Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Aug 1.
Published in final edited form as: Crim Behav Ment Health. 2020 Jun 2;30(4):210–220. doi: 10.1002/cbm.2156

Identifying and predicting criminal career profiles from adolescence to age 39

Bo-Kyung Elizabeth Kim 1, Amanda B Gilman 2, Kevin P Tan 3, Rick Kosterman 4, Jennifer A Bailey 5, Richard F Catalano 6, J David Hawkins 7
PMCID: PMC7704554  NIHMSID: NIHMS1611331  PMID: 32488935

Abstract

Background

Few longitudinal studies are capable of identifying criminal career profiles using both self-report and official court data beyond the 30s.

Aims

The current study aims to identify criminal career profiles across three developmental periods using self-report data, validate these profiles with official court records, and determine early childhood predictors.

Method

Data came from the Seattle Social Development Project (n=808). Latent Class Analysis was used to examine criminal careers from self-reported data during adolescence (ages 14–18), early adulthood (ages 21–27), and middle adulthood (ages 30–39). Official court records were used to validate the classes. Childhood risk and promotive factors measured at ages 11–12 were used to predict classes.

Findings

Findings revealed four career classes: Nonoffending (35.6%), adolescence-limited (33.2%), adult desister (18.3%), and life-course/persistent (12.9%). Official court records are consistent with the description of the classes. Early life school and family environments as well as having antisocial beliefs and friends differentiate membership across the classes.

Conclusion

The results of this study, with a gender-balanced and racially diverse sample, bolster the current criminal career knowledge by examining multiple developmental periods into the 30s using both self-report and official court data.

Keywords: criminal career, offending profiles, life-course-persistent offending, early risk factors, promotive factors, protective factors

1. INTRODUCTION

Criminal career research is well established in criminology (DeLisi & Piquero, 2011; Farrington, 1986; Steffensmeier, Allan, Harer, & Streifel, 1989; Tittle & Grasmick, 1997), with much attention paid to identifying and defining subgroups of offender populations over the life course (Loeber & Hay, 1997; Nagin, 2005; Piquero, 2008). One of the seminal works includes Moffitt’s (1993) taxonomies of crime and delinquency identifying adolescence-limited and life-course-persistent offending. Other scholars have identified additional criminal career profiles, including nonoffenders and various types of desisters (e.g., Laub & Sampson, 2003; van der Geest, Blokland, & Bijleveld, 2009). Findings of these studies have been limited, however, either due to including a single developmental period (e.g., adolescence only) or relying on official court records only, which do not capture all offending behaviour (Carkin & Tracy, 2015). This study uses longitudinal data across adolescence, early adulthood, and middle adulthood and examines both self-report and official court record data to identify profiles of criminal careers over the life course.

This paper addresses three research questions. First, what are the profiles of offending behaviour across developmental periods in different age spans between the ages of 12 and 39? Most adolescents experiment with risk-taking behaviours (Steinberg & Morris, 2001) and grow out of problem behaviours as they get older. This is consistent with Moffitt’s (1993) identification of adolescence-limited offenders as well as with observations of the “age-crime curve” in which offending behaviours peak in adolescence and decline into adulthood (Blumstein & Cohen, 1987). Research has found, however, that a small group of individuals engage in problem behaviours through adolescence and into adulthood (Moffitt, 1993, 1997). Some researchers have suggested that all offenders eventually desist from crime (Gottfredson & Hirschi, 1990; Piquero, Daigle, Gibson, Piquero, & Tibbetts, 2007; Sampson & Laub, 2003), although the timing of desistance varies (see also Farrington, 2019). This study seeks to identify profiles of offending behaviour using self-report data and to advance understanding on how criminal career trajectories develop and vary over the life course.

Second, how do the profiles based on self-reported data compare to official court records? Few studies have used both self-report and official court records to examine offending profiles (see also Jolliffe, Farrington, Piquero, Loeber, & Hill, 2017; Jolliffe, Farrington, Piquero, MacLeod, & van de Weijer, 2017). Many studies have relied only on official records (e.g., police contact, arrest, conviction), potentially missing some offending behaviours (Carkin & Tracy, 2015). In this study, we use official court charges. Other studies have relied solely on self-report data. While researchers have critiqued self-report data as unreliable, a previous study found that self-report data and official court data were comparable (Farrington, Ttofi, Crago, & Coid, 2014; Gilman, Hill, Kim, et al., 2014). In this study we use both self-report data and official court records to identify and validate offending trajectories. Furthermore, we add a qualitative dimension to these offending behaviours by categorizing offending behaviours by severity (i.e., low, moderate, high in self-report; misdemeanour or felony in official record). Much extant literature has been limited to measures of specific offence types (not qualitative categories) and prevalence (Carkin & Tracy, 2015).

