Sage Journals: Discover world-class research

Abstract

The Structured Assessment of Protective Factors for Violence Risk-Youth Version (SAPROF-YV; de Vries Robbé et al., 2015) was designed specifically to assess strengths as a complement to risk assessment tools. We retrospectively examined its reliability and validity in 305 Canadian community-sentenced youth, both in the overall sample and in male and female, and Black and White, subgroups. In all groups, the total score had strong internal consistency, inter-rater reliability, and convergent validity, and significantly predicted general recidivism at 3-year fixed follow-up. The SAPROF-YV showed incremental validity over the YLS/CMI only in Black youth. In the total sample, a moderation effect was identified whereby strengths were protective at lower levels of risk but not for moderate or high risk youth. The SAPROF-YV shows promising reliability and validity; however, more research is needed before clear guidance can be provided regarding the use of this measure in clinical practice.

Keywords

protective factors recidivism youth risk assessment criminal justice system validity

Understanding and Assessing Youth Strengths in a Criminal Justice Context

Assessment is a core component of effective practice with youth involved in the criminal justice system. Sound assessment provides information regarding risk to reoffend and guides case management, including the intensity and nature of intervention intended to prevent reoffending. In many jurisdictions in North America and internationally (Cullen & Gilbert, 2012), correctional policies and practices—including assessment—are guided by the Risk–Need–Responsivity (RNR) framework, a theoretically informed, evidence-based rehabilitation model (Bonta & Andrews, 2017). Until recently, assessment in criminal justice contexts has focused predominantly on risk factors that are strongly and directly associated with recidivism (Gagnier, 2013; Singh, Desmarais, Hurducas, et al., 2014). For youth, these include antisocial or pro-criminal attitudes, antisocial personality features, associating with antisocial peers, substance abuse, family dysfunction, challenges in education/employment, and inappropriate use of leisure time (Bonta & Andrews, 2017; Gendreau et al., 1996). However, there is growing interest in the assessment of strengths in addition to risk. The inclusion of strengths is posited to improve the accuracy of assessments, enrich intervention planning, and enhance youth engagement in intervention (de Ruiter & Nicholls, 2011; de Vries Robbé & Willis, 2017; Singh, Desmarais, Sellers, et al., 2014). Furthermore, the RNR framework has explicitly highlighted the importance of strengths or protective factors for more than two decades. For example, both versions of the YLS/CMI manual (risk/needs tool based on RNR framework, Hoge & Andrews, 2002, 2011) articulate the importance of strengths as responsivity factors, with the more recent version including direct scoring of strengths within each area of criminogenic need. Furthermore, in more recent discussions of incorporating strengths within the RNR framework, it is noted that strengths should be assessed not only to enhance treatment effects but also to enhance prediction (Bonta & Andrews, 2017).

While strengths can be understood as personal and/or environmental characteristics that are associated with positive outcomes, there is conceptual diversity and debate regarding what constitutes a strength. Some scholars conceptualize strengths as the opposite pole of a risk factor (e.g., not having antisocial peers; McAra & McVie, 2016; Stouthamer-Loeber et al., 2002). Others contend that strengths must provide unique information not captured in the assessment of risk factors (Borum et al., 2006; de Vogel et al., 2009). Empirically, risk and strength measures have been found to be moderately to highly negatively correlated (de Vries Robbé et al., 2011; Kleeven et al., 2022; Viljoen et al., 2020), which calls into question their conceptual distinctiveness from risk. In addition, the evidence regarding the unique contribution of strengths in assessments (e.g., of likelihood of reoffending or future violence) is mixed (Wanamaker et al., 2018).

Scholars have suggested that the brief nature of the strength measures that are part of widely used risk assessment tools may account for their lack of incremental prediction when included alongside risk measures (e.g., Viljoen et al., 2020), which signals that how strengths are measured is important, both conceptually and empirically. Several widely used youth risk assessment tools include a small number of strength items (e.g., six items on the Structured Assessment of Violence Risk in Youth (SAVRY) Borum et al., 2006; seven items on the Youth Level of Service/Case Management Inventory YLS/CMI; Hoge & Andrews, 2011) and while a summing of these strengths items is typically associated with lower rates of reoffense, evidence for whether the score predicts risk for reoffense over and above risk scores is mixed (e.g., Gagnier, 2013; Shepherd et al., 2016; Viljoen et al., 2020). Other tools include items that are each considered for both risk and strength potential (e.g., the Short-Term Assessment of Risk and Treatability—Adolescent Version START-AV; Viljoen et al., 2014; the Youth Assessment and Screening Instrument YASI; Orbis Partners, 2000); again, empirical support for the incremental validity of the strength scales in predicting risk for reoffense is scant, although potentially promising (e.g., Jones et al., 2016). However, few tools have been developed specifically to assess strengths.

The SAPROF-YV: A Dedicated Measure of Youth Strengths in the Context of Violence and Offending

The Structured Assessment of Protective Factors for Violence Risk—Youth Version (SAPROF-YV; de Vries Robbé et al., 2015), an adaptation of the adult version of this measure, is intended for the sole purpose of assessing strengths and has been proposed as a more robust and comprehensive tool compared with risk-focused instruments that include only a handful of strength items. The SAPROF-YV has 16 items grouped into four strength domains: resilience (social competence, coping, self-control, and perseverance); motivational (future orientation, motivation for treatment, attitudes toward agreements and conditions, and medication); relational (school/work, leisure activities, parents/guardians, peers, and other supportive relationships); and external (pedagogical climate, professional care, and court order). The tool is intended to be used in conjunction with a risk-focused measure, such as the SAVRY or YLS/CMI, to help determine risk of violence or offending and to inform intervention plans.

In terms of psychometric properties, strong inter-rater reliability has been reported in several studies (Chu et al., 2020; de Vries Robbé et al., 2015; Kleeven et al., 2022; Li et al., 2019; Lovatt et al., 2022). Correlations between the SAPROF-YV and other strength measures provide evidence for convergent validity. Moderate to strong positive correlations have been reported in a number of studies, including with the SAVRY protective scale (e.g., r = .85, Christiansen et al., 2021; .63 and .89, de Vries Robbé et al., 2015; .65, Kleeven et al., 2022; .62, Koh et al., 2022; .80, Lovatt et al., 2022). Interestingly, notwithstanding the argument that strengths provide unique information that cannot be inferred from knowledge of risks, negative relationships between the SAPROF-YV and risk measures have also been proffered as evidence of its convergent validity. Studies consistently report strong negative correlations with SAVRY Risk total scores (e.g., −.76, Christiansen et al., 2021; −.59 and −.60, de Vries Robbé et al., 2015; −.61, Kleeven et al., 2022; −.65, Koh et al., 2022; −.83, Lovatt et al., 2022) as well as other risk measures (e.g., with the VRS-YV: −.63; Koh et al., 2022; −.77, Lovatt et al., 2022) and with the YLS/CMI (−.40; Chu et al., 2020).

The literature on strength measures in general—and the SAPROF-YV in particular—has grown over the past several years, with a number of recent studies focusing specifically on predictive validity and examining different aspects of how strengths function, including whether strengths have an independent, direct effect on outcome (i.e., serve as “promotive” factors; Farrington et al., 2016), whether strengths add incremental predictive validity over risk, and whether strengths moderate, or buffer, the effect of risk on outcome (i.e., serve as “protective” factors; Farrington et al., 2016). Outcomes in these studies include binary measures of any (overall) recidivism, violent recidivism, and nonviolent recidivism, with some studies focusing on the presence of reoffense and others focusing on the absence of reoffense.

Evidence that strengths function as promotive factors is available from a number of recent studies in which the predictive validity of the SAPROF-YV was assessed using receiver operating characteristic (ROC) analysis. Area under the curves (AUCs) reported in these studies vary in size (Rice & Harris, 2005), including large effects (AUC = .80 for any 2-year recidivism in a sample of community-sentenced Singaporean youth; Chu et al., 2020) and medium effects (e.g., AUCs = .65-.70 for any, violent, and nonviolent recidivism at follow-up periods of 6, 12, and 24 months after discharge from a Dutch juvenile justice institution; Kleeven et al., 2022, and AUCs = .65-.68 for any, violent, and nonviolent 3-year recidivism in a Canadian youth sample serving community or custodial sentences; Lovatt et al., 2022). Similarly, de Ruigh et al. (2021) found moderate predictive validity for absence of violent recidivism (AUC = .64) and strong predictive validity for absence of any recidivism (AUC = .72) at 1-year follow-up in a Dutch youth sample. In contrast, Koh et al. (2022) found that the SAPROF-YV was a small predictor of any, violent, and nonviolent recidivism at both 1- and 3-year follow-ups (AUCs = .58-.62) in a sample of Singaporean youth serving custodial or community sentences. Similarly, in a sample of Canadian youth, Goodwin et al. (2022) reported that the SAPROF-YV evidenced a small association with any (AUC = .58) and violent recidivism (AUC = .56) using a 3-year follow-up.

