Presented to the Annual Conference at the University of Leeds, 2-6 April 1990 “Simulation as a tool in the measurement of human performance”

Publisher Website
Google Scholar
Abstract
The paper outlines and discusses issues in the present-day development and application of simulations of complex tasks. It is noted that this development is observed in basic as well as in applied research, resulting in a trend away from the investigation of elementary properties of human functioning in favour of the analysis of complete tasks. One of the problems in simulation concerns appropriate output measures (section 2). Some recent sophisticated measures of performance are derivatives of more elementary measures of time and accuracy, but they still facilitate new insights into the measurement of performance. In addition, recent views are discussed of verbal reports as potential measures of cognitive skills. It is noted that the issue of performance measurement has been severely neglected and ignored in many applications of simulators. The paper also focuses on theoretical models (section 3). It is concluded that models for complete tasks face the problem of excessive complexity, thus inhibiting detailed parametric analysis. In this respect the analysis of elementary properties has a clear advantage. A strategy of back-to-back experimentation, originally proposed by Gopher and Sanders (1984), is reiterated and an example of a back-to-back study is described. Section 4 of the paper is devoted to the issue of validation with major emphasis on the possibilities and limitations of the transfer paradigm and the measurement of direct correspondence. The final section (5) is devoted to some selective remarks about the three main areas where simulators are applied, namely training, equipment design, and personnel assessment. The paper ends by summarizing some emerging trends and issues for future research.