Measurement issues in assessing employee performance: A generalizability theory approach

摘要

Increasingly, organizations are assessing employee performance through the use of rating instruments employed in the context of varied data collection strategies. For example, the focus may be on obtaining multiple perspectives regarding employee performance (360{degree} evaluation). From the standpoint of evaluating managers, upward assessments and ``peer to peer`` evaluations are perhaps two of the more common examples of such a multiple perspective approach. Unfortunately, it is probably fair to say that the increased interest and use of such data collection strategies has not been accompanied by a corresponding interest in addressing both validity and reliability concerns that have traditionally been associated with other forms of employee assessment (e.g., testing, assessment centers, structured interviews). As a consequence, many organizations may be basing decisions upon information collected under less than ideal measurement conditions. To the extent that such conditions produce unreliable measurements, the process may be both dysfunctional to the organization and/or unfair to the individual(s) being evaluated. Conversely, the establishment of reliable and valid measurement processes may in itself support the utilization of results in pursuit of organizational goals and enhance the credibility of the measurement process (see McEvoy (1990), who found the acceptance of subordinate ratings to be related to perceived accuracy and fairness of the measurement process). The present paper discusses a recent ``peer to peer`` evaluation conducted in our organization. The intent is to focus on the design of the study and present a Generalizability Theory (GT) approach to assessing the overall quality of the data collection strategy, along with suggestions for improving future designs. 9 refs., 3 tabs.