Evidence for incremental predictive validity is mixed. Kleeven et al. (2022) reported that that the SAPROF-YV total score added predictive power over the SAVRY risk factors for any, violent, and nonviolent recidivism at 6-month follow-up in their sample of Dutch youth with a history of violent behavior. In contrast, de Ruigh et al. (2021) found that the SAPROF-YV total score added predictive validity to SAVRY risk factors for any reoffending but not for violent reoffending at 1-year follow-up. Koh et al. (2022) found that the SAPROF-YV did not add to the ability of the SAVRY or the VRS-YV to predict any, violent, or nonviolent reoffending in a sample of Singaporean youth sentenced to probation or juvenile detention. Similarly, in a Canadian sample of youth, Lovatt et al. (2022) found that the SAPROF-YV did not contribute incremental predictive ability over the SAVRY or the VRS-YV to subsequent convictions for any, violent, and nonviolent offenses. Turning to another widely used risk assessment tool for youth recidivism, Chu et al. (2020) reported that the SAPROF-YV significantly improved predictive accuracy for any recidivism over and above the YLS/CMI risk score at 2-year follow-up. In contrast, Goodwin et al. (2022) reported that the SAPROF-YV did not significantly improve predictive accuracy for any or violent recidivism over and above the YLS/CMI at a 3-year follow-up (in males or females).

In addressing whether strengths moderate the relationship between risk and recidivism, Chu et al. (2020) found the relationship between the SAPROF-YV total and any recidivism did not vary across YLS/CMI risk scores. Several studies have examined this question with other strength measures; while some have found evidence for a protective effect (de Vries Robbé et al., 2013; Jones et al., 2016; Wanamaker et al., 2018), others have not (Dolan & Rennie, 2008; Goodwin et al., 2022; Schmidt et al., 2011; Viljoen et al., 2020).

The Importance of Examining Validity Across Groups

In the criminal justice context, validity of assessment tools across subgroups, including gender and race/ethnicity, has become an increasingly pressing issue in policy and research (Kitzmiller et al., 2022; Lovatt et al., 2022). The consequences associated with assessment can be extremely serious, including inappropriate labeling, failure to accurately identify and provide service for relevant criminogenic needs (Kitzmiller et al., 2022), and disproportionately negative outcomes for racialized groups that lead to continued overrepresentation in corrections systems (Lovatt et al., 2022). As a result, there has been increased focus on examining the validity of risk assessment tools in females as well as males, and across ethno-racial groups. While the psychometric properties of the SAPROF-YV have been examined in youth samples from several different countries (primarily Canada, the Netherlands, and Singapore), the samples were all or mostly male and in all but two, subgroups were not compared. Lovatt et al. (2022) compared the predictive validity of the SAPROF-YV (in addition to the SAVRY and VRS-YV) for males and females, and for Indigenous and nonIndigenous Canadian justice system-impacted youth referred by the court or their probation officer to a mental health facility for specialized assessment and intervention. AUCs for any 3-year recidivism were .72 for males and .64 for females, .65 for Indigenous youth, and .75 for nonIndigenous youth, suggesting that the SAPROF-YV better predicts nonrecidivism for White and male youth. Similarly, in a sample of Canadian justice-impacted youth Goodwin et al. (2022) reported that the SAPROF-YV predicted any recidivism (AUC = .68) and violent recidivism (AUC = .60) for boys but not for girls (AUCs = .53-.54); a 3-year fixed follow-up period was used.

The Present Study

In the current study, we retrospectively reviewed existing forensic clinical data to provide a validation of the SAPROF-YV in an ethnically diverse Canadian sample of justice-impacted youth on community supervision, comparing males and females, and Black and White youth (the two ethno-racial groups with sufficient sample size for group comparisons). First, we examined the SAPROF-YV’s inter-rater reliability, internal consistency, and convergent validity with the SAVRY and YLS/CMI. Then, we tested its structural validity in the sample as a whole; to our knowledge this is the first study to empirically examine the proposed four-domain structure of the measure. Notwithstanding that the SAPROF-YV is a Structured Professional Judgment (SPJ)—as opposed to an actuarial—tool, exploring its factor structure can inform its theoretical conceptualization and clinical use. Next, we analyzed its predictive validity for the total sample, and by subgroup (males and females, Black and White youth), including whether strengths measured by the SAPROF-YV operate via a promotive effect (direct relationship to recidivism), provide incremental predictive validity when considered alongside risks, or serve as protective factors (i.e., moderate the relationship between risks and recidivism).

Consistent with the studies reported above, we expected to find strong inter-rater reliability and convergent validity. We did not make any predictions about the factor structure of the SAPROF-YV given the lack of prior research. Evidencing a direct—or promotive—effect, we hypothesized that the SAPROF-YV total score would discriminate between recidivists and nonrecidivists and significantly predict recidivism, consistent with prior research (Chu et al., 2020; de Ruigh et al., 2021; Kleeven et al., 2022; Koh et al., 2022) and that it would predict recidivism better for males than for females (Goodwin et al., 2022; Lovatt et al., 2022). Given the lack of clear theoretical or empirical (Chu et al., 2020; de Ruigh et al., 2021; Kleeven et al., 2022; Koh et al., 2022; Lovatt et al., 2022) justification, we did not hypothesize that the SAPROF-YV would provide incremental validity to a measure of risk (the YLS/CMI) in predicting recidivism. Similarly, given the mixed findings from previous studies, no hypothesis was made as to whether the SAPROF-YV would function as a protective factor—that is, would moderate the relationship between risk and recidivism. We also made no hypotheses in terms of whether the SAPROF-YV would perform better for White or Black youth.

Method

Participants

Participants were 305 justice-impacted youth (58% male) between the ages of 12 and 19 seen for assessment in the forensic service of a mental health agency in a Canadian city between October 2001 and 2017 for court-ordered assessments to assist with sentencing. Data from youth who consented for clinical information to be used in research (84% of clients seen over this time period) were included in the study, which received institutional ethics approval. The sample included all consenting female youth and a random sample of male youth for whom data were available over the same assessment period; there were no exclusion criteria. The sample was ethnically diverse (see Table 1). Approximately, two-thirds of youth were charged with a violent offense, which is consistent with their status as a forensic sample; 27% were charged with a nonviolent offense, and a small minority were charged with a sexual offense. As shown in Table 1, females were more likely than males to be charged with a violent offense, whereas males were more likely to be charged with a sexual offense. The reoffense rate was higher for males than females.

Table 1:

Demographic, Criminal History, Risk, Strength, and Recidivism Characteristics for Males, Females and Total Sample

Variable	Total (N = 305)	Females (n = 128)	Males (n = 177)	Test statistic	Effect size
Variable	M (SD)	M (SD)	M (SD)	T	D
Age (12-19 years)	16.3 (1.5)	16.1 (1.4)	16.3 (1.5)	1.16	.14
Days to recidivism	429.2 (278.3)	437.5 (304.3)	425.8 (268.7)	−0.23	.04
YLS/CMI total (0-39)	20.8 (9.2)	21.8 (8.5)	20.1 (9.5)	−1.56	.18
SAPROF-YV total^a (1-23)	8.5 (5.0)	7.9 (4.9)	8.9 (5.0)	1.78	.21
SAVRY protective total (1-6)	2.6 (1.6)	2.5 (1.6)	2.8 (1.7)	1.74	.20
	% (n)	% (n)	% (n)	χ²	φ
Race/ethnicity				10.16	.19
Black	32.6 (95)	26.4 (33)	37.3 (62)
White	35.1 (102)	40.0 (50)	31.3 (52)
West Asian	5.2 (15)	5.6 (7)	4.8 (8)
Indigenous	4.8 (14)	6.4 (8)	3.6 (6)
South-east Asian	3.4 (10)	4.0 (5)	3.0 (5)
South/central American	2.7 (8)	4.0 (5)	1.8 (3)
East Asian	2.4 (7)	2.4 (3)	2.4 (4)
South Asian	1.4 (4)	0.0 (0)	2.4 (4)
Mixed race/ethnicity	5.8 (17)	5.6 (7)	6.0 (10)
Other	6.5 (19)	5.6 (7)	7.2 (12)
Index offense				9.43**	.18
Violent (nonsexual)	60.5 (184)	65.9 (85)	55.9 (99)
Nonviolent	27.3 (83)	27.6 (35)	27.1 (48)
Sexual	12.2 (37)	5.5 (7)	16.9 (30)
Recidivism—yes	52.1 (139)	43.5 (40)	56.6 (99)	4.14*	−.13
Recidivism type				.86	.08
Violent (nonsexual)	62.0 (85)	61.5 (24)	62.2 (61)
Nonviolent	36.5 (50)	28.5 (15)	35.7 (35)
Sexual	1.5 (2)	0.0 (0)	2.0 (2)

Note. Cohen’s d value of .2 is considered a small effect, .5 a medium effect, and .8 a large effect (Cohen, 1988). Phi (φ) value of .1 is considered a small effect, .3 a medium effect, and .5 a large effect (Cohen, 1988). YLS/CMI = Youth Level of Service/Case Management Inventory; SAPROF-YV = Structured Assessment of Protective Factors for Violence Risk-Youth Version; SAVRY = Structured Assessment of Violence Risk in Youth.

