronic offenders but not low/moderate offenders (e.g., Odgers et al., 2008).

Regarding the third conceptual issue, many childhood or adolescent risk factors associated with offending in adolescence will have only a distal effect on offending in adulthood. Specifically, the strength of the relationship between risk factor and offending tends to decrease over time because the risk factor does not follow the individual in lock-step (Chung et al., 2002; Losel & Bender, 2003). In other words, the effect of the risk factor does not remain stable over the life course because the risk factor is not necessarily present/stable at each developmental stage. Traditional criminogenic risk factors over time become only distally related to offending. For example, poor parental attachment and other family adversities have less of an effect on adults because adults are not as reliant on their parents as children or adolescents (e.g., Chung et al., 2002). Thus, studies that attempt to identify characteristics of individuals that are specific to certain offending trajectories would benefit from the incorporation of risk factors that are measured in adolescence and remain relatively stable across time. One such risk factor is psychopathy, which is moderately stable over the life course (Forth, Kosson, & Hare, 2003; Lynam, Caspi, Moffitt, Loeber, & Stouthamer-Loeber, 2007; Vachon, Lynam, Loeber, & Stouthamer-Loeber, 2012) and is also the type of developmental risk factor hypothesized to differentiate chronic offenders from low/moderate offenders (Dyck, Campbell, Schmidt, & Wershler, 2013; Frick, 2009; Moffitt, 1993; 2006).

Adolescence, Psychopathic Personality Disturbance, and Offending

Psychopathy is a personality disorder that is characterized by deficits in behavioral and interpersonal domains (Cleckley, 1976). Behaviorally, psychopaths are impulsive and risk taking and engage in a variety of behaviors, often criminal, in order to satisfy sensation-seeking drives. Interpersonally, psychopaths exhibit characteristics of grandiosity, manipulation, callousness, a lack of empathy, and a parasitic orientation that impacts their relationships with others (Lynam, 1996). In terms of measurement of psychopathy in adolescence, the Psychopathy Checklist: Youth Version (PCL:YV) is considered the gold standard; it has a high degree of reliability and validity and its twenty items are considered appropriate indicators of symptoms of psychopathy in adolescence (Edens & Campbell, 2007; Edens, Skeem, Cruise, & Cauffman, 2001). The twenty items of the PCL:YV have been separated into different factor structures. Most studies support either a parceled four-factor model (Forth et al., 2003) or a three-factor model (Cooke & Michie, 2001). The four-factor model includes an Interpersonal factor (items: glibness, grandiosity, pathological lying, manipulative), an Affective factor (items: lacks remorse, shallow affect, lacks empathy, failure to accept responsibility), a Lifestyle factor (items: boredom, impulsivity, irresponsibility, parasitic orientation, lacks realistic goals), and an Antisocial factor (items: poor anger control, early behavioral problems, juvenile delinquency, revocation of conditional release, criminal versatility). The three-factor model simply excludes the Antisocial factor, based on concerns surrounding the use of prior criminal behavior to predict future criminal behavior (Cooke & Michie, 2001).

Studies using the PCL:YV or other measures of psychopathy have indicated that features of adolescent psychopathy are reliable across different populations (e.g., community and incarcerated samples), different ethnicities, and gender (Pechorro et al., 2013; Vachon et al., 2012). Psychopathy has also been identified as one of the strongest individual-level predictors of: general offending, time until recidivism, early onset of offending, persistent offending, and criminal career index measures (Corrado, Vincent, Hart, & Cohen, 2004; DeLisi et al., in press(a); DeLisi et al., in press(b); Gretton et al., 2004; Hare, 1996; Hare, 2001; Salekin, 2008; Vaughn & DeLisi, 2008; Vaughn, Howard, & DeLisi, 2008; Vincent, Odgers, McCormick, & Corrado, 2008). The robustness of psychopathy as a predictor of these different offending outcomes has been demonstrated in previous research that compared measures of psychopathy to criminogenic factors. Flexon and Meldrum’s (2013) study of a community sample of adolescents found that scores on a measure of callous-unemotional traits were significantly and substantively predictive of violent behavior even when controlling for traditional criminogenic variables, including low self-control and delinquent peers. In another study, DeLisi and colleagues (in press(b)) found that adolescents with high levels of psychopathy had an earlier onset of offending that was not mediated by moral disengagement.

Psychopathy measures for youth have also been found to be prospectively associated with criminal recidivism. Based on a ten-year follow-up period of 133 youth referred for a court-ordered mental health assessment, Schmidt, Campbell, and Houlding (2011) found that the scores of the PCL:YV were more strongly associated with general recidivism and technical violations compared to those of the Youth Level of Service/Case Management Inventory (YLS/CMI) and the Structured Assessment of Violence Risk in Youth (SAVRY). The findings also highlighted PCL:YV scores were more strongly associated with violent recidivism compared to those of the YLS/CMI. In other words, the scores of the PCL:YV might be capturing more individual differences associated with persistent juvenile offending relative to the YLS/CMI. These findings are interesting given that the YLS/CMI requires risk assessors to consider the presence of psychopathic traits (e.g., Hoge, 2010). This may suggest, among other things that the YLS/CMI, which was originally developed for adults and then adapted for youth, might include factors that obscure the prediction of reoffending. Although Welsh, Schmidt, McKinnon, Chattha, and Meyers (2008) used the same sample described in Schmidt et al. (2011), they found that over a three year period the SAVRY outperformed the PCL:YV’s ability to predict violence. In contrast, Schmidt et al. (2011) found that the PCL:YV and SAVRY did not differ in their ability to predict violence. The disparate findings between these two studies may be related to differences in the length of the time period observed. It is possible that the SAVRY is a stronger predictor of violence over the short term, whereas the PCL:YV is a stronger predictor of violence over the long term because of the stability of traits of psychopathy (e.g., Forth et al., 2003; Lynam et al., 2007; Vachon et al., 2012).

More recently, psychopathy has received greater attention in criminal career research because of the hypothesized proximal relationship between psychopathy and criminal behavior across the life course (Ribeiro da Silva et al., 2012; Schmidt et al., 2011; Vaughn & DeLisi, 2008). In their review of recent criminal career research, DeLisi and Piquero (2011) emphasized that criminal career measures can be linked to biosocial development, including personality disorders and psychopathy in particular. Moreover, DeLisi and Piquero (2011) speculated that because the size of the population of individuals with psychopathy mirrored the population of individuals who were the most chronic offenders, it was possible that these two groups were comprised of more or less the same individuals. However, few studies have actually examined the relationship between psychopathy and offending from adolescence to adulthood, and as such whether chronic offenders present a clinical profile suggestive of the presence of psychopathic traits or psychopathy is unclear. Based on data from the Cambridge Study in Delinquent Development, Piquero and colleagues (2012) found that psychopathy scores, measured in adulthood, retrospectively predicted belonging to a more frequent offending trajectory. Similarly, Dyck and colleagues (2013) measured symptoms of psychopathy in adolescence and examined offending frequency2 from age 12 to 23 in a sample of male (n = 80) and female (n = 46) adolescent offenders. Adolescents with moderate or high symptoms of psychopathy were more frequent and versatile offenders compared to the low-symptom group. The frequency and versatility of offending are conceptually similar to behavioral measures within the ‘Antisocial’ factor of psychopathy described by the PCL:YV. In other words, having a high rate and versatile criminal career is consistent with traits of psychopathy contained within the Antisocial factor of the PCL:YV.

