The Past Is Prologue: Criminal Specialization Continuity in the Delinquent Career

Abstract

The versatility/specialization debate in criminology has important theoretical, research, and juvenile/criminal justice ramifications. Although offenders are mostly versatile, there is important evidence of specialization, but much of this evidence is derived from highly technical statistical approaches. Drawing on data from a cohort of serious delinquents committed to the California Youth Authority, logistic regression models revealed robust evidence for criminal specialization net the effects of behavioral and demographic controls. Prior homicide was associated with a 1,467% increased likelihood of being currently adjudicated for a homicide offense. Similar prior–current involvement in robbery (294% increased likelihood), aggravated assault (200%), burglary (148%), and drug sales (736%) was found. Logistic regression with odds ratios provides intuitive, valuable estimates of specialization in offending whereby prior involvement in a specific form of delinquency dramatically increases the likelihood of current involvement in the same form of crime.

Keywords

specialization versatility generality of offending delinquency criminal career offense behavior

Whether criminal offenders are versatile or specialized in their offending patterns is a question of significant theoretical and applied importance. At the theoretical level, the versatility/specialization debate informs conceptualizations of criminal offenders and the potential processes that drive criminal conduct. For theorists who attribute crime to a relatively coherent disposition or set of traits, such as self-control/self-regulation (Baumeister & Heatherton, 1996; Finkel & Hall, 2018; Gottfredson & Hirschi, 1990), temperament (Buss & Plomin, 1975; DeLisi & Vaughn, 2014; Schwartz, Snidman, & Kagan, 1996), neuropsychological functioning (Moffitt, 1993, 2018), or psychopathy (DeLisi, 2016; Hare, 1996; Lynam, 1996), criminal versatility is expected because differing opportunity structures will interact with the individual’s underlying antisocial tendencies and drive various manifestations of conduct problems and crime. In this way, general propensity theorists would predict that serious offenders have lengthy juvenile and criminal records comprised of different offenses. For theorists who instead focus on contextual factors, interactional dynamics, and social processes, such as the social psychological mechanisms of general strain theory (Agnew, 1992), coercive family processes (Patterson, 1986), social information processing theory (Crick & Dodge, 1994), or social learning (Burgess & Akers, 1966), there is greater potential for specialized offending that reflects the influence of current situational contexts, the dynamism of peer associations, or other state-dependent effects.

Versatility/specialization also has important juvenile justice implications. In the event that youth engage in general or specific forms of delinquency, correctional interventions can specifically target criminogenic needs that relate to that conduct. For instance, Lai, Zeng, and Chu (2016) studied nearly 4,000 serious offenders and examined whether those who committed only violent delinquent offenses were different from youth who committed violent and other types of delinquency. They found that more general delinquents, those who committed violence and other types, had earlier starting delinquent careers, evinced more risk factors, and had more criminogenic needs than more specialized violent youth. Indeed, policies who focus on specific types of offenders, such as sexual offenders, are in part predicated on the notion they are disproportionately likely to specialize in sexual aggression (Jennings & Perez, 2018; Lin & Simon, 2016; Miethe, Olson, & Mitchell, 2006; Simon, 2000).¹

In the event that delinquents are versatile and thus evince a general antisocial disposition populated by broad-based risk factors, comprehensive risk assessment and case management tools, such as the level of service/Case Management Inventory (Andrews, Bonta, & Wormith, 2000), and comprehensive correctional interventions, such as multisystemic therapy and functional family therapy (Baglivio, Jackowski, Greenwald, & Wolff, 2014) can be used. In practice, correctional sanctions employ a host of conditions (e.g., monitored sobriety, no contact with victims, home visits, mental health counseling, employment monitoring) that are designed to modify the behavior of the offender toward conventional, prosocial conduct. In this regard, it is the general, antisocial disposition and behaviors that are addressed. Thus, the versatility/specialization debate has important implications for academic criminology and applied juvenile/criminal justice alike.

Versatility and (Not or) Specialization in Offending

Like many criminological debates, the versatility/specialization dialogue has historically been cast in zero-sum terms in which offenders were characterized as either versatile or specialized. This is unfortunate because offending trajectories and inspection of individual criminal records usually showed evidence for both trends in offending. For example, an offender with 40 career arrests for 10 different offense types is indicative of versatility, but if 20 of those arrests are for the same crime, that is indicative of specialization. Nevertheless, scholars occasionally took rather strident, polarized views in favor of specialization or versatility. For instance, Gottfredson and Hirschi (1990, p. 91) famously quipped, “In spite of years of tireless research motivated by a belief in specialization, no credible evidence of specialization has been reported. In fact, the evidence of offender versatility is overwhelming.” Indeed, the preponderance of offender’s official criminal history contains a broad diversity of charges spanning violent, property, drug, traffic, and regulatory domains, and numerous studies support criminal versatility. The empirical support derives from a variety of data sources including the Pittsburgh Youth Study (van Wijk et al., 2005), the Texas Youth Commission data (Trulson, Haerle, Caudill, & DeLisi, 2016), the Cambridge Study in Delinquent Development (Piquero, Farrington, Jennings, Diamond, & Craig, 2012), the Federal Bureau of Investigation computerized criminal history file (Blumstein & Cohen, 1979), the California Youth Authority (Armstrong & Britt, 2004), the Dunedin Multidisciplinary Health and Development Study (Lynam, Piquero, & Moffitt, 2004), and the Philadelphia Collaborative Perinatal Project (Piquero, 2000).

In their study of institutionalized sexual offenders, Weinrott and Saylor (1991) examined the self-reported and official criminal histories of 37 men who were convicted of rape, 67 men who were convicted of child molestation, and 21 men who were convicted of incest. The rapists reported an astounding 11,277 nonsexual crimes, and the median number of nonsexual crimes per rapist was 136.5. These offenses included arson, burglary, child abuse, domestic violence, drug selling, extortion, forgery, kidnapping, theft, robbery, and many others. The child molesters reported 8,219 nonsexual crimes with a median of 50 and the incest offenders reported 2,080 nonsexual crimes with a median of 99. Other studies have similarly shown that sexual offenders—often considered to have unique psychopathology and paraphilic disorders that lend themselves to specializing in sexual crime—engage in multitudinous forms of crime spanning sexual offending, nonsexual violent offending, property offending, and other miscellaneous offending (Cale, Lussier, & Proulx, 2009; Cale, Smallbone, Rayment-McHugh, & Dowling, 2016; DeLisi et al., 2017; Drury et al., 2017; Lussier, 2005; Smallbone, Wheaton, & Hourigan, 2003). Moreover, studies of putative “white-collar” offenders (Benson & Moore,1992; van Onna, van der Geest, Huisman, & Denkers, 2014; Walters & Geyer, 2004), “burglary” offenders (Vaughn, DeLisi, Beaver, & Howard, 2008), “firesetting or arson” offenders (Baglivio, Wolff, DeLisi, Vaughn, & Piquero, 2017; Vaughn et al., 2010), “dating/intimate partner violence” offenders (Bouffard, Wright, Muftić, & Bouffard, 2008; Bouffard & Zedaker, 2016), “DUI” offenders (Hallstone, 2014), “drug” offenders (Amemiya, Vanderhei, & Monahan, 2017; DeLisi, 2003; Jennings, Zgoba, Donner, Henderson, & Tewksbury, 2014; Vaughn, Salas-Wright, DeLisi, Shook, & Terzis, 2015; Walters, 2016), and “homicide” offenders (Caudill & Trulson, 2016; McCuish, Cale, & Corrado, 2018; Vaughn, DeLisi, Beaver, & Howard, 2009; Wright, Pratt, & DeLisi, 2009) similarly revealed strong evidence of generalized involvement in crime indicated by versatile criminal history. In sum, to the extent there is specialization in the criminal career, it is that most offenders specialize in versatility.