Not including Item 8 (Medication).

Measures

Strength Measures

SAPROF-YV

The SAPROF-YV (de Vries Robbé et al., 2015) comprised 16 strength items scored as either absent (0), partially present (1), or fully present (2), with higher scores representing more strengths. The manual outlines the type of information that would qualify a strength item as absent, partially present, or fully present. For example, the item “Parents/Guardians” would be coded as absent if the youth was described as estranged from parent figures (e.g., living in a group home), as having a neutral or negative relationship with parents/guardians, or as having a positive relationship with an antisocial parent/guardian. The item would be coded as fully present if the available information depicted the youth having a positive relationship (e.g., warm, supportive relationship involving adequate supervision) with at least one prosocial parent/guardian, even if they also had neutral or negative relations with other caregivers, and if this report was corroborated by all sources. The item would be coded as partially present if the youth was reported to have a positive relationship with a prosocial parent/guardian by one source but this was refuted by another source (e.g., parent reports what sounds like a positive relationship in credible terms but the youth reports not finding the relationship to be supportive). Following the procedure of previous studies (e.g., Christiansen et al., 2021; Chu et al., 2020), we summed items to create a total strength score; we did not retrospectively score an SPJ rating. The psychometric properties of the SAPROF-YV are described in the literature review above.

Structured Assessment of Violence Risk in Youth

The SAVRY (Borum et al., 2006) comprised 24 risk and six protective items designed to assess violence risk in 12- to 18-year-old adolescents. Only the protective factors scale was used in this study. Protective items, coded as present or absent, include prosocial involvement, strong social support, strong attachments and bonds, positive attitude toward intervention and authority, strong commitment to school, and resilient personality traits. The protective items were summed to create a protective total score, in line with previous studies (Lodewijks et al., 2010; Viljoen et al., 2020). We did not create an SPJ rating retrospectively.

The SAVRY has established reliability and validity, including strong internal consistency of subscales and total score (Borum et al., 2010), excellent inter-rater reliability for scoring of individual items, robust positive correlations between its risk total score other established measures of risk, such as the YLS/CMI (Borum et al., 2006), negative correlations between its strength total score and the YLS/CMI (Borum et al., 2006), and positive correlations of the strength scale with the SAPROF-YV total (de Vries Robbé et al., 2015). Meta-analyses have shown the SAVRY risk total exhibits moderate to strong predictive validity for any and violent recidivism (Olver et al., 2009; Singh et al., 2011). There is also evidence for good predictive validity of the SAVRY protective scale for any recidivism (Kleeven et al., 2022; Koh et al., 2022; Lovatt et al., 2022), with mixed evidence with respect to violent recidivism (Kleeven et al., 2022; Lovatt et al., 2022).

Risk Measure

The Youth Level of Service/Case Management Inventory (YLS/CMI) (Hoge & Andrews, 2002) is a standardized instrument used to assess the criminogenic needs and risk to reoffend of adolescents 12-18 years of age. It consists of 42 yes/no items divided into eight domains: Offense History, Family Circumstances/Parenting, Education/Employment, Peer Relations, Substance Abuse, Leisure/Recreation, Personality/Behavior, and Attitudes/Orientation. A total risk score is derived from summing all items.

The YLS/CMI was completed by the assessing clinician at the time of the assessment. Inter-rater reliability for the total score is high, with correlations among clinicians ranging from .80 to .98 (average r = .93; Vitopoulos et al., 2019). The YLS/CMI has demonstrated moderate to strong internal consistency for the subscales and total score (Rowe, 2002; Schmidt et al., 2005). Convergent validity is evidenced through correlations between YLS/CMI total score and broad and narrow band scores on the Child Behavior Checklist (Schmidt et al., 2005). The tool has also shown predictive validity for both general and violent reoffending, and number of subsequent offenses and time elapsed prior to a new offense (e.g., Olver et al., 2009, 2014; Peterson-Badali et al., 2015; Viljoen et al., 2020; Vitopoulos et al., 2012).

Recidivism

Information on recidivism was obtained from the Royal Canadian Mounted Police (RCMP) Canadian Police Information Center (CPIC) database. Any recidivism (yes/no) was defined as one or more new convictions that occurred within a 3-year fixed period following the youth’s assessment; violent recidivism included convictions in the same timeframe for offenses against the person (e.g., robbery, assault). Criminal records were also reviewed against the offense history summary in youths’ clinical reports to exclude new offenses that were committed prior to their assessment but that resulted in convictions following their assessment due to the time it takes for cases to make their way through the court system. For 41 cases, no recidivism data were provided. As the result of a provision in Canada’s youth criminal justice legislation, a CPIC record for a youth with a previous conviction is expunged following a specified period during which no new offense is committed. In this case, a records request would be returned with no information for that youth. It was also the case that recidivism data were collected as part of a larger program of research; therefore, requests for recidivism data were not made for all youth in this study.

Procedure

During their forensic assessment, youth—and typically a parent—met with a clinician (psychologist or psychiatrist and a social worker) to complete an in-depth clinical forensic interview (7-10 hours in total); youth and parents also typically completed a variety of questionnaires and standardized tests. Clinicians also interviewed additional collateral sources of information (e.g., probation officers, school staff, and previous service providers). Clinicians produced a comprehensive report, which was typically 25 to 35 pages in length focused on psychosocial history, criminogenic needs, cognitive and academic functioning, reoffense risk, mental health, and strengths. In this study, the first author reviewed the comprehensive reports to gather information on demographics, offense history, charges leading to referral for assessment, and YLS/CMI scores, and to retrospectively code for strengths using the SAPROF-YV and the SAVRY protective factors scale. For both instruments, the first author was trained to score the tools by one of the study’s co-authors, a youth forensic psychologist and head of the clinic at which the assessments were conducted—and who received formal training on a variety of adult and youth risk assessment tools, including the SAPROF and the SAVRY. The forensic psychologist undertook a detailed review of the training manuals for the SAPROF, the SAPROF-YV and the SAVRY with the first author. All coding adhered to the manual instructions and agreement was established through coding a number of practice cases. Inter-rater reliability between the first author and a second, similarly trained, student coder is reported in the “Results” section. All coders were blind to the outcome (reoffending) when coding the SAPROF-YV and SAVRY. Total scores (the sum of all items) rather than SPJ ratings were used for both the SAPROF-YV and the SAVRY. There is evidence that both SPJ ratings and total scores on SPJ measures are equally effective at predicting recidivism in justice system-impacted youth (Dolan & Rennie, 2008; Schmidt et al., 2011) and this approach is consistent with previous research (Lovatt et al., 2022; Shepherd et al., 2016; Viljoen et al., 2020).

Analytic Plan

Descriptive statistics on risk and strength variables were provided for the total sample, males, females, Black youth, and White youth. Descriptive statistics were further disaggregated within the female and male subsamples as a function race (i.e., Black females, White females, Black males, White males). In terms of race/ethnicity, the only subgroups large enough for inferential analyses were Black and White youth. Within this subsample, a multivariate analysis of variance (MANOVA) was run to explore potential gender and race differences in scores on the SAPROF-YV, SAVRY protective total, and YLS/CMI scores. For the total sample (as well as separately for males and females, and for Black and White youth), inter-rater reliability and internal consistency were examined using intra-class coefficients (ICCs) and Cronbach’s alpha, respectively.

The internal structure of the SAPROF-YV was examined with both confirmatory factor analysis (CFA) based on polychoric correlations using the weighted least squares means and variance (WLSMV) estimator and exploratory factor analysis (EFA) with principal axis factoring extraction using Promax for the oblique rotation, using Stata 15 (StataCorp, 2017). Factor solutions were considered potentially appropriate if they had eigenvalues greater than 1.0 (Kaiser, 1958). The deletion of items was considered for factors with loadings less than .40 (Pett et al., 2003), cross-loadings that exceeded .32 on two or more factors (Tabachnick & Fidell, 2007), or communalities below .40 and uniqueness values above .60 (Costello & Osborne, 2005); theoretical conformity was also considered when evaluating models (Sellbom & Tellegen, 2019). For the total sample (as well as separately for males and females, and for Black and White youth), correlations between the SAPROF-YV and the SAVRY protective total and the YLS/CMI total scores were run to examine convergent validity.

