Simulation Performance & Predictor Variables: Are We Looking In The Wrong Places To Measure The Right Learning?

Abstract

This paper examines the relationship between predictor and performance variables in a typical computerized simulation. Most simulations require the users to make a set of predictions on various performance variables. The question is whether there is any relationship between what users predict and how they ultimately perform. The authors collected data on several predictor variables and compared them to a performance variable for one semester of simulation play for two groups of undergraduate business majors at a mid-sized college. The authors also surveyed a group of business faculty who typically use simulations in their classes to ascertain their views on the relationship between predictor variables and performance variables. The authors found that as simulation play progressed, those student teams which had less variance between their predicted sales and actual sales tended to have better simulation performance than those teams which had a greater variance between their sales predications and actual sales results. In terms of the survey of instructors who use computerized simulations, one of the more interesting results was that while most respondents stated that they asked their students to establish predictor variables; few indicated what purpose this served or why they asked students to establish predictor variables. More research is needed to both understand why simulation instructors ask users to formally establish predictions and what role predictor variable have in helping students with performance.