Skip to main content
SearchLoginLogin or Signup

A prospective study of offending patterns of youth homicide offenders into adulthood: An examination of offending trajectories and the crime-mix post-homicide

Published onJan 01, 2018
A prospective study of offending patterns of youth homicide offenders into adulthood: An examination of offending trajectories and the crime-mix post-homicide


Although youth homicide offenders (YHOs) are portrayed as a group that warrants considerable attention from the justice system because of their high likelihood of future offending, little is known about this group’s offending trajectories and the nature of post-homicide offenses in adulthood. These questions were investigated using a sample of male and female YHOs (n = 26), violent youth non-homicide offenders (VYNHOs; n = 358), and non-violent youth non-homicide offenders (NVYNHOs; n =139), all of whom were followed prospectively into adulthood. First, the prevalence of adult recidivism did not vary across the three groups. Second, YHOs were more frequent offenders prior to their homicide offense than after their homicide offense, and when they did offend post-homicide, it was typically a non-serious crime. Third, YHOs did not differ from other offenders in their association to a specific offending trajectory. These findings are discussed in the context of assessment and treatment of serious and violent youth.

Keywords: crime-mix; criminal careers; homicide; recidivism; violence; youth

A prospective study of offending patterns of youth homicide offenders into adulthood: An examination of offending trajectories and the crime-mix post-homicide

A prevailing assumption is that youth homicide offenders (YHOs) represent the most serious type of young offender, even at the ‘deep end’ (Mulvey et al., 2004) of the justice system. The fact they have committed the most serious crimes, combined with the high costs associated with homicide offenses (DeLisi et al., 2010), has prompted political leaders and policymakers to support efforts to identify childhood and early-adolescent risk factors that help predict which youth will be involved in a homicide offense (DeLisi, Piquero, & Cardwell, 2014; Lee, 2013). However, there is minimal evidence of meaningful distinctions in the developmental risk factor profiles of YHOs and other serious and violent youth offenders that would allow for reliable predictions of risk for involvement in a homicide offense (DeLisi et al., 2014; Farrington, Loeber, & Berg, 2012; Loeber & Farrington, 2011). For example, although some studies have identified certain risk factors that were more common among homicide offenders compared to other offenders, false positives occurred in upwards of 90% of cases (see Loeber et al., 2005; Loeber & Farrington, 2011). In other words, developmental risk factors common among young homicide offenders also tend to be common for other serious and violent youth (Loeber & Farrington, 2011). Being a YHO is too rare, and the risk factors associated with YHOs are too common among other serious and violent offenders to reliably designate particular youth as being at-risk for a homicide offense.

The assumption that YHOs have a more serious risk factor profile compared to other serious and violent youth is problematic because of the situational and often unplanned nature of most homicide offenses (e.g., Felson & Steadman, 1983). Many homicide offenses arise in contexts that are similar to other crimes such as assaults but only differ in the outcome to the victim. Furthermore, even in cases of ‘planned’ homicide offenses where intent to murder is present, the premature development of executive functions combined with lower social cognitive skills common for adolescents mean that many of them do not fully understand the consequences of their actions (e.g., Larden, Melin, Holst, & Langstrom, 2005). This is particularly true for adolescents with neuropsychological deficits (Barriga, Sullivan-Cosetti, & Gibbs, 2009). Nonetheless, the penalties and interventions associated with homicide offenses are severe and may lead to negative consequences that result from social processes such as labeling.

A neglected dimension of research on YHOs is the nature and extent of offending patterns post-homicide beyond strictly recidivism. Although involvement in the juvenile justice system is a key precursor to involvement in the adult criminal justice system (Gilman, Hill, & Hawkins, 2015), especially for serious and violent youth (Trulson, Haerle, DeLisi, & Marquart, 2011), a key question is whether and how a homicide offense in youth impacts the unfolding of offending in adulthood. Given that developmental risk factors are not likely sufficient to meaningfully differentiate most YHOs from those involved in serious non-lethal violence (DeLisi et al., 2014), a prospective approach was used to examine YHO and youth non-homicide offender (YNHO) involvement in new crimes as they transitioned through the early stages of adulthood. First, we examined changes in criminal versatility, or the crime-mix, pre- and post-homicide among YHOs who were interviewed in adolescence as part of the Incarcerated Serious and Violent Young Offender Study (ISVYOS). Second, we compared offending trajectories measured from ages 12-28 across male and female YHOs (n = 26), violent youth non-homicide offenders (VYNHOs; n = 358), and non-violent youth non-homicide offenders (NVYNHOs; n = 139). Below, the current state of knowledge concerning the offending patterns of young homicide offenders is reviewed.

Offending patterns of young homicide offenders

Studies have produced mixed results in terms of comparisons of criminal histories prior to a young person’s involvement in homicide. This is likely related, at least in part, to the timing in adolescence of the homicide offense. For example, DiCataldo and Everett (2008) found that YNHOs had a higher frequency of delinquency and violence in their criminal histories compared to YHOs. On the other hand, they found that YHOs were slightly younger (approximately one year) than YNHOs at the age of their first violent conviction. Using data from the Pittsburgh Youth Study, Loeber et al. (2005) observed that virtually all (31 of 33) of their young male homicide offenders (defined as participants that committed a homicide up to age 26) were previously involved in violent crimes. However, using the same data, Farrington et al. (2012) found that homicide offenders were less likely to be chronic offenders relative to other offenders. Using retrospective data on 455 habitual adult male offenders, DeLisi, Hochstetler, Jones-Johnson, Caudill, and Marquart (2011) observed that chronic offending, measured by arrests, was not associated with prior homicide offending. Similarly, in their study of the population of YHOs in England and Wales, Rodway et al. (2011) found that only 52% of YHOs were previously convicted. One possible explanation that can be drawn from these findings is that young homicide offenders spend long periods of time incarcerated, which reduces their offending opportunities and impacts the unfolding of offending trajectories.

The lengthy sentences served by YHOs also make it challenging to examine offending patterns following their homicide offense (Heide, Solomon, Sellers, & Chan, 2011). The few studies that examined post-homicide offending patterns of YHOs typically relied on recidivism outcomes, most often measured either as a reconviction or return to custody for any offense. Hagan (1997) compared the prevalence of recidivism between YHOs (n = 20) and YNHOs (n = 20) over a follow-up period of at least five years post-release. Using a return to prison as the definition of reoffending, no differences were evident in the prevalence of recidivism between the groups; 60% of YHOs reoffended during the follow-up period compared to 65% of YNHOs. However, those convicted of first degree murder were never released from custody during the follow-up period. Considering that those involved in first degree murder may be the most serious offenders, the prevalence of recidivism among YHOs may have been underestimated.

