迭代算法A与Q/Hampel方法在能力验证结果分析中的比较

能力验证计划实施中,要按照科学的统计方法分析能力验证参加者的测定结果,并给出公平合理的评价。通常优先采用对离群值相对不敏感的稳健统计方法进行能力验证结果分析。通过实际检测数据计算实例,对比了迭代算法A与Q/Hampel两种稳健统计方法的计算结果。研究表明:两种稳健统计方法计算的稳健平均值基本相同,计算的稳健标准差存在区别,但对各实验室检测能力评价一致。能力验证结果中存在局部众数时,迭代算法A和Q/Hampel方法计算所得稳健平均值比值为1.03,而两者稳健标准差比值达到了2.32,Q/Hampel方法比迭代算法A表现出更好的稳健可靠性。

Comparison of iteration algorithm A and Q/Hampel method used in result analysis of proficiency testing

Scientific statistics methods have been used in analysis of measured results submitted by proficiency testing participants for reasonable evaluation. The robust statistics methods that are resistant to outliers are generally preferred. The computation results of two robust statistics methods, iteration algorithm A and Q/Hampel, are compared based on computation examples of actual test data. The computation results show that the mean of robust statistics methods are basically consistent, while there is difference in the robust standard deviation between iteration algorithm A and Q/Hampel, then the judgement to laboratory testing ability is consistent. When it comes to minor modes, the ratio of the robust mean of iteration algorithm A to the robust mean of Q/Hampel is 1.03, while the ratio of robust standard deviation between these two methods is 2.32, which means Q/Hampel method shows better robust reliability for the resistance to minor modes.