Systematic Distortions in Memory-Based Behavior Ratings and Performance Evaluations. Consequences for Rating Accuracy

There is evidence that memory-based evaluations and ratings of behavior are subject to systematic distortion, which results in high levels of halo. This suggests that memory-based ratings should be less accurate than ratings collected under conditions that minimize demands on memory. To test this hypothesis, data were obtained from two different undergraduate samples (ns = 46 and 36) in which participants rated four videotapes of graduate student lecturers. Half of the tapes were rated immediately after they were viewed (immediate rating condition); participants returned the following day and rated the remaining tapes from memory (delayed rating condition). Memory-based behavior ratings and performance evaluations showed higher intercorrelations (more halo) than did ratings that were collected immediately after viewing the ratee's performance. However, ratings were systematically more accurate in the delayed rating condition than in the immediate rating condition. We argue that under certain conditions, raters may depend on their general impressions of ratees rather than on their memory for specific details, and that these schematic evaluations may preserve a greater proportion of valid information, as compared with irrelevant detail, than is available immediately after observing ratee behavior.