Three kinds of rating statements, trait-oriented, worker-oriented, and task-oriented were evaluated in a context permitting the comparisons to be made in terms of criteria external to the ratings. One hundred twenty Air Force noncommissioned officers assigned to seminar grouPs of 13 or 14 were involved. No evidence of superiority was found for any of the three sets although significant correlations with various external criteria were obtained in all three experimental conditions. Significant differences were also found among the three rating sub-groups comprising each of the three treatment groups although these rating sub-groups were assigned randomly to the three treatment groups. The importance of controlling for group effects in peer group studies was noted. (Author/MH)