In a larger study from the Netherlands, Vries and Liem (2011) examined the prevalence of recidivism among a sample of 137 YHOs. Like the study by Hagan (1997), approximately 60% of YHOs reoffended, though the follow-up period from one to 16 years. Of those followed at least 10 years, nearly three-quarters (71%) reoffended. In effect, the prevalence of recidivism increased with the length of follow-up period, with the prevalence of recidivism nearly doubling between the first and fifth year of follow-up in the study by Vries and Liem (2011). Differences in time until recidivism among YHOs reflected differences in the frequency of offending for this group. YHO recidivists averaged nearly eight new crimes after their homicide, with a range of 1-42 new offenses. Seventy-five percent of recidivists committed at least two new crimes. In effect, among recidivists, most were multiple recidivists. Taken together, a strict emphasis on recidivism outcomes may mask substantial within-group heterogeneity in the offending patterns of this group. Other studies have produced comparable estimates; in a sample of 59 YHOs given adult sentences for their involvement in murder, manslaughter, or attempted murder, Heide et al. (2001) indicated that 60% of YHOs reoffended (defined as a subsequent incarceration), most of whom reoffended in the first three years following their release from custody.

Similar patterns were also observed in a recent and large American study that involved YHOs. Trulson, Haerle, Caudill, and DeLisi (2016) examined the prevalence and timing of recidivism among a sample of approximately 1,400 felony youth offenders, including 277 YHOs. In terms of prevalence, approximately two-thirds (64%) of YHOs were considered recidivists and the prevalence of recidivism among YHOs did not statistically differ from other youth in the sample. Of those followed for at least five years post-release into the community, almost two-thirds (approximately 60%) of YHOs were re-arrested at some point over the five-year period; one half of whom rearrested in the first three years post-release. They also identified characteristics of YHOs that increased the risk of recidivism. On the one hand, being male, African American ethnicity, having a history of institutional assaults, and a shorter incarceration period were all related to an increased likelihood of reoffending post-release among YHOs. On the other hand, participation in a specialized treatment program for YHOs significantly reduced the risk of recidivism. Given that the sample consisted of particularly serious and violent youth, and that, overall, three quarters of the overall sample went on to commit subsequent felony (i.e., serious) offenses post-release, these findings are important because they highlight potential within-YHO variability in the severity of future offending patterns.

In another effort to capture within-YHO variability, Khachatryan, Heide, and Hummel (2016a) examined the prevalence of recidivism across two types of YHOs followed over 33 years. The first type was YHOs involved in homicide during the commission of another crime and the second were those involved in homicide as part of some on-going conflict with the victim. They found minimal differences between the groups on a variety of parameters. These two groups did not differ in terms of age at homicide offense, presence of a prior criminal offense, presence of previous violence, age of onset of general offending, number of prior arrests, and time incarcerated for their homicide offense. Like the study by Vries and Liem (2011), these YHOs were incarcerated for approximately eight years before release. The prevalence of recidivism (defined as a re-arrest) was high (88% of the full sample), over half of the sample reoffended violently, and nearly 40% were re-arrested at least seven times. Clearly, YHOs were not a homogenous group in terms of their recidivism patterns, as the frequency of re-arrest ranged from 0-30 for general offending and 0-23 for violent offending specifically (Khachatryan et al., 2016a).

Recognizing that general recidivism outcomes are quite prevalent among serious offenders, in another study Caudill and Trulson (2016) examined the seriousness (i.e., felony compared to non-felony) of post-homicide offenses. Among a sample of 221 YHOs followed ten years after release from custody, 58% committed a subsequent felony offense with approximately 70% of recidivists reoffending within the first two years of release. The authors observed that lengthier periods of incarceration were associated with a decreased likelihood of felony (i.e., serious) recidivism post-release.

Across studies, differences in the definition of recidivism and the length of follow-up period, failure to control for exposure time in some studies, and strict focus on a reoffense as a post-release outcome of YHOs inhibits a more detailed understanding of their adult offending patterns. Given the prevalence of recidivism among YHOs increases with the length of the follow-up period, a reoffense, whether measured by subsequent arrests, convictions, or incarceration alone is not an optimal measure of seriousness and gives the impression that all YHOs remain at risk of reoffending for long periods of time. Indeed, some studies have already shown that YHO recidivists vary in the frequency and seriousness of reoffending (e.g., Caudill & Trulson, 2016; Khachatryan et al., 2016b; Trulson et al., 2016; Vries & Liem, 2011). From a risk-needs-responsivity perspective (Andrews & Bonta, 2010), making distinctions in the level of severity of YHOs has important implications for the efficacy of criminal justice system intervention strategies. From this perspective, the assumption that all YHOs will become prolific or serious adult offenders can result in increasing the risk that would-be nonrecidivists or YHOs characterized by a low-rate offending pattern will develop into more frequent and serious adult offenders. In effect, understanding offending trajectories and the crime-mix post-homicide among YHOs has important implications for which YHOs warrant more intensive intervention from the justice system (Liem, 2013) and the allocation of additional resources for this group (DeLisi et al., 2014).

Study aims

For various arms of the criminal justice system, whether YHOs are likely to recidivate, how often they recidivate, and the types of crimes that they commit after their release from custody has implications for the development of policy concerning this group (Liem, 2013). As noted by others (e.g., Hagan, 1997; Khachatryan et al., 2016a), apart from the handful of studies discussed above, empirical descriptions of the post-incarceration behaviors of YHOs is limited compared to research on the retrospective, pre-homicide characteristics and behaviors of this group. This is particularly true when considering measures of offending beyond strictly recidivism outcomes. The current study addressed whether the offending patterns of male and female YHOs differed from VYNHOs and NVYNHOs to better understand whether YHOs represent a distinct category of offender warranting particular attention from the justice system. To accomplish this, three research questions were posed:

(1) Are there differences in the prevalence of adult recidivism across YHOs, VYNHOs, and NVYNHOs;

(2) Are there differences in the crime-mix of YHOs prior to versus after their homicide offense; and,

(3) Are YHOs associated with a specific offending trajectory that differs from that of other serious and violent youth?



Data for the current study were derived from the Incarcerated Serious and Violent Young Offender Study (ISVYOS), which was conducted in British Columbia, Canada and has been ongoing since 1998. As part of this study, male and female youth between ages twelve and nineteen were interviewed in open and secure custody facilities within the Greater Vancouver Regional District and surrounding areas. The focus of the current study is only on youth from the first wave of data collection that were followed prospectively into adulthood (n = 523). The initial sample included 527 youth, but criminal records were not found for 0.8% (n = 4) of participants. Indigenous persons are overrepresented in detention facilities in Canada. Likewise, the percentage of Indigenous participants in the current study (23.9%) was substantially higher than within the general population of British Columbia (6.2%; Statistics Canada, 2013). The sample is by no means representative of all youth involved in crime in the province of British Columbia. However, it is representative of offenders in the Canadian youth justice system sentenced to detention and of youth with more serious patterns of offending. Youth included in the sample were convicted of, on average, 12.17 offenses (SD = 8.74) between age 12 and 17 and averaged about one year incarcerated during this same period (353.99 days; SD = 294.92). Additional descriptive information about the sample is presented in Table 1.

