Interactive simulation

In recent years a significant amount of effort has gone into making simulation languages friendlier for the decision-maker and easier to use for the simulation analyst. One area where these efforts are beginning to show results is the area of interactive simulations, in which the user can observe and modify a simulation model as it is running. However, there are hidden perils in interpreting the results of the data output from an interactive simulation. This paper highlight some of these dangers, pointing out some statistical pitfalls with the traditional end-of-run post-simulation analysis. Some suggestions are made about the directions future research might take to provide more meaningful and statistically valid end-of-run results.<<ETX>>