The initial study was based on responses of 103 examinees on an objective classroom test. Results indicated that response changes increased item difficulties as well as biserial and point biserial correlations. Test reliability was not decreased. This implies that not all students benefit from changing initial responses. The personal biserial correlation may be useful in identifying individuals who profit from response changes. The study was replicated on 239 students taking a standardized achievement test. (Author/MV)