To examine predictive validity and to test whether strengths operate via a promotive (direct) effect, for the total sample—and for the subsamples of Black vs. White youth and female vs. male youth—we conducted ROC analyses with recidivism as the outcome. Next, we ran hierarchical logistic regressions to examine the incremental predictive validity of the SAPROF-YV over the YLS/CMI; YLS/CMI scores were entered in the first block, and the SAPROF-YV was added to the second block. Analyses were conducted separately for the total sample, Black youth, White youth, female youth and male youth. To test for a protective (moderation) effect of strengths on the relationship between risk and recidivism, a risk by strength interaction term was added in the second block of the regression for the total sample. Subgroup sample sizes were too small to reliably conduct moderation regressions.

Results

Descriptive Statistics

Table 2 provides the descriptive statistics (Ms and SDs) for the SAPROF-YV, SAVRY protective, and YLS/CMI total scores. Mean SAPROF-YV and SAVRY total scores fell in the bottom third of their respective ranges, overall. YLS/CMI scores fell within the moderate risk range for both males and females (Hoge & Andrews, 2011). There were no gender differences for the SAPROF-YV total (males: M = 8.9, SD = 5.0; females: M = 7.9, SD = 4.9; t(303) = 1.74, p = .08, Cohens d = .21), SAVRY protective total (males: M = 2.8, SD = 1.6; females: M = 2.4, SD = 1.6; t(303) = 1.74, p = .08, Cohen’s d = .20), or YLS/CMI total scores (males: M = 20.1, SD = 9.5; females: M = 21.8, SD = 8.5; t(300) = -1.56, p = .12, Cohen’s d = −.18). A separate MANOVA based on the subsample of youth identified as Black or White (with YLS/CMI, SAPROF-YV, and SAVRY total scores entered as DVs and race (Black or White) and gender entered as IVs) revealed no overall significant differences: Wilks’ lambda tests for gender (F(3) = 1.39, p = .25), race (F(3) = 0.25, p = .99), and the gender by race interaction (F(3) = .61, p = .61). There was a significant main effect for gender on the SAPROF-YV score (F(1) = 4.18, p = .04), with females scoring lower than males; however, the effect was very small (partial eta squared = .022).

Overall, youth who reoffended had lower SAPROF-YV scores than youth who did not (10.23 vs. 7.18, t(266) = 5.12, p < .001). This finding replicated in the subsamples: males (10.95 vs. 7.48 t(174) = 4.76, p < .001), females (9.21 vs. 6.43, t(90) = 2.68, p = .009), Black youth (10.53 vs. 7.76, t(84) = 2.71, p = .008), and White youth (9.91 vs. 6.57 t(89) = 3.25, p = .002).

Table 2:

Mean SAPROF-YV Total, SAVRY Protective Total, and YLS/CMI Total Scores as a Function of Gender and Race

Measure	Total sample (N = 194)	Females (n = 80)			Males (n = 114)
Measure	M (SD)	Black (n = 31)M (SD)	White (n = 49)M (SD)	All femalesM (SD)	Black (n = 62)M (SD)	White (n = 52)M (SD)	All malesM (SD)
SAPROF-YV total	8.3 (4.9)	7.1 (4.2)	7.7 (4.4)	7.5 (4.3)	9.3 (4.9)	8.5 (5.5)	8.9 (5.2)
SAVRY protective total	2.7 (1.5)	2.4 (1.5)	2.6 (1.4)	2.5 (1.4)	3.0 (1.5)	2.8 (1.7)	2.9 (1.6)
YLS/CMI total	21.4 (9.1)	22.4 (6.5)	22.6 (9.0)	22.5 (8.0)	20.7 (8.8)	20.6 (10.7)	20.7 (9.7)

Note. SAPROF-YV = Structured Assessment of Protective Factors for Violence Risk-Youth Version; SAVRY = Structured Assessment of Violence Risk in Youth; YLS/CMI = Youth Level of Service/Case Management Inventory.

Reliability

To evaluate inter-rater reliability, a trained research assistant, blind to outcome, independently coded the SAPROF-YV and SAVRY for approximately 20% of cases in the sample. Substantial inter-rater agreement was found for both measures, using ICC, two-way mixed effects model, single measure, absolute agreement. The SAVRY total score (n = 60) mean ICC = .86, 95% CI [.77, .91]; the SAPROV-YV total score (n = 51) mean ICC = .96, 95% CI [.94, .98]. The SAPROF-YV total score had adequate internal consistency: α = .79 for the total sample, .80 for males, .80 for females, .80 for Black youth, and .79 for White youth. However, the internal consistency coefficients for three of the four subscales fell below the acceptable α of .70 (Nunnally, 1978): resilience = .76, motivation = .65, relational = .33, and external = .36. Inter-rater reliability was not available for YLS/CMI scores in this sample given that the YLS/CMI scores were obtained directly from clinical files rather than coded retrospectively for research purposes.

Structural Validity

To examine the internal structure of the SAPROF-YV, we ran a CFA based on the resilience, motivational, relational, and external subscales. Item 8 (medication) was omitted because it was coded “not applicable” for 72% of the sample. The solution did not converge, and no output was provided, indicating that this structure did not hold in our sample. Other factor solutions were explored using EFA with principal axis factoring as an extraction method and Promax oblique rotation. When unrestricted, the EFA produced a three-factor solution, accounting for 72% of the variance in scores, with three eigenvalues greater than 1 (5.7, 1.5, and 1.3) (see Supplemental Table S1, available in the online version of this article). However, one factor included fewer than three items (the suggested minimum to constitute a factor; Costello & Osborne, 2005), the solution contained four cross loaded items, and the three factors did not make sense theoretically when considering the content of the scale item groupings. We repeated the EFA, forcing a two-factor solution that accounted for 61% of the variance (see Supplemental Table S2, available in the online version of this article). While the two factors made some theoretical sense, there were issues around uniqueness values and cross loadings, and the internal consistency of the second factor was quite poor (α = .36). We ran a third EFA with principal axis factoring extraction, forcing a one-factor solution that accounted for 48% of the variance (see Supplemental Table S3, available in the online version of this article). Three items had factor loadings below .4 (motivation for treatment; professional care; and court order). When the EFA was re-run with these items removed, the factor accounted for 68.5% of the variance (see Supplemental Table S4, available in the online version of this article) and showed good internal consistency (alphas of .84 for the total sample, .83 for males, .81 for females, .80 for Black youth and .79 for White youth). These results suggest that the SAPROF-YV total score held together as a construct but that the subscales did not. Thus, we used the full SAPROF-YV scale (minus the medication item) for all subsequent analyses.

Convergent Validity

Support for convergent validity was demonstrated by correlations in the predicted direction between the SAPROF-YV and the SAVRY protective score (r = .77 [305], p < .001) as well as the YLS/CMI total (r = −.64 [302], p < .001) for the total sample. Similar results were obtained when males and females and Black and White youth were analyzed separately. Correlations between the SAPROF-YV and the SAVRY protective score were: r = .75 [128], p < .001 for females, r = .78 [177], p < .001 for males, r = .80 [95], p < .001 for Black youth and r = .75 [102], p < .001 for White youth. Correlations between the SAPROF-YV and the YLS/CMI were: r = −.61 [125], p < .001 for females, r = −.66 [177], p < .001 for males, r = −.53 [93], p < .001 for Black youth and r = −.68 [101], p < .001 for White youth.

Predictive Validity

Table 3 shows the AUC values for the SAPROF-YV, SAVRY Protective, and YLS/CMI scores for predicting any recidivism for the total sample, as well as separately for males and females, and for Black and White youth. Across the samples, the SAPROF-YV predicted reoffending significantly with a medium effect size (Rice & Harris, 2005) and with equal strength between Black and White youth. However, the SAPROF-YV illustrated a slight advantage in males vs. females. The SAVRY protective scale significantly predicted reoffense with medium strength for males and White youth, and with a small but significant AUC for the total sample. The SAVRY protective scale was not a significant predictor of reoffending for females or Black youth. Finally, the YLS/CMI predicted any reoffending with large AUCs for males and White youth, and medium AUCs for females and the full sample. Of note, the YLS/CMI predicted recidivism at chance levels for Black youth, and there was a significant difference in AUCs between Black and White youth (see Table 3). Thus, the SAPROF-YV demonstrated medium sized promotive (direct) effects across all subsamples whereas the SAVRY had only medium sized promotive effects for males and White youth; it did not have a promotive effect for females or Black youth.

