Self–other agreement in job performance ratings: A meta-analytic test of a process model.

Abstract

This meta-analysis explores agreement in self- and supervisory ratings of job performance (k = 128 independent samples). It suggests a 3-stage model of the rating process and reviews the empirical evidence for the relevance of each of these 3 stages to an understanding of agreement in ratings. The proposed 3-stage model serves as the guiding rationale for the examination of an extensive set of variables that moderate rater agreement. Results are reported for 2 indicators of rater agreement (correlational and mean-level agreement). Self-supervisor ratings yielded an overall correlation of .22 (rho = .34; k = 115; n = 37,752). Position characteristics and the use of nonjudgmental performance indicators were the main moderators. Leniency in self-ratings is indicated by higher mean levels of self-ratings compared with supervisory ratings. Within Western samples, performance self-ratings showed leniency (d = 0.32, Delta = .49; k = 89; n = 35,417) dependent on contextual features, scale format, and scale content.