Validation for Performance Measurement in the Task of Fault Management

Abstract

The application of a part-task nuclear simulator to the measurement of performance in the task of fault management was studied. Specifically, the design of the simulator was evaluated for internal and external validity. External validation required confirmation that the task presented by the simulator had the essential elements of the real task. In the context of this particular study, internal validation required confirmation that performance on the simulator represented an accurate and fair measure of a subject's understanding of fundamental principles.

The requirements for external and internal validity were found to be in conflict. Performance on the simulator was not an accurate measure of fundamental understanding because the task was realistic. However, it was concluded that a part-task simulator does provide an effective method of gathering information on human performance.