Classifying Juvenile Offenders According to Risk of Recidivism

• Author: Steven H. Day,Craig S. Schwalbe,Mark W. Fraser,Valerie Cooley
• Date: 01 June 2006
• Published date: 01 June 2006
• DOI: 10.1177/0093854806286451
• Subject Matter: Articles

CLASSIFYING JUVENILE

OFFENDERS ACCORDING

TO RISK OF RECIDIVISM

Predictive Validity,

Race/Ethnicity, and Gender

CRAIG S. SCHWALBE

Columbia University

MARK W. FRASER

STEVEN H. DAY

VALERIE COOLEY

University of North Carolina at Chapel Hill

Structured risk assessment instruments are increasingly used in juvenile justice systems to

support judicial decision making. They help juvenile justice authorities identify youths with a

higher likelihood of repeat delinquency and reduce discretion associated with disposition deci-

sion making. To be effective, these instruments should be accurate across diverse populations.

This study describes the predictive validity of the North Carolina Assessment of Risk (NCAR)

in a sample of 9,534 adjudicated juveniles in North Carolina. Results show the predictive

validity of the NCAR to differ by gender and race/ethnicity. Closer inspection reveals that risk

factors for recidivism differed according to demographic group and that brief risk assessment

instruments such as the NCAR leave other risk factors unmeasured. The results support the

utility of risk assessment for juvenile justice decision making and suggest strategies to

improve the validity of risk assessment for all offender groups.

Keywords: risk assessment; juvenile justice; juvenile court; delinquency; structured decision

making

305

CRIMINAL JUSTICE AND BEHAVIOR, Vol. 33 No. 3,June 2006 305-324

DOI: 10.1177/0093854806286451

© 2006 American Association for Correctional and Forensic Psychology

AUTHORS’NOTE: The authors would like to thank Catherine Dudley, Donn Hargrove, the

staff of the North Carolina Department of Juvenile Justice and Delinquency Prevention, and

the Governor’s Crime Commission for their assistance in completing this project. Correspon-

dence concerning this article should be addressed to Craig S. Schwalbe, Columbia University

School of Social Work, 1255 Amsterdam Avenue, New York, NY 10027; e-mail: css2109@

columbia.edu.

CJB286451.qxd 4/5/2006 4:52 PM Page 305

Structured risk assessment instruments have been adopted widely

by juvenile justice systems across the United States and Canada

(Hoge, 2002). These instruments are intended to reduce, in part, indi-

vidual discretionary biases that are often associated with court

decision making and case dispositions. Studies have shown that deci-

sions regarding juveniles with comparable instant offenses and

prior records often vary by factors such as race/ethnicity and gender

(Chesney-Lind & Shelden, 1998; Pope & Feyerherm, 1995). Risk

instruments are designed to reduce racial, ethnic, and gender dispari-

ties and biases by increasing the consistency of assessment through a

structured process. In practice, the result should be an increase in the

reliability or consistency of case decisions made by juvenile justice

officials responsible for intake, disposition, release, and other deci-

sions (Baird, Wagner, Healy, & Johnson, 1999; Schwalbe, Fraser,

Day, & Arnold, 2004). Thus, coupled with needs assessment, sen-

tencing guidelines, and other reforms, risk assessment is an important

element of a larger strategy to reduce both racial and gender dispari-

ties in the treatment of offenders by the juvenile justice system.

Although the salutatory effects of increasing consistency are well

noted, the goal of promoting racial, ethnic, and gender equity (i.e.,

reducing disparities based on race/ethnicity and gender) may remain

unfulfilled if the predictive validity of risk assessment instruments

differ by gender or race/ethnicity. Substantial differences in predic-

tive validity across demographic groups may inadvertently introduce

systematic biases rather than neutralize them. Indeed, risk assess-

ment instruments with differential validity by race/ethnicity or gender

could do more harm than good if the authority of standardized mea-

surement seemed to verify existing biases. To estimate the extent of

this problem, this study investigated the predictive validity of one

brief risk assessment instrument, the North Carolina Assessment of

Risk (NCAR).

RISK ASSESSMENT IN JUVENILE JUSTICE

The purpose of risk assessment in juvenile justice is to predict future

offending. Risk assessment instruments do this by measuring diverse

risk factors for recidivism. Two extensive empirical reviews inform on

this effort. Lipsey and Derzon (1998) reviewed 34 longitudinal studies

306 CRIMINAL JUSTICE AND BEHAVIOR

CJB286451.qxd 4/5/2006 4:52 PM Page 306