Table 3:

ROC Analyses: SAPROF-YV, SAVRY Protective Scale, and YLS/CMI Total Scores Predicting Any Recidivism Across Groups

Variable	Females (n = 128)		Males (n = 177)		Z-score group difference	Black youth (n = 86)		White youth (n = 91)		Z-score group difference	Total
Variable	AUC	95% CI	AUC	95% CI	Z-score group difference	AUC	95% CI	AUC	95% CI	Z-score group difference	AUC	95% CI
SAPROF-YV strength total	.64*	[.52, .75]	.68***	[.60, .77]	0.69*	.66**	[.54, .78]	.67**	[56, .78]	−0.12	.66***	[.59, .72]
SAVRY protective total	.57	[.45, .69]	.64**	[.55, .72	0.87	.55	[.42, .67]	.64*	[.53, .76]	−1.11	.60**	[.53, .67]
YLS/CMI risk total	.64*	[.53, .76]	.75***	[.67, .82]	1.47	.56	[.43, .69]	.78***	[.69, .88]	−2.67**	.70***	[.64, .77]

Note. An AUC value of .56 is considered a small effect, .64 a medium effect, and .71 a large effect (Rice & Harris, 2005). The option “smaller test result indicates more positive test” was selected in SPSS for the strength-based measures. AUC = area under the curve; ROC = receiver operating characteristic; SAPROF-YV = Structured Assessment of Protective Factors for Violence Risk-Youth Version; SAVRY = Structured Assessment of Violence Risk in Youth; YLS/CMI = Youth Level of Service/Case Management Inventory; CI = confidence interval.

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

When violent recidivism was the outcome, for the total sample, the SAPROF-YV had a small but significant effect for reoffending (AUC = .58, p = .05, 95% CI [.51, .65]), the SAVRY protective score did not significantly predict reoffense (AUC = .53, p = .43, 95% CI [.46, .61]), and the YLS/CMI moderately predicted reoffense (AUC = .64, p < .001, 95% CI [.57, .71]). Because of the small number of violent recidivists in the sample as a whole (see Table 1), it was not possible to conduct separate subgroup analyses.

To test for incremental predictive validity of the SAPROF-YV alongside YLS/CMI total risk, we ran hierarchical logistic regressions (YLS/CMI entered into Block 1, SAPROV-YV entered into Block 2) for the total sample and the subgroups of male and female youth, and Black and White youth (see Table 4). We also included corresponding AUC values for Block 1 and Block 2 (using the predicted probability values). Across all Block 1 analyses, the YLS/CMI total risk score was a significant individual predictor of any recidivism in all samples except Black youth. Interestingly, across all Block 2 analyses, the only time the SAPROV-YV evidenced incremental predictive validity over the YLS/CMI was for the subsample of Black youth, though the confidence interval did contain 1. It is noteworthy that, in Block 2, neither the SAPROF-YV nor the YLS/CMI significantly predicted reoffense for female youth.

Table 4:

Logistic Regressions: Incremental Associations of Strength (SAPROF-YV) and Risk (YLS/CMI) For Any Recidivism Across Groups

Effect	B (SE)	Wald’s χ²	OR	[95% CI]	AUC (SE)	[95% CI]
Total sample (N = 265)
Block 1					.70 (0.03)	[.64, .77]
YLS/CMI	.09 (0.02)	31.67***	1.09	[1.06, 1.12]
Block 2					.70 (0.03)	[.64, .77]
SAPROF-YV	−.04 (0.03)	1.61	−0.96	[0.90, 1.02]
YLS/CMI	.07 (0.02)	14.55***	1.08	[1.04, 1.12]
Male youth (n = 176)
Block 1					.75 (0.04)	[.67, .83]
YLS/CMI	.11 (0.02)	31.18***	1.12	[1.08, 1.16]
Block 2					.75 (0.04)	[.67, .82]
SAPROF-YV	−.03 (0.04)	0.57	−0.97	[0.89, 1.05]
YLS/CMI	.10 (0.03)	16.81***	1.11	[1.05, 1.16]
Female youth (n = 89)
Block 1					.64 (0.06)	[.53, .76]
YLS/CMI	.06 (0.03)	4.74*	1.06	[1.01, 1.12]
Block 2					.64 (0.06)	[.52,.75]
SAPROF-YV	−.08 (0.06)	1.66	−0.93	[0.82, 1.04]
YLS/CMI	.03 (0.03)	1.02	1.03	[0.97, 1.10]
Black youth (n = 84)
Block 1					.56 (0.07)	[.43, .69]
YLS/CMI	.03 (0.03)	1.45	1.03	[0.98, 1.09]
Block 2					.66 (.06)	[.54, .77]
SAPROF-YV	−.12 (0.06)	4.04*	0.89	[.79, 1.00]
YLS/CMI	.00 (0.03)	0.01	1.00	[0.94, 1.06]
White youth (n = 90)
Block 1					.78 (.05)	[.69, .88]
YLS/CMI	.12 (0.03)	17.36***	1.13	[1.07, 1.20]
Block 2					.79 (.05)	[.69, .88]
SAPROF-YV	.02 (0.07)	0.10	1.02	[0.90, 1.16]
YLS/CMI	.13 (0.04)	11.76***	1.14	[1.06, 1.23]

Note. AUC = area under the curve; SAPROF-YV = Structured Assessment of Protective Factors for Violence Risk—Youth Version Strength Total; YLS/CMI = Youth Level of Service/Case Management Inventory Risk Total; CI = confidence interval.

p <.05. ***p < .001.

Andrew F. Hayes Process Macro Version 4.1 (SPSS) was used to test whether the total SAPROF-YV score would evidence a protective effect against YLS/CMI scores in the prediction of any reoffending. We could only perform this analysis in the total sample due to sample size restrictions. A significant interaction between YLS/CMI and SAPROF-YV scores emerged (b = .0113, SE = .0038, Z = 2.96, p = .003. As Figure 1 shows, strengths appeared to reduce recidivism in lower risk cases, but strengths had no impact on moderate risk cases. Counter-intuitively, strengths appeared to increase recidivism among the higher risk cases. However, further examination revealed that this finding was attributable to very few high scoring YLS/CMI cases (over 30) who also scored high on the SAPROV-YV (13.84 or more: 1 SD or higher). In this data set, only one person scored over 30 on the YLS/CMI and 13.84 or more on the SAPROV-YV; it was this person who reoffended, which accounts for the observed result.

Figure 1:

Interaction Between YLS/CMI Total and SAPROF-YV Total for the Total Sample (N = 265)

Discussion

To our knowledge, the SAPROF-YV is the only tool created solely to assess strengths among justice system-impacted youth. Results from this study lend support to the reliability and validity of this tool for the assessment of strengths among justice-impacted youth, while also highlighting some areas of departure from the structure as proposed by its authors. Subgroup analyses—in which male and female youth and Black and White youth were examined separately—indicated some areas of similarity across groups as well as distinct findings, which suggest that youth cannot be treated as a homogeneous group.

The SAPROF-YV had high internal consistency. Evidence of strong inter-rater reliability is consistent with findings of recent studies (Chu et al., 2020; de Vries Robbé et al., 2015; Kleeven et al., 2022; Li et al., 2019; Lovatt et al., 2022). Support for the tool’s convergent validity comes from a strong positive correlation between the SAPROF-YV and SAVRY strength total score, which is consistent with the findings of several studies (Christiansen et al., 2021; de Vries Robbé et al., 2015; Kleeven et al., 2022; Koh et al., 2022; Lovatt et al., 2022). The strong negative correlation with the YLS/CMI total risk score is consistent with Chu et al.’s (2020) finding and with the robust negative relationship between the SAPROF-YV and other risk measures, including the SAVRY (Christiansen et al., 2021; de Vries Robbé et al., 2015; Kleeven et al., 2022; Koh et al., 2022; Lovatt et al., 2022) and VRS-YV (Koh et al., 2022; Lovatt et al., 2022). Taken together, the results also add to evidence supporting the conception of strengths falling on a single risk-strength dimension rather than as a construct independent of risk.

This is the first study to examine the structural validity of the SAPROF-YV. CFA and EFA analyses indicated that a one factor solution best fit the data, and suggested the removal of three items—motivation for treatment, professional care, and court order—in addition to medication (which was not applicable for the majority of the sample). Li et al. (2019) also reported low variability in the medication, professional care, and court order items and excluded these items from their analyses. Lack of support for the structural validity of the four subscales is interesting in light of previous evidence of concurrent validity (e.g., Lovatt et al., 2022) and predictive validity for general recidivism (e.g., Kleeven et al., 2022; Lovatt et al., 2022), though psychometric support has been weaker for the External subscale. While the factor structure of an SPJ assessment tool does not speak directly to its validity, further examination of the SAPROF-YV’s structure with larger samples will allow for subgroup analyses and can help evaluate the usefulness of each of the SAPROF-YV items, as not all items may be equally useful in assessing youth’s strengths.

