HOW A REGRESSION ARTIFACT CAN MAKE ANY DELINQUENCY INTERVENTION PROGRAM LOOK EFFECTIVE (FROM EVALUATION STUDIES REVIEW ANNUAL, V 4, 1979, BY LEE SECHREST, ET AL)

• NCJ Number: 63160
• Author(s): R MCCLEARY; A C GORDON; D MCDOWALL; M D MALTZ
• Date Published: 1979
• Length: 27 pages
• Grant Number(s): LEAA-99-0073
• Sponsoring Agency: Illinois Dept of Corrections; US Dept of Justice; US Dept of Justice NIJ Pub
• Corporate Author: Sage Publications, Inc
• Publication Format: Document
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

THE FINDINGS OF RESEARCH ON DELINQUENCY INTERVENTION PROGRAMS ARE CHALLENGED BY THE SUGGESTION THAT THE APPARENTLY SUCCESSFUL EFFECTS OF THE INTERVENTION MIGHT INDICATE ONLY A REGRESSION ARTIFACT.

Abstract

IN THE REANALYSIS OF DATA FROM THE RESEARCH OF THE UNIFIED DELINQUENCY INTERVENTION SERVICES (UDIS) STUDY, THE RECIDIVISM DATA WERE FOUND TO BE UNSUPPORTABLE, NOT 'UNAMBIGUOUS' AS THE UDIS STUDY CLAIMED. ALTHOUGH A SMALL PORTION OF THE EFFECT DEPICTED IN THE UDIS STATISTICS MAY BE DUE TO CORRECTIONAL INTERVENTION, THE LARGEST PART OF THE EFFECT IS LIKELY ATTRIBUTABLE TO A METHODOLOGICAL ARTIFACT. EXTENSIVE EXAMINATION OF THE INITIAL RESEARCH AND SUBSEQUENT INQUIRY REVEAL HOW A REGRESSION ARTIFACT CAN GIVE A DELINQUENCY INTERVENTION PROGRAM THE APPEARANCE OF EFFECTIVENESS. SUCH A PHENOMENON IS A REASONABLE AND VIABLE ALTERNATIVE HYPOTHESIS FOR THE EFFECT REPORT BY UDIS. ALTHOUGH THE INITIAL STUDY POSSESSES COMMONSENSE APPEAL, IN THE REANALYSIS NO UNAMBIGUOUS EVIDENCE WAS FOUND OF AN EFFECT OF INTERVENTION NOR OF ANY SPECIFIC DETERRENCE PHENOMENON. THE MAJOR BASES OF THE INITIAL FINDINGS ARE NOT SUPPORTED BY THE SUBSEQUENT INDICATION OF NO RELATIONSHIP BETWEEN THE LENGTH OF TIME SERVED IN PROGRAMS AND INSTITUTIONS AND SUBSEQUENT DELINQUENCY. TO RULE OUT A REGRESSION ARTIFACT IN THEIR DATA, THE STUDY SHOULD HAVE EXAMINED THE RELATIONSHIP BETWEEN PRE- AND POSTINTERVENTION ARREST RATES AT THE LEVEL OF THE INDIVIDUAL DELINQUENT. BY EXAMINING ONLY THE AGGREGATE STATISTIC, THE REPORT MISSES A GREAT AMOUNT OF INFORMATION AND IS THE VICTIM OF AN ECOLOGICAL FALLACY. REANALYSIS OF THE DATA INDICATES THAT THE BEFORE AND AFTER EFFECTS ARE ATTRIBUTABLE TO THE SUBSAMPLE OF DELINQUENTS WITH ABNORMALLY HIGH PREINTERVENTION ARREST RATES. TABULAR DATA, FOOTNOTES, AND REFERENCES ARE PROVIDED.