【摘要】 目的检验实验室间测量准确度,探讨评定小规模实验室检测能力的分析方法。方法选择国内4家有资质的实验室,对强筋、中筋和弱筋3类不同品质类型样品8个面粉品质参数进行重复测定。结果不同实验室品质参数存在准确度差异,实验室C中筋样品蛋白质含量和弱筋样品吸水量、实验室D强筋样品吸水量和中筋样品形成时间的重复性标准差为离群值,达到柯克伦检验1%显著水平;实验室B弱筋样品拉伸面积、实验室C中筋样品蛋白质含量测定值为离群值,达到格拉布斯检验1%显著水平。初始能力评价结果表明,实验室B和实验室C出现系统性偏离,其中实验室B在弱筋样品拉伸面积,实验室C在中筋样品蛋白质含量测定结果"不满意"。结论小规模试验检测能力评定,应结合多种统计分析方法判定离群值,初始能力评价采用经典法更为合适,稳健统计法和分割样品设计可作为实验室能力评价参考。更多还原

【Abstract】 Objective The study was to investigate the accuracy and precision of testing results of interlaboratory comparison, and to assess laboratory testing capabilities. Methods Eight quality parameters of a strong gluten flour sample, a medium gluten flour sample and a weak gluten flour sample were tested by four certified laboratories. Results It was showed that the accuracy and precision of testing results of each laboratory was different. The repeatability standard deviations of 2 laboratories showed to be greater than the critical value of Cochran’s test( α=1%), which were laboratory C with protein content of weak gluten sample and water absorption of weak gluten sample, and laboratory D with water absorption of strong gluten sample and development time of medium gluten sample. The test statistics of 2 laboratories were greater than the critical value of Grubbs’ s test( α=1%), which were laboratory B with extension area of weak gluten sample and laboratory C with protein content of medium gluten sample. They also showed to be outlier and unsatisfied, which were analyzed and appraised by adopting classical statistics and robust statistics respectively. Conclusion It is adoptable to use classical statistics for analyzing and appraising the testing results when the data were limited, and to combine with Cochran’s test and Grubbs test method to get rid of the outliers The robust statistics and splite-sample experiment can be used as a reference.更多还原