Third, what early life risk or protective/promotive factors predict group membership in identified criminal career profiles? Previous research has identified several risk factors for engaging in criminal careers, especially for those who become life-course-persistent offenders (Whitten, McGee, Homel, Farrington, & Ttofi, 2018). Many have found that exposure to more risk factors increases the likelihood of chronic and persistent offending (e.g., Hawkins et al., 1998; Krohn, Lizotte, Bushway, Schmidt, & Phillips, 2014; Loeber & Farrington, 2000). Some studies have identified potentially malleable risk factors for offending that have implications for prevention efforts, including school and family problems, antisociality, antisocial peers, problem behaviours, and mental health issues (Baglivio, Wolff, Piquero, & Epps, 2015; Carkin & Tracy, 2015; Cox, Kochol, & Hedlund, 2018; Monahan, Steinberg, & Cauffman, 2009). Consistent with the literature, we assess the most-studied risk factors—antisocial peers, antisociality as indicated by antisocial beliefs, and mental health issues as indicated by internalizing problems—from the early years as potential risks of membership in the criminal career profiles identified in this study. While assessing risk is important, it is also important to examine environmental protective factors that prevent individuals from engaging in delinquent and criminal behaviours over the life course (e.g., Kim, Gilman, Hill, & Hawkins, 2016), especially severe, chronic, and persistent offending. We, thus, assess the extent to which positive characteristics of family and school environments in early life—instead of family and school problems as risk indicated by the literature—appear to inhibit or protect against persistent involvement in offending.

2. METHODS

2.1. Data and sample description

Data come from the Seattle Social Development Project (SSDP), a longitudinal study that examines the development of positive and problem behaviours over the life course. In 1985, all fifth-grade students in 18 participating elementary schools serving children from high-crime neighbourhoods in the Seattle Public School District were recruited for the longitudinal study (due to mandated school bussing at the time, however, the sample was not limited to children from high-crime neighbourhoods). Of this eligible population, 808 students and their parents (76.7%) agreed to participate in the longitudinal study. Participating students were 47% European American, 26% African American, 22% Asian American, and 5% Native American; 388 (49%) were female. About 52% of participants were from low-income families as indicated by eligibility for the federal free school lunch programme. The participants and their parents were surveyed annually from age 11 through 16 and participants only were interviewed starting at age 18. Participants were interviewed again at ages 21, 24, 27, 30, 33, 35, and 39. Participants’ official court records from all Washington State courts from eighth grade through age 33 were also collected and used in the current study.

2.2. Measures

Self-reported criminal behaviour

Indicators used to examine life-course offending behaviour span approximately 27 years and include 11 survey waves. We categorized waves into three developmental periods: adolescence (ages 14, 15, 16, and 18), early adulthood (ages 21, 24, and 27), and middle adulthood (ages 30, 33, 35, and 39). We created nine dichotomous indicators to capture offending severity across the three periods: low severity offending in adolescence (0/1), early adulthood (0/1), and middle adulthood (0/1); moderate offending in adolescence (0/1), early adulthood (0/1), and middle adulthood (0/1); and serious offending in adolescence (0/1), early adulthood (0/1), and middle adulthood (0/1).

This method of categorizing criminal behaviour has been employed in other studies using the SSDP sample (e.g., Chung, Hill, Hawkins, Gilchrist, & Nagin, 2002; Gilman, Hill, & Hawkins, 2015; Herrenkohl et al., 2003; Kim et al., 2016). Measures were not identical across developmental periods, as minor changes were made to the survey over time and some questions became not applicable as participants aged (e.g., asking about hitting a teacher). However, there was considerable consistency, and questions measuring each level of offending severity were available in all three periods. Measures were created from questions that asked the participants about past-year behaviours.

Low-level offending included items such as picking a fight and stealing something worth less than five dollars. Four low-level offending items were available in each of the four waves of adolescence and the three waves of early adulthood, while only one to three of the four items were available in middle adulthood depending on the survey wave. Moderate offending items included hitting someone with the idea of hurting them, damaging and destroying property, and stealing something worth less than 50 dollars. Six moderate-level offending items were available in all four waves of adolescence; four to five of the six items were available in early adulthood, depending on the wave; and three to four of the six items were available in middle adulthood, depending on the wave. High-level offending items included using weapons or force to get money, selling drugs, and stealing something worth more than 50 dollars. Five to six high-level offending items were available in adolescence, depending on the wave; six items were available in each of the three waves of early adulthood; and three to six items were available in middle adulthood, depending on the wave.

