Validating and comparing simulation models using resampling

Statistical resampling methods provide a simple, versatile and effective means of comparing and validating simulation models. The purpose of this paper is to introduce the framework that the author has found very effective in tackling a wide range of practical situations. Although the methods of analysis discussed are available through the more specialist statistical packages, the main message of this paper is that resampling is actually easily implemented on spreadsheets and does not need specialist knowledge to use effectively. The commonly occurring problems of input modelling, model validation and model comparison in simulation are considered in detail to illustrate how they can be tackled using resampling.