Self-report, pre/post testing is a frequently employed measure of therapeutic change. To investigate whether expectation of change might be an alternative explanation to the scale shift explanation of response shift bias in a self-report measure, a two-session assertiveness training intervention for college women was evaluated under manipulated conditions of expectancy or no expectancy for change. A no-treatment control group was also included. Comparison of Rathus Assertiveness Inventory pretest and retrospective pretest scores (i.e., reports of pretest status obtained at time of posttest) demonstrated a reliable response shift under the condition of expectancy for change, but not under the no expectancy for change condition. These results support a self-justification rather than scale shift explanation of response shift bias. Use of retrospective pretests to evaluate therapeutic intervention is not indicated. (Author/JAC)