Statistical analysis of steady-state simulations: survey of recent progress

For non-terminating simulations with steady-state responses the following alternative analysis techniques are evaluated: (i) replicated runs (ii) approximately independent subruns or batches (iii) renewal or regenerative analysis. Two general techniques for reducing nonnormality and bias are presented: grouping (or batching) and jackknifing. Applications of the various techniques are discussed. Besides the estimation of the mean, the estimation of variances and quantiles is presented.