Measuring response-shift bias in social research

Abstract

In 1976 Golembiewskl, Billingsley, and Yeager made a substantial advance in the evaluative methodology when they identified three types of change which can result when self-reports of change are used. These were labelled alpha, beta, and gamma changes. Whereas alpha change implies that true change has occurred as a result of intervention effects, beta and gamma change can be seen as forms of response-shift bias which could threaten the internal validity of a study. In the present study the effectiveness of an assertiveness training course was evaluated with a particular view to identifying and measuring the presence of intervention effects and response-shift effects. This was achieved by making use of pretest and posttest data as well as a retrospective pre-measure. The results of the study provided evidence for the success of the training programme as the influence of both gamma and beta change could be discounted. It is asserted that self-report measures of change need to be more fully scrutinized than merely subjecting them to the conventional pretest/posttest studies as these do not allow the necessary flexibility to detect the occurrence of beta and gamma change.