Estimating a finite population mean in presence of outliers under stratified random sampling

Abstract

Estimation of finite population parameters in the presence of outliers has been studied by many researchers in Sample Survey Theory. Outliers are known to skew the estimator of the population parameters resulting in bias of estimators. To improve on efficiency of the mean estimator, we propose a modified estimator of the population mean in stratified random sampling which incorporates both the sample mean and weighted mean of each stratum. Simulation studies are performed to compare the estimators obtained using the proposed approach and their rival estimators under stratified random sampling technique. The proposed estimator of the mean demonstrated improved coverage rates compared to rival estimators, by having lower variance and competitive Mean Squared Error in certain cases. KEYWORDS: Outliers, Stratified random sampling, Skewed.