--Insert Table 1 about Here--


The purpose of the ISVYOS was to obtain information on risk factors associated with various criminal career parameters. Self-report interviews were conducted and file-based information was collected on a sample of youth incarcerated in various detention centers throughout British Columbia. The British Columbia Ministry of Child and Family Development acts as the legal guardian to all youth in custody and their consent allowed the research team to ask incarcerated youth if they wished to participate in the study. Youth were eligible to participate in the study if all the following criteria were met: (1) were English-speaking, (2) demonstrated an understanding of interview questions (e.g., had no noticeably severe learning disability), and (3) were willing to provide accurate information. Data concerning refusal rates were not collected during the entire course of the study, but in the time that such data were collected, approximately 5% refused. Research assistants (RAs) interviewed youth in an isolated interview room to help ensure confidentiality. All participants were read and given a copy of an information sheet which explained the purpose of the study, how information would be collected (i.e., interview and file information), and that all information would be kept confidential unless the participant made a direct threat against themselves or someone else. To improve the reliability of self-reported information, RAs accessed case management files, which contained presentence reports and other information, to help detect discrepancies between interview responses and official records.


Measures of offending. Offending was measured using data from British Columbia Corrections’ computerized system, Corrections Network (CORNET), which contains information pertaining to each offender’s movement in and out of custody as well as the exact criminal offense, date of conviction, and sentence type. Criminal convictions were coded for the entire sample from age 12 to age 28. The start of the follow-up period, age 12, represents the minimum age of criminal responsibility in Canada. From that point on, every criminal charge that resulted in a conviction was coded. The average number of charges that resulted in conviction between ages 12-28 was 22.50 (SD = 17.28). The median number of convictions was 18, highlighting that the high mean number of convictions found was not an artifact of a small subgroup of individuals (i.e., chronic offenders). Although it was possible for offenders to commit new offenses while outside of the province, the current study only had access to records of offenses committed within the province of British Columbia. During the follow-up period, 26 offenders died (5.0% of the sample). CORNET does not include criminal history information for crimes committed outside the province of British Columbia, and 21 participants (4.0%) moved outside of the province during the study period. For these participants, convictions after the age of death or age of move were coded as missing rather than as ‘zero’. Although official data has the disadvantage of underestimating offending, the current study is focused on more serious crimes that are less likely to be self-reported (e.g., rape, homicide, assault; see Stouthamer-Loeber, Loeber, Stallings & Lacourse, 2008) and this also minimizes memory recall bias issues.

In line with prior studies (e.g., DeLisi et al., 2014; Hagan, 1997), to be categorized as a YHO (n = 26), an offender must have been charged with first degree murder, second degree murder, manslaughter, or attempted murder. Attempted murder was included in this conceptualization of homicide because of its fundamental similarity to murder. Indeed, Shumaker and McKee (2001) noted that young persons charged with attempted murder did not differ from other homicide offenders regarding demographic characteristics, familial background, use of co-offenders, and prior offending behavior. Similarly, Myers, Scott, Burgess, and Burgess (1995) asserted that progression from attempted murder to murder was primarily the result of chance. Reiterating this assertion, Heide et al. (2001) noted that what often distinguished attempted murder from murder was the health of the victim at the time of the offense (also see DeLisi et al., 2014; Hagan, 1997 for a similar rationale). Although ages 12-17 represents the definition of a young offender in Canada, homicide charges through age 19 were used to define YHOs because of growing recognition that adulthood begins later than initially conceptualized (e.g., Arnett, 2000). This approach was similar to prior definitions of the age range of YHOs (e.g., Gerard, Jackson, Chou, Whitfield, & Browne, 2014). Excluding YHOs that died, moved, or were never released (n = 6), the group was followed for an average of 10.35 years (SD = 2.32; range of 4-14) after being released from custody. VYNHOs (n = 358) were defined as participants convicted of a violent offense between ages 12-19; all other participants were NVYNHOs (n = 139).

Analytic strategy

The current study examined whether criminal career parameters varied between YHOs, VYNHOs, and NVYNHOs. This comparison begins with an examination of traditional offending outcomes (e.g., recidivism) across the three groups of offenders. Based on public and policy makers’ concerns about YHOs following their release (Liem, 2013), YHO offending patterns up to and including their homicide offense were compared to their offending patterns after their homicide offense. Specific interest was in the qualitative nature of YHO offending patterns following their reentry to the community. To examine this, seven offense categories were used to create a variable representing the crime-mix of YHOs: violent, property, violation of court orders, weapon, miscellaneous, drug, and sexual offenses. Given that YHOs committed their homicide offenses at different ages, the length of time after the homicide offense and time before the homicide offense varied across the subsample. As such, rather than examine total crimes pre- versus post-homicide, the analysis concerned average number of each crime type committed per year prior to and including the homicide versus the average number of each crime type committed following their homicide offense. This allowed for an interpretation of whether the crime-mix of YHOs changed after their homicide offense. Thus, the crime-mix reflects the relative importance of seven different crime types and whether more serious crimes became more prominent in the post-homicide criminal careers of YHOs.

Following this line of analysis, Proc TRAJ for SAS 9.4 (Jones & Nagin, 2007) was used to conduct semi-parametric group-based modeling (SPGM), to identify the number and shape of offending trajectories, measured from ages 12-28, that best fit the sample. Very importantly, the SPGM analysis in the current study included a measure of exposure time, number of days in the community for each year of age, to control for time at risk. Piquero et al. (2001) illustrated that desistance within their sample of youth from the California Youth Authority study was substantially overestimated when studies did not account for time incarcerated. The current study used Piquero et al.’s (2001) original formula to calculate exposure time for each participant at each person-period observation:

Exposureji = 1 - (Number of Days Incarcerated/367)

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

Number of days incarcerated was divided by 367 as (a) exposure must be a non-zero value and (b) offenders, on occasion, spent 366 days incarcerated due to leap years.