Official crime record

Eight types of criminal charges were used to create dichotomous indicators of official misdemeanour (0/1) and felony (0/1) offending in each of the three periods: property destruction (misdemeanour and felony), burglary (felony), theft/larceny (misdemeanour and felony), auto theft (felony), assault (misdemeanour and felony), robbery (felony), drug delivery (misdemeanour and felony), and drug possession (misdemeanour and felony). These official records cover the 12-month period leading up to the survey date, so as to mirror self-reported data. Official criminal data were not collected in the final two waves of the study. Thus, the two indicators of official criminal charges in middle adulthood are reflective of only age 30 and 33 survey waves.

Early risk factors

Early life risk factors during ages 11–12 were used for proper time ordering. Items were standardized and averaged into scales for each grade. Composites across ages 11 and 12 were created for each risk factor. The antisocial belief scale included six items (e.g., okay to cheat, okay to take without asking others if I can get away with it) (alpha = 0.70–0.73). Antisocial peers included three to four items (e.g., best friend gets in trouble with teacher) (alpha = 0.59–0.60). Conduct problems included 11 items (e.g., cruel to animals, attacks others, destroys other’s things) based on parent report (alpha = 0.77–0.79). Internalizing problems included 16 items with anxiety and depression-related behaviours (e.g., nervous/high strung, fear of bad thoughts, complains of loneliness, harms self or tries suicide) based on teacher report (alpha = 0.75–0.78).

Early promotive factors

Like risk factors, early life promotive factors during ages 11 and 12 were used (see Kim et al., 2016). A composite measure of positive family environment was created using family management as well as rewards, involvement, and bonding scales (alpha = 0.44–0.65). A composite measure of positive school environment was created using prosocial opportunities, rewards, involvement, and bonding scales (alpha = 0.52–0.79).

Demographic variables

Self-reported sex (0 = female, 1 = male) and race/ethnicity (European American, African American, Asian American, and Native American) were used as control variables. We also included receipt of free/reduced-price lunch during ages 10–12 as a measure of socioeconomic status.

2.3. Analysis

Latent class analysis (LCA) was used to identify distinct offending profiles across adolescence, early adulthood, and middle adulthood. LCA was conducted using Mplus version 8.1 (Muthén & Muthén, 2018). LCA is an exploratory method in which the number of profiles in the sample are not known a priori; it compares a one-class model (k) with an increasing k+1 model using fit statistics. Multiple-fit indices such as Bayesian information criteria (BIC), sample size adjusted BIC, and the Lo-Mendell-Rubin likelihood-ratio test (LMR-LRT) were used to assess the number of profiles that best fit the data. The profiles selected for the final model were also determined based on substantive relevance to those identified in the extant literature (e.g., Gottfredson & Hirschi, 1990; Laub & Sampson, 2003; Moffitt, 1997).

To validate the careers derived from self-reports and their corresponding official records, the Auxiliary (bch) command was used to estimate the mean incidences of offending. This approach uses posterior probability-based multiple imputations and pairwise chi-square equality tests to examine mean differences in offending across the careers. For ease of interpretation, we reported these means as percentages due to the dichotomous nature of the variables. To compare how early childhood predictors (e.g., antisocial beliefs, antisocial peers, and positive school environment) predict membership in criminal careers, a three-step multinomial logistic regression was used, with the normative criminal profile as the reference group (Asparouhov & Muthén, 2014). This three-step modelling process takes into account the inherent probabilistic uncertainties in the classification of criminal careers. Gender, race, and socioeconomic status were included as covariates in the regression models. Missing data for court records ranged from 9.3% (early adulthood) to 21.7% (late adulthood) for participants who moved out of Washington State. Missing data was handled using full information maximum likelihood.

