Criminogenic Need Profiles Among Substance-Using Justice-Involved Youth

Abstract

The relationship between juvenile offending and substance use is well-documented. Understanding this relationship in the context of other criminogenic needs could lead to more effective treatment programming, in an effort to reduce future justice system involvement. This study retrospectively classified 276 substance-using youth in the justice system into unique profiles based on the nature and severity of self-reported substance use and criminogenic needs. We tested competing latent profile analysis (LPA) models with a variable number of profiles. A four-profile model was optimal and included: (a) clinical drug and alcohol use with high criminogenic needs; (b) borderline-clinical drug use with low criminogenic needs; (c) clinical drug use with high criminogenic needs; and (d) clinical drug use with low to moderate criminogenic needs. Profiles demonstrated unique patterns of demographic and clinical factors, index offense, and rates of recidivism. Clinical implications for justice-involved youth with substance abuse are discussed, particularly related to treatment needs/services.

Keywords

substance use youth assessment criminogenic need recidivism

Justice-involved youth are more likely to have mental health problems compared to the general adolescent population (Colins et al., 2010; Wasserman et al., 2003), with past research suggesting that up to 70% of youth in the justice system have a diagnosable psychiatric disorder (Colins et al., 2010; Fazel et al., 2008). Moreover, within the youth justice population, both females and males with any mental health disorder have demonstrated significantly greater odds (i.e., 1.8–4.1) of having a co-occurring substance use disorder (SUD) diagnosis (Teplin et al., 2003), with rates of up to 76% observed (Schubert et al., 2011). A wealth of research has documented a relationship between substance use and juvenile offending, and reductions in substance use may play a key role in reducing justice system involvement (Kazemian et al., 2009; Stoolmiller & Blechman, 2005). However, there is limited research examining the unique criminogenic risk/needs of justice-involved youth with varying levels/patterns of substance use, which may be an important consideration for service delivery and treatment uptake.

Studies have shown that SUDs are associated with a range of negative outcomes in justice-involved youth, such as high school dropout, family-related problems, and risky sexual behavior (Chassin, 2008; Reifman et al., 1998). Within the youth justice population, younger age at first alcohol/drug use is a risk factor and a strong predictor of recidivism among adolescents who have committed serious offenses (Benda et al., 2001). Moreover, in a sample of probationers in Washington State, youth with problematic substance use presented with more criminogenic needs and had less protective factors than youth who reported no drug use in the 6 months prior (van der Put et al., 2014). Such research highlights the unique needs demonstrated by justice-involved youth with substance use concerns.

Decisions regarding intervention are informed by evaluating risk and needs relevant to a youth’s criminal activity (Hoge, 2001). A widely used model for assessment and case management of justice-involved youth is the Risk-Need-Responsivity (RNR) framework (Andrews & Bonta, 2010). Within the RNR framework, an individual’s risk of future offending is determined by identifying and responding to their criminogenic needs (i.e., risk factors amenable to change, including substance abuse and others). Responsivity is determined by evaluating individual characteristics that may impact the effectiveness of intervention, such as cognitive ability and motivation (Andrews & Bonta, 2010; Peterson-Badali et al., 2015). Risk/need factors are categorized into eight domains, which can be divided into the “big four” and the “modest four” (Andrews et al., 2012). The big four factors include history of criminal behavior, as well as antisocial personality, antisocial attitudes, and antisocial peers, and are the factors most strongly associated with recidivism. The modest four include other criminogenic needs that are modestly related to recidivism (e.g., education/employment, family circumstances, leisure/recreation, and substance use). From a clinical perspective, substance use is a mental health need; within the RNR framework, substance use is also a criminogenic need in the youth justice population, and therefore is often a target for clinical and forensic intervention (Chassin, 2008).

Characterizing the criminogenic risk/need profiles that are prevalent among substance-using youth in the justice system could be helpful in understanding the relationship between substance use and ongoing criminal behavior, and thereby focus and tailor intervention efforts to address this serious issue that is relevant for mental health and justice outcomes. However, research of this nature is limited within the adolescent and emerging adult populations. Indeed, youth with different patterns of criminogenic risks/needs may respond differently to intervention. As such, a better understanding of the heterogeneity of this population is important for developing strategies to more effectively address their needs. Such research has been undertaken in the adult justice population. A recent study of incarcerated Canadian men with an identified severe substance use need found five typologies to characterize men convicted of a crime, that were differentiated according to substance-using behavior and other criminogenic needs (Ternes et al., 2019). Findings from this study suggested that social supports and employment/education may be particularly important targets for intervention among substance-using men in the justice system. That is, the group least likely to reoffend included individuals with positive social supports, while another group, which was identified as having stable employment/education, had low levels of recidivism, despite a high number of needs overall. As such, the authors highlighted that substance use severity should not be examined in isolation when classifying incarcerated individuals in need of treatment, and noted that the pattern of other criminogenic needs should weigh into recommendations. To our knowledge, there is no existing literature of this nature focused on justice-involved youth. That is, no studies have examined whether youth involved in the justice system, who present with identified substance use needs, can be classified into unique profiles on the basis of their substance use behavior and other criminogenic needs.

Understanding the relationship between substance use and criminal behavior in youth, in the context of other criminogenic needs, is important because it could lead to more effective treatment programming and/or could increase treatment engagement, thereby addressing a serious mental health issue and reducing the risk of further justice system involvement. No study has explored the presence of profiles in substance-using justice-involved youth on the basis of their substance use severity and criminogenic needs. Therefore, the first objective of this study was to classify justice-involved youth with an identified substance use need into profiles, based on the severity of their substance use behavior and the severity of their other criminogenic needs. The second objective was to investigate whether these profiles could be differentiated based on demographic characteristics, type of offense, and rates of recidivism.

Method

Sample

Data from a cohort consisting of 276 youth (237 males, 39 females) ranging from 12 to 20 years of age at the time of assessment (all youth were < 18 years old at the time of their index offense) were retrospectively examined in this study. All youth were referred for court-ordered assessments between 2010 and 2019 to the youth justice clinic of a mental health agency in a large Canadian city, to assist in disposition decisions at court. Assessments included evaluations of risk-needs (detailed below), mental health, cognitive/academic functioning, and details related to the young person’s psychosocial history and offense. All assessments were conducted by a multidisciplinary team, which included professionals in social work, psychology, and psychiatry. Given the focus of this study, youth with an identified substance use need on the Youth Level of Service/Case Management Inventory (YLS/CMI; see below) were included in the sample. Only clients for whom consent was obtained to use clinical information for research purposes were included in the study; 84% of clients approached consented. Research Ethics Board approval for this study was obtained prior to the study commencement.

Measures

Risk to Reoffend and Criminogenic Needs

The Youth Level of Service/Case Management Inventory 2.0 (YLS/CMI; Hoge, 2001) is a standardized forensic tool used to assess risk to reoffend, criminogenic needs, and responsivity factors in youth aged 12 to 18 years. This measure assesses youth in eight criminogenic needs domains (i.e., criminal history, family circumstances/parenting, education/employment, peer relations, substance abuse, leisure/recreation, personality/behavior, and attitudes/orientation). Items within each domain are summed to obtain domain scores, which are then provided with a categorical specifier (i.e., low, moderate, and high). Strong internal consistency has been reported for most subscales of the YLS/CMI (Schmidt et al., 2005; Vitopoulos et al., 2012), as well as moderate to strong concurrent validity with scores on the Child Behavior Checklist (Schmidt et al., 2005) and the Youth Self Report (Skilling & Sorge, 2014). Predictive validity is also strong, with total scores on the YLS/CMI correlating with number of subsequent reoffenses (yes/no) in both males and females (Olver et al., 2014). Notably, low, moderate, and high scores within the Substance Use domain, are 0, 1 to 2, and 3 to 5, respectively. For this study, only youth who were rated as at least a 1 (i.e., moderate) in the Substance Use domain were included in the analysis (i.e., score of 1–5).

