In an earlier study by Sprangers and Hoogstraten, a bogus-pipeline induction did remove response-style effects in the self-reported pretest. Response-shift bias, defined as a significant mean difference between conventional pre- and retrospective preratings, was consequently eliminated. It was concluded that response-style effects in the pretesting are a likely cause of response-shift bias. The present experiment was designed to examine whether these results are stable and generalizable to a different educational training. The present replication made use of the same bogus-pipeline procedure. The experimental training was a First Aid instructional film. Subjects were 53 freshmen in psychology who were fulfilling a course requirement. Contrary to expectation, a bogus-pipeline induction did not lower self-reported preratings. A response-shift did not occur in the bogus-pipeline or in the non-bogus-pipeline conditions. It was concluded that a construct not susceptible to removal of response-style effects is not susceptible to response-shift bias either. Results are consonant with the response-style basis for response-shift bias and show no contradiction of the former study. Data furthermore show that the administration of an objective pretest had no effect on subsequent objective posttreatment scores.