U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

PROBATION PREDICTION MODELS (FROM CORRECTIONS IN THE COMMUNITY - ALTERNATIVES TO IMPRISONMENT - SELECTED READINGS, 2D ED., 1978 BY G G KILLINGER AND P F CROMWELL, JR - SEE NCJ-45910)

NCJ Number
45914
Author(s)
ANON
Date Published
1978
Length
10 pages
Annotation
THE VALIDITY AND PREDICTIVE POWERS OF 8 STATISTICAL MODELS ARE TESTED BY APPLYING THE MODELS TO 900 CLOSED PROBATION CASES FROM 3 COUNTIES IN ARIZONA, OREGON, AND PENNSYLVANIA.
Abstract
THE MODELS TESTED ARE THE OREGON FORM, THE NEWARK FORM, FOUR MODELS DEVELOPED BY CALIFORNIA CORRECTION AGENCIES ON THE BASIS OF PAROLEE DATA, AND THE SALIENT FACTORS FORM, DEVELOPED FOR USE BY THE FEDERAL BOARD OF PAROLE. TEST DATA ARE FROM MARICOPA COUNTY, ARIZ.; MULTNOMAH, COUNTY, OREG.; AND PHILADELPHIA COUNTY, PA. THE MODELS APPARENTLY ARE TRANSFERABLE FROM ONE LOCATION TO ANOTHER. THREE WERE VALID IN ALL THREE COUNTIES. THE OTHER FIVE WERE VALID OR PROBABLY VALID IN AT LEAST ONE COUNTY. THE MODELS MAY BE USEFUL IN PROBATION DECISIONMAKING. IN ONE COUNTY, 93 PERCENT OF PROBATIONERS WITH HIGH MODEL SCORES (INDICATING HIGH POTENTIAL FOR SUCCESS) WERE SUCCESSFUL, WHEREAS ONLY 17 PERCENT OF THOSE WITH LOW MODEL SCORES WERE SUCCESSFUL. MODEL SCORES APPEAR TO BE USEFUL IN DETERMINING SUPERVISION LEVELS FOR PROBATIONERS. IN ONE LOCATION, THE ACTUAL FAILURE RATE FOR PROBATIONERS RECEIVING MINIMUM SUPERVISION WAS 35 PERCENT, WHEREAS ONLY 15 PERCENT OF PROBATIONERS SELECTED BY A MODEL FOR MINIMUM SUPERVISION FAILED. MODELS WERE ALSO SUCCESSFUL IN SELECTING PROBATIONERS FOR EARLY RELEASE. THE MAJOR ADVANTAGES OF MODELS ARE THEIR OBJECTIVITY, THEIR EFFICIENCY, AND THEIR POTENTIAL RELATIVE TO ASSESSING PAST EXPERIENCE SYSTEMATICALLY. THE ADVANTAGES OF HUMAN JUDGMENT RELATE TO SUCH FACTORS AS COMPASSION AND INTUITION. WHILE THE PREDICTIVE POWERS OF THE MODELS TESTED ARE NOT PERFECT, THEY APPEAR TO BE GREATER THAN THOSE OF PROBATION OFFICIALS. BETTER ACCURACY COULD BE OBTAINED IF BOTH MODEL SCORES AND HUMAN EVALUATIONS WERE USED IN PROBATION DECISIONMAKING. SUGGESTIONS FOR FURTHER RESEARCH ARE OFFERED. SUPPORTING DATA ARE INCLUDED. (LKM)

Downloads

No download available

Availability