Substance Use

All youth completed self-report questionnaires to assess the level/severity of their alcohol and/or drug use. Measures included the Alcohol Use Disorders Identification Test (AUDIT; (Saunders et al., 1993) and the Drug Abuse Screening Test—Adolescents (DAST-A; Martino et al., 2000). The AUDIT is a self-report measure developed by the World Health Organization to identify hazardous and harmful alcohol use (clinical cutoff score ≥8). While originally designed to be used in primary care contexts, the AUDIT has been validated in other health care and community settings (Lima et al., 2005) and has been shown to have high internal consistency in both adult and adolescent populations (Meneses-Gaya et al., 2009; Rumpf et al., 2013). Research has also supported the psychometric properties of the AUDIT and its moderate ability to identify hazardous drinking in samples of justice-involved females and males (El-Bassel et al., 1998; Thomas et al., 2014). There is additional support for the reliability of its score in samples of male and female justice-involved youth (e.g., O’Hagan et al., 2019).

The DAST-A is a screening measure for problematic drug use in the last year, derived from the original adult version of the measure (clinical cutoff score > 6; borderline-clinical cutoff score 3–6; Yudko et al., 2007). Strong discriminant validity and internal consistency have been noted, and the DAST-A has also displayed strong concurrent validity with Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association, 1994) drug-related disorders (Martino et al., 2000). The internal consistency and aspects of validity of the DAST have been supported in samples of adults and youth in the justice system (e.g., O’Hagan et al., 2019; Saltstone et al., 1994).

Psychiatric Diagnosis

Structured diagnostic clinical interviews were conducted using the MINI Neuropsychiatric Interview for Children (M.I.N.I Kid) and psychoeducational assessments were conducted on a case-by-case basis using the Weschler scales (Wechsler Adult Intelligence Scale—Fourth Ed./Wechsler Intelligence Scale for Children—Fifth Ed., and the Wechsler Individual Achievement Test—Third Ed.; Wechsler, 2008, 2009, 2014). All diagnoses were made by the staff psychologists and/or psychiatrists. For the purposes of this study, diagnoses were classified as (a) SUD, (b) Learning Disability (LD); (c) Attention Deficit Hyperactivity Disorder (ADHD); and (d) mood and/or anxiety disorders (i.e., one or more depressive and/or anxiety disorder diagnoses; e.g., Depressive Mood Disorder and Generalized Anxiety Disorder).

Social Desirability

The Paulhus Deception Scale (PDS; Paulhus, 1998) is a self-report measurement of social desirability with two scales: Impression Management and Self-Deceptive Enhancement. Impression Management measures the degree of inflated self-descriptions; elevated scores are suggestive of manipulation and deception. Self-Deceptive Enhancement measures an unconscious personality bias toward favorable responses; elevated scores are suggestive of narcissism and overconfidence. The PDS is typically administered in situations where respondents may purposefully self-enhance, such as in the context of a court-ordered assessment. Strong validity and internal consistency for the PDS have been noted in the general population, college students, and justice-involved individuals, with Cronbach’s alpha ranging from .70 to .86 (Paulhus, 1998). The PDS was included in this study, given that self-reported substance use was a variable of interest in generating the profiles, and impression management could have impacted responses.

Recidivism

Recidivism data were gathered from a national police criminal record database and defined as whether or not a youth was convicted of one or more new offenses following disposition after their assessment over a 3-year follow-up period. Of note, reoffenses that took place within 3 months of assessment were not considered recidivism, given the potential lag time between conviction and entry into the federal database, as well as to allow sufficient time for probation services to commence. Recidivism data were only available up to 2016 and was only examined for youth who reoffended within 3 years postassessment. Therefore, only a subsample of youth had this information available (143 youths; 51.8% of sample).

Analysis

LPA was conducted to investigate substance use profiles among justice-involved youth. LPA analyses were conducted using R (R Core Team, 2010) and the tidyLPA package (Rosenberg et al., 2018); follow-up analyses were conducted using IBM SPSS Statistics version 27 (IBM Corp. Released, 2020). LPA is a statistical approach that sorts individuals into groups with qualitatively different profiles based on shared variance among individual scores on selected measures (Ferguson & Hull, 2018; Marsh et al., 2009). This approach assumes that there are meaningful and shared patterns of behavior that exist in this population based on substance use and criminogenic need scores (Merz & Roesch, 2011). In keeping with the approach used by Ternes et al. (2019) for their adult sample, the following variables were included in the model: DAST-A total score, AUDIT total score, YLS/CMI criminogenic need domain scores for criminal history, family circumstances/parenting, education/employment, peer relations, leisure/recreation, personality/behavior, and attitudes/orientation. The YLS/CMI domain score for substance abuse was excluded from the LPA, as all subjects in the study had substance use concerns, which is in keeping with the methods used by Ternes et al. (2019). Self-reported substance use and YLS/CMI risks/needs were the only variables included in the LPA because we were interested in whether these variables uniquely contributed to distinct profiles in substance-using justice-involved youth.

Consistent with standard practice, a varying number of groups were explored (1–5 models; Ferguson & Hull, 2018; Marsh et al., 2009). Model fit was evaluated using the bootstrap likelihood ratio test (BLRT), Akaike information criterion (AIC), Bayesian information criterion (BIC), and entropy, as well as the interpretability of the latent profiles (Muthén & Muthén, 2000; Nylund et al., 2007). Finally, each profile was described by examining the means and proportions of the variables included in the model. Gender differences were explored by conducting this analysis with and without females included, to determine if females contribute any unique effect to the profiles that were generated.

To examine the profiles descriptively, frequency distributions as well as means and standard deviations were calculated, where appropriate. To examine differences between profiles, one-way analyses of variance (ANOVA) and Pearson chi-square tests were conducted, where appropriate. No control variables were included in these analyses, given we had no underlying theoretical justifications to do so and overall PDS scores were not clinically significant (see below). As described below, the LPA analyses were conducted with and without female participants to examine whether all participants could be together in these analyses. To determine the extent of the relationship between variables, Cramer’s V analyses were conducted following Pearson chi-square tests. Post hoc comparisons (Bonferroni correction; .05/6 = p < .008) were conducted for the ANOVA analyses. For the Pearson chi-square tests, post hoc comparisons (Bonferroni correction; p < .008) were examined by comparing column proportions (z-test) in SPSS. Descriptive data included demographic and clinical variables (i.e., age, sex, ethnicity, SUD diagnosis), index offense, and recidivism data (i.e., reoffence, nature of first reoffence).

Results

Full Sample Characteristics

The sample consisted of ethnically diverse youth (37.7% black; 27.5% white; 8.7% Asian; 19.6% other Youth of Color) aged 12 to 20 years (mean age = 16.8 ± 1.19 years). Of the total sample, 85.8% of the youth were male. With respect to index offense, 54.3% of youth were charged with a violent but not sexual offense (e.g., homicide, assault, and armed robbery), 23.6% with nonviolent (e.g., theft, break and enter, possession of an illegal substance), and 5.8% with a sexual offense. Approximately half of the youth (i.e., 51%) were rated at least High on the YLS/CMI for risk of recidivism. Recidivism data were available for 143 youths, 58% of whom reoffended within the 3-year follow-up period, postassessment.

