Sequential Confidence Intervals for the Mean of a Subpopulation of a Finite Population

Abstract

Sequential confidence intervals for the parameters of a subpopulation of a finite population are studied. Procedures are suggested using the mean, median, and trimmed means; these procedures are shown to be more efficient than their counterparts which are based on independent, identically distributed observations from infinite populations. The major technique used is to “linearize” the estimates, yielding as a simple consequence proofs of central limit theorems. A Monte Carlo study shows that the small-sample performance is very good.