The underlying assumption of the professional override when applied to offender risk assessment is simple. After completing a structured, statistically based or actuarial assessment on a specific individual, the professional who is conducting the assessment may have what is believed to be additional, relevant information that was not taken into consideration when conducting the structured assessment. This occurs when the information is not relevant to the scoring of any of the instrument’s items. The question then is, can or should the professional make some kind of adjustment to the statistically derived results of the assessment? This is known as the professional override. This article describes the background behind the professional override, its position in the history of offender risk assessment development, and the elusive quest to demonstrate its capacity to improve upon a structured, statistically based offender risk assessment.
Applying a professional override to a statistical prediction scheme represents one attempt to reconcile a long-standing dispute between statistical/ actuarial and clinical/professional judgment. The competition was kicked off in earnest with the classic work of Paul Meehl in 1954. However, since 1958, psychologists and criminologists, such as Robert Holt and Donald Gottfredson, have been advocating for clinically and statistically oriented researchers and clinicians to abandon their competitive attitudes and work more collaboratively.
In 1990, Donald Andrews, James Bonta, and Robert Hoge asserted that there were four key principles that contribute to the delivery of effective correctional treatment for offenders. Three of them became known as the Risk-Need-Responsivity model of offender intervention. The fourth principle, professional override, relates to the responsibility of professionals to go beyond the calculated application of Risk-Need- Responsivity when making decisions as to how to treat an offender in his or her particular circumstances. As such, it introduces an element of clinical judgment, also known as professional discretion, in the decision-making process. Perhaps the most common application of the override principle is in the context of a structured, statistically based offender risk assessment.
Because it is essentially a clinical judgment, the override bears some resemblance to, but is not quite the same as, the structured professional judgment approach to offender risk assessment. While use of the override is considered an add-on to an existing assessment, structured professional judgment assessments are based on stand-alone assessment instruments, and while structured professional judgment assessments are standardized in the sense that they comprise specific risk items (often 20), the override knows no bounds in terms of content, or risk factors, that it may consider. Yet they are similar in that both are fundamentally subjective in their approach to decision-making about offender risk.
Use of the Professional Override in Offender Risk Assessment
There is well-documented evidence, summarized in meta-analyses by William Grove and colleagues, that statistically based predictions are superior to clinical judgment in fields as diverse as education, health, financial success, and criminality. Yet there remains a sense that criminal justice professionals may contribute their own insights to the prediction of offender behavior, at least as an adjunct to statistically based predictions. Every offender is unique in many ways, and clinical professionals are experienced with working on a one-to-one basis with offenders’ uniqueness. Hence, it seems natural to give professionals an opportunity to contribute their own particular insights to the understanding of an offender and the prediction of his or her future behavior at the idiographic or individual level. It is assumed, either correctly or incorrectly, that override decisions are based on sound theory and research on offending behavior. Andrews and colleagues also believed that research on the consequences of overrides would lead to new insights into offending behavior and its amelioration.
Two of the most popular offender risk assessment instruments in North America allow for the professional override. Versions of the Level of Service Inventory have incorporated the professional override in their administration protocol. Although the Level of Service Inventory–Revised manual provides cutoff scores to denote various levels of risk, it declares that these cutoff scores should be used only as guidelines in conjunction with “good sound judgment by experienced professionals” (p. 14). The more recent Level of Service/Case Management Inventory (LS/CMI) and its youth version both formalize the override function by building it into the assessment protocol. Assessors are explicitly instructed to indicate whether or not they wish to exercise the override option and if so, why, in their professional opinion, its use is warranted. They may then assign a higher or lower risk level to the offender than what was derived from the numerical score on the assessment.
Another popular instrument, the Correctional Offender Management Profiling for Alternative Sanctions, also allows the criminal justice professional to make a final judgment regarding an offender’s risk level. Correctional Offender Management Profiling for Alternative Sanctions software includes a Screener Judgment tab, on which several questions are asked that allow assessors to introduce their own overall conclusions about an offender’s risk. They are asked to consider aggravating factors (e.g., resistant to treatment), which may render an offender higher risk than what was determined by the statistical analysis, and mitigating factors (e.g., being crime free for many years), which may render an offender lower risk. Correctional Offender Management Profiling for Alternative Sanctions authors expect that assessors will override offenders to a different risk level than what was assigned statistically in 8–15% of assessed cases.
Some sexual offender risk assessment instruments also include override features. The Minnesota Sex Offender Screening Tool–Revised provides eight guidelines for assessors to increase sexual offenders’ risk level and one guideline to reduce it. The Static-99 initially gave clinicians considerable discretion to use the override but has since tightened that position in accordance with research described in the following section.