Thus far, however, the impact of adolescent psychopathy on offending trajectories has only been hypothesized (van der Geest et al., 2009).3 This is despite Farrington’s (2005) call for the incorporation of measures of psychopathy with developmental life course research as well as the need for research that examines whether there are etiological differences in the developmental profiles of high rate persistent offenders and offenders in other criminal trajectories (see also Piquero et al., 2012). Adolescents with strong symptoms of psychopathic personality disorder are expected to engage in persistent offending because, unlike other risk factors that are measured in adolescence, symptoms of psychopathy remain relatively stable from adolescence to adulthood and therefore should be more proximally related to offending across the life course (Forth, Hart, & Hare, 1990; Hare, 2001; Lynam et al., 2007; Obradovic, Pardini, Long, & Loeber, 2007; Salihovic, Ozdemir, & Kerr, 2013; Vachon et al., 2012). Continuing to examine long-term stability is critical for assessing the predictive value of psychopathy (Ribeiro da Silva, Rijo, & Salekin, 2012).

Aims of Study

The purpose of the current study was to address the lack of research that has examined the association between psychopathy and different offense-based trajectories. Although a number of studies have examined psychopathy’s association with recidivism and offending frequency, these studies have not examined the development of offending over time; something that can be explored with SPGM. Individuals following chronic offending trajectories and individuals following less active offending trajectories were compared in relation to scores on the PCL:YV. The individuals assessed using the PCL:YV in the current study had all been incarcerated in open and secure custody facilities in British Columbia, Canada between 1998 and 2001, which follows the recommendation by Piquero et al. (2012) that offense trajectory studies assess psychopathy in adolescence and within a high-risk sample. Most research on criminal trajectories has relied on community-based samples (e.g., Piquero, 2008), and thus the generalizability of the current study is not as broad. However, this limitation was mitigated by the inclusion of a larger percent of the sample that were frequent offenders, which is needed in order to examine whether risk factor differences can be found between moderate and frequent offending trajectories.4



The first wave of data collection as part of the [Name of Study Withheld for Blind Review] ran between 1998 and 2001. Over this period a total of 507 adolescent offenders completed an interview while they were incarcerated at one of five custody settings in the Lower Mainland and surrounding areas of British Columbia, Canada. Of the 507 offenders, 323 had adequate file and interview information that permitted completion of the PCL:YV. Criminal histories were coded for 307 of the 323 offenders.5 The sample used was very specific (Canadian incarcerated adolescent offenders), which could limit generalizability. For example, although only 4.9% of the population of British Columbia self-identifies as Aboriginal, approximately 25% of offenders in the current study self-identified as Aboriginal. The over-representation of Aboriginal offenders is dissimilar from most incarcerated samples in the United States (e.g., Teplin et al., 2013), although the over-representation of Aboriginal offenders in this sample mirrors the over-representation of Black and Hispanic offenders in incarcerated samples found in the United States. Additionally, because all offenders were incarcerated at the time of their interview, the sample in the current study could have differed from other juvenile offenders who received a less severe non-custody based sentencing option (e.g., probation).

All male (n = 243) and female (n = 64) criminal convictions were coded from age 12, the age of criminal responsibility in Canada, to age 28. The majority of offenders were Caucasian (61.1%), 23.8% of offenders were Aboriginal, and 15.1% of offenders belonged to various other minority groups (Hispanic, Middle Eastern, Asian, and African Canadian). At the time youth were interviewed, ages ranged from twelve to nineteen, with a mean age of 16.35 (SD = 1.3). The average pro-rated PCL:YV score for offenders was 21.2 (SD = 6.4). The average number of convictions was 24.2 (SD = 18.8) and the average number of violent convictions was 2.8 (SD = 2.6). The mean number of total months spent in custody for offenders was 39.6 (SD = 39.5). During the study period, eleven (3.6%) offenders died and nine (2.9 %) offenders moved outside the province. Following the examples in prior studies (Eggleston, Laub, & Sampson, 2004; Livingston, Stewart, Allard, & Ogilvie, 2008; van der Geest et al., 2009), convictions occurring after the date of death or move between provinces were coded as missing rather than as ‘zero’.6


Psychopathy Checklist: Youth Version (Forth et al., 2003).7 The PCL:YV is a symptom rating scale that is coded using information from a 60-90 minute semi-structured interview and a review of file-based collateral information, including information on the offender’s family environment, substance use, and physical and mental health. The PCL:YV rating scale ranges from 0-2 (0 = item does not apply; 1 = item applies somewhat; 2 = item definitely applies). The twenty items comprising the PCL:YV were identified as the fundamental personality and behavioral traits believed to represent the construct of psychopathy in adolescence. The twenty items are summed to provide a score out of forty. Although there is no diagnostic score to categorically define adolescents who are psychopathic versus non-psychopathic, scores of thirty or higher are typically considered indicative of psychopathy-related personality disturbance. The 20 items comprising the PCL:YV represent different facets of the underlying psychopathy construct. Both the parceled four factor model and Cooke and Michie’s (2001) three factor model were examined in the current study. Pro-rated PCL:YV total scores and factor scores for the sample are presented in Table 1.8 Inter-rater reliability was not conducted in this particular study; however, Vincent (unpublished doctoral dissertation) evaluated inter-rater reliability in a subsample of 30 randomly selected cases and the intraclass correlation coefficient was high (ICC1 = 0.92).

--Insert Table One about Here--

Criminal Convictions. Convictions were based on client records uploaded to the Corrections Network (CORNET), which contained information pertaining to each offender’s movements in and out of custody as well as the exact criminal offense, date of conviction, and sentence type. To facilitate group-based trajectory modeling, every criminal charge that resulted in a conviction was coded from age twelve to twenty-eight. Figure 1 displays the mean number of convictions at each age for males and females. Although the number of convictions appears to peak in adolescence and begin to quickly desist, the number of months in custody for males remains relatively stable from adolescence to adulthood, defined as age 18 and beyond, which highlights the importance of accounting for exposure time when analyzing offending trajectories.

Exposure Time. Using information uploaded to CORNET, each date of admission and date of release from custody was recorded for each offender. Syntax was written for SPSS IBM version 18.0 so that the amount of time spent in custody over the duration of each year of age could be identified for each offender. Exposure time was measured to control for the amount of time that offenders would be unable to commit any offenses due to the lack of opportunity created by incarceration (Nagin, 2004; van der Geest et al., 2009). A measure of exposure time should be especially critical for all studies using an offender or at-risk sample given the substantial amount of time spent in custody by these two populations, the former in particular (see Eggleston et al., 2004). If prior offending trajectory studies failed to control for exposure time, chronic offenders with lengthy incarceration periods could have been misclassified as non-chronic offenders. In turn, this may be a partial explanation for why prior studies have found it challenging to identify differences between moderate and chronic offenders. Skardhamar’s (2010) comment that the shape and peak of chronic offending trajectories differed across studies may be related to some studies accounting for exposure time whereas others did not. Total time spent in custody was also calculated to examine the association between offending trajectory and length of incarceration.

