Because it is difficult to use biochemical tests to assess drug use among school-aged subjects, researchers have tried to develop accurate self-reporting measures. The authors report that a technique known as the retrospective pretest is a helpful method to measure changes in attitude and behavior pursuant to a substance abuse education course. The technique requires that subjects respond to each question on a self-report twice. First, they respond in reference to their current perception of themselves. Then, they answer the same questions again, this time in reference to how they perceived themselves to be before they took part in the substance abuse education program. The authors report that if the two tests are administered close together in time, the results are less likely to show response-shift bias. These conclusions were based upon tests administered to three classes of seventh grade students at an elementary school located in a housing project in Chicago. A control group was given the traditional pretest and posttest after the end of the course. The retrospective pretest/posttest technique revealed no significant changes in drug use, while a traditional pretest/posttest indicated significant increases in tobacco usage. The researchers suggest that this paradoxical finding might result from response shift bias. 1 table and 18 references. (Author abstract modified)