Confidence Intervals for the Quantiles of a Finite Population: Simple Random and Stratified Simple Random Sampling

Abstract

Summary Given a finite population consisting of N units, it is desired to have a confidence interval for the (t/N)th quantile, Y (t), of the population. Assuming simple random sampling, a confidence interval procedure is suggested and several expressions for the associated confidence coefficient are given. With stratified simple random sampling, for L = 2 strata, three confidence interval procedures are suggested. The associated confidence coefficients are derived under each of two (illustrative) assumptions about the nature of the stratification. (These assumptions refer to a probabilistic knowledge about the stratification.) The procedures are then compared using data from several finite populations.