With respect to social desirability, scores on the Impression Management and Self-Deceptive Enhancement subscales of the PDS did not surpass cutoffs for invalid responding, and as such, youths’ self-report data were considered a valid reflection of their functioning. In terms of substance use behavior, 64.5% of youth reported clinical levels of drug use and 25.4% reported problematic alcohol consumption (as per DAST-A and AUDIT ratings, respectively). Overall, 16.7% of youth in this sample were clinically diagnosed with an SUD.

As an exploratory analysis, males and females were compared (ANOVA) on self-report substance use (i.e., DAST-A/AUDIT scores) and YLS/CMI domain scores. Females demonstrated higher AUDIT total scores compared to males, 8.23 ± 8.82 vs. 5.82 ± 5.98, respectively; F(1) = 4.67, p = .032, and higher domain scores on the YLS/CMI Personality/Behavior domain (4.28 ± 1.64 vs. 3.30 ± 1.86, respectively; F[1] = 9.63, p = .002). No additional differences were noted between males and females. Moreover, males and females (Pearson chi-square) did not differ with respect to rates of recidivism (yes/no).

Profiles of Justice-Involved Youth Engaging in Substance Use

Model indices of fit from the LPA are presented in Table 1. Based on these results, descriptive data were obtained for the four-profile and five-profile models. Examination of the means of each of the selected variables for both models suggested that the four-profile model was optimal, as the five-profile model produced a fifth profile that was very similar to Profile 4 (described below). Two separate analyses were conducted on the four-profile model, both including and excluding females. The selected variables demonstrated equivalent patterns across profiles in both analyses, and as such, females and males were included in the sample together and are presented in the following results.

Table 1:

Indices of Fit in Five Models as Derived From Latent Profile Analysis

Model	BLRT p value	AIC	BIC	Entropy
One profile	n/a	10,217.86	10,283.02	1.00
Two profile	.01	9,868.29	9,969.66	0.79
Three profile	.01	9,789.20	9,926.78	0.76
Four profile^a	.01	9,636.17	9,809.95	0.82
Five profile^a	.01	9,597.77	9,807.75	0.80

Note. BLRT = bootstrap likelihood ratio test; AIC = Akaike information criterion; BIC = Bayesian information criterion

Models examined for means of profile variables

As highlighted above, all youth in this study were rated on the YLS/CMI substance use domain as engaging in some substance use. The means and standard deviations for each of the YLS/CMI domains and the DAST-A and AUDIT total scores are included in Table 2. Proportions for the level of need in each domain are described in Table 3 and clinical and demographic variables of the profiles are described in Table 4.

Table 2:

Means (SD) of Criminogenic Risks/Needs and Self-Reported Substance Use Across Profiles

Substance use & YLS/CMI domain score	Clinical drug and alcohol use with high criminogenic needs(Profile 1)	Borderline-clinical drug use with low criminogenic needs(Profile 2)	Clinical drug use with high criminogenic needs(Profile 3)	Clinical drug use with low to moderate criminogenic needs(Profile 4)	Full sample	F test
Substance use & YLS/CMI domain score	n = 18	n = 38	n = 108	n = 112	n = 276	F test
Substance Use mean scores
DAST-A total score	16.83 (5.39)	6.89 (3.46)	9.69 (4.59)	7.67 (3.55)	8.95 (4.71)	F(3) = 30.1, p < .001^a,b,c,d,f
AUDIT total score	25.06 (6.44)	4.53 (3.25)	5.54 (4.16)	4.29 (3.88)	6.16 (6.49)	F(3) = 136.8, p < .001^a,b,c
YLS/CMI mean scores
Total risk	27.61 (5.16)	10.03 (3.62)	27.43 (3.97)	18.41 (3.14)	21.38 (7.20)	F(3) = 256.8, p < .001^a,c,d,e,f
Criminal history	2.39 (1.69)	0.39 (0.72)	2.16 (1.61)	1.03 (1.21)	1.47 (1.52)	F(3) = 23.7, p < .001^a,c,d,f
Family	4.50 (0.99)	1.50 (1.41)	3.73 (1.28)	2.78 (1.25)	3.09 (1.51)	F(3) = 38.7, p < .001^a,c,d,e,f
Education/employment	4.22 (1.83)	2.34 (1.32)	4.92 (1.43)	2.88 (1.33)	3.69 (1.75)	F(3) = 52.3, p < .001^a,c,d,f
Peer relations	3.39 (0.92)	1.26 (0.83)	3.45 (0.86)	3.01 (1.02)	2.97 (1.16)	F(3) = 54.1, p < .001^a,d,e,f
Substance abuse	4.28 (1.02)	1.87 (0.96)	3.10 (1.32)	2.88 (1.30)	2.92 (1.35)	F(3) = 12.7, p < .001^a,b,c,d,e
Leisure/recreation	2.11 (0.58)	0.84 (0.82)	2.31 (0.50)	2.08 (0.56)	2.00 (0.75)	F(3) = 61.1, p < .001^a,d,e,
Personality/behavior	4.56 (1.62)	1.32 (1.49)	4.84 (1.24)	2.63 (1.27)	3.44 (1.86)	F(3) = 92.5, p < .001^a,c,d,e,f
Attitudes/orientation	2.17 (1.30)	0.50 (0.73)	2.92 (1.35)	1.13 (1.00)	1.81 (1.48)	F(3) = 64.9, p < .001^a,c,d,f

Note. Post hoc tests (p < .008): a = group difference in profile 1 vs. 2; b = profile 1 vs. 3; c = profile 1 vs. 4; d = profile 2 vs. 3; e = profile 2 vs. 4; f = profile 3 vs. 4. DASTA = Drug Abuse Screening Test for Adolescents; AUDIT = Alcohol Use Disorders Identification Test; YLS/CMI = Youth Level of Service/Case Management Inventory, 2nd Ed.

Table 3:

Categorical Proportions of YLS/CMI Domain Scores in the Four-Profile Model

	Clinical drug and alcohol use with high criminogenic needs(Profile 1)		Borderline-clinical drug use with low criminogenic needs(Profile 2)		Clinical drug use with high criminogenic needs(Profile 3)		Clinical drug use with low to moderate criminogenic needs(Profile 4)		Full sample		χ²	V ^a
YLS/CMI domain score	n = 18		n = 38		n = 108		n = 112		N = 276				%	n	%	n	%	n	%	n	%	n
YLS/CMI domain score	YLS/CMI Total Score													χ²(9) = 304.9, p < .001	0.61	Low	0.0	0	47.4	18	0.0	0
0.0	0	6.9	19			Moderate	5.6	1	52.6	20	2.8	3
83.0	93	42.0	116			High	66.7	12	0.0	0	80.6	87
17.0	19	42.8	118			Very high	27.8	5	0.0	0	16.7	18
0.0	0	8.3	23			Criminal history
		χ²(6) = 50.00, p < .001	0.30	Low	22.2	4	71.1	27	19.4	21	45.5	51	37.3
103			Moderate	27.8	5	26.3	10	43.5	47	39.3	44	38.4
106			High	50.0	9	2.6	1	37.0	40	15.2	17	24.3
67			Family circumstances/parenting
	χ²(6) = 79.64, p < .001	0.38	Low	0.0	0	76.3	29	15.7	17	42.0	47	33.7
93			Moderate	50.0	9	18.4	7	54.6	59	52.7	59	48.6
134			High	50.0	9	5.3	2	29.6	32	5.4	6	17.8
49			Education/employment
	χ²(6) = 85.25, p < .001	0.39	Low	5.6	1	7.9	3	0.0	0	1.8	2	2.2
6			Moderate	16.7	3	73.7	28	18.5	20	67.0	75	45.7
126			High	77.8	14	18.4	7	81.5	88	31.3	35	52.2
144			Peer relations
	χ²(6) = 149.01, p < .001	0.52	Low	0.0	0	60.5	23	0.9	1	2.7	3	9.8
27			Moderate	33.3	6	39.5	15	30.6	33	50.0	56	39.9
110			High	66.7	12	0.0	0	68.5	74	47.3	53	50.4
139			Substance abuse
	χ²(6) = 33.97, p < .001	0.35	Low	0.0	0	0.0	0	0.0	0	0.0	0	0.0
0			Moderate	5.6	1	76.3	29	30.6	33	40.2	45	39.1
108			High	94.4	17	23.7	9	69.4	75	59.8	67	60.9
168			Leisure/recreation
	χ²(6) = 137.07, p < .001	0.50	Low	0.0	0	42.1	16	0.0	0	0.9	1	6.2
17			Moderate	11.1	2	31.6	12	1.9	2	8.9	10	9.4
26			High	88.9	16	26.3	10	98.1	106	90.2	101	84.4
233			Personality/behavior
	χ²(6) = 162.22, p < .001	0.54	Low	0.0	0	39.5	15	0.0	0	4.5	5	7.2
20			Moderate	44.4	8	57.9	22	35.2	38	88.4	99	60.5
167			High	55.6	10	2.6	1	64.8	70	7.1	8	32.2
89			Attitudes/orientation
	χ²(6) = 104.36, p < .001	0.44	Low	11.1	2	60.5	23	1.9	2	32.1	36	22.8
63			Moderate	72.2	13	39.5	15	62.0	67	67.0	75	61.6
170			High	16.7	3	0.0	0	36.1	39	0.9	1	15.6
43

