Reducing Contextual Bias In Performance Appraisal

University Of South Carolina At Columbia, Columbia SC

Investigators

Abstract

Abstract *** 0011132 Douglas H. Wedell There is widespread use of assessment tools in our society. Performance appraisal is one such tool that serves as a basis for important personnel decisions, such as allocation of raises or recommendation for promotion or termination. Because performance appraisal strongly relies on subjective judgment, it is susceptible to systematic biases or distortions. One such bias that has been repeatedly demonstrated in performance appraisal is contextual contrast. This bias consists of the tendency to rate a person's work performance higher after assessing other employees with poorer work records than after assessing those with better work records. Such contextual dependency represents a serious source of error in assessment, threatening its reliability and validity. This project explores different procedures designed to reduce or eliminate contextual contrast in performance appraisal and thereby create more reliable and valid assessment techniques. Materials consist of performance profiles of hypothetical employees that include information on several facets of their work records. In each experiment, the contrast bias is induced by having participants evaluate different sets of materials prior to assessing a core set of materials. The aim of the research is to develop assessment procedures that will produce the same judgments of core materials across different contexts. These procedures include manipulating category labels, types of judgments, and training methods. Experiments will test the degree to which including clear examples for the different category levels will reduce contextual bias. Additional experiments will test the efficacy of two alternative judgment tasks, comparative and similarity judgment, for eliminating contrast effects. Finally, another series of experiments explores the potential for different training regimens to eliminate contextual bias. The aim of this research is to develop procedures that could be implemented in business, education, and government to produce more reliable and valid assessments. ***