Title: Using full ranking information in ranked set and judgment post-stratification sampling designs
Authors: Mahdi Salehi - University of Pretoria (South Africa) [presenting]
Bardia PanahBehagh - Kharazmi University (Iran)
Mohammad Salehi - Qatar University (Qatar)
Abstract: Ranked set sampling (RSS) and judgment post-stratification sampling (JPS) are two efficient sampling designs where one can rank units without full measurement based on visual inspection or some auxiliary variables. In all of the different available variants of RSS and JPS, many sets are selected to rank the units, but the information of just one set will be used for ranking each unit. Even for those versions of RSS using multi-ranker, one utilizes only the information from one set. In contrast to all the previous methods, we propose two approaches for improving the existing methods of RSS and JPS by employing all the selected sets to gain more information about the rank of units to be measured. Indeed, the information of all the sets will be used for all the selected units to be measured; consequently, any improper information of some potential outlier sets can be adjusted by the other sets. The mean estimators are developed. Using a relatively comprehensive simulation study, some other well-known ranked based competitors, including the median and the extreme RSS plans with their new versions constructed based on the introduced designs are compared as well. The results show that new designs lead to more efficient estimators than the ordinary counterparts' estimators for all considered cases.