3. RESULTS

3.1. Profiles of offending behaviours

Based on the fit indices (Table 1), the best fitting model ranges from three to six classes. To select the final model, we compared the item probability plots of all models with existing profiles found in the extant literature. The four-class solution was selected as it reflected unique profiles that are consistent with those in the literature: a normative/nonoffending (35.6%), an adolescence-limited (33.2%), an adult desister (18.3%), and a life-course-persistent (12.9%) profile. In the model for the three-class solution, the life-course-persistent profile, which was subsumed in the adult desister profile, did not emerge. The five-class model reflected two profiles that were similar to the life-course-persistent profile, and the differentiation was not substantially meaningful. Figure 1 depicts the four criminal profiles identified in this study. The normative profile had the lowest probabilities of offending from adolescence through middle adulthood. The adolescence-limited profile reported the highest probabilities in adolescence, which decreased in early and middle adulthood. The adult desister profile had high probabilities of offending in adolescence and early adulthood that sharply decreased in middle adulthood. Finally, the life-course persistent profile had the highest probabilities of offending from adolescence through middle adulthood.

Table 1.

Model fit indices for latent class analysis

No. of classes −2LL AIC BIC aBIC Entropy N of smallest class LMRT p value BLRT p value
1 −3422.694 6863.388 6905.539 6876.959 NA NA NA NA
2 −2955.768 5949.536 6038.520 5978.185 0.773 289 0.001 0.001
3 −2849.745 5757.491 5893.308 5801.217 0.716 207 0.001 0.001
4 −2820.828 5719.655 5902.306 5778.460 0.725 93 0.246 0.001
5 −2803.577 5705.155 5934.640 5779.038 0.745 34 0.146 0.001
6 −2792.321 5702.641 5978.960 5791.602 0.755 13 0.286 0.060
Open in a new tab

Note: Best fitting indices are highlighted in bold.

Abbreviations: −2LL = negative 2 log likelihood; AIC = Akaike Information Criteria; BIC = Bayesian Information Criteria; aBIC = sample size adjusted Bayesian Information Criteria; LMR = Lo-Mendell-Rubin adjusted likelihood ratio test; BLRT = Bootstrapped likelihood ratio test.

Figure 1.

Figure 1.

Open in a new tab

Criminal Career Profiles over the Life Course

3.2. Comparison of self-report profiles with official criminal records

Consistent throughout all developmental periods, individuals in the life-course-persistent profile showed the highest percentage of serious offending in their official records compared to the other profiles (Table 2): 38% had an official felony offence during adolescence and 34% during early adulthood. Conversely, the normative/nonoffending profile showed the least amount of offending according to official court data. There were no significant differences in official records of offending between the adolescence-limited and adult desister profiles.

Table 2.

Comparison of demographics and official crime records across profiles

  (A) Normative (35.6%) (B) Adolescence-limited (33.2%) (C) Adult desister (18.3%) (D) Life-course-persistent (12.9%) Sig test (p ≤ 0.05)
Male 34.1 46.6 73.9 76.1 B > A
          C > B,A
          D > A,B
White 44.1 54.8 46.2 38.5 B > D,A
Black 16.7 26.2 32.1 38.8 C > A
          D > A,B
Native 3.8 5.8 3.1 10.8  
Asian 35.4 13.2 18.5 11.9 A > B,C,D
FRL   57.7 46.2 58.8  
Official misdemeanour adolescence 7.8 32.2 29.6 55.0 D > A,B,C
B,C > A
Official felony adolescence 1.6 13.3 15.7 38.0 D > A,B,C
B,C > A
Official misdemeanour early adulthood 2.2 7.4 15.3 33.9 C > A
D > A,B
Official misdemeanour middle adulthood 0.2 3.7 7.1 17.6 D > A,B
Official felony early adulthood 3.2 2.4 9.4 34.4 D > A,B,C
Official felony middle adulthood 0.1 0.5 3.3 11.8 D > A,B
Open in a new tab

Note: For ease of interpretation, due to the dichotomous nature of these variables, we reported these as percentages instead of means.

Abbreviations: Sig. = Significance; FRL = Free/Reduced-price Lunch.

3.3. Prediction of group membership by early life risk and promotive factors

Comparing the early life risk and promotive factors (Table 3), positive school and family environment, as well as antisocial beliefs and friends differentiated membership across the profiles. A higher level of positive school environment was associated with lower odds of membership in the adolescence-limited group when compared to the normative/nonoffending group (OR: 0.58). Similarly, a higher level of positive family environment was associated with lower odds of membership in the life-course-persistent group compared to the normative group (OR: 0.31). A higher level of antisocial beliefs was associated with higher odds of membership in the adolescence-limited group (OR: 2.05) and in the life-course-persistent group (OR: 3.57). Having more antisocial friends was associated with higher odds of membership in the adolescence-limited (OR: 1.94), adult desister (OR: 2.56), and the life-course-persistent groups (OR: 3.78).

Table 3.