Note. YLS/CMI = Youth Level of Service/Case Management Inventory, 2^nd Ed.; ^aCramer’s V, where appropriate.

Table 4:

Demographic and Clinical (DSM) Variables by Profile

Demographic/Clinical variable	Clinical drug and alcohol use with high criminogenic needs(Profile 1)		Borderline-clinical drug use with low criminogenic needs(Profile 2)		Clinical drug use with high criminogenic needs(Profile 3)		Clinical drug use with low to moderate criminogenic needs(Profile 4)		Full sample		F test/χ²	V
Demographic/Clinical variable	n = 18		n = 38		n = 108		n = 112		N = 276		F test/χ²	V
Age (years; M[SD])	17 (0.97)		17.2 (1.41)		16.5 (1.23)		16.9 (1.04)		16.8 (1.19)		F(3) = 4.76, p = .003
Sex (M: F)	13:5		33:5		93:15		98:14		237:39		χ² (3) = 3.04, p = .39
	%	n	%	n	%	n	%	n	%	n
Ethnicity											χ² (33) = 41.67, p = .14	0.23
Black	16.7	3	28.6	8	35.2	38	49.1	55	37.7	104
White	33.3	6	53.6	15	27.8	30	22.3	25	27.5	76
Asian	5.6	1	10.7	3	8.3	9	9.8	11	8.7	24
Other	44.4	8	42.9	12	28.7	31	18.8	21	19.6	54
DSM diagnosis
Substance	27.8	5	10.7	3	19.4	21	15.2	17	16.7	46	χ² (3) = 4.48, p = .21	0.13
ADHD	33.3	6	7.1	2	17.6	19	6.3	7	12.3	34	χ² (3) = 13.34, p = .004^a,c	0.35
LD	11.1	2	3.6	1	12.0	13	8.9	10	9.4	26	χ² (3) = 2.86, p = .41	0.16
Mood and/or anxiety	5.6	1	17.9	5	6.5	7	12.5	14	9.8	27	χ² (3) = 13.79, p = .032^d	0.25
DASTA level											χ² (6) = 18.91, p = .004^a,d	0.19
Nonclinical	0.0	0	0.0	0	1.9	2	0.9	1	1.1	3
Borderline	5.6	1	55.3	21	26.9	29	39.3	44	34.4	95
Clinical	94.4	17	44.7	17	71.3	77	59.8	67	64.5	178
AUDIT level											χ² (3) = 60.74, p < .001^a,b,c	0.47
Nonclinical	0.0	0	89.5	34	74.1	80	82.1	92	74.6	206
Clinical	100.0	18	10.5	4	25.9	28	17.9	20	25.4	70

Note. Post hoc tests (p < .008): a = group difference in profile 1 vs. 2; b = profile 1 vs. 3; c = profile 1 vs. 4; d = profile 2 vs. 3; e = profile 2 vs. 4; f = profile 3 vs. 4. DSM = Diagnostic and Statistical Manual of Mental Disorders—5th Ed.; ADHD = Attention Deficit Hyperactivity Disorder; LD = Learning Disability; DASTA = Drug Abuse Screening Test for Adolescents; AUDIT = Alcohol Use Disorders Identification Test; V = Cramer’s V, where appropriate.

Profile 1 (n = 18; 6.5% of sample) was defined as clinical drug and alcohol use with high criminogenic needs. This group was characterized by clinical levels of both drug and alcohol use (as measured by the DAST-A and AUDIT questionnaires), and most youth had a high criminogenic need rating in the substance use domain of the YLS/CMI (i.e., score of 3–5 out of 5; 94.4%). Notably, this was the only group of the four profiles that demonstrated alcohol use levels at a high-enough level to be considered clinically abusive on the AUDIT. Moreover, the majority of youth in this group demonstrated high needs in all other YLS/CMI criminogenic need domains, with exception of Attitudes/Orientation (72.2% rated as moderate need).

Profile 2 (n = 38; 13.8% of sample) was defined as borderline-clinical drug use with low criminogenic needs. This was the only group to demonstrate mainly borderline-clinical levels of drug use (scores on the DAST-A elevated but not above clinical cut off), as well as the only group to show mostly moderate ratings for substance use on the YLS/CMI (i.e., score of only 1–2 out of 5; 76.3%). In addition, most criminogenic need domains on the YLS/CMI were rated as low for this group, with exception to Education/Employment and Personality/Behavior (73.7% and 57.9% rated as moderate need, respectively). Of note, this was the only group among all four profiles to have the majority of youth rated as low need on Peer Relations, Leisure/Recreation, and Attitudes/Orientation.

Profile 3 (n = 108; 39.1% of sample) was defined as clinical drug use with high criminogenic needs. This group demonstrated clinical drug use (DAST-A) but nonclinical alcohol use (AUDIT), as well as high ratings in the substance use domain of the YLS/CMI (69.4%). Across all domains, needs on the YLS/CMI were rated as moderate to high, notably in the areas of Education/Employment, Peer Relations, Leisure/Recreation, and Personality/Behavior (each >60% rated as high need).

Profile 4 (n = 112; 40.6% of sample) was defined as clinical drug use with low to moderate criminogenic needs. Similar to Profile 3, this group demonstrated clinical drug use (DAST-A) and nonclinical alcohol use (AUDIT), with most of the sample given a high rating for overall substance use on the YLS/CMI (59.8%). In contrast to Profile 3, it was characterized by low to moderate needs in all YLS/CMI domains, with exception to Leisure/Recreation (90.2% rated as high need).

