GB/T 14838-2009 in English

《GB/T 14838-2009橡胶与橡胶制品 试验方法标准精密度的确定》由TC35（全国橡胶与橡胶制品标准化技术委员会）归口，主管部门为中国石油和化学工业联合会。
本标准为通过实验室间试验方案(ITP)确定橡胶制造工业和炭黑工业所使用的试验方法标准精密度提供指南。本标准采用GB/T6379中基本单向分析的方差计算法则,尽可能使GB/T6379的术语和定义不与过去历史相矛盾,不与这两个行业中精密度的确定程序相矛盾。虽然本标准没有确定偏倚,但它是理解精密度确定的一个基本概念。本标准不包括准确度和精度的确定,GB/T6379中包含这两个概念。
本标准提供了两种精密度的确定方法,它们被称为试图消除或大幅度降低离群值影响的稳健统计方法。第一种方法是一水平精密度程序,适用于橡胶制造工业的所有试验方法,第二种方法则是一水平精密度程序的一个特殊情况,称为二水平精密度程序,适用于炭黑工业试验。两者使用相同的均匀水平实验设计和曼德尔�韬��胪臣品ǘ郧痹诘睦肴褐稻�密度数据库进行检查。然而,它们采用略有不同的剔除程序处理作为离群值的不协调数据值。二水平精密度程序在每个数据库单元或材料/实验室组合的复制数量方面是独特的。

Scope

This standard is adopted by the interlaboratory test program. (ITP) provides guidance for determining the standard precision of test methods used by the rubber manufacturing industry and the carbon black industry. This standard adopts the variance calculation rule of the basic one-way analysis in GB/T 6379, so that the terms and definitions of GB/T 6379 do not contradict the past history as much as possible, and do not contradict the precision determination procedures in these two industries. Although bias is not identified in this standard, it is a fundamental concept for understanding the determination of precision. This standard does not include the determination of accuracy and precision, these two concepts are included in GB/T 6379. This International Standard provides two methods of determining precision, which are referred to as "robust" statistical methods that attempt to eliminate or substantially reduce the effects of outliers. The first method is the "one-level precision" procedure, which is applicable to all test methods in the rubber manufacturing industry, and the second method is a special case of the one-level precision procedure, called the "two-level precision" procedure, which is applicable to In the carbon black industry test. Both use the same uniform-level experimental design and Mandel h-sum walk statistics for potential outlier precision database checks. However, they employ slightly different rejection procedures for incongruous data values that are outliers. The "two-level precision" procedure is unique in terms of the number of replicates per database unit or material/laboratory combination.