Although the evidence is clear that offenders are mostly versatile, there is also nontrivial evidence that offenders sometimes specialize in committing specific forms of crime, or specific categories of crime, such as property, drug, white-collar, or violent.² According to Blumstein, Cohen, Das, and Moitra (1988, p. 317),

In the traditional view of criminal careers, the offender ‘samples’ a fairly wide variety of offenses, during the early phases and then converges on those that he finds most appropriate to his taste and skills. This would suggest fairly general switching among crime types through the first few arrests…with later arrests becoming more specialized in particular crime types.

Indeed, prior work has shown that older offenders are more likely than juvenile delinquents to specialize (Fox & Farrington, 2016; Piquero, Oster, Mazerolle, Brame, & Dean, 1999; Steffensmeier & Ulmer, 2005; Tumminello, Edling, Liljeros, Mantegna, & Sarnecki, 2013) presumably supporting Blumstein et al.’s explanation about the unfolding of specialization from a broader versatility pattern.

In psychiatry and the forensic sciences, there is evidence of offenders that very narrowly specialize and seemingly fixate on specific forms of criminal conduct. In these cases, the offending is compulsive in nature and appears to satisfy a catathymic struggle within the individual that is comorbid with extensive psychopathology (e.g., Grant & Won, 2007; Karakasi, Vasilikos, Voultsos, Vlachaki, & Pavlidis, 2017; Kirsch, Simeon, Berlin, & Hollander, 2015; Meloy, 2000, 2010). Lindberg, Holi, Tani, and Virkkunen (2005), for example, reported evidence of “pure” arsonists whose entire criminal record was comprised of arson arrest charges.³ These pure offenders were disproportionately intellectually disabled and/or psychotic. Mattek and Hanson (2018) conducted a clinical case report of a male pedophile who was civilly committed at nearly 100 years of age and whose entire official criminal record is comprised of sexual assaults against children. There was also unofficial, self-reported evidence of exponentially more sexual assaults against children, but no evidence of any other forms of criminal activity. Moreover, the offender was otherwise high functioning in terms of his socioeconomic performance and community involvement.

In criminology, there is also empirical precedence that within an individual’s delinquent/criminal career, there are periods of relatively focused involvement in specific types of crime (McGloin, Sullivan, Piquero, & Pratt, 2007; Sullivan, McGloin, Pratt, & Piquero, 2006; Steffensmeier & Ulmer, 2005). Evidence for offending specialization couched in a broader pattern of offending versatility comes from several data sources including the Montreal Two-Sample Longitudinal Study (Tzoumakis, Lussier, Le Blanc, & Davies, 2013), the Drug Abuse Treatment Outcome Studies (Farabee, Joshi, & Anglin, 2001), and the Massachusetts Treatment Center for Sexually Dangerous Persons (Harris, Smallbone, Dennison, & Knight, 2009). As a whole, these studies demonstrate how various life circumstances, events, and relationship changes shape and refine offending. For example, there is evidence that offenders who are drug dependent and initiate their criminal career after the onset of their addiction are more likely to specialize in property offending and avoid predatory violent crimes (Farabee et al., 2001). Additional specialization has also been reported among a variety of subgroups, including female fraud offenders (Tumminello et al., 2013), serial burglars (Fox & Farrington, 2016), property crime offenders (Ha & Andresen, 2017), drug offenders (Adams & Pizarro, 2014), juvenile property offenders (Farrington, Snyder, & Finnegan, 1988), and sexual aggressors of children (Lussier, LeBlanc, & Proulx, 2005). In other words, specialization research has shown that demographic variations as well as the timing and progression of the offending career are linked to the offender’s narrowing focus on specific forms of antisocial conduct.

Methodological and Analytical Considerations

To date, versatility/specialization research has been highly technical. A variety of analytical techniques including conditional quasi-symmetry modeling (Britt, 1996), smallest space analysis (Trojan & Salfati, 2016; Youngs, Ioannou, & Eagles, 2016), Markov chain analysis (Stander, Farrington, Hill, & Altham, 1989), network analysis (Tumminello et al., 2013), and simultaneous quantile regression (DeLisi et al., 2011) and statistics including the forward specialization coefficient (Farrington et al., 1988), Jaccard’s coefficient (Fox & Farrington, 2016), offense specialization coefficient (DeLisi et al., 2011), and diversity index (McGloin et al., 2007; Wright, Pratt, & DeLisi, 2008) have been used to study versatility and specialization. Unfortunately, the highly quantitative nature of this research has likely obscured some of the qualitative complexities that exist between versatile and specialized offending (Sullivan, McGloin, Ray, & Caudy, 2009; Wright & Bouffard, 2016). For instance, serial homicide offenders are often understood as specialized offenders who focus on murder; however, these offenders usually commit kidnapping, rape, armed robbery, theft, and other crimes during the course of their homicidal conduct (Wright et al., 2008). However, those general criminal acts are likely secondary to the offender’s primary criminal charge which is murder, while the primary motive or choice of crime remains unclear. Prior investigators have similarly noted the interrelation between specialized and versatile offending. For instance, in their study of sexual offenders, Weinrott and Saylor (1991, p. 297) observed,

It could be argued that many of the nonsex crimes occurred in conjunction with a sex crime or an attempt. If that were true, then the notion of specialization might still be valid. After all, auto theft, burglary, providing drugs, kidnapping, assault, drunkenness, and many other offenses could be part of a man’s sexual modus operandi.

Although statistical effects tell one story, the substantive interplay between types of criminal offending potentially tells another.

Prior research has also disproportionately focused attention on the association between consecutive criminal events, such as arrests or convictions. For instance, the forward specialization coefficient presents scores ranging from 0 that indicates complete versatility to 1 that indicates complete specialization. The pros and cons of this approach have been debated considerably (e.g., Britt, 1996; Farrington et al., 1988; Paternoster, Brame, Piquero, Mazerolle, & Dean, 1998); however, focusing on consecutive criminal events obscures larger trends toward specialization and versatility that are derived from an offender’s entire criminal record, which might contain hundreds of police contacts/arrests or adjudications/convictions. Modeling prior criminal offenses or offense types (e.g., violent, property, drug) as predictors of subsequent criminal offenses is a more intuitive, straightforward manner to study the specialization in offending. To the extent that offenders’ criminal career shows evidence of specialization, similar prior offenses should have significant associations with subsequent arrests/convictions for the same offense (e.g., earlier burglaries predict current burglary). To the extent that the offender’s criminal career shows evidence of versatility, prior offenses should have significant associations with subsequent arrests/convictions for other offenses (e.g. burglary, fraud, arson, rape, robbery predict burglary).

Fortunately, prior research has taken such an approach. Using data from over two decades of offenders in the United Kingdom, Soothill, Francis, and Liu (2008) studied 45,915 persons convicted of arson, 5,774 offenders convicted of blackmail, 7,291 offenders convicted of kidnapping, and 9,816 offenders convicted of threats to kill. Although they found that offenders engaged in multiple forms of crime and were thus versatile overall, they also found that the subsequent reconviction prevalence of each form of crime (arson, blackmail, kidnapping, and threats to kill) was exponentially higher among those who had been previously convicted of the same offense. In other words, prior arson was strongly associated with subsequent arson, prior blackmail was strongly associated with subsequent blackmail, prior kidnapping was strongly associated with subsequent kidnapping, and prior threats to kill was strongly associated with subsequent threats to kill.

Substantively similar findings exist using data from the United States. Based on data from the 1958 Philadelphia Birth Cohort Study, Baker, Falco Metcalfe, and Jennings (2013) found that prior violent offending was significantly associated with subsequent violent offending, prior property offending was significantly associated with subsequent property offending, prior drug offending was significantly associated with subsequent drug offending, and prior other offending was significantly associated with subsequent other offending. These effects were variable across offender trajectories, and the greatest continuity of offending—and presumably specialization—was shown for the most severe subgroups, the low-rate and high-rate chronic offenders.

