Response-shift bias occurs when participants' initial constructs, such as self-efficacy in teaching science, are incomplete because they do not fully conceptualize something they have yet to experience. This study examines whether elementary pre-service teachers can consistently evaluate constructs such as self-efficacy and attitudes toward science throughout an elementary methods course. After the administration of traditional pre-tests, retrospective pre-tests, and post-tests, this study examined whether a response-shift bias consistently occurred in scales indicating science teaching self-efficacy (as measured by the Science Teaching Efficacy Beliefs Instrument-B), attitudes toward science (modified Attitudes Toward Science Inventory), and relevancy of science (Changes in Attitudes about the Relevancy of Science). Results indicate that a significant response-shift bias occurred in the scales relating to self-efficacy, confidence, and attitudes toward science; while no response-shift bias occurred in scales relating to outcome expectancy, value, and relevancy of science. Our data provide evidence that response-shifts naturally discriminate among different constructs and that participants successfully and consistently reported certain constructs over time which may be of interest for program evaluators and self-efficacy researchers. This research could have implications for program evaluators and researchers who need to measure program impacts on pre-service teachers with limited science-teaching experiences.