Recent attempts to reduce internal invalidity in studies employing pretest/posttest self-report indices of improvement have included the refinement of methodologies employing retrospective reports of pre-treatment states. The present study investigated the operation of social desirability and impression management response bias on such retrospective measures. The results do not support the hypothesis of greater bias on retrospective measurement and, in fact, are in a direction that might suggest an interpretation of reduced bias on such measures. The results also continue to support superior validity of retrospective over traditional pretest/posttest indices of improvement following treatment.A science progresses by constantly revising, updating, and improving its research methodologies. Campbell and Stanley (1963), in a landmark work, analyzed the strengths and weaknesses of various experimental and quasi-experimental designs. One of their conclusions was that &dquo;true&dquo; experimental designs controlled for all potential sources of internal invalidity. A recent have demonstrated an instrumentation-related source of internal invalidity in some true experi-