MATHEMATICS OF BEHAVIORAL CHANGE - RECIDIVISM AND CONSTRUCT VALIDITY

• NCJ Number: 56272
• Journal: EVALUATION QUARTERLY Volume: 1 Issue: 3 Dated: (AUGUST 1977) Pages: 421-438
• Author(s): M D MALTZ; R MCCLEARY
• Date Published: 1977
• Length: 18 pages
• Corporate Author: Sage Publications, Inc
• Publication Format: Article
• Publication Type: Program/Project Evaluation
• Language: English
• Country: United States of America

Annotation

THIS PAPER DESCRIBES A METHOD OF CALCULATING THE FAILURE RATE OF PROGRAMS INVOLVING BEHAVIOR CHANGE. IT OVERCOMES A THREAT TO CONSTRUCT VALIDITY APPARENT IN SOME MODELS AND CAN ESTIMATE RATES OF FAILURE AND SUCCESS FRACTIONS.

Abstract

CRITICISMS OF USING RECIDIVISM TO MEASURE THE EFFECTIVENESS OF OFFENDER TREATMENT PROGRAMS RELATE PRIMARILY TO INTERNAL AND EXTERNAL VALIDITY. THE PROPORTION OF FAILURES IN A PROGRAM AT A GIVEN POINT IN TIME IS THE ESTIMATE OF THE RECIDIVISM RATE, AND TIME IS THUS HELD CONSTANT. THIS WAY OF HANDLING TIME LEADS TO VARIOUS PROBLEMS IN VALIDITY, WITH THREATS TO CONSTRUCT VALIDITY DEFINED AS POOR FITS BETWEEN THEORETICAL AND OPERATIONAL DEFINITIONS OF TERMS. IN THE CASE OF RECIDIVISM, THE GENERALLY USED OPERATIONALIZATION IS BASED ON ASSUMPTIONS IMPLICIT IN THE CLASSICAL EXPERIMENTAL PARADIGM USED TO EVALUATE OFFENDER TREATMENT PROGRAMS. THIS PARADIGM CONSIDERS TREATMENT AND OBSERVATION OR MEASUREMENT OF THE EFFECT OF TREATMENT OVER TIME. IT ASSUMES THAT BOTH EXPERIMENTAL AND CONTROL GROUPS ARE RETURNING TO PRISON OR FAILING IN SOME WAY, AT THE SAME RATE WITH RESPECT TO TIME. SUCH AN ASSUMPTION WHICH IS NOT SAFE UNDER MOST CIRCUMSTANCES CAN BE RELAXED IF RECIDIVISM IS REDEFINED TO ACCOUNT FOR TIME DEPENDENCY. THE FAILURE RATE DEFINITION OF RECIDIVISM, HOWEVER, REQUIRES THAT RESEARCHERS ASSUME ALL PAROLEES (EXPERIMENTALS AND CONTROLS ALIKE) WILL INEVITABLY FAIL. THE PROPOSED METHOD OF CALCULATING RECIDIVISM ALLOWS THE ESTIMATION OF TWO PARAMETERS, THE FAILURE RATE AND THE ULTIMATE PERCENTAGE OF PERSONS WHO RECIDIVATE (THE RECIDIVISM) LEVEL). THE METHOD IS BASED ON THE STOLLMACK-HARRIS FAILURE RATE CONSTRUCT OF RECIDIVISM, WHICH IS A MAJOR IMPROVEMENT OVER THE BINOMIAL ESTIMATE OF RECIDIVISM FROZEN AT A SINGLE POINT IN TIME. THE SIGNIFICANT ADVANTAGE OF THE PROPOSED METHOD IS THAT IT ALLOWS FOR COMPARISONS OF PROGRAMS THAT PREVIOUSLY COULD NOT BE COMPARED WITHOUT HAVING TO USE LESS INFORMATION THAN WAS AVAILABLE (E.G., A PROGRAM HAVING 9 MONTHS OF DATA CAN BE COMPARED TO ONE HAVING 16 MONTHS OF DATA USING ALL AVAILABLE DATA). EQUATIONS AND A LIST OF REFERENCES ARE INCLUDED. (DEP)