MINIMAX RANKED SET SAMPLING

Abstract

In this article, a new sampling scheme based on ranked data is proposed. The main idea of the new sampling procedure is to produce a more flexible, reliable, cost-efficient design than the classical ranked set sampling and simple random sampling techniques. Accordingly, the new sampling scheme minimizes the number of wasted measurement units with high efficiency performances in estimating the population mean. Moreover, as different set sizes are used then the sample mean expected to be biased, to solve this problem an information theoretic weighted mean estimator is proposed. It is found that the weighted mean is more accurate and more efficient than the standard one. and the sample mean based a simple random sampling technique. Both estimators, weighted and un-weighted outperform the simple random sampling scheme in estimating the population mean. A real data set for estimating the average cohort percentage for survivals to age 65 in Jordan from 1960 to 2015 is used to illustrate the proposed sampling scheme.