Early life risk and promotive factors across profiles

  Means (standard error)
   Odds ratio (95% CI)
  (A) Normative (35.6%) (B) Adolescence-limited (33.2%) (C) Adult desister (18.3%) (D) Life-course-persistent (12.9%) Sig test (p ≤ 0.05) Adolescence-limited vs. normative Adult desister vs. normative Life-course-persistent vs. normative
Positive school environment 0.068
(0.03)		 −0.030
(0.03)		 −0.070 (0.05) −0.074
(0.06)		 A > B,C,D 0.58 *
(0.34 – 0.97)		 0.46
(0.21 – 1.02)		 0.46
(0.21 – 1.04)
Positive family environment 0.056
(0.03)		 0.002 (0.03) −0.011 (0.06) −0.159
(0.07)		 A,B > D 0.71
(0.43 – 1.17)		 0.66
(0.32 – 1.39)		 0.31 *
(0.16 - 0.60)
Antisocial belief 0.205
(0.03)		 0.441
(0.05)		 0.353 (0.07) 0.741
(0.08)		 D > A,B,C
B > A		 2.05 *
(1.31 – 3.21)		 1.27
(0.72 – 2.25)		 3.57 *
(2.05 – 6.21)
Antisocial friends −0.212
(0.03)		 0.005 (0.05) 0.185 (0.09) 0.398
(0.10)		 B,C,D > A
D > B		 1.94 *
(1.18 – 3.19)		 2.56 *
(1.47 – 4.46)		 3.78 *
(2.17 – 6.57)
Conduct problems −0.040
(0.04)		 0.042 (0.04) −0.057 (0.06) 0.157
(0.07)		 D > A,C 1.16
(0.49 – 2.71)		 0.86
(0.40 – 1.86)		 1.33
(0.59 – 3.02)
Internalizing problems −0.053
(0.03)		 0.074 (0.04) −0.072 (0.05) 0.135
(0.07)		 D > A,C
B > A,C		 1.48
(0.89 – 2.46)		 0.35
(0.02 – 6.54)		 1.70
(0.92 – 3.15)
Open in a new tab

Note:

(a)

Odds ratio models included covariates for gender, race/ethnicity, and free/reduced-priced lunch.

(b) * p ≤ 0.05.

(c)

(c) Early life risk and promotive factor measures are on a standardized scale (range -1 to 1).

4. DISCUSSION

These findings advance the knowledge on criminal careers by using longitudinal data across multiple developmental periods including both self-reported and official court data. In addressing the first research question (what are the profiles of offending behaviour across developmental periods in different age spans between the ages of 12 and 39?), consistent with Moffitt’s (1997) taxonomy, we identified adolescence-limited and life-course-persistent offending profiles. Individuals in the adolescence-limited profile engaged in offences in adolescence but indicated little to no engagement in offending behaviours across early to middle adulthood. Individuals in the life-course-persistent offending profile displayed offending behaviours in adolescence and persisted in moderate to serious offences in early and middle adulthood. Also consistent with the literature (e.g., Laub & Sampson, 2003; Wiesner & Capaldi, 2003), we identified normative/nonoffending and adult desister profiles. A small portion of individuals in the normative/nonoffending profile reportedly engaged in minor and moderate offences during adolescence but indicated little to no offending behaviours in adulthood. Through adolescence and early adulthood, the adult desister profile closely resembled the life-course-persistent offending profile; however, individuals in the adult desister profile drastically declined in offending behaviours during middle adulthood.

With respect to our second research question (how do the profiles based on self-reported data compare to official court records?), these profiles based on self-report data were largely validated by official court data, indicating the lowest proportion of individuals with official records in the normative/nonoffending category and the highest proportion in the life-course-persistent category. While Carkin and Tracy (2015) argued that self-report data are highly unreliable, our findings suggest that, when compared to official court record, self-report data produce results similar to official records. This is consistent with our previous analyses of SSDP data (Gilman, Hill, Kim, et al., 2014).

With regard to the third research question (what early life risk or protective/promotive factors predict group membership in identified criminal career profiles?), we found that having more antisocial peers in early life consistently predicted a significantly greater likelihood of membership in the three criminal career profiles compared to the normative/nonoffending profile. Individuals in the life-course-persistent profile exhibited significantly higher levels of antisocial beliefs in elementary school compared to those in all other profiles. Individuals in the adolescence-limited and life-course-persistent profiles had significantly higher levels of internalizing problems in elementary school compared to those in the normative/nonoffending and adult desister profiles. Findings on these risk factors are consistent with previous studies (e.g., Whitten et al., 2018) and have important implications for practice. Prevention efforts should focus on increasing prosocial peer networks within schools and help young people engage with positive adults who can foster prosocial rather than antisocial beliefs. Findings regarding positive school and family environments strengthen this conclusion; individuals in the normative/nonoffending profile reported the highest level of positive family and school environments.

