Averages of performance measures from stochastic simulation models are gneerally calculated as interval estimates based on the t-or z-statistic. However, this approach is based on the assumption that the measure of interest is normally distributed across replications but some doubt has been cast upon the accuracy of this assertion. A more recently developed analysis technique known as the bootstrap has been suggested as an alternative to the traditional parametric method since t makes no assumptions about the pupulation distribution of the sample.
展开▼