Interval estimation using replication/deletion and MSER truncation

This paper addresses the construction of a consistent interval estimator for the steady-state mean within a replication/deletion framework for output analysis when MSER truncation is applied. Because the MSER truncation point is a random variable, the truncated output sequences for each replication typically are unequal in length. A weighting scheme is applied to the replication means to correct for unequal sample sizes, as is standard in ANOVA. A numerical example is provided to illustrate the procedure and consequences.