Current Focus

Building on the approach of Soothill et al. (2008) and Baker et al. (2013), the current study examined versatility/specialization by modeling the effects of prior criminal offending and demographic controls on current homicide, robbery, assault, violent offense, burglary, and drug offense. This approach allows us to assess specialization in the event that prior offense predicts the same subsequent offense and versatility in the event that multiple prior offenses predict a subsequent offense.

Method

Participants and Procedures

Data were derived from the full sample, publicly available data file (n = 813) of serious delinquents committed to the California Department of Corrections and Rehabilitation California Youth Authority between 1997 and 1999. The participants were originally studied to assess mental health problems among the institutionalized delinquent population (Haapanen & Steiner, 2003, 2006). The California Youth Authority has since changed its name to the Division of Juvenile Justice, and these data have been used by prior investigators to study a variety of topics including personality disorders and psychopathology among serious delinquents (Kaszynski et al., 2014), diagnostic issues relating to psychiatric distress among serious delinquents (McCoy, Vaughn, Maynard, & Salas-Wright, 2014), and institutional misconduct among serious delinquents (DeLisi et al., 2010) among others. Moreover, in terms of the severity and seriousness of their delinquent history, these data are consistent with other samples of serious, violent, and chronic or “deep-end” juvenile delinquents (e.g., Abram et al., 2017; Baglivio & Epps, 2016; Caudill & Trulson, 2016; Fox & DeLisi, 2018; Trulson, Caudill, Haerle, & DeLisi, 2012; Trulson, DeLisi, Caudill, Belshaw, & Marquart, 2010).

Measures

Dependent variables

Six dichotomous dependent variables indicating the youth’s current adjudicated offense were used: current homicide offense (comprised of murder 1, murder 2, manslaughter, vehicular manslaughter, and attempted murder; 96.3% no, 3.7% yes), current robbery (comprised of enhanced [armed] robbery, robbery, and carjacking; 77.4% no, 22.6% yes), current assault (comprised of aggravated assault, assault and battery, domestic violence, miscellaneous assault, and assault while discharge or display weapon) offense (76.8% no, 23.2% yes), current violent offense (48% no, 52% yes), current burglary (comprised of burglary 1, burglary 2, and attempted burglary) offense (83.2% no, 16.8% yes), and current drug (possession or sale of narcotics, possession or sale of marijuana, and possession or sale of dangerous drugs offense; 94% no, 6% yes). The offense-specific dependent variables for homicide, robbery, assault, burglary, and drug were mutually exclusive. The current violent offense variables included the aforementioned offenses for homicide, robbery, and assault in addition to rape, child molestation, sodomy, perversion, and extortion/kidnapping. The dependent variables are the youth’s instant offense that resulted in placement in the Division of Juvenile Justice (formerly the California Youth Authority).

Independent variables

Nine independent variables spanning prior delinquent offending prior to the instant offense and sociodemographic characteristics were used. These included prior homicide (M = .14, standard deviation [SD] = .48, range = 0–6), prior robbery (M = .67, SD = 1.14, range = 0–10), prior aggravated assault (M = .58, SD = 1.00, range = 0–7), prior simple assault (M = 1.50, SD = 1.86, range = 0-15), prior burglary (M = .92, SD = 1.53, range = 0–17), and prior drug sales (M = .92, SD = 1.53, range = 0–7). Prior work has shown that demographic features are important predictors of versatility/specialization (Armstrong & Britt, 2004; Guerette, Stenius, & McGloin, 2005; Jennings et al., 2014), these included age (M = 16.88, SD = 1.10, range = 12.6–20.4 years), sex (81.2% male, 18.8% female, male = 1, female = 2), and race (17.2% White, 82.8% non-White, 0 = non-White, 1 = White).

Sensitivity independent variables

In sensitivity analyses, three summary prior offending variables spanning total offending (M = 8.75, SD = 5.51, range = 1–34), violent offending (M = 2.93, SD = 2.67, range = 0–18), and property offending (M = 2.83, SD = 3.09, range = 0–21) were used to examine whether the evidence of versatility/specialization was dependent on assorted specifications of total offending (Armstrong, 2008a, 2008b).

Analytical approach

Due to the binary measurement of the dependent variables, logistic regression with bootstrapped standard errors and 95% confidence intervals using 50 replications was used. In all models, the dependent variable was regressed on the nine independent variables (shown in Tables 1 –6). In the sensitivity analyses, all models were executed with the same covariates plus the three summary offending controls (total, violent, and property). Prior research (Brame, Paternoster, & Bushway, 2004; Capaldi & Patterson, 1996; Trojan & Salfati, 2016; Wiesner, Yoerger, & Capaldi, 2018) has shown that versatility/specialization is potentially confounded with total offending so it is important to empirically control for these effects.

Table 1.

Logistic Regression Model for Current Homicide Offense.

Variable	OR	BSE	z	95% CI
Prior homicide	15.67	10.51	4.10***	[4.21, 58.31]
Prior robbery	0.49	0.44	−0.80	[.08, 2.84]
Prior aggravated assault	0.36	0.23	−1.57	[.10, 1.29]
Prior simple assault	0.74	0.25	−0.90	[.39, 1.42]
Prior burglary	0.82	0.24	−0.70	[.46, 1.44]
Prior drug sales	1.02	0.35	0.06	[.52, 2.01]
Sex	1.33	0.72	0.53	[.46, 3.84]
Race	0.66	1.07	−0.26	[.03/16.07]
Age	1.07	0.30	0.24	[.62, 1.85]
Wald χ²	38.73***
Pseudo R ²	0.403

Note. BSE = bootstrapped standard error; CI = confidence interval; OR = odds ratio. ***p < .001. **p < .01. *p < .05.

Table 2.

Logistic Regression Model for Current Robbery Offense.

Variable	OR	BSE	z	95% CI
Prior homicide	0.48	0.33	−1.16	[.14, 1.65]
Prior robbery	3.94	0.56	9.64***	[2.98, 5.21]
Prior aggravated assault	0.78	0.15	−1.33	[.54, 1.13]
Prior simple assault	0.89	0.07	−1.57	[.77, 1.03]
Prior burglary	0.74	0.07	−3.12**	[.61, .89]
Prior drug sales	0.68	0.16	−1.60	[.42, 1.09]
Sex	1.43	0.32	1.58	[.92, 2.22]
Race	0.77	0.20	−1.02	[.46, 1.28]
Age	0.86	0.08	−1.59	[.72, 1.04]
Wald χ²	111.02***
Pseudo R ²	0.286

Note. BSE = bootstrapped standard error; CI = confidence interval; OR = odds ratio. ***p < .001. **p < .01. *p < .05.

Table 3.

Logistic Regression Model for Current Assault Offense.

Variable	OR	BSE	z	95% CI
Prior homicide	1.48	0.47	1.23	[.79, 2.77]
Prior robbery	0.51	0.08	−4.56***	[.38, .68]
Prior aggravated assault	3.0	0.47	7.07***	[2.21, 4.07]
Prior simple assault	1.10	0.05	2.12*	[1.01, 1.21]
Prior burglary	0.62	0.07	−4.28***	[.50, .77]
Prior drug sales	0.30	0.13	−2.88**	[.13, .68]
Sex	0.74	0.19	−1.20	[.45, 1.21]
Race	0.78	0.20	−0.95	[.46, 1.30]
Age	0.93	0.07	−0.87	[.80, 1.09]
Wald χ²	91.5***
Pseudo R²	0.251

Note. BSE = bootstrapped standard error; CI = confidence interval; OR = odds ratio. ***p < .001. **p < .01. *p < .05.