Clinical and Demographic Variables of Each Profile

Clinical and demographic variables of each profile are summarized in Table 4. After correcting for multiple comparisons, groups did not differ with respect to age. No differences in male:female ratio was noted in any groups. With respect to DSM diagnoses, overall, the groups did not differ on whether an SUD was given, with the majority of youth not receiving this diagnosis (i.e., <30% in all groups). Groups also did not differ with respect to diagnosis of an LD. Groups did differ with respect to ADHD and mood/anxiety diagnoses; the Clinical Drug and Alcohol Use with High Criminogenic Need group (Profile 1) demonstrated a significantly higher proportion of participants with an ADHD (33.3%) as compared to Profiles 2% and 4 (7.1% and 6.3%, respectively). The borderline-clinical drug use with low criminogenic needs group (Profile 2) demonstrated the highest proportion of mood and/or anxiety disorders (including depression, anxiety, and bipolar disorder; 17.9%) and was significantly higher than that in Profile 3 (12.5%).

Index Criminal Charge and Recidivism Characteristics of Each Profile

Data regarding the index criminal charge and recidivism characteristics for each profile are presented in Table 5. Considering the youths’ major offense type, violent (nonsexual) offense was noted as the major offense in all groups (i.e., >50% of youth in each group), although there were differences between groups related to the specific types of violent offenses committed. Notably, the highest proportion of sexual offenses was noted in the borderline-clinical drug use with low criminogenic needs group (Profile 2; 32.1%), significantly higher than those in Profiles 3 and 4.

Table 5:

Criminal Charge and Recidivism Characteristics by Profile

	Clinical drug and alcohol use with high criminogenic needs(Profile 1)		Borderline-clinical drug use with low criminogenic needs(Profile 2)		Clinical drug use with high criminogenic needs(Profile 3)		Clinical drug use with low to moderate criminogenic needs(Profile 4)		Full sample		χ²	V
Criminal charge/Recidivism characteristic	n = 18		n = 38		n = 108		n = 112		N = 276				%	n	%	n	%	n	%	n	%	n
Criminal charge/Recidivism characteristic	Index offense													χ²(6) = 34.76, p < .001^d,e	0.29
Nonviolent	22.2	4	14.3	4	29.6	32	23.2	26	23.6	65
Violent (nonsexual)	55.6	10	53.6	15	57.4	62	56.3	64	54.3	150
Sexual	11.1	2	32.1	9	2.8	3	1.8	2	5.8	16
Missing	11.1	2	35.7	10	10.2	11	17.9	20	15.6	43
	n = 11		n = 18		n = 67		n = 47		N = 143
	%	n	%	n	%	n	%	n	%	n
Recidivism—Yes	63.6	7	16.7	3	68.7	46	55.3	26	58.0	83	χ²(3) = 18.87, p < .001^d,e	0.38
Reoffence type (n = 83)											χ²(48) = 68.0, p = .33	0.53
Nonviolent	28.6	2	0.0	0	26.1	12	7.7	2	19.2	16
Violent (nonsexual)	57.1	4	33.3	1	43.5	20	53.8	14	47.0	39
Sexual	0.0	0	0.0	0	0.0	0	0.0	0	0.0	0
Administrative	14.3	1	66.6	2	15.2	7	26.9	7	20.4	17
Other	0.0	0	0.0	0	15.2	7	11.5	3	12.0	10

Note: V = Cramer’s V, where appropriate; Recidivism = data from 2016 record search (n = 143); Reoffence type = in youth who reoffended; note that reoffence type was missing from one youth.

Post Hoc tests (p < .008): a = group difference in profile 1 vs. 2; b = profile 1 vs. 3; c = profile 1 vs. 4; d = profile 2 vs. 3; e = profile 2 vs. 4; f = profile 3 vs. 4.

Recidivism data were collected for 143 youth from this sample within 3 years postassessment. Of this subsample, 83 youth reoffended (58%). Youth in the borderline-clinical drug use with low criminogenic needs group (Profile 2) were significantly less likely (16.7%) to reoffend as compared to Profiles 3 and 4 (but not significantly different from Profile 1), with most offenses being administrative in nature (e.g., failure to comply with conditions of probation; 66.6%). Notably, more than half of the youth in the other groups reoffended. Youth in the clinical drug and alcohol use with high criminogenic needs group (Profile 1; 57.1%), the clinical drug use with high criminogenic needs group (Profile 3; 68.7%), and the clinical drug use with low to moderate criminogenic needs (Profile 4; 55.3%) did not significantly differ from one another with respect to reoffence within 3 years postassessment. Moreover, groups did not significantly differ with respect to type of reoffence committed.

Discussion

This study aimed to classify substance-using youth involved in the justice system into unique profiles based on the nature and severity of their self-reported substance use and clinician-rated criminogenic needs. Substance use was chosen as a factor of interest, as it is one of the most common mental health problems in the youth justice population and is also a criminogenic risk/need related to justice-system involvement and recidivism (Schubert et al., 2011; Teplin et al., 2002, 2003). The primary aim of this study was to better understand the heterogeneity and patterns of criminogenic needs for justice-involved youth who use substances, to better understand how to effectively prioritize and address their needs. A four-profile model best described substance-using justice-involved youth in this sample; each profile demonstrated unique patterns of demographic and clinical factors, index offense, and rates of recidivism. Both males and females were included in this analysis, as the exclusion of females did not change the profile pattern.

Our four-profile model was in some contrast to the results from the study examining adults (Ternes et al., 2019), which classified incarcerated Canadian adult males into five clusters. Nonetheless, similar patterns of substance use were noted between the adult and youth justice populations. That is, similar to Ternes et al. (2019), this study did not observe a profile characterizing alcohol use only. Indeed, only polysubstance use (i.e., drug and alcohol) and drug use were noted among the groups. Importantly, there were some notable differences between the youth and adult profiles. For instance, Ternes et al. (2019) observed a profile of adult incarcerated males that included individuals convicted primarily of drug-related offenses. No unique profile of drug offenses was noted in the youth sample. Moreover, Ternes et al. (2019) noted multiple profiles with polysubstance use, with one group demonstrating very high criminogenic needs, and the other exhibiting education and employment as a strength. This study observed only one polysubstance profile which included youth with high needs in nearly all domains (similar to the high-risk adult group). Finally, the adult sample included only incarcerated males with severe substance use that had resulted in an SUD diagnosis, while this study included youth with borderline-clinical substance use (self-reported). In this study, youth with moderate to severe substance use concerns on the YLS were included in the sample, regardless of SUD diagnosis. As such, the difference in sampling criteria may, in part, account for the differences in profiles between youth and adults. Nonetheless, all youth in this sample were flagged by a clinician as having at least a moderate concern related to substance use behavior.

Of the four profiles, the borderline-clinical drug use with low needs group (Profile 2) presented with significant drug use on the DAST-A, but the score on the drug use measure was not in the clinical range because, generally speaking, while drug use was endorsed, associated adverse consequences were not endorsed on this self-report measure. Notably, youth in this group had relatively lower criminogenic needs in nearly all domains of the YLS/CMI, compared to the other profiles. Indeed, this was the only group among the four profiles in which the majority of youth were rated low in the domains of peer relations, leisure/recreation, and attitudes/orientation (i.e., all other groups were rated in the moderate to high ranges in these domains). This finding suggests that positive social functioning may be protective against negative outcomes, such as clinical levels of substance use or adverse consequences associated with substance use, as noted in previous research (van der Put et al., 2014). Alternatively, borderline-clinical drug use may protect against engaging deeply with antisocial peers and using leisure time unproductively. These findings are consistent with those from van der Put (2014) where youth who abstained from using substances had a higher number of protective factors (e.g., prosocial relationships, positive family circumstances, etc.).

