A Comprehensive Approach to Poststratification for Subpopulation Estimation in Stratified Sampling Design

This paper addresses the problem of subpopulation parameter estimation within the framework of stratified sampling. Instead of relying on additional sampling, we introduce three novel estimators for subpopulation means and totals that leverage poststratification to facilitate effective estimation. These estimators are designed to account for subpopulation heterogeneity and variability. A comprehensive simulation study under various distributional assumptions evaluates the statistical properties of the proposed estimators, including bias and mean squared error. The results provide insights into the statistical behavior of these methods under different conditions and demonstrate their applicability using real-world data from the U.S. Census of Agriculture.