四分位法和迭代法对数据分散的能力验证检测数据统计分析结果的比较

能力验证是指利用实验室间比对,按照预先制定的准则评价参加者能力的活动。四分位法和迭代法均是分析能力验证数据的稳健统计分析方法,可以用来计算数据目标标准偏差,评价实验室上报数据是否合格。通过分析多项检测能力验证数据,比较了两种稳健统计分析方法的计算结果。分析数据的结果表明,当上报数据的标准化四分位距与基于经验模型的再现性标准差之比(H值)大于2或稳健变异系数CV大于0.05时,各实验室间检测能力相差较大,能力验证数据分布较分散,迭代计算次数大于1次,标准化四分位距与迭代法计算的稳健标准差比值大于1,四分位法放宽了能力验证的评价标准。

Statistical analysis results comparison of quartile and iteration used in proficiency testing for data dispersion inspection

Proficiency testing is an action to evaluate the ability of participants according to the pre-specified criterions based on the inter-laboratory comparison. Quartile and iteration are two robust statistical analysis methods dealing with the data of proficiency testing, which can be used to calculate the target standard deviation and evaluate whether the laboratory data are qualified. The calculation results of two robust statistical methods were compared by analyzing the multi-data sources of proficiency testing. The analysis results showed that the inter-laboratory detection ability exhibited significant difference and the data of proficiency testing distributed dispersedly when the ratio of normalization interquartile range of laboratory data to the reproducibility standard deviation (H value) based on empirical model was larger than 2 or the robust coefficient of variation (CV) was larger than 0.05. At this time, the number of iterative calculation was more than 1 and the ratio of normalization interquartile range to robust standard deviation was larger than 1. In other words, the quartile method broadened the evaluation standard of proficiency testing.