It is worth noting that most youth in this sample were not formally diagnosed with an SUD. However, among the profiles that demonstrated levels of drug use that were clearly in the clinical range (as measured by self-report questionnaire), all were noted as having recidivism rates above 50% (ranging from 55%–68%). As such, regardless of diagnostic status, problematic substance use as a criminogenic need is potentially an important consideration for treatment. Indeed, there are mixed results in the literature regarding the relationship between SUD diagnosis and recidivism, with some noting a strong relationship (Hoeve et al., 2013; McReynolds et al., 2010; Schubert et al., 2011) and others not (Copeland et al., 2007). These findings, along with this study, reinforce that treatment recommendations should take into consideration problematic substance use, regardless of SUD diagnosis.

Violent offense was noted as the most common index offense for each profile in this study, consistent with previous findings in the youth justice population, particularly for youth with substance abuse concerns (Lai et al., 2016). This finding was also noted in the adult profiles described by Ternes et al. (2019), with exception to the profile that characterized drug-related offenses only. In this study, it was notable that youth in the borderline-clinical drug use with low criminogenic needs group (Profile 2) had the highest rates of index sexual offenses. This finding is consistent with previous research, which has noted that youth with sexual offending behavior may present with lower criminogenic needs and lower risk for recidivism as compared to nonsexual offending behavior (Caldwell, 2010; Rojas & Olver, 2020). This group also had high levels of mood and/or anxiety disorder diagnoses; a finding which has been noted among youth with sexual offending histories in prior research (Hunter et al., 2003). Thus, as other researchers have noted, youth with sexual offenses may require different clinical services from youth convicted of other violent offenses, such as treatment focused on internalizing disorders.

When the three profiles with clinical ratings of substance use were examined (i.e., Profiles 1, 3, and 4), the clinical drug use with low to moderate criminogenic needs group (Profile 4) included potential protective factors against recidivism, as it was the only group where the majority of youth were rated moderate in the education/employment and personality/behavior domains. This finding highlights the contribution that vocational interventions (e.g., career counseling) could have in mitigating recidivism risk, as was noted from the profiles identified for adult incarcerated males by Ternes et al. (2019). It also underscores the importance of personality differences that impact decision-making with respect to engaging in future antisocial activity (Nigel et al., 2018; van Goozen et al., 2007). For instance, personality characteristics such as inflated self-esteem, poor frustration tolerance, and inadequate feelings of guilt, are traits within the YLS/CMI that have been related to increased risk of recidivism in justice-involved youth (Onifade et al., 2008).

The highest rates of recidivism were noted in the clinical drug use with high criminogenic needs group (Profile 3) and the clinical drug and alcohol use with high criminogenic needs group (Profile 1). Notably, the primary factor distinguishing these two groups was their pattern of substance use (i.e., the presence of alcohol use). Indeed, while both groups demonstrated similar total YLS/CMI scores, differences were only noted with respect to the Substance Abuse domain. That is, the clinical drug and alcohol use with high criminogenic needs group (Profile 1) demonstrated significantly higher scores on the Substance Abuse domain, which makes sense, given problematic substance use across both alcohol and drug use. This group also demonstrated higher rates of ADHD diagnosis (when compared to Profiles 2 and 4), supporting previous research that noted an association between comorbid SUD and externalizing disorders with recidivism (McReynolds et al., 2010). Other studies have also noted associations between hyperactivity and later convictions in males involved in the justice system (Elander et al., 2000; Schubert et al., 2011). The contribution of ADHD symptomatology may in part explain increased impulsive behaviors in this population, such as substance use and criminal activity.

As highlighted above, substance-using youth can be meaningfully grouped into profiles that have unique patterns of substance use and criminogenic needs (as outlined in the RNR framework). These profiles are associated with meaningful differences in clinical presentation and in the rates of reoffending, which suggests attendance to these factors in the allocation of treatment is essential. For example, the intensity of the intervention should increase with increasing risk scores (risk principle), and so, based on our results, Profile 2 youth would require fewer intensive supports overall compared to the other three profiles. Furthermore, the targets of programming should be determined based on the combination of criminogenic needs (need principle) identified that vary across profiles. The current results suggest that attending to “substance use” as a general criminogenic need in isolation is insufficient, as it fails to consider important information pertaining to the nature of substance-using behavior and other unique criminogenic needs, which will likely require individualized or specific treatment. The results are consistent with the large body of literature indicating that adhering to the RNR principles reduces recidivism (Andrews & Bonta, 2010; Brogan et al., 2015; Vitopoulos et al., 2012). Indeed, enrolling youth in treatments that are not specific to their unique needs may in fact negatively impact their rehabilitation and increase the likelihood of continued involvement with the law (Brogan et al., 2015).

The profiles identified herein can be further examined through the third principle within the RNR framework—responsivity. For example, youth with clinically elevated polysubstance or drug use (Profiles 1 and 3, respectively) presented with high criminogenic risk/needs in most domains, as well as higher rates of ADHD diagnosis and higher risk for recidivism. As such, more intensive intervention is likely necessary for these groups of youth, and the interventions used may need to take into account high rates of inattention and impulsive behavior in designing the treatment sessions (i.e., shorter sessions, sessions earlier in the day, and fidget toys available during sessions). A potential responsivity concern related to youth, who present with borderline clinically elevated drug use (Profile 2), is that this group also demonstrated the highest rates of internalizing disorders and may benefit from interventions that target mood/anxiety before or concurrent with treatment focused on criminogenic needs.

This study is not without limitations. First, this study only included youth living within a large urban Canadian city. As such, further studies of this nature in other jurisdictions across Canada should be conducted, to confirm the generalizability of the current results. Moreover, the nature of this sample included youth rated as moderate to high risk to reoffend with mostly violent offenses; as such, these results may not generalize to a lower risk group of justice-involved youth. In addition, the sample of girls was quite small, and as such, we were not able to assess for potentially unique characteristics of females in each profile. While the inclusion of girls in the analysis did not affect the overall profiles, they may have unique targets for intervention. Indeed, when males and females were compared in the full sample, females demonstrated higher self-reported alcohol use, and higher scores on the YLS/CMI Personality/Behavior domain, warranting further study into their unique profiles with a larger female sample. Finally, in keeping with the RNR framework, this study addressed the risks and needs of each profile; however, we were not able to examine responsivity factors among the youth in this sample. Responsivity considerations are essential, such as ensuring evidence-based treatments are being used and that a youth’s characteristics that could play a role in treatment engagement and completion (e.g., cognitive abilities and motivation) are being addressed. Indeed, responsivity factors should play a key role in treatment decisions because they have been shown to impact treatment compliance and effectiveness (Andrews & Bonta, 2010; Dowden & Andrews, 1999; Vitopoulos et al., 2012).

Taken together, the results of this study highlight the complexity and heterogeneity of clinical/criminogenic needs for justice-involved youth with substance use concerns. This conclusion is consistent with previous research that has noted greater criminogenic needs in youth with substance abuse (Chassin, 2008). Results indicate that justice-involved youth with substance abuse identified on the YLS/CMI can be statistically classified into four distinct profiles based on their criminogenic needs and self-reported substance use. Compared to the study by Ternes et al. (2019) which examined adult incarcerated males, justice-involved youth demonstrated a unique profile distribution that distinguished them from adults, although there were similarities. This overall finding reinforces the need to develop strategies following criminogenic risk/needs assessments that address the unique needs of youth in the justice system. Results also highlight important clinical implications, given the high prevalence of substance abuse in the youth justice population. Intervention for substance-using behavior was warranted for all youth identified in this sample; however, the intensity of this intervention need varied. Importantly, recommended interventions for other criminogenic risk/needs also varied across the four groups studied herein, and thus, the pattern of criminogenic needs of each group should weigh into treatment assignment.

Footnotes

Acknowledgements

The authors thank the staff at the Youth Justice Assessment Clinic at the Centre for Addiction and Mental Health for their contributions to the clinical data collected for this work, as well as the youth who contributed to this research.

