This study was designed to (1) examine the effect of instrument format on evaluators' ability to make valid ratings of teaching performance on a specified criterion, (2) examine the effect of instrument format on the variance of evaluator ratings on a specified criterion, (3) examine the effect of instrument format on evaluators' ability to identify targets for improvement and reinforcement, and (4) on one of the instruments, examine differences in evaluators' ratings when using a point scale versus a continuous scale. Two instruments were designed for this study--a graphic response mode and two forms of a format which required the ratings of eight indicators (descriptors of effective performance on the criterion) prior to rating the criterion--continuous scale and point scale. One hundred five administrators attending a Teacher Performance Evaluation workshop in Illinois in June of 1985, provided the data for the study. They rated a videotaped teaching performance on the criterion "Communicates Effectively with Students" and identified teaching behaviors to target for imporvement and for reinforcement. All Ratings and responses were compared to expert panel's ratings and responses;The study yielded four important findings. (1) The instrument format which forced evaluators to rate indicators before rating the criterion led to ratings significantly closer to an expert panel rating than did the graphic response mode instrument format. (2) There was significantly less variance in the ratings of evaluators who used the instrument format which required ratings of indicators than the format which did not. (3) Neither instrument format helped evaluators identify teaching behaviors to target for improvement or reinforcement. (4) Many evaluators who used both forms of the forced indicator rating format did use the continuous scale to change ratings.