This paper presents novel estimators for a judgment post-stratified (JPS) sample, which combine the ranking information from different methods or rankers. A JPS sample divides the units in the original simple random sample (SRS) into several ranking groups based on the relative positions (ranks) of the units in their individual small comparison sets. Ranks in the comparison sets may be assigned with several different ranking procedures. When considered separately, each ranking method leads to a different JPS sample estimator of the population mean or total. Here we introduce equally or unequally weighted estimators, which combine the ranking information from multiple sources. The unequal weights utilize the standard errors of the individual ranking methods estimators. The weighted estimators provide a substantial improvement over an SRS estimator and a JPS estimator based on a single ranking method. The new estimators are applied to crop establishment phenotypic data from an agricultural field experiment.

Supplementary materials accompanying this paper appear online.

本文提出了一种新的估计器，用于判断后分层（JPS）样本，该样本结合了来自不同方法或排序器的排序信息。JPS样本根据原始简单随机样本（SRS）中各个单元在各自的小型比较集中的相对位置（等级）将其划分为几个排名组。可以用几种不同的排名程序来分配比较集中的排名。当单独考虑时，每种排序方法会导致总体平均值或总数的不同JPS样本估计量。在这里，我们介绍了相等或不相等的加权估计量，它们结合了来自多个来源的排名信息。不相等的权重利用了各个排名方法估计量的标准误差。加权估计器相对于基于单个排序方法的SRS估计器和JPS估计器提供了实质性的改进。新的估算器应用于来自农田试验的作物建立表型数据。

本文随附的补充材料在网上显示。