Offender-type and offending outcomes

In Table 2, offending parameters were compared across the three offender-types. In adolescence (ages 12-17), YHOs and NVYNHOs averaged significantly fewer convictions compared to VYNHOs (p < .001). With respect to time incarcerated, YHOs and VYNHOs averaged a significantly greater number of days incarcerated compared to NVYNHOs. In adulthood (ages 18-28), YHOs averaged significantly fewer convictions compared to VYNHOs (p < .01) but spent a significantly greater number of days incarcerated compared to both VYNHOs and NVYNHOs (p < .001). Individuals incurring a new conviction between ages 20-28 were defined as adult recidivists. The lower bound of adulthood was defined as age 20 rather than age 18 to ensure that there was no overlap between receiving a charge or conviction for a homicide offense between ages 18-19 and the measure of recidivism in adulthood. As shown in Table 2, excluding study participants that died and did not recidivate (n = 6) or moved and did not recidivate (n = 2), offender-type was unrelated (p > .05) to the prevalence of recidivism in adulthood, with YHOs (70.8%) and VYNHOs (74.6%) trending towards a higher prevalence of recidivism compared to NVYNHOs (65.0%). After excluding from the recidivism analyses the group of YHOs that were never released from custody and did not recidivate1 (n = 4), there was a marginally significant association between offender-type and recidivism (χ2 (2) = 6.0, p = .05). Specifically, YHOs had the highest prevalence of recidivism in adulthood (84.2%, n = 16) compared to VYNHOs NVYNHOs. In other words, once exposure time/opportunity was better accounted for, YHOs trended towards a greater likelihood of recidivism compared to other offenders.

To explore this relationship further, a logistic regression analysis was performed with the latter recidivism outcome regressed on offender-type, gender, and ethnicity included as covariates. The overall model was significant (-2LL = 569.18, χ2 (4) = 24.44, p < .001), but offender-type was unrelated to recidivism. Being male increased the odds of recidivism by a factor of 2.72 (p < .001). At the bivariate level, when another chi-square analysis was performed after removing from the sample all female YHOs (n = 6), VYNHOs, (n = 77), and NVYNHOs (n = 36) the association observed between offender-type and any of the abovementioned measures of adult recidivism disappeared.

--Insert Table 2 about Here--

Pre- and post-homicide offending patterns

The aggregate crime-mix of YHOs was measured up to and including each YHO’s homicide offense and then measured for each YHO following their homicide offense (see Table 3). This comparison did not include YHOs that were never released from custody following their conviction for their homicide offense. A paired-samples t-test was used to compare frequency of convictions across both measurement periods. After their homicide offense, YHOs averaged 0.43 total convictions per person-period observation (i.e., per year), an offense rate that was significantly lower (p < .05) compared to average total convictions per person-period observation in the period leading up to and including their homicide offense. Overall, a significant (p < .05) and decreasing pattern of offending frequency was observed for total crimes, violent crimes, and property crimes. In inspecting the average frequency of each conviction type included in Table 3, when YHOs did commit new offenses following release, they tended to commit relatively minor crimes such as violations of court orders and miscellaneous offenses (e.g., drinking and driving, dangerous driving). The crime-mix of YHOs is depicted in Figure 1 to help visually illustrate differences in pre- and post-release offending patterns. Importantly, in comparing the pre- and post-incarceration crime-mix of YHOs, time incarcerated was not accounted for. However, as shown in Table 3, there was no significant difference in number of days incarcerated per person-period observation across the two time-periods.

--Insert Figure 1 and Table 3 about Here--

Offending trajectories

With recidivism in adulthood common for all offender-types, offending trajectories were examined to help capture within-group differences in offending patterns among the sample. Model identification was the first stage of the SPGM analysis, which involved specification of the number and shape of the offending trajectories that best fit the data. A zero-inflated Poisson (ZIP) model with quadratic functional form was used to estimate the distribution of the offending trajectories. Bayesian Information Criteria (BIC) values are typically used to identify the number of trajectories that best fit the data, with values closer to zero indicating improved fit. A four-group solution showed a BIC value closer to zero (-16106.83) compared to both a three-group model (-16245.12) and a five-group model (-17918.25). Jeffrey’s scale of the evidence of the Bayes factor was used to validate the retention of the four-group solution. Using the formula e BICi – BICj, where values of Bij greater than ten indicate strong evidence for model ‘i’ (see Nagin, 2005), there was strong evidence for a four-group model over a five-group model (Bij >10) but not for a three-group model over a four-group model (Bij <10). Parameters of the four-group model are shown in Table 4. Classification accuracy was determined by the average posterior probability of accurately assigning individuals to a particular trajectory, and values were high for each trajectory (range of .94-.99). Odds of correct classification (OCC) provided additional confidence that individuals were assigned to the appropriate trajectory, and was calculated as:

OCCg = (AvePPg/ (1-AvePPg)) / (∏g/ (1-∏g))

where ∏g is the estimated size of group g (see Skardhamar, 2010).

--Insert Table 4 about Here--

OCC values for each of the four trajectories were greater than five (range of 18.81-98.54), which indicates high classification accuracy (Skardhamar, 2010). Based on their shape, the trajectories (see Figure 2) were labeled: adolescent limited (40.9% of the sample; n = 214), stable low (18.7%; n = 98), high rate desisting (HRD; 18.7%; n = 98), and high rate persisting (HRP; 21.6%; n = 113). For all four trajectories, offending peaked in late adolescence. From ages 12-17, the trajectories resembled one another in shape but not in height. After this period, however, each trajectory took on a unique shape. As its name implies, the adolescent limited trajectory reached a near-zero level of offending shortly into adulthood. In contrast, the stable low trajectory showed no decline in level of offending from age 18 onward. The HRD trajectory reached a near-zero level of offending that was similar to the adolescent limited trajectory, but did so at a much later age period. Finally, although the HRP trajectory showed a declining frequency of offending from ages 15 to 28, this trajectory averaged the greatest number of convictions at each person-period observation relative to all other trajectories.

--Insert Figure 2 about Here--

Offending trajectories and sample characteristics. At the bivariate level, ethnicity (χ2 (6) = 22.42, p < .01) and gender (χ2 (3) = 42.66, p < .001) were both associated with trajectory-group, with non-Indigenous minorities least likely to be associated with a chronic offending trajectory compared to other ethnic groups and males more likely to be associated with a chronic offending trajectory compared to females. Using Fisher’s exact test, as shown in Table 4, offender-type (YHO, VYNHO, NVYNHO) was marginally related to trajectory association (χ2 (6) = 12.38, p = .06). The prevalence of each offender-type within each trajectory is shown at the bottom of Table 4. In total, only approximately 27% of YHOs were in the two chronic offending trajectories, compared to approximately 50% of VYNHOs and 33% of VYNHOs. Over half of all YHOs were in the adolescent limited trajectory. From ages 12-17 and ages 18-28, this trajectory averaged the lowest frequency of offending and averaged the least amount of time spent incarcerated of the four trajectories identified (see Table 4). To look more closely at the relationship between offender-type and trajectory association, specific comparisons were made between YHOs and YNHOs across each of the four trajectories. As shown in Table 5, there was no statistically significant difference between the prevalence of YHOs and YNHOs within each trajectory.


In prior studies, minimal differences have been observed in the prevalence of recidivism between YHO and YNHO (e.g., Hagan, 1997), as well as between types of YHOs (Khachatryan et al. 2016a). Furthermore, the bulk of studies on this issue demonstrate that the prevalence of recidivism among YHOs increases with the length of follow-up period. The current study examined three key questions to shed further light on the post-homicide offending patterns YHO: (1) are their evident differences in the adult offending outcomes of YHOs, VYNHOs, and NVYNHOs; (2) are there differences in the severity of offending patterns of YHOs prior to and after their homicide offense; and, (3) are YHOs associated with a specific offending trajectory that differs from that of other serious and violent youth?