Structure aside, as expected, youth who reoffended had lower mean scores on the SAPROF-YV, both in the total sample and in the subgroups. Moreover, ROC analyses lent support to the predictive validity of the SAPROF-YV in relation to any recidivism. Our findings were consistent with the moderate relationships reported by de Ruigh et al. (2021), Kleeven et al. (2022) and Lovatt et al. (2022). They contrasted somewhat with Chu et al. (2020)’s large AUC and with Goodwin et al. (2022) and Koh et al. (2022), who found that the SAPROF-YV was a more modest predictor of any recidivism. The SAPROF-YV was a weak predictor of violent recidivism, which is consistent with Goodwin et al. (2022) and Koh et al. (2022) but contrasts with de Ruigh et al. (2021), Kleeven et al. (2022), and Lovatt et al. (2022), who found the tool to be a moderate predictor of violent recidivism. That said, consistent with our findings, other studies have found the tool to better predict any recidivism than violent recidivism (de Ruigh et al., 2021; Lovatt et al., 2022).

Researchers have proposed that the addition of strengths will improve the accuracy of risk assessment but the evidence for this assertion is mixed, with some studies finding the SAPROF-YV to contribute incrementally over risk measures, including the SAVRY (de Ruigh et al., 2021; Kleeven et al., 2022) and YLS/CMI (Chu et al., 2020), while others have found no incremental contribution of the SAPROF-YV over various risk measures (Goodwin et al., 2022; Koh et al., 2022; Lovatt et al., 2022). In the current study, in the overall sample—as well as the subsamples of males, females, and White youth—the SAPROF-YV was not a unique predictor of general recidivism alongside the YLS/CMI. However, in Black youth (for whom the YLS/CMI was a poor predictor of recidivism), the SAPROF-YV did contribute significantly to the prediction of reoffending. These results are consistent with the robust negative correlations (in the mid- to high: .6 range) between the SAPROF-YV and YLS/CMI discussed earlier, which suggest sizable overlap in what is being measured by the two instruments. It is noteworthy that the correlation was smaller—though still sizable—in the subgroup of Black youth (−.58). This finding suggests that there was somewhat less overlap in what was being measured between the SAPROF-YV and the YLS/CMI in this group, which is also consistent with the finding that the SAPROF-YV was a significant predictor of recidivism alongside the YLS/CMI for Black youth. These results also speak to the question of how strengths can best be conceptualized. The finding that, in most groups, strengths did not provide additional predictive power with respect to reoffending adds evidence to the argument that strengths and risks are poles of a continuum.

To explore whether strengths can be conceptualized as protective factors, we examined whether strengths would moderate the relationship between risk and reoffending. Our results indicated a protective effect only among lower risk youth; strengths were associated with reduced recidivism in this group, but not in moderate or high risk youth. Indeed, in the high risk group, it appeared that strengths increased the likelihood of recidivism, but this result is unreliable due to the tiny number of high risk/high strength youth in the sample. Shepherd et al. (2016) found a similar result among the lower risk youth in their sample of Australian males in detention: the SAVRY protective factors score was associated with nonreoffending, but the effect did not extend to higher risk youth. In contrast to the current findings, a protective effect was identified by Jones et al. (2016), who found that high levels of strength (on the YASI) improved outcomes for high-risk individuals, but not for low-risk youth.

Aside from the theoretical question of whether strengths can be understood as protective factors, the moderation finding is relevant for treatment planning: lower risk youth might benefit from incorporating strength-focused treatment into interventions that target criminogenic needs, while for higher risk youth, it may be that less weight should be afforded to strength factors until more immediate criminogenic needs have been addressed, or at the very least, strengths should not be the focus of service provision without an equally strong focus on criminogenic needs. Further research is needed to examine whether such differential practice would positively impact treatment success and reoffending outcomes. Unfortunately, the few studies examining moderation effects have produced inconsistent results (e.g., Chu et al., 2020; de Vries Robbé et al., 2013; Dolan & Rennie, 2008; Goodwin et al., 2022; Jones et al., 2016; Schmidt et al., 2011; Viljoen et al., 2020; Wanamaker et al., 2018). Thus, more research is needed before we can begin to understand how strength information can or should be incorporated into risk assessment and service planning.

Our results also highlight the importance of examining how assessment tools function in specific sub-groups of justice system-impacted individuals. Attention to the importance of gender has grown substantially over the last 15 years, and—more recently—researchers have begun to examine ethno-racial groups separately (e.g., Lovatt et al., 2022). When potentially high-stake decisions are made on the basis of these assessments, it is incumbent on clinicians to ensure that the instruments they use are reliable and valid in the population from which their clients come, which requires that researchers establish an adequate base of evidence. With respect to the functioning of the SAPROF-YV in male and female youth, and Black and White youth, there was evidence of strong convergent validity across groups. With respect to predictive validity, we have already touched on the fact that the SAPROF-YV (and the YLS/CMI and SAVRY) functioned differently in Black youth than in White youth. While the SAPROF-YV predicted reoffending with equal strength between Black and White youth, the YLS/CMI predicted recidivism at chance levels for Black youth, a finding that may be limited to this particular sample but that is nonetheless concerning. It was only in Black youth that the SAPROF-YV evidenced incremental predictive validity over the YLS/CMI—presumably because the latter was such a poor predictor of recidivism. In terms of gender differences, the AUCs for males and females were similar with respect to general recidivism, which contrasts with the findings of Goodwin et al. (2022) and Lovatt et al. (2022) that the SAPROF-YV had stronger predictive validity for male youth than female youth. When we examined the incremental predictive validity of the SAPROF-YV for female youth in the model that included both the SAPROF-YV and the YLS/CMI, neither measure significantly predicted reoffending. Together with these other recent studies, our results indicate a continued need to examine the SAPROF-YV (as well as other measures of risk and strength) within subgroups of youth. This need is particularly important for racialized and marginalized groups (e.g., Indigenous youth; Lovatt et al., 2022) who are overrepresented in the criminal justice system.

Results should be considered in light of the following limitations. First, the use of retrospective chart-review, while common for the investigation of new measures (Campbell et al., 2009; Gearing et al., 2006), makes it impossible to discern whether the absence of information about specific strength items represents the true absence of those characteristics or simply reflects missing information. Moreover, the SAPROF-YV was coded for a baseline score only. Given the dynamic nature of the items, regularly updated assessments (i.e., every 6 months) may improve the predictive validity of the measure. Reassessment of dynamic risk factors improves prediction (Lloyd et al., 2020); therefore, future research would benefit from evaluating strengths at multiple time points. In addition, because the SAPROF-YV and SAVRY were not completed as part of youth’s assessments, we do not have SPJ measures associated with these tools, instead analyzing total strength scores. While there is evidence of good agreement of SPJ ratings and scores (Dolan & Rennie, 2008; Schmidt et al., 2011), future studies would benefit from examining both types of measures in validating the SAPROF-YV. Similarly, we did not have data on the YLS/CMI strengths scale. In future research, it will be useful to examine this scale—along with the SAVRY strengths—in comparison with the SAPROF-YV, which is a lengthier, standalone measure of strengths.

While the analysis of subgroups—males and females, and Black and White youth—is a strength of the study, small subgroup sizes meant that we could not do separate incremental validity and moderation analyses on these subgroups, as well as to examine race and gender intersectionally; consistent with our above discussion of subgroup analyses, these are important directions for future studies. In addition, the number of youth with a violent reoffense precluded our ability to conduct ROC and regression analyses with violent reoffending as the outcome. In future research, it will be important to continue examining the role of the SAPROF-YV, in combination with a risk measure, as a predictor of violent reoffending as well as general recidivism.

The results from this study provide support for the psychometric properties of the SAPROF-YV, including strong inter-rater reliability and internal consistency, as well as evidence for convergent validity and predictive validity for general reoffending. These findings indicate that this measure can be used to reliably and validly to assess strengths among justice-impacted youth. Further research is needed to build on these results; in particular, it will be important for the moderation analysis to be investigated in additional samples before clear recommendations can be made regarding the use of the SAPROF-YV to enhance current risk assessment procedures. Our understanding of strengths and their relationship to recidivism is exceedingly underdeveloped compared with our understanding of the associations between risk/criminogenic needs and reoffending (Dickens & O’Shea, 2017). It is vital that further research help elucidate how strengths operate in relation to risk and reoffending to guide changes to policy and practice.

In addition to, or perhaps instead of, contributing to risk prediction, the authors of the SAPROF-YV have also suggested that this tool might enhance rehabilitation service planning and provision by highlighting areas of strength that can be used to increase motivation or engagement in services (de Vries Robbé et al., 2015). This aim aligns with the specific responsivity principle of the RNR framework (Bonta & Andrews, 2017). However, while it is a compelling idea, there is a dearth of research examining this position. An empirical investigation into the role of strengths in case management and treatment provision may help augment our understanding of effective rehabilitation and the cessation of offending (de Vries Robbé & Willis, 2017) and is therefore an important avenue for future research. Existing research suggests that probation officers find risk assessment tools that include strength factors (e.g., the SAVRY) helpful for guiding service and supervision recommendations (Guy et al., 2014) and, importantly, that such tools lead to a greater consideration of strength factors in relation to service recommendations (Vincent et al., 2012). There is preliminary evidence that strength factors function as specific responsivity factors (Finseth et al., 2022), that using strength factors to inform rehabilitation is valuable (Singh, Desmarais, Sellers et al., 2014), and that improvements in strength factors over time is linked to reduced recidivism and positive community outcomes (Coupland & Olver, 2020), but additional research is needed to further advance our understanding of the benefits of assessing strengths in the youth justice context.

