Alaska localized LSI-R violence and recidivism assessment: Development and validation of localized general and violent reconviction scores for the Alaska Level of Service Inventory-Revised (Doc. #2026-11-1001)

Abstract

This report examines the predictive performance of Alaska’s Level of Service Inventory-Revised (LSI-R) and the development of locally weighted risk assessment scores for general and violent reconviction. Conducted as part of the Bureau of Justice Assistance-funded Adult Treatment Court Training Initiative, the study sought to improve risk assessment practices for individuals with histories of violent offending who may be eligible for treatment courts. Initial validation of the LSI-R revealed weak predictive accuracy for general reconviction, moderate accuracy for felony and violent reconviction, and substantial overprediction of recidivism risk. Using existing LSI-R items and Alaska-specific assessment and administrative data, researchers developed separate General and Violent Reconviction scores that demonstrated strong predictive accuracy and substantially improved calibration. The two scores also identified distinct risk factors, with violent reconviction most strongly associated with social isolation, mental health problems, and substance use, while general reconviction was driven primarily by criminal history, substance use, and social isolation. Predictive performance remained strong across sex and racial subgroups. Findings demonstrate that locally weighted risk assessments can improve prediction of general and violent reconviction without requiring additional assessment items or changes to existing data collection procedures. The resulting tools offer Alaska probation agencies and treatment courts more accurate, individualized information to inform supervision, treatment planning, and eligibility decisions.

Type
Publication
Advancing Research in Corrections Lab

Suggested Citation

Ursino, J., & Hamilton, Z. (2026). Alaska localized LSI-R violence and recidivism assessment: Development and validation of localized general and violent reconviction scores for the Alaska Level of Service Inventory-Revised (Doc. #2026-11-1001). Advancing Research in Corrections Lab, University of Nebraska Omaha.

John Ursino
John Ursino
Research Specialist

My research focuses on advancing evidence-based practices in criminal justice. Primary areas of research include evaluating and constructing risk-need assessments to enhance decision-making processes, addressing challenges related to prison and jail crowding, and exploring correctional system reforms.

Zachary Hamilton
Zachary Hamilton
Professor

My research centers on innovation in risk and needs assessment development.