Thirty-nine American and 78 Turkish preservice elementary teachers' personal science teaching efficacy (PSTE) beliefs were investigated during science methods courses with standard and retrospective pre-post testing methods. Significant differences in the PSTE gain scores, which indicate the changes in the mean PSTE scores from standard/retrospective pretests to the posttest, were found between the standard and retrospective measurements in both samples. Significant differences between the standard and retrospectively measured gain scores were detected among all subgroups under study, which were formed by participants' PSTE levels and gender. It has been concluded that the differences between the standard and retrospectively measured PSTE gain scores are due to the difference in the nature of these measurement methods and can be seen in most research samples in educational studies around the world. The findings of this study suggest that the response-shift bias should be considered as a common threat to validity for research studies measuring self-efficacy beliefs with the standard pre-post testing method.