Table 4.

Logistic Regression Model for Current Violent Offense.

Variable	OR	BSE	z	95% CI
Prior homicide	8.97	4.86	4.05***	[3.10, 25.96]
Prior robbery	2.52	0.32	7.23***	[1.96, 3.24]
Prior aggravated assault	2.41	0.43	4.90***	[1.69, 3.42]
Prior simple assault	1.00	0.05	0.08	[.92, 1.10]
Prior burglary	0.57	0.04	−7.46***	[.50, .66]
Prior drug sales	0.42	0.11	−3.24***	[.25, .71]
Sex	1.27	0.34	0.87	[.74, 2.16]
Race	0.66	0.12	−2.30*	[.47, .94]
Age	0.91	0.07	−1.19	[.78, 1.06]
Wald χ²	168.85***
Pseudo R ²	0.295

Note. BSE = bootstrapped standard error; CI = confidence interval; OR = odds ratio. ***p < .001. **p < .01. *p < .05.

Table 5.

Logistic Regression Model for Current Burglary Offense.

Variable	OR	BSE	z	95% CI
Prior homicide	0.09	0.04	−5.82***	[.04, .21]
Prior robbery	0.72	0.09	−2.66**	[.56, .92]
Prior aggravated assault	0.77	0.19	−1.04	[.47, 1.26]
Prior simple assault	0.93	0.06	−1.06	[.81, 1.06]
Prior burglary	2.48	0.39	5.77***	[1.82, 3.37]
Prior drug sales	0.76	0.20	−1.08	[.45, 1.26]
Sex	0.85	0.29	−0.47	[.43, 1.68]
Race	1.25	0.32	0.89	[.76, 2.06]
Age	0.97	0.09	−0.35	[.81, 1.16]
Wald χ²	132.94***
Pseudo R ²	0.296

Note. BSE = bootstrapped standard error; CI = confidence interval; OR = odds ratio. ***p < .001. **p < .01. *p < .05.

Table 6.

Logistic Regression Model for Current Drug Offense.

Variable	OR	BSE	z	95% CI
Prior homicide	Dropped from model, predicted failure perfectly
Prior robbery	0.65	0.30	−0.92	[.26, 1.62]
Prior aggravated assault	0.49	0.13	−2.59**	[.29, .84]
Prior simple assault	1.18	0.11	1.74	[.98, 1.43]
Prior burglary	0.31	0.09	−4.03***	[.17, .54]
Prior drug sales	8.36	2.85	6.24***	[4.29, 16.29]
Sex	2.21	0.89	1.97*	[1.01, 4.86]
Race	1.00	0.57	0.00	[.33, 3.06]
Age	2.17	0.50	3.35***	[1.38, 3.41]
Wald χ²	61.74***
Pseudo R ²	0.451

Note. BSE = bootstrapped standard error; CI = confidence interval; OR = odds ratio.***p < .001. **p < .01. *p < .05.

Findings

Logistic Regression Model for Current Homicide Offense

As shown in Table 1, prior homicide offending was a powerful predictor of current homicide offense (odds ratio [OR] = 15.67, z = 4.10, p < .001). None of the other offense types including robbery, aggravated assault, simple assault, burglary, or drug sales were significantly associated with current homicide offense. Null effects were also found for sex, race, and age.

Logistic Regression Model for Current Robbery Offense

As shown in Table 2, prior robbery offending was a powerful predictor of current robbery offense (OR = 3.94, z = 9.64, p < .001). One other significant effect was found. Prior burglary offending (OR = 0.74, z = −3.12, p < .001) was negatively associated with current robbery offense. None of the remaining offense types or demographics were significantly associated with current robbery offending.

Logistic Regression Model for Current Assault Offense

As shown in Table 3, two forms of prior assault offending were significantly associated with current assault offending including prior aggravated assault (OR = 3.0, z = 7.07, p < .001) and prior simple assault (OR = 1.10, z = 2.12, p < .05). Three other significant effects were found. Prior robbery (OR = 0.51, z = −4.56, p < .001), prior burglary (OR = 0.62, z = −4.28, p < .001), and prior drug sales (OR = 0.30, z = −2.88, p < .01) were inversely associated with current assault offense. Null effects were found for prior homicide, sex, race, and age.

Logistic Regression Model for Current Violent Offense

As shown in Table 4, multiple significant effects were found to be associated with current violent offense. Prior homicide (OR = 8.97, z = 4.05, p < .001), prior robbery (OR = 2.52, z = 7.23, p < .001), and prior aggravated assault (OR = 2.41, z = 4.90, p < .001) were positively associated with current violent offense. In contrast, prior burglary (OR = 0.57, z = −7.46, p < .001) and prior drug sales (OR = 0.42, z = −3.24, p < .001) were negatively associated with current violent offense. Whites (OR = 0.66, z = −2.30, p < .05) were less likely to have a current violent offense than non-Whites.

Logistic Regression Model for Current Burglary Offense

As shown in Table 5, prior burglary offending was a powerful predictor of current burglary offense (OR = 2.48, z = 5.77, p < .001). Two other significant effects were found. Both prior homicide (OR = 0.09, z = −5.82, p < .001) and prior robbery (OR =0.72, z = −2.66, p < .01) were negatively associated with current burglary offense. None of the remaining covariates were significantly associated with current burglary offense.

Logistic Regression Model for Current Drug Offense

As shown in Table 6, prior drug sales were a powerful predictor of current drug offense (OR = 8.36, z = 6.24, p < .001). Several other significant effects were also found. Prior aggravated assault (OR = 0.49, z = −2.59, p < .01) and prior burglary (OR = 0.31, z = −4.03, p < .001) were negatively associated with current drug offense, whereas females (OR = 2.21, z = 1.97, p < .05) and older wards (OR = 2.17, z = 3.35, p < .001) were positively associated with current drug offense.

Sensitivity Analyses

To check the robustness of the logistic regression models, additional models were estimated using the same prior and current offense (e.g., for homicide, robbery, assault, burglary, and drug offenses) and summary prior measures for total offending, total violent offending, and total property offending as independent variables along with demographic covariates. For current homicide offense, prior homicide offending maintained a powerful effect (OR = 17.54, z = 4.54, p < .001), but neither total prior offending, total prior violent offending, nor total prior property offending was associated with current homicide offense (overall model Wald χ² = 35.09, p < .001, pseudo R ² = .413). For current robbery offense, prior robbery offending maintained a powerful effect (OR = 4.52, z = 9.10, p < .001) on current robbery offense, but neither total prior offending, total prior violent offending, nor total prior property offending was associated with current robbery offense (overall model Wald χ² = 89.76, p < .001, pseudo R ² = .270). For current assault offense, both prior aggravated assault (OR = 4.35, z = 6.71, p < .001) and simple assault (OR = 1.72, z = 4.12, p < .001) were significantly associated. In addition, total prior violent offending (OR = 0.68, z = −3.53, p < .001) and total prior property offending (OR = 0.87, z = −2.44, p < .02) were negatively associated with current assault offense (overall model Wald χ² = 69.99, p < .001, pseudo R ² = .203). For current burglary offense, prior burglary offending maintained a powerful effect (OR = 2.41, z = 3.79, p < .001) and prior violent offending (OR = 0.86, z = −2.17, p < .03) was negatively associated. Null effects were found for prior total offending and prior property offending (overall model Wald χ² = 31.60, p < .001, pseudo R ² = .274). For current drug offense, prior drug sales (OR = 6.85, z = 6.87, p < .001) maintained a strong positive association and prior property offending (OR = 0.78, z = −1.97, p < .05) was negatively associated. Null effects were found for total prior offending and prior violent offending (overall model Wald χ² = 57.16, p < .001, pseudo R ² = .331).

