Blind Spots in Assessing and Predicting Key Dimensions of Job Performance: An Examination of Supervisor and Self-Ratings

Abstract

Measuring job performance is fundamental to validating selection procedures. However, job performance is multi-dimensional, and common measurement practices vary across its dimensions. While task performance is typically measured by supervisor ratings, counterproductive work behaviors are often measured by employee self-ratings, and there is no standard measurement practice for measuring organizational citizenship behaviors. Drawing from research on multi-rater models, this study proposes that the unique perspectives of supervisors and employees can both provide insights into employees’ job performance and sets out to understand for which performance dimensions their combination is most valuable when validating selection procedures. Using latent models, it examines the extent to which supervisor and self-ratings contain shared and unique information about job performance dimensions, and how two common selection predictors (conscientiousness and general mental ability) predict this shared and unique information. Results show that consensus between supervisor ratings and employee self-ratings is modest across performance dimensions and that conscientiousness shows stronger relations with unique employee perspectives, whereas general mental ability is more strongly related to unique supervisor perspectives. The present findings have implications for the measurement of performance and the interpretation of criterion-related validities of selection procedures.