Employees’ Reactions to Algorithmic Performance Evaluation: Threat of Evaluation Bias and Objectivity

Abstract

Listen

Translate article

ABSTRACT Although algorithms are increasingly used for performance evaluation purposes, prior algorithm research shows that individuals tend to trust human decisions more than algorithmic decisions in a subjective domain such as performance evaluation that requires human skills. I conduct an experiment to examine whether the effect of evaluator type (algorithm versus human supervisor) on employees’ perceived trustworthiness of performance evaluations hinges on evaluation bias (positive versus negative) and further investigate whether this interaction effect is mediated by perceived objectivity. I find that employees’ perceived trustworthiness shifts from human evaluations to algorithmic evaluations when negative evaluation bias exists. Further, this reduced algorithm aversion is stronger and significant for people who are vulnerable to evaluation bias because people perceive algorithms as making objective evaluation decisions. Overall, results of this study suggest that the threat of evaluation bias reduces algorithm aversion because of the higher perceived objectivity of algorithmic decisions. Data Availability: Data are available upon request.