Discussion For Three Kinds Of Generalized Ranked Set Sampling

This article studies design optimized question of three kind of generalized ranked set sampling method that is the random selection ranked set sampling, generally generalized ranked set sampling and extreme ranked set samplingâ…¡.First, research random selection ranked set sampling (MRSS), its application in the index distribution and the logarithmic distribution, obtained under two distribution MRSS average value estimate variance. Then, under the index distribution, this method compares with the simple random sampling (SRS); under the logarithmic distribution, the R-C world of mortals which estimated with the average value compares, obtained the MRSS method most superior design proposal separately. Had proven under certain situations, the MRSS average value estimator is more practical than SRS, is more effective.Next, proposed that one kind of new generalized ranked set sampling (GRSS), compared with best liner unbiased estimate (BLUE), has given the sample uniform distribution quantile estimate formula which this sampling method produces. Has given smallest variance unbiased estimator, the simple estimator and the balanced ranked set estimate formula based on this new method and the literature method, calculates the corresponding variance and the correlation coefficient, had proven when estimate uniform distribution quantile, this new method surpasses in the literature the method.In final, against a special case of k-tuple ranked set samples, extreme ranked set sampling 2 on the sign ranked test, investigate the exact null distribution and size big asymptotic distribution, thought the comparison of Pitman asymptotic relative efficiency of ERSS2 versus SRS, when the ranking cost is not ignore it will be shown that in small sample, ERSS2 better than RSS when test the median, while in the big sample, ERSS2 only better than SRS.