Inmate Self-Report Data: A Study of Reliability

Abstract

The present study examines the reliability of self-report data inadvertently collected from 11 adult inmates in a study of a house arrest program. Data were collected with a self-administered questionnaire that was given twice within a four-month period. Correlations of 92 items, divided into categories of "Information," "Expectations," and "Affections," showed wide variation in reliability. Complete agreement on the two occasions was found for only 24 items, which were primarily in the "Information" category. Correlations across the 92 items were also computed for each subject, and these correlations also showed wide variation in reliability. Although no subject gave exactly the same responses on all 92 items, greater reliability was again found in the "Information" category. In addition, 3 of the subjects gave particularly unreliable responses. These results, together with the lack of existing literature on the topic, suggest a need to conduct more research on reliability of inmates in order to develop more reliable measures as well as to detect unreliable subjects.