It should also be noted that some instruments, including the LS/CMI and the U.S. Federal Post Conviction Risk Assessment, allow for an administrative or policy override. These are cases when the override feature is exercised, not for clinician or professional reasons, but for policy reasons determined by the correctional agency. It may be applied to sexual offenders, mentally disordered offenders, and those who, for various reasons, are high-profile offenders.
Research on Use of the Professional Override in Offender Risk Assessment
As a precursor to the professional override in offender risk assessment, some early efforts were made to amalgamate actuarial risk assessment with human judgment. One study examined whether some combination of the Correctional Service of Canada’s risk assessment tool, the Statistical Information on Recidivism Scale, might dovetail with decisions by the National Parole Board to arrive at a better overall prediction of offender recidivism. By combining independent Parole Board decisions with offenders’ Statistical Information on Recidivism scores, a modest improvement in the prediction of offenders’ recidivism was found over predictions based on the Statistical Information on Recidivism score alone. However, most of the subsequent override research has been less encouraging. This research falls into two categories, studies that have examined the override with general risk scales on general offender populations and studies that have examined its use with sexual offender instruments on samples of sexual offenders.
When correctional workers were first given the opportunity to apply the professional override to a group of probationers and provincial prisoners using a draft version of the LS/CMI (Level of Service Inventory: Ontario Revision), the professional override was used in only 3% of the cases, with increases and decreases in risk being approximately equal. After the override option was given, the final risk level correlated slightly higher with recidivism than the initial risk level. When the LS/CMI was implemented on a systemwide basis in Ontario, subsequent studies documented much greater use of the override (12–17%), primarily to increase risk and disproportionately on sexual offenders, and found it had a negative impact on predictive validity.
Research on the use of the override with the STATIC-99 has found that assessors could not agree on which cases to use the override, and in which direction to do so, beyond a chance level of agreement. Moreover, its use decreased the predictive validity of the instrument. This led STATIC-99 authors to suggest using the override only in conjunction with other statistically based instruments. Nonetheless, field studies have found the override has been applied to the STATIC-99 in as much as one third of the assessments and often without the support of another statistical tool. Not surprisingly, such practice again reduced the predictive validity of the instrument.
Not to be confused with the professional override, research after the initial publication of the STATIC-99 led its authors to provide directions to modify the risk assessment of sexual offenders based on factors such as age. However, these adjusted actuarial estimates of risk are strictly statistical modifications of assessed risk and therefore not professional overrides. They have also generated some degree of controversy.
Research on the Minnesota Sex Offender Screening Tool–Revised has found a surprisingly high use of the override (from three quarters to almost all cases). The effect of the override was to move offenders, both high and low risk, toward the median risk group. The impact of this practice was to reduce the predictive validity of the instrument slightly.
Future of the Professional Override in Offender Risk Assessment
Empirical evidence for the incremental predictive validity of the professional override remains elusive. It is quite possible that part of the difficulty comes from the very nature of professional discretion. Specifically, its variability and complexity defies easy quantification and systematic investigation. While some have come to deny any role for professional discretion in offender risk assessment, others continue to advocate for its controlled use as part of a long-standing effort to improve the prediction of offending behavior. Most would agree that any effort to apply the override to field settings would require ongoing monitoring and research to direct and guide such a practice.
References:
- Andrews, D. A., & Bonta, J. (1995). The Level of Service Inventory-Revised: User’s manual. Toronto, Canada: Multi-Health Systems.
- Andrews, D. A., Bonta, J., & Wormith, J. S. (2004). The Level of Service/Case Management Inventory (LS/ CMI). Toronto, Canada: Multi-Health Systems.
- Girard, L., & Wormith, J. S. (2004). The predictive validity of the Level of Service Inventory-Ontario revision on general and violent recidivism among various offender groups. Criminal Justice and Behavior, 31, 150–181. doi:10.1037/a0035080
- Grove, W. M., Zald, D. H., Lebow, B. S., Snitz, B. E., & Nelson, C. (2000). Clinical versus mechanical prediction: A meta-analysis. Psychological Assessment, 12, 19–30. doi:10.1037//1040-3590.12.1.19
- Meehl, P. E. (1954). Clinical versus statistical predictions: A theoretical analysis and a review of the evidence. Minneapolis: University of Minnesota Press.
- Wormith, J. S., & Goldstone, C. S. (1984). The clinical and statistical prediction of recidivism. Criminal Justice and Behavior, 11, 3–34. Retrieved from https://doi.org/10.1177/0093854884011001001
- Wormith, J. S., Hogg, S., & Guzzo, L. (2012). The predictive validity of a general risk/need assessment inventory on sexual offender recidivism and an exploration of the professional override. Criminal Justice and Behavior, 39, 1511–1538. doi:10.1177/ 0093854812455741