VALIDITY OF SELF-ASSESSMENT OF WORK PERFORMANCE - A STUDY IN INDUSTRIAL-RESEARCH AND DEVELOPMENT

Based on a literature review, the psychometric criteria (reliability, validity, rating tendencies) as well as the (assumed) moderators of the validity of self-assessment are presented. A closer examination shows that most existing studies on the validity of self-assessment did not refer to work performance. Therefore, the quality of self-assessed work performance was investigated in a sample of scientists and engineers in industrial R & D. Especially, metaanalytic results were used in order to improve the conditions for higher validity of self-assessment (anonymity, social comparison instruction, information that self-assessment will be validated). In the present study, no leniency effect was found in self-assessment compared to supervisory's assessment. In agreement with earlier results, self-assessments have lower standard deviations and less halo than supervisory's assessments. The correlation (validity) between self- and supervisory's assessment is r = .42 (n = 144). Further moderator analysis confirmed the robustness of this result. Finally, explanations for the limited validity of self-assessment are discussed.