--Insert Figure One about Here--


Informed consent was first provided by the British Columbia Ministry of Child and Family Development (MCFD). MCFD serves as the legal guardian to all youth in custody. Consent from MCFD allowed this project to approach all youth in various custody centres throughout the province during the study period. Research assistants (RAs) approached subjects at their unit within the custody centre and asked if they wanted to participate in a research study for [Name of University Withheld for Blind Review]. Approximately five percent of youth declined to participate. If subjects indicated they wished to participate, RAs brought the subject to an isolated interview room away from their living unit, other youth, and custody staff. All subjects were read and given a copy of an information sheet which explained the purpose of the study, how information would be collected (e.g. interview and file information), and that all information would be kept confidential by law unless the subject made a direct threat against themselves or someone else. Youth who agreed to participate in the study were asked to sign a consent form signifying that they had been read and understood the details of the study that had been provided in the information sheet. Once youth were interviewed, trained RAs collaborated file and interview information to score the PCL:YV.

Analytic Strategy

Nagin and Land (1993) introduced semi-parametric group based modeling (SPGM) as an analytic technique that was suitable for measuring trajectories of offending over substantial periods of time. Unlike cluster analysis and other grouping methods, SPGM does not identify groups ex ante (Nagin, 2005). Instead, this method allows distinct developmental trajectory groups to emerge from the data, rather than assume their existence (Nagin, 2005). Although this method has been widely used (Piquero, 2008), it has also recently been criticized as a technique that is not suitable for identifying evidence for a taxonomy (e.g., Skardhamar, 2010). The validity of Skardhamar’s critique has been questioned (see Brame, Paternoster, & Piquero, 2012 for a response) and in the current study it does not seem to apply, given the purpose of this study is not to provide evidence of a taxonomy. In this study, the SPGM method was used to identify the offending trajectories of 307 individuals over a period of 17 years. Piquero et al.’s (2001) formula for calculating exposure was used; however, following the example in van der Geest et al. (2009), Piquero et al.’s (2001) formula was adjusted to avoid high standard errors and improbable rates of offending. van der Geest et al.’s (2009) formula constrained the minimum exposure value to 0.5. In effect, if an offender spent one year in custody, they were coded as spending only half of a year in custody. This was appropriate for a sample from the Netherlands, where sentences tend to be more lenient than in other countries (Blokland, Nagin, & Nieuwbeerta, 2005). In the current study, an average of 13 offenders spent the full year in custody during any given year. A minimum exposure time of 0.5 would likely have overrepresented the length of time that many offenders spent in the community. As such, van der Geest et al.’s (2009) formula for exposure time was adapted so that the minimum exposure time would be approximately 0.2. In effect, spending 12 months in custody would be minimized to nine months in custody. The exposure time formula was estimated the same way at each age. The formula for exposure was:

Exposureji = 1 - (Number of Days Incarcerated/4559)

where j is the respondent and i is the year of observation.

Analyses were conducted in SAS 9.3 using the Proc TRAJ add-on developed by Jones and colleagues (2001) (see also Jones & Nagin, 2007). The current study used the zero-inflated Poisson (ZIP) model to estimate the distribution of the offending trajectories. The ZIP model is most commonly used to examine criminal careers because it accounts for periods of criminal inactivity that are particularly common as individuals reach adulthood (Nagin, 2005). Bayesian Information Criteria (BIC) values were used to identify the number of offending trajectories that best represented the data. BIC is the most commonly used option for determining model selection because BIC rewards parsimony by penalizing the addition of more groups to the model (Nagin, 2005). Larger BIC values generally indicate an improvement in model fit. In addition, the Bayes factor approximation should also be examined to determine whether the difference in BIC values between two models is substantive (Nagin, 2005). SPGM allows for the inclusion of multiple risk factors that predict the probability of trajectory group membership. Due to the imperfect classification accuracy of the group-based method, the association between risk factors and group membership is estimated simultaneously with the trajectories so that the uncertainty in trajectory membership is accounted for. In SPGM, multinomial logistic regression is used, and the reference group typically refers to the lowest trajectory (i.e., the least serious) (Nagin, 2005; van Domburgh et al., 2008).


Model Identification and Description

A four group quadratic model resulted in a BIC value of -8087.96. This four group model was retained because the BIC value from this model was closer to zero when compared to both a three group quadratic model (-8197.61) and a five group quadratic model (-8112.04). The four group solution with quadratic functional form was retained over a four group solution with cubic functional form because the BIC values in the former model were closer to zero. Jeffrey’s scale of the evidence of the Bayes factor confirmed that there was strong evidence for a four group solution.10 The parameters of the four group model are outlined in Table 2. Classification accuracy based on the average posterior probability of accurately assigning individuals to a particular trajectory was high (94%), which helped substantiate the appropriateness of a four group model. Additionally, as indicated in Table 2, the odds of correct classification (OCC)11 was at least fifteen for each of the four groups, which again indicates good model fit (Nagin, 2005) and is a more conservative test than simply relying on average posterior probabilities (Skardhamar, 2010). The four trajectories, presented in Figure 2, were labeled adolescence-limited (AL) (27.5%), explosive-onset fast desister (EOFD) (30.6%), high rate slow desister (HRSD) (14.6%) and high frequency chronic (HFC) (27.3%). Based on Figure 2, the EOFD and AL trajectories appear parallel. However, using the confidence interval extension developed by Jones and Nagin (2007), there was very little overlap between the confidence intervals, which suggests that these two trajectories are unique. The characteristics of each trajectory are described below and summarized in Table 3.

--Insert Table Two about Here--

--Insert Figure Two about Here--

Adolescence Limited. Adolescence-limited (AL) offenders comprised 27.5% of the sample. For this group, offending typically began at age 13, peaked at 16, and reached a near-zero rate by age 20. AL offenders also had the lowest mean rate of offending of all groups at 8.11(SD = 5.8) convictions. AL offenders had the lowest mean number of violent convictions at 1.5 (SD = 1.3). Of all trajectories, the AL trajectory had the lowest ratio of males to females. Half of all female offenders (53.1%) were in the AL trajectory, indicating that the majority of female offenders who are incarcerated in adolescence will limit their offending to adolescence. In contrast, only 22.2% of male offenders were in the AL trajectory. Offenders in the AL trajectory averaged the lowest score on both the four factor (17.5, SD = 6.0) and three factor (11.5, SD = 4.8) models of the PCL:YV compared to the other three trajectories.

Explosive-Onset Fast Desister. Explosive Onset Fast Desisters (EOFDs) comprised 30.6% of the sample. For this group, offending typically began at age 13, peaked at 15, and reached a low rate (less than 0.5 offenses per year) by age 22. Unlike AL offenders, the offenders in this group continued to offend, albeit at a low rate, into their late twenties. Offenders in this group averaged 14.7 (SD = 8.9) convictions over their criminal career and 2.3 (SD = 2.1) violent convictions. Just below one third of both males (29.6%) and females (31.3%) were members of this group. EOFDs scored similarly to AL offenders on both the four and three factor models of the PCL:YV (17.9 (SD = 5.3) and 11.5 (SD = 3.0), respectively).

High Rate Slow Desister. High-rate slow desisters (HRSDs) comprised 14.6% of the sample. For offenders in this group, their criminal career typically began at age twelve, which is the youngest age at which offenders in Canada can be held legally responsible. Offending in this trajectory peaked at age 17 at an average approximate rate of 5.5 offenses. At this point, the rate of offending declined at each year until a leveling-off point from age 25 to 28. Offenders in this group averaged 42.3 (SD = 12.6) criminal convictions and 4.2 (SD = 3.3) violent convictions over their criminal career. There was only one female offender in the HRSD trajectory. In total, 16.9% of males and 1.6% of females belong to this trajectory. HRSDs had the highest scores on both the three and four factor model of psychopathy (23.1 (SD = 4.6) and 14.4 (SD = 4.0), respectively).

