Abstract
Simulation experiments of stochastic systems provide system-performance point estimators that are corrupted by sampling error. We discuss ideas for point-estimator reporting to intuitively reflect the sampling error while retaining a valid foundation. This research first discusses some shortcomings of confidence intervals, the classical approach adopted by most simulation textbooks. This research focus on the standard error to measure sampling error, and on its informal use to determine which digits in the point estimate are meaningful. In particular, we consider the concept of a swing digit, the digit in the point estimator whose distribution would be improved by extending the simulation experiment. We discuss an approach for reporting the point estimator. The approach is to print only the meaningful digits of the point estimator.