Use of Valid Actuarial Assessments of Risk and Needs

Understanding the Evidence and Its Implications for
Risk-Informed Release Decision-Making

With the emergence of evidence-based practices in corrections and parole, the Risk/Need/ Responsivity (R/N/R) construct has become a common part of the vernacular (Andrews and Bonta, 2010). Risk is the most discussed and widely used aspect of the construct, and risk assessment is by far the component most commonly applied in practice. Despite the widespread use of the construct and the process, there remains a deficit in terms of policymaker and practitioner understanding of risk and how it is best utilized in correctional and parole decisionmaking.

Risk assessment using actuarial-based instruments is an evidence-based practice for gauging the relative likelihood of a given offender recidivating. Prudent decision-makers are right to ask questions. What are the implications of using risk assessment? How is the likelihood of recidivism determined? How accurate is it? How reliable is it? How do we best use it in release decision-making?

Risk assessment is not individual prediction. It is the development of probabilities for groups of offenders based on characteristics which are related to the behavior of interest, in this case recidivism. These probability statements are developed for groups of individuals with certain profiles of characteristics. As a result, the risk assessment cannot identify which individual offenders in a group will recidivate, but rather only approximately how many within that group, expressed in rates or percentages, are likely to recidivate. While this falls short of what decision-makers would like – reliable individual predictions – these probabilities are extremely helpful for decisionmakers who handle dozens, hundred, or thousands of cases over the course of a year.  The information is invaluable in determining how to manage offenders—both in determining who should be required to complete programming prior to parole release—and who should NOT have such requirements; and also in determining the timing of release for individuals eligible for parole. 

What does "low risk" mean? In this context, it means that an offender is part of a group of offenders that has a low probability of reoffending, for example 10%. With this information, decision-makers know that nine of ten offenders in this group will not recidivate.

As this discussion indicates, low risk does not mean there is no risk at all.  Using actuarial instruments will result in decision errors – assuming no recidivism and then the offender commits a new crime. The frequency of these errors can be reduced by collecting, analyzing and applying other information in the decision process. This offender-specific information can help to more fully understand the offender, his/her background, behaviors and ultimately the risk of reoffending. By their nature, actuarial risk assessments are limited in the number of items of information, or variables, they contain. The variables in the risk assessment instrument are those which explain the greatest amount of recidivism for the largest number of offenders. These instruments do not and cannot include every relevant piece of information about every, or even one, offender. Information that provides greater detail concerning an individual offender can be found in that offender's criminal history, institutional conduct, or social history reports, and of course this type of information can be of tremendous value to decision makers.    

Considering both the objective risk assessment information and the offender-specific information that is available is critical to formulating an appropriate decision.  Factors not incorporated in the risk assessment instrument itself, but providing further insight into risk, should be carefully considered by the decision-makers. If, in their professional judgment, the impact of this additional information is such that it outweighs the risk assessment, then the decision maker may well arrive at a different judgment regarding risk.  In fact, the research demonstrates that actuarial instruments, when supplemented by professional judgment, produce even higher levels of decision accuracy than the instrument alone (Dawes, Faust, and Meehl 1989). Andrews and Bonta incorporated this concept termed professional discretion as one of their principles of effective correctional treatment (Andrews and Bonta 2010, pp. 47, 52).

Of course, as parole Board members make dozens, hundreds, or thousands of decisions, risk assessment instruments will provide valuable support—and research demonstrates that valid and reliable assessment instruments out-perform unstructured judgment in terms of their accuracy of prediction for groups of offenders.  So it would be expected that, in the majority of cases, the objective risk assessment will prove valuable and will be used to guide parole Board members decisions.


< Background and Context

Continued >