High Frequency Chronic. High frequency chronics (HFCs) comprised 27.3% of the sample. For this group, offending typically began at age 13, peaked at 16, and reached a leveling off point from age 18 to 28 at a rate of approximately 2.5 convictions per year. Offenders in this group averaged 42.4 (SD = 15.2) convictions and 3.9 (SD = 2.8) violent convictions over their criminal career. Approximately nine out of every ten offenders in this group were male. The largest proportion of male offenders was in this group (31.3%). In contrast, only 14.1% of females belonged to this trajectory. HFCs had the second highest scores on both the three and four factor measures of psychopathy (21.3 (SD = 5.6) and 13.6 (SD = 4.6), respectively).

--Insert Table Three about Here--

Psychopathy and Criminal Trajectories

There was a significant association between offending trajectory and psychopathy category. Specifically, individuals in the HRSD trajectory were five times more likely than individuals in the AL and EOFD trajectories to have a score of thirty or higher on the PCL:YV. Individuals in the HFC trajectory were twice as likely as individuals in the AL and EOFD trajectories to have a score of thirty or higher. From these initial analyses, higher psychopathy scores were expected to be associated with the most serious criminal trajectories. Three separate multinomial logistic regression models were run, each looking at a different factor structure of psychopathy (see Table 4). The AL trajectory was treated as the reference group because this group had the lowest rate of offending. Due to the low base rate of females in the HFC and HRSD trajectory groups, gender was not controlled for in any of the models.12 Results from the different factor structure analyses (Models One and Two) and the individual dimensions of psychopathy (Model Three) are discussed separately.

--Insert Table Four about Here--

Factor Structure Analyses. The first model in Table 4 indicates that, controlling for ethnicity, the four factor model of psychopathy was significantly and positively related to both the HFC and the HRSD trajectory groups compared to the AL trajectory group. The focus within model two was on the relationship between Cooke and Michie’s (2001) three factor model and criminal trajectory group, controlling for ethnicity. Higher scores on the PCL:YV’s three factor model were associated with the HFC and HRSD trajectory groups compared to the

Dimension Analyses. In model three, each of the factors from the four factor model were entered. Among the four dimensions of psychopathy examined, only the Antisocial factor (F4) was significantly related to offending trajectories. In this case, scores on the Antisocial factor were positively and significantly associated with both the HFC and HRSD trajectories (relative to the AL trajectory). The other three factors, Interpersonal (F1), Affective (F2), and Lifestyle (F3) were not significantly related to the offending trajectories found. In effect, the three factor model was significantly associated with the HFC and HRSD trajectories yet the individual factors in this model were not significant. Thus, the cumulative effect of F1, F2, and F3 significantly differentiated offending trajectories. Individually, however, they were insignificant.13


Empirical relevance of trajectory groups is dependent on the extent to which the nature of each group differs from one another (van der Geest et al., 2009). The results of the SPGM analysis in the current study revealed four unique offending trajectories: an adolescence-limited (AL) offending trajectory, an explosive-onset fast desister (EOFD) trajectory, a high-rate slow desister (HRSD) trajectory, and a high frequency chronic (HFC) trajectory. It appeared that two of the groups (HRSD and HFC) represented different life-course persistent offending trajectories and the other two groups (EOFD and AL) represented adolescent-limited or early adulthood-limited offending trajectories. Critics of the SPGM approach (e.g., Skardhamar, 2010) may argue that the four trajectories representing two meta-trajectories is a function of atheoretical decision making when it comes to model selection. A more plausible explanation is that life course persistent offending is a broad category and that there may be different trajectories leading to the same persistent adult offending outcome. Further, life course offending may be true for two groups, but one group may still offend in adulthood at a higher rate. As evidence of this latter explanation, in the current study the HFC trajectory had a later onset of offending compared to the HRSD trajectory; yet the HFC trajectory indicated a higher rate of offending in adulthood (i.e., offending after age 18). Individuals described by the HRSD and HFC trajectories may experience similar risk factors, but these risk factors may manifest earlier for those in the HRSD trajectory, which helps to explain their earlier onset of offending. Specific to psychopathy, differences in trajectory shape may reflect differences between individuals with primary versus secondary psychopathy. Primary psychopaths are believed to inherit psychopathic traits and thus an earlier onset of antisocial behavior should be expected (i.e., the HRSD group). Secondary psychopaths are believed to acquire psychopathic traits through environmental insults, and thus onset of antisocial behavior may be delayed until acquisition of such traits (i.e., the HFC group) (see Skeem, Johansson, Andershed, Kerr, & Louden, 2007).

The number and shape of trajectories identified in the current study is also typical of most studies that have utilized offender samples (Jennings & Reingle, 2012; Piquero, 2008). Specifically, SPGM studies using offender-based samples have commonly identified an early and persistent criminal trajectory (i.e., HFCs in the current study) and a trajectory that peaks in mid-adolescence and reaches a near-zero level by early adulthood (i.e., ALs in the current study) (Piquero, 2008). Also consistent with prior offending trajectory research was the finding that females comprised only a small percentage of the HRSD (2.4%) and HFC (10.6%) trajectories (see Fergusson & Horwood, 2002) which meant that identifying differences between chronic female offenders and other female offenders remains challenging (e.g., Andersson & Torstensson-Levander, 2013). A unique finding in the current study was that the trajectory with the earliest average age of onset of offending (HRSDs) was not also the trajectory with the highest rate of offending in adulthood (the HFC group had the highest rate of offending from age 18 to 28). The HRSD and HFC trajectories provided a clear example of the utility of SPGM. Despite averaging an equal number of convictions (42), the shapes of the trajectories of these two groups are quite different (Figure 2). In terms of similarities between these two groups, offenders assigned to the HRSD and HFC trajectories (classification accuracy was 94%) spent more time in custody, had a greater number of convictions, a greater number of violent convictions, and were more likely to be male compared to the AL and EOFD trajectories.

The prevalence of chronic offenders in the current study (HFC = 27.3%, HRSD = 14.6%) differed dramatically from the ‘severe 5%’ group of chronic offenders found in the recent work of Vaughn and colleagues (2011; see also Vaughn, Salas-Wright, DeLisi, & Maynard, in press). However, the disparate prevalences are perhaps better understood as a function of different sampling strategies rather than as two groups composed of different individuals. The Vaughn et al. (2011; in press) studies relied on a nationally representative and generalizable sample that provided confirmation of a severe 5% group of offenders that differed from non-offenders. Expanding on these findings, the current study relied on a large group of chronic offenders that was sufficient for detecting differences between chronic and moderate offender groups that had previously remained elusive in trajectory studies (e.g., Piquero, 2008) and latent class models (e.g., Andersson & Torstensson-Levander, 2013).

