On the totality of ranked set sampling

摘要

The mean of the usual balanced ranked set sample (BRSS), as introduced by McIntyre [Aust. J. Agric. Res. 3 (1952) 385], is more efficient as an estimator of the population mean than the mean of a simple random sample (SRS). Hence in this regard, BRSS is considered as a non-parametric procedure. However, for estimating other parameters of the population, the BRSS is not as universal (non-parametric) as in the case of the mean of the population. In this paper, it is shown that, under some suitable conditions, any reasonable estimator of any parameter of a population based on a SRS, can be always dominated by the corresponding estimator using some RSS plans.