Interaction with a visual interactive simulation, and the effect of cognitive style

Despite the considerable folklore surrounding Visual Interactive Simulation (VIS), there is little published empirical evidence concerning how users interact with models, what displays they require, and how well they perform with them. This paper reports an initial experiment into user interaction with a VIS. Twenty-five subjects were asked to ‘solve’ a service capacity and allocation problem using a VIS model that provided three separate displays: an animation, a listing, and a dynamically changing graphic. Use of the model was monitored by the simulation program. Compared to a known optimum, performance with the VIS was mediocre, and there was little evidence that any particular pattern of usage affected performance, although performance and absolute time to solve the problem were significantly correlated. Those who performed well, however, generally had higher confidence in their solution. Further, there was a marked preference for displays, with the listing hardly used, but with users’ having a strong preference for either the animation or the graphic. Users’ cognitive style was measured, using the Myers-Briggs instrument, and there is some weak evidence that preference for display type can be partially explained by cognitive style.