An important caveat of any study that examines the association between specific risk factors and trajectory group membership is that, even if some risk factor increases the likelihood of membership in a particular group, not all offenders with that risk factor will be members of that particular group (Nagin & Tremblay, 2005). Thus, although higher scores on the PCL:YV were associated with the HRSD and HFC offending trajectories compared to the AL trajectory, not all individuals with high scores on the PCL:YV were guaranteed to follow either the HRSD or HFC trajectory. It may not be the case that all or the vast majority of chronic offenders are also the individuals who comprise the population of individuals with high symptoms of psychopathy (e.g., DeLisi & Piquero, 2011). For example, in their evaluation of Moffitt’s (1993) developmental taxonomy, Fairchild, Goozen, Calder, and Goodyer (2013) noted that distinct qualitative differences in the personality profiles of chronic and adolescent-limited offenders do not appear to exist, though the personality profiles of the chronic group tend to be more severe. This is important from a risk assessment perspective because it cautions against making an assumption that all adolescent offenders scoring high on the PCL:YV will continue to offend throughout adulthood.

Implications for Offender Assessment

Effectively responding to the small group of offenders who are responsible for the majority of all crime remains challenging because these offenders typically have personality features, such as psychopathy, that pose significant barriers to successful intervention and treatment (e.g., Caldwell, Skeem, Salekin, & van Rybroek, 2006). Early childhood intervention programs represent an alternative to the more reactive-based treatment and incarceration approaches that have received less than favorable outcomes with adolescent populations (e.g., Frick, 2009; Frick & Ellis, 1999). The effectiveness of early intervention programs can potentially be improved through assessments that help to identify the appropriate program for individuals with features of psychopathy. This consideration must be balanced with concerns that have been raised over the appropriateness of labeling a child or adolescent as a psychopath (Edens, Skeem, Cruise, & Cauffman, 2001; Hart, Watt, & Vincent, 2002). The most successful early interventions for the types of individuals in the HFC and HRSD trajectory groups require an awareness of risk factors from multiple domains (Frick & White, 2008). One instrument that has been designed specifically for serious and violent offenders and can aid in promoting the identification of risk factors from multiple domains is the Cracow instrument (see Corrado, Roesch, Hart, & Gierowski, 2002). The Cracow instrument includes risk factors that have been identified to be important at different developmental stages that will allow for individual-specific interventions at the individual, familial, and community level (e.g., Corrado et al., 2002; Lussier, Corrado, Healey, Tzoumakis, & Deslauriers-Varin, 2011). Interventions for offenders are particularly important given that this group is also more likely to experience health problems that contribute to public health costs (Vaughn, Salas-Wright, DeLisi, & Piquero, in press).

Limitations and Future Research

Many studies do not follow sample members passed their thirties (Piquero, 2008), and this was one of them. Blokland et al. (2005) found that individuals that followed a high-rate persistent trajectory continued to offend at a substantial rate through their fifties. Had the current study continued to follow its participants further into adulthood, differences in offending frequency between the HFC and HRSD trajectory groups may have been amplified. This would have helped identify which of the two trajectories should be of the greatest concern to criminal justice professionals. Additionally, because the current study relied on a sample of individuals who had all committed crimes in adolescence, any existence of an ‘adult-onset’ trajectory group could not be identified. The likelihood of an ‘adult-onset’ trajectory group is relatively low given that virtually all adult offenders have engaged in criminal behavior in adolescence (Robins, 1978).

For all but one offender, PCL:YV assessments were conducted after the age of twelve, which was when measurement of offending trajectories began. Jones and Nagin (2007) emphasized that risk factors should be measured prior to trajectory measurement. Although the current study violated this recommendation, the use of Cooke and Michie’s (2001) three factor model (which excludes the Antisocial factor) helped to avoid tautological issues. Yet, the issue of the PCL:YV’s emphasis on delinquent, criminal, and antisocial behavior (Cooke, Michie, Hart, & Clark, 2004; Dawson, McCuish, Hart, & Corrado, 2012) remains. Future research may consider using other measures of psychopathy, such as the Comprehensive Assessment of Psychopathic Personality (CAPP). The CAPP includes 33 symptoms that are intended to encompass personality rather than antisocial characteristics of psychopathy (Cooke et al., 2004).

Although psychopathy is one of the strongest predictors of both adolescent and adult offending, not all adolescents scoring high on measures of psychopathy are chronic offenders; nor are all adolescents scoring low on measures of psychopathy non-frequent offenders. Moffitt (1993) hypothesized that the trajectories of life course persistent offenders were best explained through a combination of interpersonal deficits and negative family environment. As such, to better explain differences between chronic and non-chronic offenders, future research should incorporate (a) other interpersonal deficits in addition to psychopathy, and (b) negative familial outcomes such as abuse, substance abuse, mental health issues, and criminal behavior. The relative contribution of psychopathy, controlling for interpersonal deficits and negative familial outcomes, should be examined along with interaction and mediating effects (e.g., DeLisi et al., in press(b); Flexon & Meldrum, 2012).


Many studies have found a relationship between psychopathy and offending, and in that respect, the current study is no different. What the current study does have to offer, however, is an additional perspective on how chronic offending trajectories can be differentiated from individuals in moderate offending trajectories on the basis of scores on the PCL:YV. By presenting this perspective, three conceptual issues related to criminal trajectory research were addressed. First, using a Canadian sample of individuals who had all been incarcerated in adolescence meant that there would be sufficiently high base rates of individuals in the chronic offender trajectories. This facilitated the types of multivariate analyses needed to examine whether risk factors are differentially associated with chronic versus moderate offending trajectories. Second, compared to behavioral and attention disorders that are predominant amongst most adjudicated adolescent offenders (e.g., Forth, 1995; Gretton et al., 2004), measures of psychopathy may be a more accurate operationalization of the type of neuropsychological deficits that are expected to differentiate chronic offenders from low/moderate offenders. Third, unlike childhood risk factors that have a more temporaneous or distal impact on offending, such as parental attachment (e.g., Chung et al., 2002; Losel & Bender, 2003), symptoms of psychopathy in adolescence are relatively stable across the life course (Forth et al., 2003; Lynam et al., 2007; Obradovic et al., 2007; Salihovic et al., 2013) and thus psychopathy is more likely to be more proximally related to negative outcomes across all life stages, including adulthood (Salekin, 2008).


Andersson, F., & Torstensson-Levander, M. (2013). Adult onset offending in a Swedish female birth cohort. Journal of Criminal Justice, 41(3), 172-177.

Blokland, A. A. J., Nagin, D., & Nieuwbeerta, P. (2005). Life span offending trajectories of a Dutch conviction cohort. Criminology, 43(4), 919-954.

Blokland, A. A., & Nieuwbeerta, P. (2005). The effects of life circumstances on longitudinal trajectories of offending. Criminology43(4), 1203-1240.

Brame, R., Paternoster, R., & Piquero, A. R. (2012). Thoughts on the analysis of group-based developmental trajectories in criminology. Justice Quarterly, 29(4), 469-490.

Caldwell, M., Skeem, J., Salekin, R., & van Rybroek, G. (2006). Treatment response of adolescent offenders with psychopathy features: A 2-year follow-up. Criminal Justice and Behavior33(5), 571-596.

Chung, I. J., Hill, K. G., Hawkins, J. D., Gilchrist, L. D., & Nagin, D. S. (2002). Childhood predictors of offense trajectories. Journal of Research in Crime and Delinquency, 39(1), 60-90.

Cleckley, H. (1976). The mask of sanity (5th ed.). St. Louis, MI: Mosby.