ORCID iD

Elisea De Somma

Elisea De Somma is a PhD student in the clinical developmental psychology program within the department of psychology at York University. Her research interests include examining the neural correlates of cognitive functioning in pediatric demyelinating diseases, as well as substance use in youth within the justice system.

Jala Rizeq is a clinical research fellow at the Hospital for Sick Children in Toronto, Canada. Her research interests include measuring contaminated mindware in adolescents and young adults, as well as developing integrative trauma models, demonstrating the negative sequelae of trauma exposure.

Tracey A. Skilling is a psychologist and clinician scientist at the Center for Addiction and Mental Health and an assistant professor in the Department of Psychiatry at the University of Toronto. Her research interests include mental health issues in justice-involved youth and risk assessments in both male and female justice-involved youth from diverse backgrounds. Her current projects include studies evaluating risk assessment practices and understanding the desistance process with adolescents involved in the justice system.

References

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.).

Andrews

D. A.

Bonta

(2010). The psychology of criminal conduct (5th ed.). Routledge. https://doi.org/10.1017/CBO9781107415324.004

Andrews

D. A.

Guzzo

Raynor

Rowe

R. C.

Rettinger

L. J.

Brews

Wormith

J. S.

(2012). Are the major risk/need factors predictive of both female and male reoffending? A test with the eight domains of the Level of service/case management inventory. International Journal of Offender Therapy and Comparative Criminology, 56(1), 113–133. https://doi.org/10.1177/0306624X10395716

Benda

B. B.

Corwyn

R. F.

Toombs

N. J.

(2001). Recidivism among adolescent serious offenders: Prediction of entry into the correctional system for adults. Criminal Justice and Behavior, 28, 588–613. https://doi.org/10.1177/009385480102800503

Brogan

Haney-Caron

NeMoyer

DeMatteo

(2015). Applying the Risk-Needs-Responsivity (RNR) model to juvenile justice. Criminal Justice Review, 40(3), 277–302. https://doi.org/10.1177/0734016814567312

Caldwell

M. F.

(2010). Study characteristics and sex offender recidivism. International Journal of Offender Therapy and Comparative Criminology, 54(2), 1–16. https://doi.org/10.1177/0306624X08330016

Chassin

(2008). Juvenile justice and substance use. The Future of Children, 18(2), 165–183. https://www-jstor-org-443.web.bisu.edu.cn/stable/20179983

Colins

Vermeiren

Vreugdenhil

van den Brink

Doreleijers

Broekaert

(2010). Psychiatric disorders in detained male adolescents. The Canadian Journal of Psychiatry, 55(4), 255–263. https://doi.org/10.1177%2F070674371005500409

Copeland

W. E.

Miller-johnson

Keeler

Angold

Costello

E. J.

(2007). Childhood psychiatric disorders and young adult crime : A prospective, population-based study. American Journal of Psychiatry, 164(11), 1668–1675. https://doi.org/10.1176/appi.ajp.2007.06122026

10.

Dowden

Andrews

D. A.

(1999). What works for female offenders: A meta-analytic review. Crime and Delinquency, 45(4), 438–452. https://doi.org/10.1177/0011128799045004002

11.

Elander

Simonoff

Pickles

(2000). A longitudinal study of adolescent and adult conviction rates among children referred to psychiatric services for behavioral or emotional problems. Criminal Behaviour and Mental Health, 10(1), 40–59. https://doi.org/10.1002/cbm.342

12.

El-Bassel

Schilling

Ivanoff

Chen

Hanson

(1998). Assessing the World Health Organization’s Alcohol Use Disorder Identification Test among incarcerated women. Journal of Offender Rehabilitation, 26(3–4), 71–89. https://doi.org/10.1300/J076v26n03_04

13.

Fazel

Doll

Långström

(2008). Mental disorders among adolescents in juvenile detention and correctional facilities: A systematic review and metaregression analysis of 25 surveys. Journal of the American Academy of Child and Adolescent Psychiatry, 47(9), 1010–1019. https://doi.org/10.1097/CHI.0b013e31817eecf3

14.

Ferguson

S. L.

Hull

D. M.

(2018). Personality profiles: Using latent profile analysis to model personality typologies. Personality and Individual Differences, 122, 177–183. https://doi.org/10.1016/j.paid.2017.10.029

15.

Hoeve

McReynolds

Wasserman

G. A.

Mcmillan

(2013). The influence of mental health disorders on severity of reoffending in juveniles. Criminal Justice and Behavior, 40(3), 289–301. https://doi.org/10.1177/0093854812459639

16.

Hoge

R. D.

(2001). A case management instrument for use in juvenile justice systems. Juvenile and Family Court Journal, 52(2), 25–32. https://doi.org/10.1111/j.1755-6988.2001.tb00041.x

17.

Hunter

J. A.

Figueredo

Malamuth

N. M.

Becker

J. V.

(2003). Juvenile sex offenders : Toward the development typology. A Journal of Research and Treatment, 15(1), 27–48.

18.

IBM Corp. Released. (2020). IBM SPSS Statistics for Window. IBM Corp.

19.

Kazemian

Farrington

D. P.

Le Blanc

(2009). Can we make accurate long-term predictions about patterns of de-escalation in offending behavior? Journal of Youth and Adolescence, 38(3), 384–400. https://doi.org/10.1007/s10964-008-9338-z

20.

Lai

Zeng

Chu

C. M.

(2016). Violent and nonviolent youth offenders : Preliminary evidence on group subtypes. Youth Violence and Juvenile Justice, 14(3), 313–329. https://doi.org/10.1177/1541204015615193

21.

Lima

C. T.

Freire

A. C. C.

Silva

A. P. B.

Teixeira

R. M.

Farrell

Prince

(2005). Concurrent and construct validity of the audit in an urban Brazilian sample. Alcohol and Alcoholism, 40(6), 584–589. https://doi.org/10.1093/alcalc/agh202

22.

Marsh

H. W.

Lüdtke

Trautwein

Morin

A. J. S.

(2009). Classical latent profile analysis of academic self-concept dimensions: Synergy of person- and variable-centered approaches to theoretical models of self-concept. Structural Equation Modeling: A Multidisciplinary Journal, 16(2), 191–225. https://doi.org/10.1080/10705510902751010

23.

Martino

Grilo

C. M.

Fehon

D. C.

(2000). Development of the Drug Abuse Screening Test for Adolescents (DAST-A). Addictive Behaviors, 25(1), 57–70. https://doi.org/10.1016/S0306-4603(99)00030-1

24.

McReynolds

L. S.

Schwalbe

C. S.

Wasserman

G. A.

(2010). The contribution of psychiatric disorder to juvenile recidivism. Criminal Justice and Behavior, 37(2), 204–216. https://doi.org/10.1177/0093854809354961

25.

Meneses-Gaya

De Zuardi

A. W.

Loureiro

S. R.

Crippa

J. A. S.

(2009). Alcohol Use Disorders Identification Test (AUDIT): An updated systematic review of psychometric properties. Psychology & Neuroscience, 2(1), 83–97. https://doi.org/10.3922/j.psns.2009.1.12

26.

Merz

E. L.

Roesch

S. C.

(2011). A latent profile analysis of the Five Factor Model of personality: Modeling trait interactions. Personality and Individual Differences, 51(8), 915–919. https://doi.org/10.1016/j.paid.2011.07.022

27.

Muthén

L. K.

(2000). Integrating person-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes. Alcoholism: Clinical and Experimental Research, 24(6), 882–891. https://doi.org/10.1111/j.1530-0277.2000.tb02070.x

28.

