Measurement Issues in the Evaluation of the Effects of Juvenile Delinquency Programs - Final Report

• NCJ Number: 84968
• Author(s): M W Lipsey
• Date Published: 1981
• Length: 99 pages
• Grant Number(s): 80-IJ-CX-0036
• Sponsoring Agency: US Dept of Justice; US Dept of Justice NIJ Pub
• Sale Source: National Institute of Justice/
• Language: English
• Country: United States of America

Annotation

The report developed a measurement assessment scheme based on generalizability theory and components of variance analysis to investigate commonly used outcome (recidivism) measures in juvenile delinquency prevention programs.

Abstract

Two large studies examined a range of measures derived from police arrest histories. They included measures based primarily on number of offenses, severity of offenses, and time to next offense. All such measures had extremely low sensitivity to surrogate treatment effects in the form of a criterion contrast between groups of juveniles with clearly different delinquency levels. The large criterion contrast contributed less than 4 percent of the total measurement variance on even the best measures. Effect sizes more similar to actual likely treatment impact accounted for less than 1 percent of the total measurement variance. The insensitivity of arrest-based measures to real differences was traced to unacceptably low reliability. The result of these measurement deficiencies is that the typical evaluation study using an arrest-based outcome measure will have such low statistical power that it will be unlikely to detect a treatment effect even when one of worthwhile size exists. This problem can be solved by developing alternative measures, such as self-report delinquency measures. A small study examining several self-report measures produced disappointing results -- those variables capable of serving as treatment outcome measures did not appear to be any more responsive to the criterion contrast than the arrest-based measures. Self-report measures do, however, warrant further examination as potential treatment outcome indicators. Study results and 19 references are included. Appendixes include California Youth Authority offense seriousness codes, means and standard deviations for the various delinquency measures, the questionnaire used, and a paper describing the measurement assessment scheme (with notes and references).