Cooke, D. J., Michie, C., Hart, S. D., & Clark, D. A. (2004). Reconstructing psychopathy: Clarifying the significance of antisocial and socially deviant behavior in the diagnosis of psychopathic personality disorder. Journal of Personality Disorders, 18, 337–357.

Copas, J. B., & Tarling, R. (1986). Some methodological issues in making predictions. In J.G. Weis (Ed.), Criminal careers and career criminals (v.2), pp. 291-313.

Corrado, R.R., Roesch, R., Hart, S.D., & Gierowski, J.K. (Eds.). (2002). Multi-problem violent youth: A foundation for comparative research on needs, interventions and outcomes. Amsterdam: IOS Press.

Corrado, R. R., Vincent, G. M., Hart, S. D., & Cohen, I. M. (2004). Predictive validity of the Psychopathy Checklist: Youth Version for general and violent recidivism. Behavioral Sciences & the Law22(1), 5-22.

Day, D. M., Nielsen, J. D., Ward, A. K., Sun, Y., Rosenthal, J. S., Duchesne, T. Bevc, I., & Rossman, L. (2010). Long-term follow up of criminal activity adjudicated youth in Ontario: Identifying offence trajectories and predictors/correlates of trajectory group measure. Canadian Journal of Criminology and Criminal Justice, 54(4), 378-413.

Dawson, S., McCuish, E., Hart, S. D., & Corrado, R. R. (2012). Critical issues in the assessment of adolescent psychopathy: An illustration using two case studies. International Journal of Forensic Mental Health11(2), 63-79.

DeLisi, M., Angton, A., Vaughn, M. G., Trulson, C. R., Caudill, J. W., & Beaver, K. M. (in press(a)). Not my fault: Blame externalization is the psychopathic feature most associated with pathological delinquency among confined delinquents. International Journal of Offender Therapy and Comparative Criminology.

DeLisi, M., & Piquero, A. R. (2011). New frontiers in criminal careers research, 2000-2011: A state-of-the-art review. Journal of Criminal Justice, 39(4), 289-301.

DeLisi, M., Peters, D. J., Dansby, T., Vaughn, M. G., Shook, J. J., & Hochstetler, A. (in press(b)). Dynamics of psychopathy and moral disengagement in the etiology of crime. Youth Violence and Juvenile Justice,

Dyck, H.L., Campbell, M. A., Schmidt, F., & Wershler, J. L. (2013). Youth psychopathic traits and their impact on long-term criminal offending trajectories. Youth Violence and Juvenile Justice, 11(3), 230-248.

Edens, J. F., & Campbell, J. S. (2007). Identifying youths at risk for institutional misconduct: A meta-analytic investigation of the psychopathy checklist measures. Psychological Services, 4(1), 13.

Edens, J. F., Skeem, J. L., Cruise, K. R., & Cauffman, E. (2001). Assessment of “juvenile psychopathy” and its association with violence: a critical review. Behavioral Sciences & the Law19(1), 53-80.

Eggleston, E. P., Laub, J. H. & Sampson, R. J. (2004). Methodological sensitivities to latent class analysis of long-term criminal trajectories. Journal of Quantitative Criminology, 20(1), 1-25.

Fairchild, G., Goozen, S. H., Calder, A. J., & Goodyer, I. M. (2013). Research review: Evaluating and reformulating the developmental taxonomic theory of antisocial behaviour. Journal of Child Psychology and Psychiatry54(9), 924-940.

Farrington, D. P. (2005). The importance of child and adolescent psychopathy. Journal of Abnormal Child Psychology33(4), 489-497.

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

Fergusson, D. M., Horwood, L. J., & Nagin, D. S. (2000). Offending trajectories in a New Zealand birth cohort. Criminology, 38(2). 525-551.

Flexon, J.L. & Meldrum, R.C. (2013). Adolescent psychopathic traits and violent delinquency: Additive and nonadditive effects with key criminological variables. (2013). Youth Violence and Juvenile Justice, 11 (4), 349-369.

Forth, A.E. (1995). Psychopathy and young offenders: Prevalence, family background, and violence. Unpublished manuscript. Carelton Univesity, Ottawa, Canada.

Forth, A. E., Hart, S. D., & Hare, R. D. (1990). Assessment of psychopathy in male young offenders. Psychological Assessment: A Journal of Consulting and Clinical Psychology, 2, 342–344.

Forth, A.E., Kosson, D.S., & Hare, R. (2003). The Hare Psychopathy Checklist: Youth Version. New York, NY: Multi-Health Systems.

Frick, P.J. (2009). Extending the construct of psychopathy to youth: Implications for understanding, diagnosing, and treating antisocial children and adolescents. The Canadian Journal of Psychiatry, 54, 803-812.

Frick, P. J., & Ellis, M. (1999). Callous-unemotional traits and subtypes of conduct disorder. Clinical Child and Family Psychology Review2(3), 149-168.

Frick, P. J., & White, S. F. (2008). Research review: The importance of callous‐unemotional traits for developmental models of aggressive and antisocial behavior. Journal of Child Psychology and Psychiatry49(4), 359-375.

Gretton, H. M., Hare, R. D., & Catchpole, R. E. (2004). Psychopathy and offending from adolescence to adulthood: a 10-year follow-up. Journal of Consulting and Clinical Psychology72(4), 636.

Hare, R. D. (1996). Psychopathy: A clinical construct whose time has come. Criminal Justice and Behavior, 23, 25–54.

Hare, R. D. (2001). Psychopaths and their nature: Some implications for understanding human predatory violence. In A. Raine & J. Sanmartin (Eds.), Violence and psychopathy (pp. 5–34). New York: Kluwer Academic Publishers.

Hart, S. D., Watt, K. A., & Vincent, G. M. (2002). Commentary on Seagrave and Grisso: Impressions of the state of the art. Law and Human Behavior, 26, 241–145.

Hoge, R.D. (2012). Youth Level of Service/Case Management Inventory. In R.K. Otto & K.S. Douglas (Eds.), Handbook of violence risk assessment (pp. 81-95). New York, NY: Routledge.

Jennings, W. G., & Reingle, J. M. (2012). On the number and shape of developmental/life-course violence, aggression, and delinquency trajectories: A state-of-the-art review. Journal of Criminal Justice40(6), 472-489.

Jones, B. L., & Nagin, D. S. (2007). Advances in group-based trajectory modeling and an SAS procedure for estimating them. Sociological Methods & Research35(4), 542-571.

Jones, B. L., Nagin, D. S., & Roeder, K. (2001). A SAS procedure based on mixture models for estimating developmental trajectories. Sociological Methods & Research, 29(3), 374-393.

Landsheer, J. A. & Dijkum, C. V. (2005). Male and female delinquency trajectories from pre through middle adolescence and their continuation in late adolescence. Adolescence, 40(160), 729-748.

LeBlanc, M., & Frechette, M. (1989). Male criminal activity from childhood through youth: Multilevel and developmental perspective. New York, NY: Springer-Verlag.

LeBlanc, M., & Loeber, R. (1998). Developmental criminology updated. Crime and Justice 23, 115-198.

Livingston, M. Stewart, A., Allard, T., Ogilvie, J. (2008). Understanding juvenile offending trajectories. Australian & New Zealand Journal Of Criminology,41(3), 345-362.

Loeber, R., & LeBlanc, M. (1990). Toward a developmental criminology. Crime & Justice12, 375.