Supplemental Material

sj-docx-1-cjb-10.1177_00938548231165286 – Supplemental material for Assessment of Strengths in Criminal Justice System-Impacted Youth: A Retrospective Validation Study of the SAPROF-YV

Supplemental material, sj-docx-1-cjb-10.1177_00938548231165286 for Assessment of Strengths in Criminal Justice System-Impacted Youth: A Retrospective Validation Study of the SAPROF-YV by Sonia Finseth, Michele Peterson-badali, Shelley L. Brown and Tracey A. Skilling in Criminal Justice and Behavior

Footnotes

Authors’ Note:

This paper is dedicated to Jodi Viljoen,whose research,knowledge mobilization,community engagement,and mentorship was devoted to supporting young people’s strengths and wellbeing,and to contributing to a more just society.

Funding for this research was supported in part by a SSHRC Doctoral Scholarship to the first author.

ORCID iDs

Sonia Finseth

Michele Peterson-Badali

Shelley L. Brown

Supplemental Material

Supplemental Tables S1 to S4 are available in the online version of this article at

Sonia Finseth,PhD,is a psychologist with the Ministry of Children and Family Development of British Columbia. Her research interests include mental health issues among justice-involved youth,the evaluation of risk and strength assessment practices for youth from diverse cultural and ethnic backgrounds and understanding the desistance process of youth involved in the justice system.

Michele Peterson-Badali,PhD,is a professor in the Department of Applied Psychology and Human Development at the Ontario Institute for Studies in Education,University of Toronto. Her research focuses on children and adolescent knowledge,reasoning,perceptions,and experiences of the youth justice system;their understanding of rights;and their evolving legal capacities. Current projects focus on the implementation of the Risk–Need–Responsivity framework in community-sentenced youth,mental health and youth criminal justice,and Indigenous youth.

Shelley L. Brown,PhD,is a professor of forensic psychology within the Department of Psychology,Carleton University. She is also an associate graduate faculty member with the Department of Applied Psychology and Human Development,Ontario Institute for Studies in Education,University of Toronto. Her program of research aims to improve gender-responsive services for girls and women in the criminal justice system. Current projects focus on examining the reliability and validity of existing psychologically based assessment tools and developing new models of female criminal behavior using integrated theoretical and methodological approaches.

Tracey A. Skilling,PhD,is a psychologist and Associate Scientist at The Center for Addiction and Mental Health and an assistant professor in the Departments of Psychiatry and Applied Psychology and Human Development 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

Bonta

Andrews

D. A.

(2017). The psychology of criminal conduct (6th ed.). Routledge.

Borum

Bartel

Forth

(2006). Manual for the structured assessment of violence risk in youth. University of South Florida.

Borum

Lodewijks

Bartel

P. A.

Forth

A. E.

(2010). Structured Assessment of Violence Risk in Youth (SAVRY). In Douglas

Otto

(Eds.), Handbook of violence risk assessment (pp. 63–80). Routledge.

Campbell

M. A.

French

Gendreau

(2009). The prediction of violence in adult offenders: A meta-analytic comparison of instruments and methods of assessment. Criminal Justice and Behavior, 36, 567–590. https://doi.org/10.1177/0093854809333610

Christiansen

A. K.

Viljoen

J. L.

Fuller

E. K.

(2021). Examination of the Structured Assessment of Protective Factors for Violence Risk—Youth Version (SAPROF-YV) in Canadian adolescents. International Journal of Risk and Recovery, 4(2), 4–17. https://doi.org/10.15173/ijrr.v4i2.4269

Chu

C. M.

Ruby

Chng

G. S.

(2020). The utility of SAPROF-YV ratings for predicting recidivism in male youth under community supervision in Singapore. Criminal Justice and Behavior, 47(11), 1409–1427. https://doi.org/10.1177/0093854820949595

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum.

Costello

A. B.

Osborne

(2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research & Evaluation, 10, 1–9. https://doi.org/10.7275/jyj1-4868

Coupland

R. B. A.

Olver

M. E.

(2020). Assessing protective factors in treated violent offenders: Associations with recidivism reduction and positive community outcomes. Psychological Assessment, 32(5), 493–508. https://doi.org/10.1037/pas0000807

10.

Cullen

F. T.

Gilbert

K. E.

(2012). Reaffirming rehabilitation (2nd ed.). Routledge. https://doi.org/10.4324/9781315721361

11.

de Ruigh

E. L.

Kleeven

A. T. H.

Jansen

L. M. C.

de Vries Robbé

Vermeiren

R. R. J. M.

Mulder

E. A.

de Paauw

van de Ven

P. M.

Popma

. (2021). Predicting youth reoffending after incarceration: Added value of protective factors and heart rate variability. The Journal of Forensic Psychiatry & Psychology, 32(4), 449–480. https://doi.org/10.1080/14789949.2020.1860250

12.

de Ruiter

Nicholls

T. L

. (2011). Protective factors in forensic mental health: A new frontier. The International Journal of Forensic Mental Health, 10(3), 160–170. https://doi.org/10.1080/14999013.2011.600602

13.

de Vogel

de Ruiter

Bouman

de Vries Robbé

. (2009). SAPROF: Guidelines for the assessment of protective factors for violence risk [English Version]. Forum Educatief.

14.

de Vries Robbé

de Vogel

de Spa

. (2011). Protective factors for violence risk in forensic psychiatric patients: A retrospective validation study of the SAPROF. International Journal of Forensic Mental Health, 10, 178–186. https://doi.org/10.1080/14999013.2011.600232

15.

de Vries Robbé

de Vogel

Douglas

K. S

. (2013). Risk factors and protective factors: A two-sided dynamic approach to violence risk assessment. Journal of Forensic Psychiatry & Psychology, 24(4), 440–457. https://doi.org/10.1080/14789949.2013.818162

16.

de Vries Robbé

Geers

Stapel

Hilterman

de Vogel

. (2015). SAPROF youth version: Guidelines for the assessment of protective factors for violence risk in juveniles [English version]. Forum Educatief.

17.

de Vries Robbé

Willis

G. M

. (2017). Assessment of protective factors in clinical practice. Aggression and Violent Behavior, 32, 55–63. https://doi.org/10.1016/j.avb.2016.12.006

18.

Dickens

G. L.

O’Shea

L. E.

(2017). Protective factors in risk assessment schemes for adolescents in mental health and criminal justice populations: A systematic review and meta-analysis of their predictive validity. Adolescent Research Review, 3(1), 95–112. https://doi.org/10.1007/s40894-017-0062-3

19.

Dolan

M. C.

Rennie

C. E.

(2008). The structured assessment of violence risk in youth as a predictor of recidivism in a United Kingdom cohort of adolescent offenders with conduct disorder. Psychological Assessment, 20(1), 35–46. https://doi.org/10.1037/1040-3590.20.1.35

20.

Farrington

D. P.

Ttofi

M. M.

Piquero

A. R.

(2016). Risk, promotive, and protective factors in youth offending: Results from the Cambridge study in delinquent development. Journal of Criminal Justice, 45, 63–70. https://doi.org/10.1016/j.jcrimjus.2016.02.014

21.

Finseth

Peterson-Badali

Brown

S. L.

Skilling

T. A.

(2022). Promoting rehabilitation among youth on probation: An examination of strengths as specific responsivity factors. Criminal Justice and Behavior, 49(5), 745–760. https://doi.org/10.1177/00938548211039881

22.

Gagnier

K. R.

(2013). Toward a strength-based approach to risk assessments: An examination of the measurement and clinical use of information about youth strengths [Unpublished doctoral dissertation, York University].

23.

Gearing

R. E.

Mian

I. A.

Barber

Ickowicz

(2006). A methodology for conducting retrospective chart review research in child and adolescent psychiatry. Journal of the Canadian Academy of Child and Adolescent Psychiatry, 15(3), 126–134.

24.

Gendreau

Little

Goggin

(1996). A meta-analysis of the predictors of adult offender recidivism: What works! Criminology, 34(4), 575–608. https://doi.org/10.1111/j.1745-9125.1996.tb01220.x

25.

Goodwin

Brown

S. L.

Skilling

T. A.

(2022). Gender differences in the prevalence and predictive validity of protective factors in a sample of justice-involved youth. Youth Violence and Juvenile Justice, 20(3), 231–249. https://doi.org/10.1177/15412040221089235

26.

Guy

L. S.

Nelson

R. J.

Fusco-Morin

S. L.

Vincent

G. M.

(2014). What do juvenile probation officers think of using the SAVRY and YLS/CMI for case management, and do they use the instruments properly? International Journal of Forensic Mental Health, 13(3), 227–241. https://doi.org/10.1080/14999013.2014.939789

27.