In line with prior studies (e.g., DiCataldo & Everett, 2008), YHOs had fewer convictions in their criminal histories prior to their homicide compared to other serious and violent offender groups and spent comparably more time incarcerated in both adolescence and adulthood. Furthermore, up to age 28, there were no evident differences in the prevalence of recidivism between YHOs, VYNHOs, and NVYNHOs in the current study, echoing previous between-group comparisons (e.g., Hagan, 1997). Finally, the prevalence of adulthood recidivism among YHOs was high (70.8%), especially after controlling for those that moved, died, or were never released (84.2%), which was also in line with studies with comparable follow-up periods (e.g., Vries & Liem, 2011).

To explore these relationships further, we employed a measure of the crime-mix of YHOs pre- and post-homicide offense to describe the qualitative aspects of their offending patterns. Using seven different crime types, compared to the pre-homicide period, it was evident that YHOs averaged fewer crimes and were more likely to commit less serious crimes following their homicide offense. The crime-mix of YHOs revealed a tendency for this group to be involved in miscellaneous crimes (e.g., mischief, vandalism, drinking and driving) and violations of court orders (e.g., breaching curfew, violating no-contact orders) following their homicide offense. This stands in contrast to early clinical conceptions of YHOs as a homogenous and uniquely dangerous group. Early clinical research suggested that YHOs must be different from other types of offenders as their involvement in a homicide offense was evidence of a propensity for serious violence that involved a lack of sensitivity to the pain and suffering of others (Bender, 1959; Miller & Looney, 1974; Myers et al. 1995). These clinical assertions continue to characterize some descriptions of YHOs as a group of persistent violent offenders (e.g., Kelly & Totten, 2002). For example, Woodworth, Agar, and Coupland (2013) suggested that YHOs would be exposed to increasing levels of violence that would contribute to continued aggressive behavior. Although this may be the case for some, the post-release crime-mix of YHOs in the current study showed that violence was not necessarily the norm for this group. The findings are also in line with the explanation put forth by DeLisi et al. (2014) that many homicides resulted from the same types of fights that were committed by violent non-homicide offenders. Taken together with the current finding that YHOs, especially following their homicide offense, were typically involved in less serious crimes, it may be necessary to revisit the clinical notion of YHOs as group characterized by a propensity for serious violence. Nonetheless, it is also evident that key indicators not examined in the current study can potentially differentiate these more serious YHOs; YHOs with a history of severe violence prior to and while incarcerated are at an increased likelihood of serious violence post-release (Caudill & Trulson, 2016; Trulson et al., 2016).

Caudill and Trulson (2016) found a similar pattern when examining the likelihood of felony (i.e., serious) offenses of JHOs post-homicide in adulthood. Importantly, there are likely competing explanations for these patterns. On the one hand, they suggested that lengthier prison sentences and involvement in specialized treatment (see Trulson et al., 2016) reduced the likelihood of recidivism among YHOs post-release. At the same time, Baay, Liem, and Nieuwbeerta (2012) showed that lengthier prison sentences adversely affected homicide offenders’ connections to their intimate partners, the latter being an important form of social support that promotes desistance (e.g., Sampson & Laub, 1997). Taken together, these findings point to the potential for specialized treatment in custody to reduce the likelihood of recidivism and the need for support upon community re-entry to help promote desistance. Although it was not possible to investigate the impact of treatment and degree of post-release social support for YHOs in the current study, heterogeneity in long-term offending patterns (i.e., offending trajectories) of YHOs can potentially inform the allocation of resources along these lines.

Accounting for exposure time, four offending trajectories measured from ages 12-28 were identified. These trajectories included patterns of: adolescent-limited, stable-low, high-rate desisting, and high-rate persistent offending. Critically, YHOs were neither overrepresented nor underrepresented in any specific offending trajectory. Although YHOs are routinely identified as a group warranting great attention and resources from the criminal justice system compared to other serious and violent youth (DeLisi et al., 2014; Lee, 2013), the heterogeneity in their criminal careers, at least through age 28, seems to suggest that a more individualized and tailored approach to assessment and treatment is required. Although much has been learned about YHOs in terms of their personal characteristics, victim-selection, and use of weapons in the commission of their offense (e.g., Gerard et al., 2014), less is understood about the course of offending leading up to, and following, involvement in a homicide offense. Currently, a young person’s involvement in a homicide offense seems to reveal more about how the criminal justice system will respond than about how this offender will remain in contact with this system. YHOs can vary in nature and seriousness of their criminal careers. A greater appreciation of criminal career patterns (e.g., age of onset, versatility, lambda) and key risk factors (e.g., psychopathy, gang membership) is likely to be more informative of which offenders remain at a continued risk for involvement in frequent/serious offending. For example, McCuish, Corrado, Hart, and DeLisi (2015) found that youth with higher symptoms of psychopathy were more likely to show a pattern of persistent violence in adulthood. This persistent involvement in violence may increase the likelihood that any given act of violence will result in the death of a victim. Similarly, Corrado, DeLisi, McCuish, and Hart (2015) noted that higher symptoms of psychopathy would increase the likelihood of involvement in especially serious, premeditated forms of offending such as homicide.

Following from this, one aspect of homicide offending we were not able to asses in the current study relates to gang involvement and other within-group differences regarding the nature of homicide events. In the Pittsburgh Youth Study, Farrington et al. (2012) noted that 30% of young homicide offenders were gang-involved. DeLisi, Spruill, Vaughn, and Trulson (2014) found that gang-involved homicide offenders were approximately a decade younger at the age of their homicide offense compared to other homicide offenders and Trulson et al. (2012) showed that gang-affiliated YHOs were more likely to recidivate compared to other YHOs.

More broadly, Khachatryan, Heide, Rad, and Hummel (2016b) observed that YHOs that offended in a group had more pre-homicide arrests compared to solo offenders and were also more likely to reoffend following release compared to solo offenders. The fact that group offenders were more frequent offenders may be reflective of access to a larger pool of potential criminal accomplices that can help provide opportunities and collaboration for future criminal offenses. At the same time, Khachatryan et al. (2016b) reported no differences in the prevalence of violent rearrest across the two types of YHOs (i.e., group vs. solo). This suggests that if group-based YHOs did have access to more opportunities to offend, these opportunities were not necessarily for violent crimes. A contrasting finding was observed by Trulson et al. (2012) using retrospective data on serious violent offenders (n = 1,804), including a subsample that were involved in gang-related homicide. Gang-affiliated YHOs were more likely to recidivate in general and were more likely to commit a new serious (i.e., felony) offense compared to other serious and violent offenders, even after controlling for a variety of demographic characteristics and risk factors. In terms of frequency of offending, being a gang-affiliated YHO was unrelated to frequency of re-arrest (Trulson et al. 2012).


