A method of confidence interval construction for simulation output analysis

Abstract

A partial variance procedure for constructing confidence intervals (c.i.) for the mean of a process is proposed. The use of partial variance allows constructing an asymmetrical c.i. that accounts for the skewness of the population. The results of a Monte Carlo experiment comparing the proposed method with other methods indicate that the proposed method provides more desirable confidence intervals. That is, the proposed method provides a confidence interval with a higher probability of coverage or the same probability of coverage but with a shorter width.