Handling non-response in the presence of extreme values

نویسندگان
doi
10.22034/jirss.2026.1999131.1006
چکیده

Non-response and extreme values are two major issues faced by surveyors at the survey and estimation stage; they lead to biased and inefficient estimators. This study proposes four unbiased estimators of the finite population mean in the presence of extreme values when non-response occurs in surveys. We consider two robust ranked set sampling (RSS) designs, called improved paired RSS (IPRSS) and median RSS (MRSS) for the derivation of the proposed estimators. Expressions are provided for the variance of the proposed estimator, and it is mathematically proved that these estimators are more efficient than the Hansen naïve model. Additionally, the efficiency conditions are provided in comparison with the Bouza model of handling non-response in RSS. We conduct a comprehensive simulation study to observe the relative efficiency of the proposed estimators based on hypothetical normal data. For applications in real life, we also consider heart failure data for simulations.