水质氨氮的测定能力验证结果统计和评价分析

Results Statistics and Evaluation Analysis of Proficiency Testing for Determination of Ammonia Nitrogen in Water

本研究以水质氨氮的测定能力验证(GZSCJ-2020-03)为例对测试结果的统计方法和评价模式进行了探究。首先,采用经典统计方法、稳健统计方法、经验模型法(Horwitz函数)、自助法对不同浓度水平样品的测试数据进行独立统计分析,分别算得指定值和能力评定标准差,并用单一样品Z比分数进行评价。综合考虑能力验证目的、数据分布情况、离群比例、测试方法特性、经验参考值等因素,采用迭代算法A作为本次计划的结果统计分析方法。同时,样品对Z比分数作为一种特殊的Z比分数,也与单一样品Z比分数进行了比较。结果表明,样品对Z比分数能从不同的维度去判定数据结果,更有助于实验室异常结果分析。文中深入探讨了不同统计方法和评价模式的原理、使用场景以及优缺点,旨在为环境监测领域能力验证中结果统计评价的选择提供参考依据。

The proficiency testing(PT)for determination of ammonia nitrogen in water(GZSCJ-2020-03)was taken as an example to explore the statistical method and evaluation mode. Firstly, classical statistical method, robust statistical method, empirical model method(Horwitz function)and bootstrap method are used to independently analyze the data of samples with different concentration levels, calculate the assigned value and the standard deviation, then the specified value of results and standard deviation of ability evaluation were calculated respectively, and Z-score of single sample was used for evaluation. Considering the purpose of the PT,data distribution, outlier ratio, testing method characteristics and empirical value, iterative algorithm A was used as the statistical analysis method in this PT. At the same time, as a special Z-score, Z-score of sample pairs was also compared with single sample. The results showed that Z-score of sample pairs can evaluate the results from different dimensions, which was more helpful to search laboratory abnormal results. The principles, application scenarios, advantages and disadvantages of different statistical methods and evaluation modes were discussed deeply, in order to provide references for the selection of statistical evaluation of results in the proficiency testing of environmental monitoring.