The current study relied on a small subsample of YHOs (n = 26), all of whom were incarcerated. It is typically the case that community-based studies are more generalizable than studies using offender-based samples, and it could be argued that the current study includes a highly specific sample of offenders. Although this is true of the YNHOs in the current study, as this subsample represented only those YNHOs that were incarcerated for their crimes (i.e., a small proportion of all offenders), when it comes to studying YHOs, offender-based samples are advantageous because they are more likely to capture a fuller range of YHOs compared to community studies. Specifically, community based studies often consist of children or adolescents from higher-risk neighbourhoods or lower sociodemographic areas. Consequently, YHOs from lower-risk neighborhoods or more affluent areas are excluded from such samples. Given that the typically punitive response to homicide offenses (Hagan, 1997), it is likely that all YHOs will serve time in youth detention regardless of neighbourhood status. Therefore, sampling from a youth detention population is likely to provide the most representative sample of YHOs, even if this population does not represent the typical offender.


An important policy implication to address in future research is the impact of a YHO’s sentence in preventing future offending. Although this type of offender is likely to receive a lengthier period of incarceration to help prevent future offending, from a state dependence perspective, lengthier periods of incarceration absent of specialized treatment would decrease or inhibit the acquisition of informal social controls (e.g., Sampson & Laub, 1997) and thus might contribute to continued offending. Future research studies would benefit from considering Baay et al.’s (2012) research examining the relationship between time incarcerated and the breaking down of informal social controls. This is not to say that lengthy sentences are never justified. Balance is needed when considering the importance of protecting the public (Trulson et al., 2016) and ensuring the public’s perception of the justice system does not fall into disrepute given the grievous nature of the offense and other costs (DeLisi et al., 2010). Critically, the deterioration of informal social controls for these youth is also a consequence that jeopardizes public safety. Although it is true that not all YHOs are involved in future serious and violent offending, it is also true that some of these offenders will commit similar crimes in the future. What is clear from the current state of research is the necessity for precise assessment of risk and protective factors of YHOs, closer attention to their behavior and treatment progress while incarcerated, and the nature and extent of social support post-release. In other words, where an offender’s homicide offense is situated within their broader criminal career may be informative of their likelihood of continued offending. At this point, a homicide appears to be a distinct type of crime, but being involved in a homicide does not appear to reflect a distinct type of offender involved in more frequent or serious crimes relative to other serious and violent youth.


Andrews, D. A., & Bonta, J. (2010). The psychology of criminal conduct (5th ed.). New Providence, NJ: LexisNexis.

Arnett, J. J. (2000). Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist55, 469-480.

Baay, P., Liem, M., & Nieuwbeerta, P. (2012). “Ex-imprisoned homicide offenders: Once bitten, twice shy?” The effect of the length of imprisonment on recidivism for homicide offenders. Homicide Studies16, 259-279.

Barriga, A. Q., Sullivan‐Cosetti, M., & Gibbs, J. C. (2009). Moral cognitive correlates of empathy in juvenile delinquents. Criminal Behaviour and Mental Health19(4), 253-264.

Bender, L. (1959). Children and adolescents who have killed. American Journal of Psychiatry116, 510-513.

Caudill, J. W., & Trulson, C. R. (2016). The hazards of premature release: Recidivism outcomes of blended-sentenced juvenile homicide offenders. Journal of Criminal Justice46, 219-227.

Corrado, R. R., DeLisi, M., Hart, S. D., & McCuish, E. C. (2015). Can the causal mechanisms underlying chronic, serious, and violent offending trajectories be elucidated using the psychopathy construct? Journal of Criminal Justice43, 251-261.

DeLisi, M., Hochstetler, A., Jones-Johnson, G., Caudill, J. W., & Marquart, J. W. (2011). The road to murder: The enduring criminogenic effects of juvenile confinement among a sample of adult career criminals. Youth Violence and Juvenile Justice9, 207-221.

DeLisi, M., Piquero, A. R., & Cardwell, S. M. (2014). The unpredictability of murder juvenile homicide in the pathways to desistance study. Youth Violence and Juvenile Justice, 14, 26-42.

DeLisi, M., Spruill, J.O., Vaughn, M.G., & Trulson, C.R. (2014). Do gang members commit abnormal homicide? American Journal of Criminal Justice, 39, 125-138. 

DeLisi, M., Kosloski, A., Sween, M., Hachmeister, E., Moore, M., Drury, A. (2010).
Murder by numbers: Monetary costs imposed by a sample of homicide offenders.
Journal of Forensic Psychiatry and Psychology, 21, 501-513.

DiCataldo, F., & Everett, M. (2008). Distinguishing juvenile homicide from violent juvenile offending. International Journal of Offender Therapy and Comparative Criminology, 52, 158-174.

Farrington, D. P., Loeber, R., & Berg, M. T. (2012). Young men who kill: A prospective longitudinal examination from childhood. Homicide Studies, 16, 99-128.

Felson, R. B., & Steadman, H. J. (1983). Situational factors in disputes leading to criminal violence. Criminology21, 59-74.

Gerard, F. J., Jackson, V., Chou, S., Whitfield, K. C., & Browne, K. D. (2014). An exploration of the current knowledge on young people who kill: A systematic review. Aggression and Violent Behavior19, 559-571.

Gilman, A. B., Hill, K. G., & Hawkins, J. D. (2015). When is a youth’s debt to society paid? Examining the long-term consequences of juvenile incarceration for adult functioning. Journal of Developmental and Life Course Criminology1, 33-47.

Hagan, M. P. (1997). An analysis of adolescent perpetrators of homicide and attempted homicide upon return to the community. International Journal of Offender Therapy and Comparative Criminology41, 250-259.

Heide, K. M. (1999). Young killers: The challenge of juvenile homicide. Thousand Oaks: Sage.

Heide, K. M., Solomon, E. P., Sellers, B. G., & Chan, H. C. O. (2011). Male and female juvenile homicide offenders: An empirical analysis of US arrests by offender age. Feminist Criminology, 6, 3-31.

Heide, K. M., Spencer, E., Thompson, A., & Solomon, E. P. (2001). Who's in, who's out, and who's back: Follow‐up data on 59 juveniles incarcerated in adult prison for murder or attempted murder in the early 1980s. Behavioral Sciences & the Law19, 97-108.

Heide, K. M. (2003). Youth homicide: A review of the literature and a blueprint for action. International Journal of Offender Therapy and Comparative Criminology47, 6-36.

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

Kelly, K., & Totten, M. (2002). When children kill: A social psychological study of youth homicide. Peterborough, Ontario, Canada: Broadview Press.