Summary of Findings

To summarize, strong evidence for specialization accreted across models. Prior homicide offending was associated with a 1,467% increase likelihood of current homicide offense. Prior robbery increased the likelihood of current robbery offending by 294%. Prior aggravated assault and prior simple assault increased the likelihood of current assault offending by 200% and 10%, respectively. Prior burglary increased the likelihood of current burglary by 148%, and prior drug sales increased the likelihood of current drug offense by 736%. For current violent offense, prior homicide (797%), prior robbery (152%), and prior aggravated assault (141%) were positively associated whereas prior burglary (−43%) and prior drug sales (−58%) were negatively associated. Recall that these are the effect sizes for each offense with controls for other offenses and demographic characteristics.

Discussion

Versatility/specialization is a primary research area in criminology and one that also has direct practical application and relevance (Blumstein, Cohen, Das, & Moitra, 1988; DeLisi & Piquero, 2011; Gottfredson & Hirschi, 2016). Drawing on data from serious delinquent offenders from the California Youth Authority, the current study used a simple logistic regression approach to model the effects of prior offending on subsequent offending by offense. Across models and specifications, there was strong evidence of specialization consistent with prior research that examined broad groupings of offending types among both delinquents (Baker, Falco Metcalfe, & Jennings, 2013) and adults (Soothill, Francis, & Liu, 2008). Several issues warrant discussion.

First, a delinquent that commits a specific criminal offense is much more likely to subsequently commit the same offense suggesting that prior record can serve as a heuristic for understanding present offense conduct. This is particularly important for understanding relatively rare forms of crime, such as homicide. To commit one murder is rare (nearly 89% of the delinquents in these data have no official history of homicide offending), and the models show that once an offender has murdered, the likelihood of doing it again increases nearly 1,500% (nearly 1,700% in the sensitivity model). On this issue, we concur with prior researchers. For example, in their study of specialization and subsequent homicide scene behavior, Horning, Salfati, and Crawford (2010, p. 379, emphasis added) indicated

Classifying homicide offenders by type of prior criminal specialization relies on the idea that a history of a particular type of criminal behavior indicates a pattern that relates to a stable underlying psychology, and this in turn may influence the offenders’ behavior at a subsequent homicide crime scene.

Similar continuity between prior involvement in robbery, assault, burglary, and drug offending and subsequent involvement in the same crimes was also seen. Our findings suggest that the commission of these offenses in a delinquent career is an indicator of the prospective risk that they will occur again (also see Caudill & Trulson, 2016; Trulson et al., 2016) even within a delinquent career that is more globally characterized by involvement in an assortment of offenses.

Second, we found no offense-specific evidence of versatility where prior offending was positively associated with subsequent offending for a different crime. Indeed, there were inverse relationships for (1) prior burglary and subsequent robbery; (2) prior robbery, prior burglary, and prior drug sales and subsequent assault; (3) prior homicide and prior robbery and subsequent burglary; and (4) prior aggravated assault and prior burglary and subsequent drug offending. There was, however, evidence of broader versatility where (5) prior homicide, prior robbery, and prior aggravated assault were strongly associated with subsequent violent offending (in the same equation, prior burglary and prior drug sales were negatively associated). The strong effects for prior homicide, robbery, and aggravated assault and subsequent violence are consistent with research that casts offending specialization as “themes” of offending (Trojan & Salfati, 2016; Youngs et al., 2016), in this case a general theme involving serious interpersonal violence. For example, once convicted of a homicide-related offense, the odds of being convicted of murder in the current crime are drastically higher, but murder also predicts current assault. By thinking about delinquent offending with a themes approach, one can appreciate evidence for specialization and versatility within the same data.

Third, burglary is illustrative of an offense that correlates with reduced odds of the occurrence of other crimes. It is highly correlated in the expected direction with current burglary, negatively correlated with the other current offense crimes, and as an outcome is negatively correlated with other prior crimes. This might indicate any number of things, but one possibility is that burglary is indicative of a “type,” perhaps indicating that a youth who commits a burglary is likely to become a persistent property offender. Such an offender may be a specialist not only in repeated crimes in the record but in the sense that goals and motives persist; therefore, the category might merit discrete empirical attention in books and articles. Indeed, it has been given specific attention in classic books about burglars, and the current results seem to justify treating it as a more discrete type than those who commit some other offenses (Shover, 1996; Wright & Decker, 1994). The burglar’s motive usually is acquisitive and certainly one reason that many select the crime is that it minimizes the potential for violent encounters. This is not to say that burglaries cannot be dangerous, but only that the burglar’s objectives, preferred method, and designs do not usually entail violence. Possibly, these preferences translate into repeated incidences of the crime rather than selecting other offenses. Most burglars intend to avoid violence and structure their crimes toward this end, albeit they often admit that they are willing to perpetrate violence if confronted (Wright & Decker, 1994).

This is only one example that reveals that there is much speculation about the reasons for specialization, often relying on suppositions about the delinquent’s mind-set. However, there is little convincing evidence of the cognitive and decision-making structures that explain either specialization or versatility. It may be that offenders specialize in certain crimes or types of crime simply because they are repelled by the characteristics, meanings associated with, and risks of other crimes in what amounts to a process of elimination.⁴ It may be that specialization exists because different crimes exhibit different desires. The current study provides simple preliminary evidence of potential places to look for threshold effects in what offenders might or might not be willing to do when confronted with a particular sort of criminal opportunity, but little more.

Fourth, a strength of the current approach, using logistic regression and a short list of controls, is its intuitiveness. The layperson, courts, and juvenile justice practitioners are interested in patterns of serious delinquency as an indicator of the potential dangerousness of an offender and often for recurrence of the same offense designated as especially harmful. For serious delinquents, judges and policy makers are interested in simple understanding of the prospective likelihood of the same crime occurring again once a youth has a given type of severe offense on their record. It is reasonable to want to know not only whether an offender is likely to be a nuisance to the public in frequency and duration of delinquent careers but also whether the particular crimes that have been committed before are likely to recur. The implications of these findings offer an insight into how juvenile justice practitioners could gear treatment programs to target specific behaviors through cognitive behavior therapy (Landenberger & Lipsey, 2005) or other specialized treatment programs intended to deter specific forms of delinquency.

Limitations

The current findings should be considered in light of study limitations. The delinquent career data were raw and reflected instant adjudication offense and prior delinquency and lacked contextual data that have been shown to be the state-dependent processes that drive periods of specialized offending in the short-term (McGloin, Sullivan, & Piquero, 2008; McGloin, Sullivan, Piquero, Blokland, & Nieuwbeerta, 2011). Moreover, the use of official records means that the differential odds of being contacted and ultimately committed for a given offense can create the artificial appearance of specialization in the absence of self-reports of delinquency. The reliance on official correctional records that are known to significantly undercount the true incidence of offending is an issue that is especially relevant to studying specialization. For instance, Lynam, Piquero, and Moffitt (2004) reported little evidence of specialization based on official records using the Dunedin birth cohort data; however, self-reports indicated evidence of violence specialization in assault and robbery. It is our hope that future research can incorporate these suggestions. To conclude, offending specialization usually emerges as offenders’ age, but compelling evidence for it was found among delinquents in these data.

Footnotes

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.

Notes

References

Abram

K. M.

Azores-Gococo

N. M.

Emanuel

K. M.

Aaby

D. A.

Welty

L. J.

Hershfield

J. A.

… Teplin

L. A.

(2017). Sex and racial/ethnic differences in positive outcomes in delinquent youth after detention: A 12-year longitudinal study. JAMA Pediatrics, 171, 123–132.