The study has limitations with regard to measurement. Data collection occurred approximately every three years in adulthood and indicators of self-reported criminal behaviour represent past-year prevalence. Without continuous assessment across the entire study period, we cannot be certain that all instances of criminal behaviour were captured. Additionally, identical questions were not asked across all waves. In some instances, questions in the survey were replaced or removed to reflect developmental changes in the sample. However, age-appropriate indicators of low, moderate, and severe offences were available in each developmental period, allowing for analyses that are possible in few longitudinal data sets.

We did not have official court data for participants who moved from Washington State. Moreover, official criminal charge data were available for only two of the four survey waves in middle adulthood. Nonetheless, relative to many other criminal career studies, inclusion of longitudinal official records in this study contributed to the understanding of criminal careers across more of the life course. Finally, reliabilities of some promotive factor measures (family and school environment) were somewhat low. We combined measures using an established measurement approach, theoretically driven (see social development model; Catalano & Hawkins, 1996) and used in other SSDP studies, seeking to maintain consistency with those prior studies (e.g., Gilman, Hill, & Hawkins, 2014; Kim et al., 2016; Kim, Oesterle, Catalano, & Hawkins, 2015).

This study advances knowledge in important ways. While many previous studies have relied on high-risk male samples (e.g., Nagin & Tremblay, 1999), this sample included a mix of high-risk and low-risk individuals as well as nearly 50% females. The study also addresses Piquero’s (2008) concern that most studies do not include offenders beyond 30 years of age and miss the distinction between those aging out of crime and persisting over the life course. Our findings distinguished these two groups during middle adulthood. By including both self-report and official court data and data on severity of offences in both self-report and official court data, the study furthers the understanding of criminal careers.

Acknowledgements

This research was supported by National Institute on Drug Abuse (NIDA) grant numbers R01DA033956, R01DA024411, and R01DA09679. Content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agency. At the time of writing the manuscript, Dr. Kim was a Scholar with the HIV/AIDS, Substance Abuse, and Trauma Training Program (HA-STTP), at the University of California, Los Angeles; supported through an award from the National Institute on Drug Abuse (R25DA035692).

Footnotes

Conflict of Interest

Authors declare no conflict of interest.

Contributor Information

Bo-Kyung Elizabeth Kim, USC Suzanne Dworak-Peck School of Social Work, University of Southern California.

Amanda B. Gilman, Washington State Center for Court Research.

Kevin P. Tan, School of Social Work, University of Illinois at Urbana-Champaign.

Rick Kosterman, Social Development Research Group, School of Social Work, University of Washington.

Jennifer A. Bailey, Social Development Research Group, School of Social Work, University of Washington.

Richard F. Catalano, Social Development Research Group, School of Social Work, University of Washington.

