Retrospective Pretest-Posttest Analysis versus Traditional Pretest-Posttest Analysis

Abstract

Traditional pretest-posttest comparisons of self-report data are distorted by response-shift bias. Administration of a retrospective pretest in lieu of the traditional pretest eliminates a form of response-shift bias which distorts the comparability of pretest-posttest measurements. The present study compared the sensitivity of a retrospective pretest-posttest measurement versus a traditional pretest-posttest measurement in detecting a treatment effect for a university stress counseling program. The substitution of the retrospective pretest for the traditional pretest as the covariate in the analysis of covariance yielded the same conclusion of no treatment effect.