Losel, F., & Bender, D. (2003). Protective factors and resilience. In D.P. Farrington & J. Coid, Early intervention of adult antisocial behavior (pp. 130-204). Cambridge, UK: Cambridge University Press.

Lussier, P., Corrado, R., Healey, J., Tzoumakis, S., & Deslauriers-Varin, N. (2011). The Cracow instrument for multi-problem violent youth: Examining the postdictive validity with a sample of preschoolers. International Journal of Child, Youth and Family Studies2(2.1), 294-329.

Lussier, P., & Davies, G. (2011). A person-oriented perspective on sexual offenders, offending trajectories, and risk of recidivism: A new challenge for policymakers, risk assessors, and actuarial prediction? Psychology, Public Policy, and Law17(4), 530.

Lynam, D. R. (1996). Early identification of chronic offenders: Who is the fledgling psychopath? Psychological Bulletin120(2), 209.

Lynam, D.R., Caspi, A., Moffitt, T.E., Loeber, R., & Stouthamer-Loeber, M. (2007). Longitudinal evidence that psychopathy scores in early adolescence predict adult psychopathy. Journal of Abnormal Psychology, 116, 155-165.

MacLennan, R. N. (1988). Correlation, base-rates, and the predictability of behaviour. Personality and Individual Differences9(3), 675-684.

Magnusson, D., & Bergman, L.R. (1988). Individual and variable-based approaches to longitudinal research in early risk factors. In M. Rutter (Ed.), Studies of psychosocial risk: The power of longitudinal data (pp. 45–61). New York, NY: Cambridge University Press.

Moffitt, T.E. (1993). "Life-course-persistent" and "adolescent-limited" antisocial behavior: A developmental taxonomy. Psychological Review 100(4), 674-701.

Moffitt, T.E. (2006). Life course persistent versus adolescence-limited antisocial behavior. In D. Cicchetti and D. Cohen (Eds.), Developmental Psychopathology. New York, NY: John Wiley & Sons.

Nagin, D. S. (2004). Response to “methodological sensitivities to latent class analysis of long-term criminal trajectories.” Journal of Quantitative Criminology, 20(1), 27-35.

Nagin, D.S. (2005). Group-based modeling of development. Cambridge, MA: Harvard University Press.

Nagin, D.S. (2010). Group-based trajectory modeling: An overview. In A.R. Piquero & D. Weisburd (Eds.), Handbook of quantitative criminology, 53-68. New York, NY: Springer.

Nagin, D. S., Farrington, D. P., & Moffitt, T. E. (2005). Life-course trajectories of different types of offenders. Criminology, 33(1), 111-139.

Nagin, D. S. & Land, K. C. (1993). Age, criminal careers and population heterogeneity: Specification and estimation of nonparametric mixed poisson model. Criminology, 31(3), 327-354.

Nagin, D., & Paternoster, R. (2000). Population heterogeneity and state dependence: State of the evidence and directions for future research. Journal of Quantitative Criminology16(2), 117-144.

Nagin, D. S. & Tremblay, R. E. (2005). Developmental trajectory groups: Fact or a useful statistical fiction? Criminology 43(4), 873-904.

Obradovic, J., Pardini, D., Long, J.D., & Loeber, R. (2007). Measuring interpersonal callousness in boys from childhood to adolescence: An examination of longitudinal invariance and temporal stability. Journal of Clinical Child and Adolescent Psychology, 36, 276–292.

Odgers, C. L., Moffit, T. E., Broadbent, J. M., Dickson, N. Hancox, R. J., Harrington, H., Poulton, R., Sears, M. R., Thomson, W. M., Caspi, A. (2008). Female and male antisocial trajectories: From childhood origins to adult outcomes. Development and Psychopathology, 20, 673-716.

Patterson, G.R., DeBaryshe, B.D., & Ramsey, E. (1989). A developmental perspective on antisocial behavior. American Psychologist44(2), 329.

Pechorro, P.S., Vieira, D.N., Poiares, C.A., Vieira, R.X., Marôco, J., Neves, S., Nunes, C. (2013). Psychopathy and behavior problems: A comparison of incarcerated male and female juvenile delinquents. International Journal of Law and Psychiatry, 36 (1), 18-22. 

Piquero, A. R. (2008). Taking stock of developmental trajectories of criminal activity over the life course. In Liberman, A. M. (Ed.), The long view of crime: a synthesis of longitudinal research, 23-53. New York, NY: Springer.

Piquero, A. R., Farrington, D. P., Nagin, D. S., & Moffitt, T. E. (2010). Trajectories of offending and their relation to life failure in late middle age: Findings from the Cambridge study in delinquent development. Journal of Research in Crime and Delinquency, 47(2), 151-173.

Ribeiro da Silva, D., Rijo, D., & Salekin, R. T. (2012). Child and adolescent psychopathy: A state-of-the-art reflection on the construct and etiological theories. Journal of Criminal Justice, 40(4), 269-277.

Robins, L.N. (1978). Sturdy childhood predictors of adult antisocial behavior: Replications from longitudinal studies. Psychological Medicine, 8, 611-622.

Salekin, R. T. (2008). Psychopathy and recidivism from mid-adolescence to young adulthood: Cumulating legal problems and limiting life opportunities. Journal of Abnormal Psychology, 117(2), 386-395.

Salihovic, S., Ozdemir, M., & Kerr, M. (2013). Trajectories of adolescent psychopathic traits. Journal of Psychopathology Behavior Assessment 1-13.

Schmidt, F., Campbell, M.A., Houlding, C. (2011).Comparative analyses of the YLS/CMI, SAVRY, and PCL:YV in adolescent offenders: A 10-year follow-up into adulthood. Youth Violence and Juvenile Justice, 9 (1), pp. 23-42.

Skardhamar, T. (2009). Reconsidering the theory on adolescent-limited and life-course persistent anti-social behaviour. British Journal of Criminology49(6), 863-878.

Skardhamar, T. (2010). Distinguishing facts and artifacts in group‐based modeling. Criminology48(1), 295-320.

Skeem, J., Johansson, P., Andershed, H., Kerr, M., & Louden, J. E. (2007). Two subtypes of psychopathic violent offenders that parallel primary and secondary variants. Journal of Abnormal Psychology116(2), 395.

Teplin, L.A., Abram, K.M., Washburn, J.J., Welty, L.J., Hershfield, J.A., & Dulcan, M.K. (2013). The Northwestern Juvenile Project: Overview. OJJDP Bulletin. Washington, DC: US Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.

Thornberry, T. P. (2004). Introduction: Some advantages of developmental and life-course perspectives for the study of crime and delinquency. In T.P. Thornberry (Ed.), Developmental theories of crime and delinquency (v.7), 1-10.

van der Geest, V., Blokland, A., & Bijleveld, C. (2009). Delinquent development in a sample of high-risk youth: Shape, content, and predictors of delinquent trajectories from age 12 to 32. Journal of Research in Crime and Delinquency, 46(2), 111-143.

van Domburgh, L., Vermeiren, R. R. J. M., Blokland, A. A. J., & Doreleijers, T. A. (2009). Delinquent development in Dutch childhood arrestees: Developmental trajectories, risk factors and co-morbidity with adverse outcomes during adolescence. Journal of Abnormal Child Psychology37(1), 93-105.

Vaughn, M. G., & DeLisi, M. (2008). Were Wolfgang's chronic offenders psychopaths? On the convergent validity between psychopathy and career criminality. Journal of Criminal Justice36(1), 33-42.