Adams

J. J.

Pizarro

J. M.

(2014). Patterns of specialization and escalation in the criminal careers of gang and non-gang homicide offenders. Criminal Justice and Behavior, 41, 237–255.

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47–88.

Amemiya

Vanderhei

Monahan

K. C.

(2017). Parsing apart the persisters: Etiological mechanisms and criminal offense patterns of moderate- and high-level persistent offenders. Development and Psychopathology, 29, 819–835.

Andrews

D. A.

Bonta

Wormith

S. J.

(2000). Level of service/case management inventory: LS/CMI. Toronto, Canada: Multi-Health Systems.

Armstrong

T. A.

(2008a). Are trends in specialization across arrests explained by changes in specialization occurring with age? Justice Quarterly, 25, 201–222.

Armstrong

T. A.

(2008b). Exploring the impact of changes in group composition on trends in specialization. Crime & Delinquency, 54, 366–389.

Armstrong

T. A.

Britt

C. L.

(2004). The effect of offender characteristics on offense specialization and escalation. Justice Quarterly, 21, 843–876.

Baglivio

M. T.

Epps

(2016). The interrelatedness of adverse childhood experiences among high-risk juvenile offenders. Youth Violence and Juvenile Justice, 14, 179–198.

10.

Baglivio

M. T.

Jackowski

Greenwald

M. A.

Wolff

K. T.

(2014). Comparison of multisystemic therapy and functional family therapy effectiveness: A multiyear statewide propensity score matching analysis of juvenile offenders. Criminal Justice and Behavior, 41, 1033–1056.

11.

Baglivio

M. T.

Wolff

K. T.

DeLisi

Vaughn

M. G.

Piquero

A. R.

(2017). Juvenile animal cruelty and firesetting behaviour. Criminal Behaviour and Mental Health, 27, 484–500.

12.

Baker

Falco Metcalfe

Jennings

W. G.

(2013). What are the odds? Predicting specialization in offending over the life course. Criminal Justice and Behavior, 40, 909–932.

13.

Baumeister

R. F.

Heatherton

T. F.

(1996). Self-regulation failure: An overview. Psychological Inquiry, 7, 1–15.

14.

Benson

M. L.

Moore

(1992). Are white-collar and common offenders the same? An empirical and theoretical critique of a recently proposed general theory of crime. Journal of Research in Crime and Delinquency, 29, 251–272.

15.

Blumstein

Cohen

(1979). Estimation of individual crime rates from arrest records. Journal of Criminal Law and Criminology, 70, 561–585.

16.

Blumstein

Cohen

Das

Moitra

S. D.

(1988). Specialization and seriousness during adult criminal careers. Journal of Quantitative Criminology, 4, 303–345.

17.

Bouffard

L. A.

Wright

K. A.

Muftić

L. R.

Bouffard

J. A.

(2008). Gender differences in specialization in intimate partner violence: Comparing the gender symmetry and violent resistance perspectives. Justice Quarterly, 25, 570–594.

18.

Bouffard

L. A.

Zedaker

S. B.

(2016). Are domestic violence offenders specialists? Answers from multiple analytic approaches. Journal of Research in Crime and Delinquency, 53, 788–813.

19.

Brame

Paternoster

Bushway

S. D.

(2004). Criminal offending frequency and offense switching. Journal of Contemporary Criminal Justice, 20, 201–214.

20.

Britt

C. L.

(1996). The measurement of specialization and escalation in the criminal career: An alternative modeling strategy. Journal of Quantitative Criminology, 12, 193–222.

21.

Browne

Finkelhor

(1986). Impact of child sexual abuse: A review of the research. Psychological Bulletin, 99, 66–77.

22.

Burgess

R. L.

Akers

R. L.

(1966). A differential association-reinforcement theory of criminal behavior. Social Problems, 14, 128–147.

23.

Buss

A. H.

Plomin

(1975). A temperament theory of personality development. Oxford, England: Wiley.

24.

Cale

Lussier

Proulx

(2009). Heterogeneity in antisocial trajectories in youth of adult sexual aggressors of women: An examination of initiation, persistence, escalation, and aggravation. Sexual Abuse: A Journal of Research and Treatment, 21, 223–248.

25.

Cale

Smallbone

Rayment-McHugh

Dowling

(2016). Offense trajectories, the unfolding of sexual and non-sexual criminal activity, and sex offense characteristics of adolescent sex offenders. Sexual Abuse: A Journal of Research and Treatment, 28, 791–812.

26.

Capaldi

D. M.

Patterson

G. R.

(1996). Can violent offenders be distinguished from frequent offenders: Prediction from childhood to adolescence. Journal of Research in Crime and Delinquency, 33, 206–231.

27.

Caudill

J. W.

Trulson

C. R.

(2016). The hazards of premature release: Recidivism outcomes of blended-sentenced juvenile homicide offenders. Journal of Criminal Justice, 46, 219–227.

28.

Crick

N. R.

Dodge

K. A.

(1994). A review and reformulation of social information-processing mechanisms in children’s social adjustment. Psychological Bulletin, 115, 74–101.

29.

DeLisi

(2003). The imprisoned nonviolent drug offender: Specialized Martyr or versatile career criminal? American Journal of Criminal Justice, 27, 167–182.

30.

DeLisi

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

31.

DeLisi

Beaver

K. M.

Wright

K. A.

Wright

J. P.

Vaughn

M. G.

Trulson

C. R.

(2011). Criminal specialization revisited: A simultaneous quantile regression approach. American Journal of Criminal Justice, 36, 73–92.

32.

DeLisi

Caudill

J. W.

Trulson

C. R.

Marquart

J. W.

Vaughn

M. G.

Beaver

K. M.

(2010). Angry inmates are violent inmates: A Poisson regression approach to youthful offenders. Journal of Forensic Psychology Practice, 10, 419–439.

33.

DeLisi

Drury

Elbert

Tahja

Caropreso

Heinrichs

(2017). Sexual sadism and criminal versatility: Does sexual sadism spillover into nonsexual crimes? Journal of Aggression, Conflict and Peace Research, 9, 2–12.

34.

DeLisi

Kosloski

A. E.

Vaughn

M. G.

Caudill

J. W.

Trulson

C. R.

(2014). Does childhood sexual abuse victimization translate into juvenile sexual offending? New evidence. Violence and Victims, 29, 620–635.

35.

DeLisi

Piquero

A. R.

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

36.

DeLisi

Vaughn

M. G.

(2014). Foundation for a temperament-based theory of antisocial behavior and criminal justice system involvement. Journal of Criminal Justice, 42, 10–25.

37.

Drury

Heinrichs

Elbert

Tahja

DeLisi

Caropreso

(2017). Adverse childhood experiences, paraphilias, and serious criminal violence among federal sex offenders. Journal of Criminal Psychology, 7, 105–119.

38.

Farabee

Joshi

Anglin

M. D.

(2001). Addiction careers and criminal specialization. Crime & Delinquency, 47, 196–220.

39.

Farrington

D. P.

Snyder

H. N.

Finnegan

T. A.

(1988). Specialization in juvenile court careers. Criminology, 26, 461–488.

40.

Finkel

E. J.

Hall

A. N.

(2018). The I³ Model: A meta-theoretical framework for understanding aggression. Current Opinion in Psychology, 19, 125–130.

41.

Forsman

Johansson

Santtila

Sandnabba

Långström

(2015). Sexually coercive behavior following childhood maltreatment. Archives of Sexual Behavior, 44, 149–156.

42.

Fox

DeLisi

(2018). From criminological heterogeneity to coherent classes: Developing a typology of juvenile sex offenders. Youth Violence and Juvenile Justice, 16, 299–318.

43.

Fox

B. H.

Farrington

D. P.

(2016). Behavioral consistency among serial burglars: Evaluating offense style specialization using three analytical approaches. Crime & Delinquency, 62, 1123–1158.

