The (Twice) Failure of the Wisconsin Risk Need Assessment in a Sample of Probationers

The Wisconsin Risk Need Assessment (WRNA) has been widely used by probation and parole agencies across the United States since its development in the 1970s. Despite its extensive adoption, this study examines its effectiveness in predicting recidivism among probationers in Texas. Previous research has shown the WRNA’s limitations, leading to a reweighted scoring system introduced in 2009. This study evaluates both the original and reweighted WRNA’s predictive accuracy using a sample of Texas probationers.

BACKGROUND

The WRNA was designed to classify offenders based on their risk of failing probation, using both static and dynamic factors. Despite its widespread use, research has consistently shown that the WRNA’s predictive utility is limited. The 2009 reweighting aimed to improve the WRNA’s accuracy, but little research has been conducted outside Wisconsin to validate these changes.

KEY FINDINGS

  1. Ineffectiveness: The Wisconsin Risk Need Assessment was ineffective at predicting recidivism among Texas male probationers. Both the original and reweighted versions showed minimal improvement and only slightly outperformed chance predictions.
  2. Misclassification: The original scale often overclassified offenders as high-risk, and reweighting reduced overclassification but increased undersupervision risks, missing services for those in need.
  3. Minimal Gains: Reweighting the risk items provided only marginal improvement in accuracy, with AUC values close to chance levels, demonstrating limited utility.
  4. Needs Assessment Issues: The needs section did not significantly predict recidivism, with many items not correlating well with reoffending.
  5. Replacement Suggested: Due to poor classification and predictive performance, the study recommends replacing the Wisconsin Risk Need Assessment with more validated tools like LSI-R or IORNS.

RECOMMENDATIONS

– Discontinue the use of the Wisconsin Risk Need Assessment due to its poor predictive performance, even after reweighting.

– Adopt more empirically validated risk/need assessment instruments such as:

– Level of Service Inventory-Revised (LSI-R): Known for strong reliability and validity in assessing offender risk and needs.

– Inventory of Offender Risk, Needs, and Strengths (IORNS): Offers a comprehensive evaluation of factors related to recidivism, including criminal orientation, substance abuse, and personal resources.

– Ensure that selected instruments are appropriately validated for the specific jurisdiction and population to enhance predictive accuracy.

– Regularly evaluate and update assessment tools to maintain their effectiveness in predicting probation outcomes and guiding resource allocation.

CONCLUSION

The findings of this study are consistent with previous research that questions the utility of the Wisconsin Risk Need Assessment. Despite attempts to improve its predictive accuracy through reweighting, the WRNA continues to fall short in effectively classifying offenders and predicting recidivism. As such, it is recommended that probation and parole agencies discontinue its use in favor of more reliable and validated assessment tools like the LSI-R and IORNS, which offer stronger support for evidence-based offender management and resource allocation.

Sample Characteristics

The study analyzed data from 194 male probationers in Texas, predominantly consisting of White (53%) and Black (35%) individuals. The majority (81%) were serving probation for misdemeanor offenses.

Predictive Validity

The original WRNA explained only 8% of the variance in recidivism, while the reweighted version accounted for 17.4%. Both versions left a significant portion of recidivism unexplained, indicating poor predictive performance.

Recidivism Rates

Forty-four percent of the sample was rearrested within five years. The reweighted WRNA reduced the high-risk category by 76%, but the classification changes did not significantly improve overall predictive accuracy.

Error Rates

The reweighted WRNA reduced the number of false positives (overclassification errors) by 95% but increased false negatives (underclassification errors) by 83%. This raises concerns about the potential undersupervision of high-risk offenders.

Comparison with Other Tools

The study suggests that other instruments like the Level of Service Inventory-Revised (LSI-R) and the Inventory of Offender Risk, Needs, and Strengths (IORNS) have demonstrated better predictive validity and should be considered as alternatives.