Reliability and Validity Issues in Evaluation Research - The Community Corrections Experience (From Criminal Justice in Minnesota Proceedings, P 5-13 - See NCJ-84559)

• NCJ Number: 84560
• Author(s): B B Cohen
• Date Published: Unknown
• Length: 9 pages
• Sale Source: National Institute of Justice/
• Language: English
• Country: United States of America
• Note: Available on microfiche as NCJ-84559

Annotation

The paper examines problems related to the reliability and validity of evaluation research, based on the author's experience in evaluating community corrections projects and correctional treatment programs.

Abstract

Most of the validity problems involved in evaluation research result from the structure of the quasi-experimental designs necessitated by the impossibility of experimental designs. New programs usually have directors who are charismatic leaders and are optimistic spokespeople for their programs. This leadership may affect the results in a way that will not be repeated once the leader is replaced by others. In addition, the Hawthorne effect, whereby research subjects may behave differently due to their designation as subjects rather than because of the program, may affect an evaluation's results. Other problems include bias in the selection of research subjects, sample mortality, instability in the programs being evaluated, lack of rigor in measurement, and lack of precision in the measures themselves. Deliberate delay in the beginning of an evaluation, better instruction of the individuals whose cooperation is required, and the triangulation of measurement techniques can alleviate some of these problems. To avoid the problems resulting from contractual constraints on criterion measures, evaluators should be given greater discretion in the selection of evaluation criteria. Twenty-seven reference notes are provided.