44.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

45.

Gottfredson

M. R.

Hirschi

(2016). The criminal career perspective as an explanation of crime and a guide to crime control policy: The view from general theories of crime. Journal of Research in Crime and Delinquency, 53, 406–419.

46.

Grant

J. E.

Won

S. K.

(2007). Clinical characteristics and psychiatric comorbidity of pyromania. The Journal of Clinical Psychiatry, 68, 1717–1722.

47.

Guerette

R. T.

Stenius

V. M.

McGloin

J. M.

(2005). Understanding offense specialization and versatility: A reapplication of the rational choice perspective. Journal of Criminal Justice, 33, 77–87.

48.

O. K.

Andresen

M. A.

(2017). Unemployment and the specialization of criminal activity: A neighborhood analysis. Journal of Criminal Justice, 48, 1–8.

49.

Haapanen

Steiner

(2003). Identifying mental health treatment needs among serious institutionalized delinquents using paper-and-pencil screening instruments, final report. Washington, DC: U.S. Department of Justice, Office of Justice Program, National Institute of Justice.

50.

Haapanen

Steiner

(2006. Assessing mental health problems among serious delinquents committed to the California Youth Authority, 1997-1999 [Computer file]. ICPSR04337-v1. Sacramento, CA: California Youth Authority [producer], 1999. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor].

51.

Hallstone

(2014). Types of crimes committed by repeat DUI offenders. Criminal Justice Studies, 27, 159–171.

52.

Hare

R. D.

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

53.

Harris

D. A.

Pedneault

Knight

R. A.

(2013). An exploration of burglary in the criminal histories of sex offenders referred for civil commitment. Psychology, Crime & Law, 19, 765–781.

54.

Harris

D. A.

Smallbone

Dennison

Knight

R. A.

(2009). Specialization and versatility in sexual offenders referred for civil commitment. Journal of Criminal Justice, 37, 37–44.

55.

Horning

A. M.

Salfati

C. G.

Crawford

(2010). Prior crime specialization and its relationship to homicide crime scene behavior type. Homicide Studies, 14, 377–399.

56.

Jennings

W. G.

Perez

N. M.

(2018). Sex offending: Empirical evidence and policy and practice. In DeLisi

Conis

P. J.

(Eds.), Violent offenders: Theory, research, policy, and practice (3rd ed., pp. 257–266). Burlington, MA: Jones & Bartlett Learning.

57.

Jennings

W. G.

Zgoba

K. M.

Donner

C. M.

Henderson

B. B.

Tewksbury

(2014). Considering specialization/versatility as an unintended collateral consequence of SORN. Journal of Criminal Justice, 42, 184–192.

58.

Jespersen

A. F.

Lalumière

M. L.

Seto

M. C.

(2009). Sexual abuse history among adult sex offenders and non-sex offenders: A meta-analysis. Child Abuse & Neglect, 33, 179–192.

59.

Karakasi

M. V.

Vasilikos

Voultsos

Vlachaki

Pavlidis

(2017). Sexual homicide: Brief review of the literature and case report involving rape, genital mutilation and human arson. Journal of Forensic and Legal Medicine, 46, 1–10.

60.

Kaszynski

Kallis

D. L.

Karnik

Soller

Hunter

Haapanen

… Steiner

(2014). Incarcerated youth with personality disorders: Prevalence, comorbidity and convergent validity. Personality and Mental Health, 8, 42–51.

61.

Kirsch

J. L.

Simeon

Berlin

Hollander

(2015). Disruptive, impulse control, and conduct disorders: Intermittent explosive disorder, kleptomania, and pyromania. In Tasman

Kay

Lieberman

J. A.

First

Riba

M. B.

(Eds.), Psychiatry, Fourth Edition (pp. 1367–1393). Hoboken, NJ: John Wiley.

62.

Lai

Zeng

Chu

C. M.

(2016). Violent and nonviolent youth offenders: Preliminary evidence on group subtypes. Youth Violence and Juvenile Justice, 14, 313–329.

63.

Landenberger

N. A.

Lipsey

M. W.

(2005). The positive effects of cognitive–behavioral programs for offenders: A meta-analysis of factors associated with effective treatment. Journal of Experimental Criminology, 1, 451–476.

64.

Langton

C. M.

Murad

Humbert

(2017). Childhood sexual abuse, attachments in childhood and adulthood, and coercive sexual behaviors in community males: Main effects and a moderating function for attachment. Sexual Abuse: A Journal of Research and Treatment, 29, 207–238.

65.

Lin

Simon

(2016). Examining specialization among sex offenders released from prison. Sexual Abuse: A Journal of Research and Treatment, 28, 253–267.

66.

Lindberg

Holi

M. M.

Tani

Virkkunen

(2005). Looking for pyromania: Characteristics of a consecutive sample of Finnish male criminals with histories of recidivist fire-setting between 1973 and 1993. BMC Psychiatry, 5, 47.

67.

Lussier

(2005). The criminal activity of sexual offenders in adulthood: Revisiting the specialization debate. Sexual abuse: A Journal of Research and Treatment, 17, 269–292.

68.

Lussier

LeBlanc

Proulx

(2005). The generality of criminal behavior: A confirmatory factor analysis of the criminal activity of sex offenders in adulthood. Journal of Criminal Justice, 33, 177–189.

69.

Lynam

D. R.

(1996). Early identification of chronic offenders: Who is the fledgling psychopath? Psychological Bulletin, 120, 209–234.

70.

Lynam

D. R.

Piquero

A. R.

Moffitt

T. E.

(2004). Specialization and the propensity to violence: Support from self-reports but not official records. Journal of Contemporary Criminal Justice, 20, 215–228.

71.

Mattek

Hanson

R. K.

(2018). Committed as a violent sexual predator in his 10th decade: A case study. Archives of Sexual Behavior, 47, 543–550.

72.

McCoy

Vaughn

M. G.

Maynard

B. R.

Salas-Wright

C. P.

(2014). Caution or warning? A validity study of the MAYSI-2 with juvenile offenders. Behavioral Sciences & the Law, 32, 508–526.

73.

McCuish

E. C.

Cale

Corrado

R. R.

(2018). A prospective study of offending patterns of youth homicide offenders into adulthood: An examination of offending trajectories and the crime mix posthomicide. Youth Violence and Juvenile Justice, 16, 18–36.

74.

McGloin

J. M.

Sullivan

C. J.

Piquero

A. R.

Blokland

Nieuwbeerta

(2011). Marriage and offending specialization: Expanding the impact of turning points and the process of desistance. European Journal of Criminology, 8, 361–376.

75.

McGloin

J. M.

Sullivan

C. J.

Piquero

A. R.

(2008). Aggregating to versatility? Transitions among offender types in the short term. British Journal of Criminology, 49, 243–264.

76.

McGloin

J. M.

Sullivan

C. J.

Piquero

A. R.

Pratt

T. C.

(2007). Local life circumstances and offending specialization/versatility: Comparing opportunity and propensity models. Journal of Research in Crime and Delinquency, 44, 321–346.

77.

Meloy

J. R.

(2000). The nature and dynamics of sexual homicide: An integrative review. Aggression and Violent Behavior, 5, 1–22.

78.

Meloy

J. R.

(2010). A catathymic infanticide. Journal of Forensic Sciences, 55, 1393–1396.

79.

Miethe

T. D.

Olson

Mitchell

(2006). Specialization and persistence in the arrest histories of sex offenders: A comparative analysis of alternative measures and offense types. Journal of Research in Crime and Delinquency, 43, 204–229.

80.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674–701.

81.

Moffitt

T. E.

(2018). Male antisocial behaviour in adolescence and beyond. Nature Human Behaviour, 2, 177–186.

82.

