Rating Performance or Asserting Status: A Test of a Social Dominance Theory of Supervisor Bias in Performance Ratings

Supervisory ratings of employees ’ performance remain the dominant means of assessing job performance in organizations despite decades of research documenting their biases. We propose that a major reason why so little progress in improving the accuracy of these ratings has been made is the theoretical mis-specification of the reasons for systematic supervisory rating bias. We extend Social Dominance Theory to explain similarity bias in supervisory ratings and test this theory against the reigning theory of homophily in a strong inference test. The contradictory hypotheses are tested against one another in a field study of 358 supervisor-subordinate dyads in ten organizations. Using hierarchical linear modeling to eliminate explanations at the organizational and supervisory levels, we find that supervisors ’ biases in rating their subordinates ’ contextual and task performance are explained by Social Dominance effects, and that none of the homophily bias hypotheses are supported. The implications of these results for addressing bias in supervisory performance ratings, as well as the theories of status in organizational behavior are discussed.