Vaughn, M. G., DeLisi, M., Gunter, T., Fu, Q., Beaver, K. M., Perron, B. E., & Howard, M. O. (2011). The severe 5%: A latent class analysis of the externalizing behavior spectrum in the United States. Journal of Criminal Justice, 39(1), 75-80.

Vaughn, M. G., Howard, M. O., & DeLisi, M. (2008). Psychopathic personality traits and delinquent careers: An empirical examination. International Journal of Law and Psychiatry31(5), 407-416.

Vaughn, M. G., Salas-Wright, C. P., DeLisi, M., & Maynard, B. R. (in press). Violence and Externalizing Behavior Among Youth in the United States: Is There a Severe 5%?. Youth Violence and Juvenile Justice.

Vaughn, M. G., Salas-Wright, C. P., Delisi, M., & Piquero, A. R. (in press). Health associations of drug-involved and criminal-justice-involved adults in the United States. Criminal Justice and Behavior.

Vachon, D.D., Lynam, D.R., Loeber, R., Stouthamer-Loeber, M. (2012). Generalizing the nomological network of psychopathy across populations differing on race and conviction status. Journal of Abnormal Psychology, 121 (1), 263-269. 

Vincent, G. M., Odgers, C. L., McCormick, A. V., & Corrado, R. R. (2008). The PCL: YV and recidivism in male and female juveniles: A follow-up into young adulthood. International Journal of Law and Psychiatry31(3), 287-296.

Ward, A.K., Day, D. M., Bevc, I., Sun, Y., Rosenthal, J. S., & Duchesne, T. (2010). Criminal trajectories and risk factors in a Canadian sample of offenders. Criminal Justice and Behavior, 37(11), 1278-1300.

Welsh, J. L., Schmidt, F., McKinnon, L., Chattha, H. K., & Meyers, J. R. (2008). A comparative study of adolescent risk assessment instruments predictive and incremental validity. Assessment15(1), 104-115.

Table 1. PCL:YV total and factor scores and reliabilities by gender

Prorated PCLY:YV scores

Boys (n = 243)

Females (n = 64)








Four-factor model

F1: Interpersonal








F2: Affective








F3: Lifestyle








F4: Antisocial








Total four-factor score (/36)








Total three-factor score (/26)








Total PCL:YV score (/40)








*p < .05

Table 2. Zero-inflated poisson quadratic model with four groups (n = 307)







n (%)

85 (27.3)

42 (14.6)

92 (30.6)

88 (27.5)

Estimated model parameters
















Model characteristicsa

Mean probability-HFC

0.94 (.12)

0.05 (.10)

0.02 (.07)

0.00 (.00)

Mean probability-EOFD

0.02 (.06)

0.02 (.08)

0.94 (.12)

0.06 (.09)

Mean probability-AL

0.00 (.00)

0.00 (.00)

0.03 (.08)

0.93 (.10)

Mean probability-HRSD

0.04 (.09)

0.94 (.12)

0.02 (.05)

0.01 (.04)






 Note. HFC = high frequency chronic, HRSD = high rate slow desister, EOFD = explosive-onset fast desister, AL = adolescence limited

a Based on a four group model.

Table 3. Trajectory groups and different individual-level characteristics (n = 307)






χ2/F, p, Φ/η2

n (%)

85 (27.3)

42 (14.6)

92 (30.6)

88 (27.5)


Offense characteristics

Age of onset

14.0 (1.3) ac

13.8 (1.3) ac

14.7 (1.5) bd

14.8 (1.6) bd

F (3) = 8.0, p < .001, η2 = .08

Peak age





Total days incarcerated

2087 (952)ac

2129 (1107)ac

627 (733)bd

451 (906)bd

F (3) = 69.2, p < .001, η2 = .42

Total convictions

42.2 (15.2) ac

42.3 (12.6) ac

14.7 (8.9) cbd

8.1 (5.8) abd

F (3) = 194.7, p < .001, η2 = .67

Violent convictions

3.9 (2.8)ac

4.2 (3.3)ac

2.3 (2.1)cbd

1.5 (1.3)abd

F (3) = 23.3, p < .001, η2 = .18

% drug-related?





χ2 (3)=79.8, p < .001, Φ=.51

% violence?





χ2 (3)=11.7, p < .01, Φ=.20

% homicide?






% with sex offense






Demographic characteristics






χ2 (3)=31.0, p < .001, Φ=.32






χ2 (3)=9.4, p < .05, Φ=.18

PCL:YV factor scores

Four factor score

21.3 (5.6)ac

23.1 (4.6)ac

17.9 (5.3)bd

17.5 (6.0)bd

F (3) = 15.3, p < .001, η2 = .13

Three factor score

13.6 (4.6)ac

14.4 (4.0)ac

11.5 (3.0)bd

11.5 (4.8)bd

F (3) = 8.2, p < .001, η2 = .07

PCL:YV Dimensions

Interpersonal factor

3.4 (2.0)

3.5 (2.2)

2.7 (2.0)

2.9 (1.9)

F (3) = 3.1, p < .05, η2 = .03

Affective factor

4.6 (2.0)

5.0 (1.6)

4.1 (1.9)

4.1 (2.1)

F (3) = 2.7, p < .05, η2 = .03

Lifestyle factor

5.6 (1.9)ac

5.9 (1.8)ac

4.7 (1.9)bd

4.5 (2.2)bd

F (3) = 8.3, p < .001, η2 = .08

Antisocial factor

7.7 (1.8) acd

8.7 (1.4)acb

6.5 (2.3) bd


F (3) = 25.4, p < .001, η2 = .16

 Note: HFC = high frequency chronic, HRSD = high rate slow desister, EOFD = explosive-onset fast desister, AL = adolescence-limited. Total convictions = youth and adult convictions.

a Significantly different from EOFD, b Significantly different from HFC, c significantly different from AL, d Significantly different from HRSD

Asymptotically F distributed

Table 4. Coefficients of Risk Factors by Trajectory Group (AL Group is Reference Group) (n = 307)

Model 1


Model 2


Model 3











Coef. (SE)

Coef. (SE)

Coef. (SE)


Coef. (SE)

Coef. (SE)

Coef. (SE)


Coef. (SE)

Coef. (SE)

Coef. (SE)














-.07 (.35)

.47 (.5)

-.5 (.35)

-.19 (.34)

.01 (.46)

-.85 (.41)

-3.18 (.79)

.95 (.50)

.46 (.35)


.12 (.03)***

.18 (.04)***












.11 (.04)**

.11 (.05)*

-.03 (.06)











.08 (.10)

.03 (1.35)

.06 (.10)








.05 (.10)

.08 (.13)

-.04 (.10)








.16 (.10)

.12 (.13)

.00 (.10)








.29 (.11) ***

.70 (.15) ***

-.01 (.10)

 Note: HFC = high frequency chronic, HRSD = high rate slow desister, EOFD = explosive-onset fast desister, AL = adolescence limited, FFM = Four factor model, TFM = Three factor model, F1 = Interpersonal factor, F2 = Affective factor, F3 = Lifestyle factor, F4 = Antisocial factor

* p < .05, ** p < .01, *** p < .001


Figure 1. Average conviction rate and incarceration length for males and females from age 12 to 28


Figure 2. Offending trajectories from age 12 to 28

No comments here
Why not start the discussion?