Khachatryan, N., Heide, K. M., & Hummel, E. V. (2016a). Recidivism patterns among two types of juvenile homicide offenders: A 30-year follow-up study. International Journal of Offender Therapy and Comparative Criminology, (online first), 1-23.

Khachatryan, N., Heide, K. M., Rad, J., & Hummel, E. V. (2016b). Post‐incarceration recidivism of lone versus group juvenile homicide offenders. Behavioral Sciences & the Law, (online first), 1-17.

Lardén, M., Melin, L., Holst, U., & Långström, N. (2006). Moral judgement, cognitive distortions and empathy in incarcerated delinquent and community control adolescents. Psychology, Crime & Law12, 453-462.

Lee, T. (2013). A tearful Michelle Obama: ‘‘Hadiya Pendleton was me, and I was her.’’ Retrieved January 13, 2017 from

Liem, M. (2013). Homicide offender recidivism: A review of the literature. Aggression and Violent Behavior18, 19-25.

Loeber, R., & Farrington, D.P. (2011). Young homicide offenders and victims: Risk factors, prediction, and prevention from childhood. New York, NY: Springer

Loeber, R., Pardini, D., Homish, D. L., Wei, E. H., Crawford, A. M., Farrington, D. P., Stouthamer-Loeber, M., Creemers, J., Koehler, S., & Rosenfeld, R. (2005). The prediction of violence and homicide in young men. Journal of Consulting and Clinical Psychology73, 1074-1088.

McCuish, E. C., Corrado, R. R., Hart, S. D., & DeLisi, M. (2015). The role of symptoms of psychopathy in persistent violence over the criminal career into full adulthood. Journal of Criminal Justice43, 345-356.

Miller, D., & Looney, J. (1974). The prediction of adolescent homicide: Episodic dyscontrol and dehumanization. American Journal of Psychoanalysis34, 187-198.

Mulvey, E. P., Steinberg, L., Fagan, J., Cauffman, E., Piquero, A. R., Chassin, L., & Losoya, S. H. (2004). Theory and research on desistance from antisocial activity among serious adolescent offenders. Youth Violence and Juvenile Justice2, 213-236.

Myers, W. C., Chan, H. C. O., Vo, E. J., & Lazarou, E. (2010). Sexual sadism, psychopathy, and recidivism in juvenile sexual murderers. Journal of Investigative Psychology and Offender Profiling7, 49-58.

Myers, W. C., Scott, K., Burgess, A. W., & Burgess, A. G. (1995). Psychopathology, biopsychosocial factors, crime characteristics, and classification of 25 homicidal youths. Journal of the American Academy of Child & Adolescent Psychiatry34, 1483-1489.

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

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., Blumstein, A., Brame, R., Haapanen, R., Mulvey, E. P., & Nagin, D. S. (2001). Assessing the impact of exposure time and incapacitation on longitudinal trajectories of criminal offending. Journal of Adolescent Research16, 54-74.

Rodway, C., Norrington-Moore, V., While, D., Hunt, I. M., Flynn, S., Swinson, N., et al. (2011). A population-based study of juvenile perpetrators of homicide in England and Wales. Journal of Adolescence34, 19-28.

Sampson, R. J., & Laub, J. H. (1997). A life-course theory of cumulative disadvantage and the stability of delinquency. In T.P. Thornberry (Ed.), Development theories of crime and delinquency (pp. 133-161). New Brunswick, NJ: Transaction Publishers.

Shumaker, D. M., & McKee, G. R. (2001). Characteristics of homicidal and violent juveniles. Violence and Victims16, 401-409.

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

Statistics Canada. (2013). British Columbia (Code 59) (table). National Household Survey (NHS) Aboriginal Population Profile. 2011 National Household Survey. Statistics Canada Catalogue no. 99-011-X2011007. Ottawa. Released November 13, 2013. (accessed January 1, 2017).

Stouthamer-Loeber, M., Loeber, R., Stallings, R., & Lacourse, E. (2008). Desistance from and persistence in offending. In R. Loeber, D.P. Farrington, M. Stouthamer-Loeber, & H.R. White (Eds.), Violence and serious theft: Development and prediction from childhood to adulthood (pp. 269-308). New York, NY: Routledge.

Trulson, C.R., Haerle, D.R., Caudill, J.W., & DeLisi, M. (2016). Lost causes: Blended sentencing, second chances, and the Texas Youth Commission. Austin, TX: University of Texas Press.

Trulson, C. R., Caudill, J. W., Haerle, D. R., & DeLisi, M. (2012). Cliqued up: The postincarceration recidivism of young gang-related homicide offenders. Criminal Justice Review37, 174-190.

Vries, A. M., & Liem, M. (2011). Recidivism of juvenile homicide offenders. Behavioral Sciences & the Law29, 483-498.

Woodworth, M., Agar, A. D., & Coupland, R. B. (2013). Characteristics of Canadian youth-perpetrated homicides. Criminal Justice and Behavior40, 1009-1026.

Tables and Figures

Table 1. Descriptive characteristics of the sample (n = 523)

n (%)

Mean (SD)

Demographic characteristics



404 (77.2)


119 (22.8)

Ethnic origin


320 (61.2)


124 (23.7)

Non-Indigenous Minority

76 (14.5)

Criminal career measures

Age of onset

14.34 (1.51)

Days in custody (12-28)

1085 (1071)

Offending frequency (12-28)

22.50 (17.27)

Nature of homicide offense


13 (2.5)


6 (1.1)

Attempted murder

7 (1.3)

Table 2. Comparison of demographic characteristics and criminal career parameters across offender-type

Offender-Type (n = 523)

YHO (n = 26)

VYNHO (n = 358)

NVYNHO (n = 139)

χ2/F, p, Φ/η2

M (SD) / n (%)

M (SD) / n (%)

M (SD) / n (%)

Demographic characteristics


20 (76.9%)

281 (78.5%)

103 (74.1%)

χ2 (2) =1.1, p = .577, Φ=.05



12 (46.2%)

223 (62.5%)

85 (62.0%)

χ2 (4) =9.9, p < .05, Φ=.14


7 (26.9%)

77 (21.6%)

40 (29.2%)

Non-Indigenous Minority

7 (26.9%)

57 (16.0%)

12 (8.8%)

Measures of offending

Age of onset

14.73 (1.40)

14.18 (1.40)c

14.69 (1.74)b

F (2) = 6.8, p < .01, η2 = .02

Offending frequency (12-17)

4.73 (5.94)b,c

13.20 (8.96)a,c

10.91 (7.74)a,b

F* (2) = 19.8, p < .001, η2 = .05

Offending frequency (18-28)

5.18 (6.57)b

11.24 (11.82)a

8.82 (11.87)

F* (2) = 6.2, p < .01, η2 = .02

Time incarcerated (12-17)

500.89 (296.91)c

380.50 (297.51)c

258.41 (262.6)a,b

F (2) = 12.5, p < .001, η2 = .04

Time incarcerated (18-28)