Nigel

S. M.

Dudeck

Otte

Knauer

Klein

Böttcher

Maaß

Vasic

Streb

(2018). Psychopathy, the Big Five and empathy as predictors of violence in a forensic sample of substance abusers. The Journal of Forensic Psychiatry & Psychology, 29(6), 882–900. https://doi.org/10.1080/14789949.2018.1439993

29.

Nylund

K. L.

Asparouhov

Muthén

B. O.

(2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14(4), 535–569. https://doi.org/10.1080/10705510701575396

30.

O’Hagan

H. R.

Brown

S. L.

Jones

N. J.

Skilling

T. A.

(2019). The reliability and validity of the measure of criminal attitudes and associates and the pride in delinquency scale in a mixed sex sample of justice-involved youth. Criminal Justice and Behavior, 46(5), 751–769. https://doi.org/10.1177%2F0093854818810459

31.

Olver

M. E.

Stockdale

K. C.

Wormith

J. S.

(2014). Thirty years of research on the level of service scales: A meta-analytic examination of predictive accuracy and sources of variability. Psychological Assessment, 26(1), 156–176. https://psycnet.apa.org/doi/10.1037/a0035080

32.

Onifade

Davidson

Livsey

Turke

Horton

Malinowski

Atkinson

Wimberly

(2008). Risk assessment: Identifying patterns of risk in young offenders with the Youth Level of Service/Case Management Inventory. Journal of Criminal Justice, 36(2), 165–173. https://doi.org/10.1016/j.jcrimjus.2008.02.006

33.

Paulhus

D. L.

(1998). Manual for the balanced inventory of desirable responding: Version 7. Multi-Health Systems.

34.

Peterson-Badali

Skilling

Haqanee

(2015). Examining implementation of risk assessment in case management for youth in the justice system. Criminal Justice and Behavior, 42(3), 304–320. https://doi.org/10.1177/0093854814549595

35.

R Core Team. (2019). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/

36.

Reifman

Barnes

G. M.

Dintcheff

B. A.

Farrell

M. P.

Uhteg

(1998). Parental and peer influences on the onset of heavier drinking among adolescents. Journal of Studies on Alcohol, 59(3), 311–317. https://doi.org/10.15288/jsa.1998.59.311

37.

Rojas

E. Y.

Olver

M. E.

(2020). Validity and reliability of the Violence Risk Scale–Youth Sexual Offense Version. Sexual Abuse, 32(7), 826–849. https://doi.org/10.1177/1079063219858064

38.

Rosenberg

J. M.

Beymer

P. N.

Anderson

D. J.

Schmidt

J. A.

(2018). tidyLPA: An R package to easily carry out latent profile analysis (LPA) using open-source or commercial software. Journal of Open Source Software, 3(30), 978. https://doi.org/10.21105/joss.00978

39.

Rumpf

Wohlert

Freyer-Adam

Grothues

Bischof

(2013). Screening questionnaires for problem drinking in adolescents: Performance of AUDIT, AUDIT-C, CRAFFT and POSIT. European Addition Research, 19(3), 121–127. https://doi.org/10.1159/000342331

40.

Saltstone

Halliwell

Hayslip

M. A.

(1994). A multivariate evaluation of the Michigan alcoholism screening test and the Drug Abuse Screening Test in a female offender population. Addictive Behaviors, 195(5), 455–462. https://doi.org/10.1016/0306-4603(94)90001-9

41.

Saunders

Assland

Babor

De La Fuente

Grant

(1993). Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO collaborative project on early detection of persons with harmful alcohol consumption-II. Addiction, 88(6), 791–804. https://doi.org/10.1111/j.1360-0443.1993.tb02093.x

42.

Schmidt

Hoge

R. D.

Gomes

(2005). Reliability and validity analyses of the youth level inventory. Criminal Justice and Behavior, 32(3), 329–344. https://doi.org/10.1177/0093854804274373

43.

Schubert

C. A.

Mulvey

E. P.

Glasheen

(2011). Influence of mental health and substance use problems and criminogenic risk on outcomes in serious juvenile offenders. Journal of the American Academy of Child & Adolescent Psychiatry, 50(9), 925–937. https://doi.org/10.1016/j.jaac.2011.06.006

44.

Skilling

T. A.

Sorge

G. B.

(2014). Measuring antisocial values and attitudes in justice-involved male youth. Criminal Justice and Behavior, 41(8), 992–1007. https://doi.org/10.1177/0093854814521415

45.

Stoolmiller

Blechman

E. A.

(2005). Substance use is a robust predictor of adolescent recidivism. Criminal Justice and Behavior, 32(3), 302–328. https://doi.org/10.1177/0093854804274372

46.

Teplin

L. A.

Abram

K. M.

McClelland

G. M.

Dulcan

M. K.

(2003). Comorbid psychiatric disorders in youth in juvenile detention. Arch Gen Psychiatry, 60(11), 1097–1108. https://doi.org/10.1001/archpsyc.60.11.1097

47.

Teplin

L. A.

Abram

K. M.

Mcclelland

G. M.

Dulcan

M. K.

Mericle

A. A.

(2002). Psychiatric disorders in youth in juvenile detention. Arch Gen Psychiatry, 59, 1133–1143. https://doi.org/10.1001/archpsyc.59.12.1133

48.

Ternes

Farrell Macdonald

Cheverie

(2019). Substance use disorder typologies of Canadian federally sentenced men: relationships with institutional behavior and postrelease success. Criminal Justice and Behavior, 46(3), 457–474. https://doi.org/10.1177/0093854818812516

49.

Thomas

Degenhardt

Alati

Kinner

(2014). Predictive validity of the AUDIT for hazardous alcohol consumption in recently released prisoners. Drug and Alcohol Dependence, 134(1), 322–329. https://doi.org/10.1016/j.drugalcdep.2013.10.019

50.

van der Put

Creemers

Hoeve

(2014). Differences between juvenile offenders with and without substance use problems in the prevalence and impact of risk and protective factors for criminal recidivism. Drug and Alcohol Dependence, 134(1), 267–274. https://doi.org/10.1016/j.drugalcdep.2013.10.012

51.

van Goozen

S. H. M.

Fairchild

Snoek

Harold

G. T.

(2007). The evidence for a neurobiological model of childhood antisocial behavior. Psychological Bulletin, 133(1), 149–182. https://doi.org/10.1037/0033-2909.133.1.149

52.

Vitopoulos

Peterson-Badali

Skilling

(2012). Examining the RNR principles in practice. Criminal Justice and Behavior, 39(8), 1025–1041. https://doi.org/10.1177/0093854812442895

53.

Wasserman

G. A.

Jensen

P. S.

S. J.

Cocozza

Trupin

Angold

Cauffman

Grisso

(2003). Mental health assessments in juvenile justice: Report on the consensus conference. Journal of the American Academy of Child and Adolescent Psychiatry, 42(7), 752–761. https://doi.org/10.1097/01.CHI.0000046873.56865.4B

54.

Wechsler

(2008). Wechsler Adult Intelligence Scale—Fourth Edition (WAIS-IV). The Psychological Corporation.

55.

Wechsler

(2009). Wechsler Individual Achievement Test (3rd ed.). The Psychological Corporation.

56.

Wechsler

(2014). Wechsler Intelligence Scale for Children (5th ed.). Pearson.

57.

Yudko

Lozhkina

Fouts

(2007). A comprehensive review of the psychometric properties of the Drug Abuse Screening Test. Journal of Substance Abuse Treatment, 32(2), 189–198. https://doi.org/10.1016/j.jsat.2006.08.002