Validation and Comparative Evaluation of Four Predictive Devices for Classifying Federal Probation Caseloads

• NCJ Number: 86625
• Author(s): J B Eaglin; P A Lombard
• Date Published: 1982
• Length: 204 pages
• Corporate Author: Federal Judicial Ctr
• Sale Source: Federal Judicial Ctr; National Institute of Justice/
• Language: English
• Country: United States of America
• Note: A Report to the Committee of the Judicial Conference of the United States on the Administration of the Probation System

Annotation

This report evaluates the comparative validity of four predictive models probation officers use to classify their caseloads in order to identify the best model for nationwide use.

Abstract

The four models studied are the California BE61A (Modified), the Revised Oregon Model, the U.S. Parole Commission's Salient Factor Score, and the U.S. District of Columbia 75 Scale. They all contain similar items (social or economic stability variables, etc.) but differ in the number of items scored. The study tested the predictive power of the four models in eight probation offices, representing a mix of regions and office size. The Oregon Model and the 75 Scale gave, respectively, the best and second-best predictions of risk. The 75 Scale would be the best national model for probation officers because the Oregon Model, although the best overall predictor, contains too many prediction elements, requires more sequential calculations, and has a number of elements sensitive to an offender's race and sex. Tables and graphs illustrate study data. Appendixes include elements of the four models, study methods, and data tables showing the distribution of outcomes by model category assignments and officer classifications. The data collection form and coding manual are included.