J. David Hawkins, Social Development Research Group, School of Social Work, University of Washington.

References

  1. Asparouhov T, & Muthén B (2014). Auxiliary variables in mixture modeling: Three-step approaches using Mplus. Structural Equation Modeling, 21, 329–341. 10.1080/10705511.2014.915181 [DOI] [Google Scholar]
  2. Baglivio MT, Wolff KT, Piquero AR, & Epps N (2015). The relationship between adverse childhood experiences (ACE) and juvenile offender trajectories in a juvenile offender sample. Journal of Criminal Justice, 43, 229–241. [Google Scholar]
  3. Blumstein A, & Cohen J (1987). Characterizing criminal careers. Science, 237, 985–991. 10.1126/science.237.4818.985 [DOI] [PubMed] [Google Scholar]
  4. Carkin D, & Tracy P (2015). Moffitt revisited: Delinquent and criminal career paths in the 1958 Philadelphia birth cohort. Journal of Law and Criminal Justice, 3, 14–39. 10.15640/jlcj.v3n1a2 [DOI] [Google Scholar]
  5. Catalano RF, & Hawkins JD (1996). The social development model: A theory of antisocial behavior In Hawkins JD (Ed.), Delinquency and crime: Current theories (pp. 149–197). New York, NY: Cambridge University Press [Google Scholar]
  6. Chung I-J, Hill KG, Hawkins JD, Gilchrist LD, & Nagin DS (2002). Childhood predictors of offense trajectories. Journal of Research in Crime and Delinquency, 39, 60–90. 10.1177/002242780203900103 [DOI] [Google Scholar]
  7. Cox SM, Kochol P, & Hedlund J (2018). The exploration of risk and protective score differences across juvenile offending career types and their effects on recidivism. Youth Violence and Juvenile Justice, 16, 77–96. 10.1177/1541204016678439 [DOI] [Google Scholar]
  8. DeLisi M, & Piquero AR (2011). New frontiers in criminal careers research, 2000–2011: A state-of-the-art review. Journal of Criminal Justice, 39, 289–301. 10.1016/j.jcrimjus.2011.05.001 [DOI] [Google Scholar]
  9. Farrington DP (1986). Age and crime In Tonry M & Morris N (Eds.), Crime and justice: An annual review of research: Vol. 7 (pp. 189–250). Chicago, IL: University of Chicago Press [Google Scholar]
  10. Farrington DP (2019). The duration of criminal careers: How many offenders do not desist up to age 61? Journal of Developmental and Life-Course Criminology, 5, 4–21. 10.1007/s40865-018-0098-5 [DOI] [Google Scholar]
  11. Farrington DP, Ttofi MM, Crago RV, & Coid JW (2014). Prevalence, frequency, onset, desistance and criminal career duration in self‐reports compared with official records. Criminal Behaviour and Mental Health, 24, 241–253. 10.1002/cbm.1930 [DOI] [PubMed] [Google Scholar]
  12. Gilman AB, Hill KG, & Hawkins JD (2014). Long-term consequences of adolescent gang membership for adult functioning. American Journal of Public Health, 104, 938–945. 10.2105/AJPH.2013.301821 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gilman AB, Hill KG, & Hawkins JD (2015). When is youths’ debt to society paid off? Examining the long-term consequences of juvenile incarceration for adult functioning. Journal of Developmental and Life-Course Criminology, 1, 33–47. 10.1007/s40865-015-0002-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gilman AB, Hill KG, Kim BKE, Nevell A, Hawkins JD, & Farrington DP (2014). Understanding the relationship between self-reported offending and official criminal charges across early adulthood. Criminal Behaviour and Mental Health. Special Issue: Criminal Careers in Self-Reports Compared with Official Records, 24, 229–240. 10.1002/cbm.1934 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gottfredson MR, & Hirschi T (1990). A general theory of crime. Stanford, CA: Stanford University Press. [Google Scholar]
  16. Hawkins JD, Herrenkohl T, Farrington DP, Brewer D, Catalano RF, & Harachi TW (1998). A review of predictors of youth violence In Loeber R & Farrington DP (Eds.), Serious and violent juvenile offenders: Risk factors and successful interventions (pp. 106–146). Thousand Oaks, CA: Sage [Google Scholar]
  17. Herrenkohl TI, Hill KG, Chung I-J, Guo J, Abbott RD, & Hawkins JD (2003). Protective factors against serious violent behavior in adolescence: A prospective study of aggressive children. Social Work Research, 27, 179–191. 10.1093/swr/27.3.179 [DOI] [Google Scholar]
