The Utility of SAPROF-YV Ratings for Predicting Recidivism in Male Youth Under Community Supervision in Singapore

• Published date: 01 November 2020
• Date: 01 November 2020
• DOI: 10.1177/0093854820949595
• Subject Matter: Articles

CRIMINAL JUSTICE AND BEHAVIOR, 2020, Vol. 47, No. 11, November 2020, 1409 –1427.

DOI: https://doi.org/10.1177/0093854820949595

Article reuse guidelines: sagepub.com/journals-permissions

© 2020 International Association for Correctional and Forensic Psychology

1409

THE UTILITY OF SAPROF-YV RATINGS FOR

PREDICTING RECIDIVISM IN MALE YOUTH

UNDER COMMUNITY SUPERVISION IN

SINGAPORE

CHI MENG CHU

National Council of Social Service

Ministry of Social and Family Development

XUEXIN XU

DONGDONG LI

National Council of Social Service

KALA RUBY

Ministry of Social and Family Development

GRACE S. CHNG

National Council of Social Service

There is bourgeoning empirical support for the usage of the Structured Assessment of Protective Factors (SAPROF) across

many jurisdictions, but there is a dearth of research on the Structured Assessment of Protective Factors for Violence Risk—

Youth Version (SAPROF-YV). This study examined (a) the predictive validity of the SAPROF-YV ratings for general

recidivism and (b) the incremental predictive validity of the SAPROF-YV ratings when used in conjunction with the Youth

Level of Service/Case Management Inventory (YLS/CMI) 2.0 ratings. Using a sample of 822 male youths who were involved

with the justice system and under community supervision in Singapore, the results showed that the SAPROF-YV total score

and final protection judgment rating were significantly predictive of general recidivism. Moreover, the SAPROF-YV total

score and final judgment rating showed incremental predictive validity over the YLS/CMI 2.0 total score and risk rating.

Overall, the results suggest that SAPROF-YV ratings are suited for assessing justice-involved youth within the Singaporean

context and can be used in conjunction with YLS/CMI 2.0 ratings for predicting recidivism.

Keywords: SAPROF; YLS/CMI 2.0; protective factors; recidivism; risk assessment

INTRODUCTION

Youth offending and desistance are criminological phenomena that have received a lot of

attention from scholars and policy makers. In the existing literature of forensic risk assess-

ment, a risk factor refers to a variable associated with increased likelihood of a negative

AUTHORS’ NOTE: The authors thank the staff of Probation and Community Rehabilitation Service for

their support of this study. The views expressed are those of the authors’ and do not represent the official

position or policies of the National Council of Social Service and the Ministry of Social and Family

Development. Correspondence concerning this article should be addressed to Chi Meng Chu, Translational

Social Research Division, National Council of Social Service, 170 Ghim Moh Road #01-02, Singapore

279621; e-mail: chu_chi_meng@ncss.gov.sg.

949595CJBXXX10.1177/0093854820949595Criminal Justice and BehaviorChu et al.

research-article2020

1410 CRIMINAL JUSTICE AND BEHAVIOR

outcome (Farrington et al., 2016). Numerous studies have documented the risk factors pre-

dicting recidivism in justice-involved youth (see e.g., Cottle et al., 2001 for a review). In

particular, the Risk Needs Responsivity (RNR) framework is well established for the risk

assessment and treatment of justice-involved youth and is widely adopted by youth reha-

bilitation services around the globe (Bonta & Andrews, 2017). The practice of assessing the

risk of (re)offending has been a predominantly deficits-based approach until recently, and

this typically involves using actuarial and/or structured professional judgment measures to

assess empirically derived risk factors to determine an overall risk classification (de Vries

Robbé & Willis, 2017).

More recently, scholars have examined protective factors that are related to why and how

people desist from (or stop) offending (e.g., Maruna, 2001), as well as what increases the

likelihood of prosocial functioning and a meaningful life. In general, protective factors refer

to all personal, social, and environmental factors that reduce the risk of future offending

behavior (de Vries Robbé, Geers, et al., 2015). A protective factor could be opposite to risk

factors (McAra & McVie, 2016; White et al., 1989), or a distinct and stand-alone factor

separated from the presence of risk factors (Borum et al., 2006). Nonetheless, research on

the utility of protective factors in assessing justice-involved youth and predicting recidi-

vism is limited as compared with the extensive extant research on risk assessment.

ASSESSING RISK OF REOFFENDING IN JUSTICE-INVOLVED YOUTH

Research studies have shown that structured risk assessment methods (i.e., actuarial and

structured clinical judgment approaches) are not only more accurate than unstructured clini-

cal judgment, but they are also preferred because of their increased transparency and reli-

ability (see Heilbrun et al., 2010, for a review). Notably, many structured risk assessment

measures perform well in identifying individuals with higher risk of violence and other

forms of offending and are deemed to be useful for informing treatment as well as manage-

ment decisions (Fazel et al., 2012).

Although there are many youth risk assessment measures, one of the most commonly

used for assessing justice-involved youth is the Youth Level of Service/Case Management

Inventory (YLS/CMI 2.0; Hoge & Andrews, 2002, 2011). Importantly, the YLS/CMI 2.0

has been extensively examined in studies of youth offending and recidivism, in both west-

ern (e.g., McGrath & Thompson, 2012; Onifade et al., 2008; Perrault et al., 2017; Rennie &

Dolan, 2010; Schmidt et al., 2011) and Asian contexts (e.g., Chu et al., 2014, 2015, 2016;

Takahashi et al., 2013). In a meta-analytic review of youth risk assessment measures,

Schwalbe (2007) found that the weighted area under curve (AUC) of YLS/CMI 2.0 ratings

for predicting juvenile recidivism was .64. As a general rule for practice, AUCs greater than

.54, .63, and .71, as well as correlation coefficients that are greater than .10, .24, and .37, are

regarded as small, moderate, and large effects, respectively (Rice & Harris, 2005).

Another meta-analysis by Olver et al. (2009) also suggested moderate to large effect

sizes of YLS/CMI 2.0 ratings in predicting general, nonviolent, and violent recidivism

(mean-weighted correlation coefficient = .32, .29, and .26, respectively). Further meta-

analyses by Schwalbe (2008), Olver et al. (2009), as well as Pusch and Holtfreter (2018)

showed that the YLS/CMI 2.0 was useful for predicting general recidivism in male (r

values = .32, .33, and .28, respectively) and female (r values = .40, .36, and .25, respec-

tively) youth. In the meta-analyses by Olver et al. (2009) and Pusch and Holtfreter (2018),