2004.82 (1482.02)b,c

733.42 (850.83)a,c

526.53 (805.16)a,b

F* (2) = 14.7, p < .001, η2 = .11

Recidivism in adulthood

17 (70.8%)

264 (74.6%)

89 (65.0%)

χ2 (2) = 4.5, p = .100, Φ=.09

Note. YHO = Youth Homicide Offender; VYNHO = Violent Youth Non-Homicide Offender; NVYNHO = Non-Violent Youth Non-Homicide Offender

* indicates asymptotically F distributed; Browne Forsythe statistic used.

a denotes significantly different from YHO; b denotes significantly different from VYNHO; c denotes significantly different from NVYNHO

Table 3. Comparing the offending patterns of youth homicide offenders before and after their incarceration for their homicide offense (n = 21)


M (SD)


M (SD)

t, p

Number of general offenses

1.03 (0.94)

0.43 (0.53)

t (20) = 2.72, p < .05

Number of violent offenses

0.38 (0.30)

0.07 (0.12)

t (20) = 4.38, p < .001

Number of property offenses

0.22 (0.33)

0.04 (0.10)

t (20) = 2.22, p < .05

Number of violation offenses

0.20 (0.34)

0.16 (0.26)

t (20) = 0.46, p = .653

Number of weapons offenses

0.04 (0.78)

0.04 (0.09)

t (20) = 0.22, p = .832

Number of miscellaneous offenses

0.14 (0.25)

0.08 (0.14)

t (20) = 1.58, p = .131

Number of drug offenses

0.03 (0.11))

0.04 (0.09)

t (20) = -0.31, p = .762

Number of sexual offenses

0.01 (0.03)

0.00 (0.00)

t (20) = 1.00, p = .329

Number of days incarcerated

115.85 (66.02)

101.00 (104.46)

t (20) = 0.71, p = .486

Note. Pre-incarceration refers to all crimes committed prior to incarceration for the homicide offense. Post-incarceration refers to all crimes committed after release from custody following the homicide offense. Paired-samples t-test used to compare pre-incarceration to post-incarceration levels of offending. Analyses excluded youth homicide offenders that were never released from custody. Number of offenses refers to crimes per year of observation as the number of years of observation varied between the pre-incarceration period and the post-incarceration period

Table 4. Fit statistics for a zero inflated poisson model of trajectories of general offending (n = 523)


Offending Trajectories


Stable Low



χ2/F, p

n (%)

214 (40.9%)

98 (18.7%)

98 (18.7%)

113 (21.6)

Estimated model parameters


-18.37 (1.16)

-1.99 (0.48)

-16.56 (0.83)

-5.69 (0.33)


2.29 (0.14)

0.27 (0.05)

2.21 (0.10)

0.77 (0.04)


-0.07 (0.00)

-0.01 (0.00)

-0.07 (0.00)

-0.02 (0.00)

Model fit characteristics

Peak age










Mean probability-AL

.99 (.05)

.01 (.03)

.02 (.09)

.00 (.00)

Mean probability-Stable Low

.01 (.03)

.95 (.11)

.02 (.05)

.01 (.05)

Mean probability-HRD

.01 (.05)

.01 (.04)

.94 (.12)

.01 (.03)

Mean probability-HRP

.00 (.00)

.03 (.09)

.02 (.09)

.98 (.07)

OCC Value





Offending characteristics

Offending frequency (12-17)

5.97 (4.29)b,c,d

10.34 (6.00)a,c,d

18.42 (6.63)a,b

20.16 (8.79)a,b

F* (3, 334.62) = 143.29, p < .001

Offending frequency (18-28)

1.44 (1.63)b,c,d

12.83 (7.03)a,c,d

8.26 (6.30)a,b,d

26.92 (11.00)a,b,c

F* (3, 245.32) = 271.28, p < .001

Time incarcerated (12-17)

186.22 (187.57)b,c,d

311.18 (228.44)a,c,d

540.24 (281.07)a,b

549.02 (318.16)a,b

F* (3, 356.63) = 67.03, p < .001

Time incarcerated (18-28)

202.11 (628.79)b,c,d

798.76 (711.55)a,c,d

461.82 (465.83)a,b,d

1932.72 (712.70)a,b,c

F* (3, 369.69) = 176.65, p < .001

Offender type

Youth Homicide Offender

14 (53.8%)

5 (19.2%)

4 (15.4%)

3 (11.5%)

χ2(6) = 12.38, p = .06

Youth Violent Offender

130 (36.3%)

70 (19.6%)

69 (19.3%)

89 (24.9%)

Youth Non-Violent Offender

70 (50.4%)

23 (16.5%)

25 (18.0%)

21 (15.1%)

Note. AL = Adolescent Limited; HRD = high rate desisting; HRC = high rate persisting; OCC = odds of correct classification. ANOVA was used to compare offending characteristics across the four trajectories. Tamhane used for all post-hoc tests as Levene’s test was violated. Fisher’s exact test used for the chi-square analysis due to two cells with expected counts less than five.

* indicates asymptotically F distributed; Browne Forsythe statistic used.

a Significantly different from AL, b Significantly different from Stable Low, c significantly different from HRD, d Significantly different from HRP.

Table 5. Comparisons between trajectory association, YHOs, and YNHOs


n (%)


n (%)

χ2, p

AL Trajectory

14 (53.8)

200 (40.2)

χ2 (1) = 1.9, p = .169

Stable Low Trajectory*

5 (19.2)

93 (18.7)

χ2 (1) = 0.0, p = .999

High Rate Desister Trajectory*

4 (15.4)

94 (18.9)

χ2 (1) = 0.2, p = .800

High Rate Persistent Trajectory

3 (11.5)

110 (22.1)

χ2 (1) = 1.6, p = .201

Note. YHO = Youth homicide offender; YNHO = Youth non-homicide offender. Comparisons across individual trajectories were also made between YHO and VYNHO and YHO and NVYNHO. None of these comparisons resulted in a statistically significant association (p > .05).

* Fisher’s exact test used for the chi-square analysis due to one cell with expected counts less than five.

Figure 1. Aggregate crime-mix showing the average frequency of conviction per year of observation for each crime type.

Note. The left-hand side reflects convictions per person-period observation up to and including the homicide offense whereas the right-hand side reflects convictions per person-period observation after the homicide offense


Figure 2. General offending trajectories measured from age 12-28 for the sample (n = 523)

Acknowledgements: This work was supported by the Social Sciences and Humanities Research Council of Canada (410-2004-1875). A version of this paper was presented as part of a panel at the 2015 ASC conference in Washington, DC. The authors would like to thank Rolf Loeber, Lia Ahonen, and Kathleen Heide for participating in our panel and for helping inspire this work. We would also like to thank Stacy Tzoumakis and Julia Mazurchuk for their helpful comments on earlier drafts of the manuscript.

No comments here
Why not start the discussion?