Paternoster

Brame

Piquero

Mazerolle

Dean

C. W.

(1998). The forward specialization coefficient: Distributional properties and subgroup differences. Journal of Quantitative Criminology, 14, 133–154.

83.

Patterson

G. R.

(1986). Performance models for antisocial boys. American Psychologist, 41, 432–444.

84.

Pedneault

Beauregard

Harris

D. A.

Knight

R. A.

(2015). Rationally irrational: The case of sexual burglary. Sexual Abuse: A Journal of Research and Treatment, 27, 376–397.

85.

Pedneault

Harris

D. A.

Knight

R. A.

(2012). Toward a typology of sexual burglary: Latent class findings. Journal of Criminal Justice, 40, 278–284.

86.

Piquero

(2000). Frequency, specialization, and violence in offending careers. Journal of Research in Crime & Delinquency, 37, 392–418.

87.

Piquero

A. R.

Farrington

D. P.

Jennings

W. G.

Diamond

Craig

(2012). Sex offenders and sex offending in the Cambridge study in delinquent development: Prevalence, frequency, specialization, recidivism, and (dis) continuity over the life-course. Journal of Crime and Justice, 35, 412–426.

88.

Piquero

Oster

R. P.

Mazerolle

Brame

Dean

C. W.

(1999). Onset age and offense specialization. Journal of Research in Crime and Delinquency, 36, 275–299.

89.

Schlesinger

L. B.

Revitch

(1999). Sexual burglaries and sexual homicide: Clinical, forensic, and investigative considerations. Journal of the American Academy of Psychiatry and the Law, 27, 227–238.

90.

Schwartz

C. E.

Snidman

Kagan

(1996). Early childhood temperament as a determinant of externalizing behavior in adolescence. Development and Psychopathology, 8, 527–537.

91.

Shover

(1996). Great pretenders: Pursuits and careers of persistent thieves. Boulder, CO: Westview Press.

92.

Simon

L. M.

(2000). An examination of the assumptions of specialization, mental disorder, and dangerousness in sex offenders. Behavioral Sciences & the Law, 18, 275–308.

93.

Smallbone

S. W.

Wheaton

Hourigan

(2003). Trait empathy and criminal versatility in sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 15, 49–60.

94.

Stander

Farrington

D. P.

Hill

Altham

P. M.

(1989). Markov chain analysis and specialization in criminal careers. British Journal of Criminology, 29, 317–335.

95.

Soothill

Francis

Liu

(2008). Does serious offending lead to homicide? Exploring the interrelationships and sequencing of serious crime. British Journal of Criminology, 48, 522–537.

96.

Steffensmeier

D. J.

Ulmer

J. T.

(2005). Confessions of a dying thief: Understanding criminal careers and illegal enterprise. New Brunswick, NJ: Transaction Books.

97.

Sullivan

C. J.

McGloin

Pratt

T. C.

Piquero

A. R.

(2006). Rethinking the “norm” of offender generality: Investigating specialization in the short-term. Criminology, 44, 199–233.

98.

Sullivan

C. J.

McGloin

J. M.

Ray

J. V.

Caudy

M. S.

(2009). Detecting specialization in offending: Comparing analytic approaches. Journal of Quantitative Criminology, 25, 419–441.

99.

Trojan

Salfati

(2016). Criminal history of homicide offenders: A multi-dimensional analysis of criminal specialization. Journal of Criminal Psychology, 6, 28–41.

100.

Trulson

C. R.

Caudill

J. W.

Haerle

D. R.

DeLisi

(2012). Cliqued up: The postincarceration recidivism of young gang-related homicide offenders. Criminal Justice Review, 37, 174–190.

101.

Trulson

C. R.

DeLisi

Caudill

J. W.

Belshaw

Marquart

J. W.

(2010). Delinquent careers behind bars. Criminal Justice Review, 35, 200–219.

102.

Trulson

C. R.

Haerle

D. R.

Caudill

J. W.

DeLisi

(2016). Lost causes: Blended sentencing, second chances, and the Texas Youth Commission. Austin: University of Texas Press.

103.

Tumminello

Edling

Liljeros

Mantegna

R. N.

Sarnecki

(2013). The phenomenology of specialization of criminal suspects. PLoS One, 8, e64703.

104.

Tzoumakis

Lussier

Blanc

M. L.

Davies

(2013). Onset, offending trajectories, and crime specialization in violence. Youth Violence and Juvenile Justice, 11, 143–164.

105.

van Onna

J. H.

van der Geest

V. R.

Huisman

Denkers

A. J.

(2014). Criminal trajectories of white-collar offenders. Journal of Research in Crime and Delinquency, 51, 759–784.

106.

van Wijk

Loeber

Vermeiren

Pardini

Bullens

Doreleijers

(2005). Violent juvenile sex offenders compared with violent juvenile nonsex offenders: Explorative findings from the Pittsburgh Youth Study. Sexual Abuse: A Journal of Research and Treatment, 17, 333–352.

107.

Vaughn

M. G.

DeLisi

Beaver

K. M.

Howard

M. O.

(2008). Toward a quantitative typology of burglars: A latent profile analysis of career offenders. Journal of Forensic Sciences, 53, 1387–1392.

108.

Vaughn

M. G.

DeLisi

Beaver

K. M.

Howard

M. O.

(2009). Multiple murder and criminal careers: A latent class analysis of multiple homicide offenders. Forensic Science International, 183, 67–73.

109.

Vaughn

M. G.

DeLisi

Wright

J. P.

Beaver

K. M.

Perron

B. E.

Howard

M. O.

(2010). Prevalence and correlates of fire-setting in the United States: Results from the national epidemiological survey on alcohol and related conditions. Comprehensive Psychiatry, 51, 217–223.

110.

Vaughn

M. G.

Salas-Wright

C. P.

DeLisi

Shook

J. J.

Terzis

(2015). A typology of drug selling among young adults in the United States. Substance Use & Misuse, 50, 403–413.

111.

Walters

G. D.

(2016). Age of crime/substance onset and crime/drug versatility as dimensions of the “worst of both worlds” effect. Criminal Justice Policy Review, 27, 668–683.

112.

Walters

G. D.

Geyer

M. D.

(2004). Criminal thinking and identity in male white-collar offenders. Criminal Justice and Behavior, 31, 263–281.

113.

Weinrott

M. R.

Saylor

(1991). Self-report of crimes committed by sex offenders. Journal of Interpersonal Violence, 6, 286–300.

114.

Widom

C. S.

Massey

(2015). A prospective examination of whether childhood sexual abuse predicts subsequent sexual offending. JAMA Pediatrics, 169, e143357–e143357.

115.

Wiesner

Yoerger

Capaldi

D. M.

(2018). Patterns and correlates of offender versatility and specialization across a 23-year span for at-risk young men. Victims & Offenders, 13, 28–47.

116.

Wright

K. A.

Bouffard

L. A.

(2016). Capturing crime: The qualitative analysis of individual cases for advancing criminological knowledge. International Journal of Offender Therapy and Comparative Criminology, 60, 123–145.

117.

Wright

K. A.

Pratt

T. C.

DeLisi

(2008). Examining offending specialization in a sample of male multiple homicide offenders. Homicide Studies, 12, 381–398.

118.

Wright

K. A.

Pratt

T. C.

DeLisi

(2009). Multiple homicide offenders: Arbitrary cut-off points and selection bias. Homicide Studies, 13, 193–199.

119.

Wright

R. T.

Decker

S. H.

(1994). Burglars on the job: Streetlife and residential break-ins. Boston, MA: Northeastern University Press.

120.

Youngs

Ioannou

Eagles

(2016). Expressive and instrumental offending: Reconciling the paradox of specialization and versatility. International Journal of Offender Therapy and Comparative Criminology, 60, 397–422.