四分位法和迭代法在建材检测能力验证中的比较

打开文本图片集

Abstract： Proficiency testing is to determine the testing or calibration capabilities of laboratories by using the

measurementcomparisonand evaluationresultsamong laboratories，andis widelyused inthe assessment，supervision

andconfirmationof thecapabilities ofvarious laboratories.In terms of proficiency testing，the statistical analysis

techniques of experimental data and the evaluation methods of reference values play a crucial role in objectively

evaluatig the testing capabilities of laboratories.At present， the main recommended robust statistical methods in the standardsarethe interquartile methodand the iterative method.In this paper，two methodsareused tocalculate the specifiedvalues and standarddeviationsof the building materialtesting proficiencytestdata inmultipledistributions. Combined with the Shapiro-Wilk normaltestof the data by SPSS software，thecalculation resultsof the two robust statisticalanalysis methods are comparedandanalyzed.The summaryshows thatwhen the distributionof the detection dataisconcentrated （the robust coeffcient of variationisless than 5% ），thedetectioncapabilitiesofeachlaboratory differlitte.Overallthesatisfactionratescalculatedbythetwomethodsaresimilar，andtherobuststandarddeviations calculatedby thequartile method are mostlysmaller than those calculated bythe iterative method.It is notable that when thesignificance level PoftheShapiro-Wilk normal testislessthan orequal toO.05andthekurtosis is too large， thesatisfactionratecalculatedbythequartilemethodissignificantlylower thanthatcalculatedbytheiterativemethod， and the deviation of the satisfaction rates obtained by the two methods is large.

Keywords： quartile method; iterative method; proficiency testing; robust statistics

0引言

能力验证是国际通行的实验室检测报告/证书有效性评价的手段，其不仅可以判断实验室是否具备检测/校准的能力，还可以识别同行机构之间的差异，增加实验室间的交流学习，为提升行业的检测能力及质量管理水平提供了信息支持。（剩余5484字）