Hoge

R. D.

Andrews

D. A.

(2002). The Youth Level of Service/Case Management Inventory manual and scoring key. Multi-Health Systems.

28.

Hoge

R. D.

Andrews

D. A.

(2011). Youth Level of Service/Case Management Inventory 2.0 (YLS/CMI 2.0): User’s manual. Multi-Health Systems.

29.

Jones

N. J.

Brown

S. L.

Robinson

Frey

(2016). Validity of the Youth Assessment and Screening Instrument: A juvenile justice tool incorporating risks, needs, and strengths. Law and Human Behavior, 40, 182–194. https://doi.org/10.1037/lhb0000170

30.

Kaiser

H. F.

(1958). The varimax criterion for analytic rotation in factor analysis. Psychometrika, 23(3), 187–200. https://doi.org/10.1007/bf02289233

31.

Kitzmiller

Hoskins

Cavanagh

(2022). Examining sex-based measurement invariance in the Youth Level of Service/Case Management Inventory. Crime & Delinquency. Advance online publication. https://doi.org/10.1177/00111287211073677

32.

Kleeven

A. T. H.

de Vries Robbe

Mulder

E. A.

Popma

(2022). Risk assessment in juvenile and young adult offenders: Predictive validity of the SAVRY and SAPROF-YV. Assessment, 29(2), 181–197. https://doi.org/10.1177/1073191120959740

33.

Koh

L. L.

Day

Klettke

Daffern

Chu

C. M.

(2022). The predictive validity of three youth violence assessment instruments: The SAVRY, VRS-YV, and SAPROF-YV. International Journal of Offender Therapy and Comparative Criminology, 66(2–3), 168–185. https://doi.org/10.1177/0306624X20970887

34.

Chu

C. M.

Zeng

Ruby

(2019). Risk and protective factors for probation success among youth offenders in Singapore. Youth Violence and Juvenile Justice, 17, 194–213. https://doi.org/10.1177/1541204018778887

35.

Lloyd

C. D.

Hanson

R. K.

Richards

D. K.

Serin

R. C.

(2020). Reassessment improves prediction of criminal recidivism: A prospective study of 3,421 individuals in New Zealand. Psychological Assessment, 32(6), 568–581. https://doi.org/10.1037/pas0000813

36.

Lodewijks

H. P.

de Ruiter

Doreleijers

T. A.

(2010). The impact of protective factors in desistance from violent reoffending. Journal of Interpersonal Violence, 25(3), 568–587. https://doi.org/10.1177/0886260509334403

37.

Lovatt

K. M.

Stockdale

K. C.

Olver

M. E.

(2022). Dynamic violence risk, protective factors, and therapeutic change in a gender and ethnoculturally diverse sample of court-adjudicated youth. Psychological Assessment, 34(2), 105–124. https://doi.org/10.1037/pas0001071

38.

McAra

McVie

(2016). Understanding youth violence: The mediating effects of gender, poverty and vulnerability. Journal of Criminal Justice, 45, 71–77. https://doi.org/10.1016/j.jcrimjus.2016.02.011

39.

Nunnally

J. C.

(1978). Psychometric theory. McGraw-Hill.

40.

Olver

M. E.

Stockdale

K. C.

Wormith

J. S.

(2009). Risk assessment with young offenders: A meta-analysis of three assessment measures. Criminal Justice and Behavior, 36, 329–353. https://doi.org/10.1177/0093854809331457

41.

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, 156–176. https://doi.org/10.1037/a0035080

42.

Orbis Partners. (2000). Youth Assessment and Screening Instrument (YASI).

43.

Peterson-Badali

Skilling

Haqanee

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

44.

Pett

M. A.

Lackey

N. R.

Sullivan

J. J.

(2003). Making sense of factor analysis: The use of factor analysis for instrument development in health care research. SAGE.

45.

Rice

Harris

(2005). Comparing effect sizes in follow-up studies: ROC area, Cohen’s. d, and r. Law and Human Behavior, 29(5), 615–620. https://doi.org/10.1007/s10979-005-6832-7

46.

Rowe

D. C.

(2002). Biology and crime. Roxbury.

47.

Schmidt

Campbell

M. A.

Houlding

(2011). Comparative analyses of the YLS/CMI, SAVRY, and PCL:YV in adolescent offenders: A 10-year follow-up into adulthood. Youth Violence and Juvenile Justice, 9, 23–42. https://doi.org/10.1177/1541204010371793

48.

Schmidt

Hoge

R. D.

Gomes

(2005). Reliability and validity analyses of the Youth Level of Service/Case Management Inventory. Criminal Justice and Behavior, 32(3), 329–344. https://doi.org/10.1177/0093854804274373

49.

Sellbom

Tellegen

(2019). Factor analysis in psychological assessment research: Common pitfalls and recommendations. Psychological Assessment, 31(12), 1428–1441. https://doi.org/10.1037/pas0000623

50.

Shepherd

S. M.

Luebbers

Ogloff

J. R.

(2016). The role of protective factors and the relationship with recidivism for high-risk young people in detention. Criminal Justice and Behavior, 43(7), 863–878. https://doi.org/10.1177/0093854815626489

51.

Singh

J. P.

Desmarais

S. L.

Hurducas

Arbach-Lucioni

Condemarin

Dean

Doyle

Folino

Godoy-Cervera

Grann

Mei Yee Ho

Large

Nielsen

L. H.

Pham

Rebocho

R. M.

Reeves

Rettenberger

de Ruiter

Seewald

Otto

R. K.

(2014). International perspectives on the practical application of violence risk assessment: A global survey of 44 countries. International Journal of Forensic Mental Health, 13, 193–206. https://doi.org/10.1080/14999013.2014.922141

52.

Singh

J. P.

Desmarais

S. L.

Sellers

B. G.

Hylton

Tirotti

Van Dorn

R. A.

(2014). From risk assessment to risk management: Matching interventions to adolescent offenders’ strengths and vulnerabilities. Child and Youth Services Review, 47, 1–9. https://doi.org/10.1016/j.childyouth.2013.09.015

53.

Singh

J. P.

Grann

Fazel

(2011). A comparative study of violence risk assessment tools: A systematic review and metaregression analysis of 68 studies involving 25,980 participants. Clinical Psychology Review, 31(3), 499–513. https://doi.org/10.1016/j.cpr.2010.11.009

54.

StataCorp. (2017). Stata statistical software: Release 15.

55.

Stouthamer-Loeber

Loeber

Wei

Farrington

D. P.

Wikström

P.-O. H.

(2002). Risk and promotive effects in the explanation of persistent serious delinquency in boys. Journal of Consulting and Clinical Psychology, 70(1), 111–123. https://doi.org/10.1037/0022-006X.70.1.111

56.

Tabachnick

B. G.

Fidell

L. S.

(2007). Using multivariate statistics (5th ed.). Allyn and Bacon.

57.

Viljoen

J. L.

Bhanwer

A. K.

Shaffer

C. S.

Douglas

K. S.

(2020). Assessing protective factors for adolescent offending: A conceptually-informed examination of the SAVRY and YLS/CMI. Assessment, 27(5), 959–975. https://doi.org/10.1177/1073191118768435

58.

Viljoen

J. L.

Nicholls

T. L.

Cruise

K. R.

Desmarais

S. L.

Webster

C. D.

, & with contributions by Douglas-Beneteau

. (2014). Short—Term Assessment of Risk and Treatability: Adolescent Version (START: AV)—User guide. Mental Health, Law, and Policy Institute.

59.

Vincent

G. M.

Paiva-Salisbury

M. L.

Cook

N. E.

Guy

L. S.

Perrault

R. T.

(2012). Impact of risk/needs assessment on juvenile probation officers’ decision making: Importance of implementation. Psychology, Public Policy, and Law, 18(4), 549–576. https://doi.org/10.1037/a0027186

60.

Vitopoulos

N. A.

Peterson-Badali

Brown

Skilling

(2019). The relationship between trauma, recidivism risk, and reoffending in male and female juvenile offenders. Journal of Child & Adolescent Trauma, 12(3), 351–364. https://doi.org/10.1007/s40653-018-0238-4

61.

Vitopoulos

N. A.

Peterson-Badali

Skilling

T. A.

(2012). The relationship between matching service to criminogenic need and recidivism in male and female youth: Examining the RNR principles in practice. Criminal Justice and Behavior, 39, 1025–1041. https://doi.org/10.1177%2F0093854812442895

62.

Wanamaker

K. A.

Jones

N. J.

Brown

S. L.

(2018). Strengths-based assessments for use with forensic populations: A critical review. The International Journal of Forensic Mental Health, 17(2), 202–221. https://doi.org/10.1080/14999013.2018.1451414

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB