NCJ Number
48244
Journal
Bulletin of the Criminological Research Department Issue: 19 Dated: (1976) Pages: 27-29
Date Published
1976
Length
3 pages
Annotation
DEVELOPMENTS IN THE CONSTRUCTION AND VALIDATION OF TABLES FOR PREDICTING LIKELIHOOD OF PROBATION SUCCESS/FAILURE AMONG JUVENILE OFFENDERS IN JAPAN ARE REPORTED.
Abstract
THE REPORT SUMMARIZES THE EXAMINATION AND VALIDATION OF TWO PREDICTION TABLES CONSTRUCTED IN AN EARLIER STUDY, AS WELL AS THE CONSTRUCTION AND VALIDATION OF ADDITIONAL TABLES IN LIGHT OF METHODOLOGICAL PROBLEMS ENCOUNTERED IN THAT STUDY. THE TABLES ARE BASED ON PROBATION PERFORMANCE DATA FOR 1,493 JUVENILES PLACED ON PROBATION FROM DECEMBER 1971 TO FEBRUARY 1972. THE SUBJECTS WERE DIVIDED RANDOMLY INTO CONSTRUCTION AND VALIDATION SAMPLES. THE FOLLOWING TABLES WERE CONSTRUCTED AND VALIDATED: A 21-ITEM TABLE BASED ON A SIMPLE POINT-SCORING METHOD; TWO 14-ITEM TABLES BASED ON A MULTIPLE CORRELATION RATIO METHOD; AND 5 DIFFERENT TABLES (9-ITEM, 13-ITEM, 26-ITEM, AND TWO 14-ITEM) BASED ON THE MULTIPLE CORRELATION METHOD. THE STUDY FOUND THAT TABLES BASED ON THE MULTIPLE CORRELATION RATIO METHOD WERE SLIGHTLY LESS PREDICTIVE THAN TABLES BASED ON THE MULTIPLE CORRELATION METHOD. THE 26-ITEM TABLE WAS FOUND TO BE PARTICULARLY PREDICTIVE, BUT THE NUMBER OF ITEMS WAS THOUGHT TO BE TOO GREAT FOR PRACTICAL PURPOSES. THE PROBLEM AND CERTAIN WEAKNESSES IN THE 26-ITEM TABLE AND IN OTHER TABLES LED TO THE RECOMMENDATION THAT THE 9-ITEM AND 13-ITEM TABLES BE GIVEN FURTHER CONSIDERATIONS. A CHART SUMMARIZING THE PREDICTIVE POWERS OF THE TABLES IS INCLUDED. TABULAR DATA AND DETAILS OF THE ANALYSIS ARE OMITTED. (LKM)