Improving self-report data : an examination of the retrospective pretest in controlling for response shift bias

Response shift bias jeopardizes the internal validity of self-report data used in a pretest-posttest fashion when evaluating the impact of treatments which are designed to change the subject's understanding of his functioning. Response shift bias may be eliminated by use of a retrospective pretest-posttest control .group design; however, retrospective* pretests have been criticized as subject to the influence of social desirability response sets and impression management. This criticism was evaluated in subjects who were assigned to either an assertiveness training treatment condition, or a wait list control condition. Half the subjects in each condition were exposed to a bogus pipeline, desigmxl to remove social desirability responding on the part of the subjects. Results show no evidence of the operation of social desirability responding on retrospective pretests with self-report scales.