Subjects within a group of nine were supposedly given different amounts of practice prior to taking a test on which performance was either related or unrelated to practice. After taking the test, each subject was given his score, his rank order in the group, and the amount of practice presumably done by each of the other group members. The subject was then given a choice of which performance score he would like to know; a second choice was also given. On both first and second choices, subjects in the related condition were more likely to choose to see the scores of others with the same amount of practice. Related condition subjects were more interested in their comparison choices and thought they had done better on the task than unrelated subjects. In addition to this confirmation of the related attributes hypothesis, there was a significant tendency toward comparison on the basis of similarity of performance per se and a significant tendency toward upward comparison.