  18. Jolliffe D, Farrington DP, Piquero AR, Loeber R, & Hill KG (2017). Systematic review of early risk factors for life-course-persistent, adolescence-limited, and late-onset offenders in prospective longitudinal studies. Aggression and Violent Behavior, 33, 15–23. 10.1016/j.avb.2017.01.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Jolliffe D, Farrington DP, Piquero AR, MacLeod JF, & van de Weijer S (2017). Prevalence of life-course-persistent, adolescence-limited, and late-onset offenders: A systematic review of prospective longitudinal studies. Aggression and Violent Behavior, 33, 4–14. 10.1016/j.avb.2017.01.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kim BKE, Gilman AB, Hill KG, & Hawkins JD (2016). Examining protective factors against violence among high-risk youth: Findings from the Seattle Social Development Project. Journal of Criminal Justice, 45, 19–25. 10.1016/j.jcrimjus.2016.02.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kim BKE, Oesterle S, Catalano RF, & Hawkins JD (2015). Change in protective factors across adolescent development. Journal of Applied Developmental Psychology, 40, 26–37. 10.1016/j.appdev.2015.04.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Krohn MD, Lizotte AJ, Bushway SD, Schmidt NM, & Phillips MD (2014). Shelter during the storm: A search for factors that protect at-risk adolescents from violence. Crime & Delinquency, 60, 379–401. 10.1177/0011128710389585 [DOI] [Google Scholar]
  23. Laub JH, & Sampson RJ (2003). Shared beginnings, divergent lives: Delinquent boys to age 70. Cambridge, MA: Harvard University Press. [Google Scholar]
  24. Loeber R, & Farrington DP (2000). Young children who commit crime: Epidemiology, developmental origins, risk factors, early interventions, and policy implications. Development and Psychopathology, 12, 737–762. 10.1017/s0954579400004107 [DOI] [PubMed] [Google Scholar]
  25. Loeber R, & Hay D (1997). Key issues in the development of aggression and violence from childhood to early adulthood. Annual Review of Psychology, 48, 371–410. 10.1146/annurev.psych.48.1.371 [DOI] [PubMed] [Google Scholar]
  26. Moffitt TE (1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674–701. [PubMed] [Google Scholar]
  27. Moffitt TE (1997). Adolescence-limited and life-course-persistent offending: A complementary pair of developmental theories In Thornberry TP (Ed.), Developmental theories of crime and delinquency. (Vol. 7, pp. 11–54). Piscataway, NJ: Transaction Publishers [Google Scholar]
  28. Monahan KC, Steinberg L, & Cauffman E (2009). Affiliation with antisocial peers, susceptibility to peer influence, and antisocial behavior during the transition to adulthood. Developmental Psychology, 45, 1520–1530. 10.1037/a0017417 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Muthén LK, & Muthén BO (2018). Mplus statistical software version 8.1 Los Angeles, CA: Muthén & Muthén. [Google Scholar]
  30. Nagin D, & Tremblay RE (1999). Trajectories of boys’ physical aggression, opposition, and hyperactivity on the path to physically violent and nonviolent juvenile delinquency. Child Development, 70, 1181–1196. 10.1111/1467-8624.00086 [DOI] [PubMed] [Google Scholar]
  31. Nagin DS (2005). Group-based modeling of development. Cambridge, MA: Harvard University Press. [Google Scholar]
  32. Piquero AR (2008). Taking stock of developmental trajectories of criminal activity over the life course In Liberman AM (Ed.), The long view of crime: A synthesis of longitudinal research (pp. 23–78). New York, NY: Springer; 10.1007/978-0-387-71165-2_2 [DOI] [Google Scholar]
  33. Piquero AR, Daigle LE, Gibson C, Piquero NL, & Tibbetts SG (2007). Are life-course-persistent offenders at risk for adverse health outcomes? Journal of Research in Crime and Delinquency, 44, 185–207. 10.1177/0022427806297739 [DOI] [Google Scholar]
  34. Sampson RJ, & Laub JH (2003). Life-course desisters? Trajectories of crime among delinquent boys followed to age 70. Criminology, 41, 555–592. 10.1111/j.1745-9125.2003.tb00997.x [DOI] [Google Scholar]
  35. Steffensmeier DJ, Allan EA, Harer MD, & Streifel C (1989). Age and the distribution of crime. American Journal of Sociology, 94, 803–831. 10.1086/229069 [DOI] [Google Scholar]
  36. Steinberg L, & Morris AS (2001). Adolescent development. Annual Review of Psychology, 52, 83–110. 10.1146/annurev.psych.52.1.83 [DOI] [PubMed] [Google Scholar]
  37. Tittle CR, & Grasmick HG (1997). Criminal behavior and age: A test of three provocative hypotheses. Journal of Criminal Law and Criminology, 88, 309–342. 10.2307/1144079 [DOI] [Google Scholar]
  38. van der Geest V, Blokland A, & Bijleveld C (2009). Delinquent development in a sample of high-risk youth: Shape, content, and predictors of delinquent trajectories from age 12 to 32. Journal of Research in Crime and Delinquency, 46, 111–143. 10.1177/0022427808331115 [DOI] [Google Scholar]
  39. Whitten T, McGee TR, Homel R, Farrington DP, & Ttofi M (2018). Comparing the criminal careers and childhood risk factors of persistent, chronic, and persistent–chronic offenders. Australian & New Zealand Journal of Criminology, 52, 151–173. 10.1177/0004865818781203 [DOI] [Google Scholar]
  40. Wiesner M, & Capaldi DM (2003). Relations of childhood and adolescent factors to offending trajectories of young men. Journal of Research in Crime and Delinquency, 40, 231–262. 10.1177/0022427803253802 [DOI] [Google Scholar]

RESOURCES