HD59 - Development of a Model Risk Assessment for Juvenile Felony Offenders


Executive Summary:
The Legislative Mandate

The 1996 Session of the Virginia General Assembly passed a joint resolution (HJR 70) that directed the Department of Juvenile Justice (DJJ) to develop a risk assessment instrument to assist judges in determining dispositions for adjudicated felony offenders. This report describes the process and results of DJJ's effort to meet that mandate.

As is the case with many other states, Virginia is facing increasing numbers of juvenile offenders - many of them serious, violent or chronic offenders - without concomitant increases in the resources needed to intervene effectively with them. One result has been limited bed space in Virginia's juvenile correctional institutions. The problem of escalating caseloads and restricted resources means that juvenile justice agencies must find ways to ensure that all available resources for delinquent youth are used in the most efficient manner. In many states, this has meant devising structured assessment and classification strategies to ensure that certain types of resources are used only for selected, and clearly defined, offender sub-populations. As the Virginia legislature has recognized, the use of risk assessment to inform dispositions can be a powerful tool for achieving this goal.

The HJR 70 Study Group

To carry out its legislative mandate, the Department of Juvenile Justice:

• convened a group of juvenile justice system officials to provide leadership, input and oversight to the risk instrument development process. This HJR 70 Study Group consisted of judges, representatives of the Virginia Sentencing Commission and the Department of Criminal Justice Services, as well as DJJ administrative, research and field staff;

• contracted with the National Council on Crime and Delinquency to provide expertise in the development of the risk assessment instrument;

• facilitated a series of meetings between the study group and NCCD; and,

• approved the attached NCCD-developed risk instrument for submission to the legislature.

Risk Assessment: Purpose and Definition

The ultimate purpose of formal risk assessment is to enhance public safety by providing decision makers with the ability to more accurately and consistently identify those adjudicated juvenile offenders who are most likely to commit subsequent offenses. Armed with this information, judges and other decision makers can then more accurately link offenders having varying degrees of risk with the most appropriate level of security or supervision. This process also promotes more efficient use of limited - and often costly - resources.

The central characteristics of formal risk assessment include:

• the use of a scale that incorporates a limited number of factors known or believed to be predictive of reoffending. Each of the factors is given a "weight" (or points). The offender's total score determines the level of risk. The risk classification leads to a custody or supervision decision based on selected rules (e.g., all juveniles scoring 15 points or higher will be placed under intensive supervision).

• research-based risk scales are derived from an analysis of the statistical relationship between youth characteristics and recidivism rates. This is an actuarial approach that is similar to that used to determine automobile insurance rates.

• the basic risk strategy is to identify sub-groups of the delinquent population that have very different recidivism rates. For example in Rhode Island, NCCD developed a risk scale in which those youth identified as livery high" risk were found to have a recidivism rate that was four times higher (82% vs. 21%) than the youth identified by the tool as "low" risk (Wagner and Wiebush 1996).

• typical risk factors identified through research include historical items such as number of prior referrals and age at first adjudication I as well as "stability" factors such as substance abuse, delinquent peers, school problems, and family dysfunction. It is important to note that most risk research has shown that the seriousness of the current offense is not predictive of (and is often inversely related to) rates of subsequent offending (Clear 1988).

DEVELOPMENT OF THE PROPOSED VIRGINIA RISK INSTRUMENT

Key Decisions

The HJR 70 study group made several important decisions that heavily influenced the risk assessment development process. The group decided that:

• the term :risk" was intended to mean risk of re-offending. This meant that the intended purpose of the Virginia scale was to classify adjudicated youth based on their likelihood of future recidivism, rather than on the seriousness of offenses already committed;

• a research-based model was preferable to one developed through a consensus-building approach (in which group members come to agreement about what factors they believe are important in assessing risk);

• there was insufficient time to conduct the research necessary to develop and validate a risk tool using the characteristics and outcomes of Virginia delinquents; (*1)

• it did not want to simply adopt an empirically-based scale developed in another state.

As a result of these several decisions, the HJR 70 study group opted to pursue the development of a "model" empirically-based scale, using data from risk research previously completed by NCCD in other sites. (*2)
_________________________________
(*1) In developing a validated risk scale, it is necessary to analyze the characteristics and outcomes of the youth in the jurisdiction for which the scale is being developed. While there are certain predictor variables that apply in most jurisdictions, previous research has also shown that each jurisdiction has some unique variables that serve to increase the predictive and classification power of the risk tool. Using the local population insures a -best fit- between the scale and the youth it is used to assess.

(*2) Over the past five years, NCCD has developed research-based juvenile risk scales in approximately 30 different jurisdictions including the states of MI, WI, NE, RI, NJ, AZ. as well as in Cuyahoga, Lucas and Travis counties.