| “z值评价-风险评估-技术核验”统计评价模型用于低样本量能力验证——以水质监测为例 |
| The Implementation of a Statistical Evaluation Model "z-Score Evaluation-Risk Assessment-Technical Verification" for Proficiency Testing with Low Sample Size:Taking Water Quality Monitoring as an Example |
| 投稿时间：2022-06-30 修订日期：2023-01-29 |
| DOI：10.19316/j.issn.1002-6002.2024.01.07 |
| 中文关键词: 能力验证 低样本量 指定值 风险评估 |
| 英文关键词:proficiency testing low sample size assigned value risk assessment |
| 基金项目:一带一路国际科学组织联盟项目(ANSO-CR-KP-2020-05);中国科学院关键技术人才项目(E1Y1880101) |
|
| 通讯作者:李红岩* 中国科学院生态环境研究中心环境水质学国家重点实验室, 北京 100085 |
| 摘要点击次数: 844 |
| 全文下载次数: 299 |
| 中文摘要: |
| 为解决低样本量下的能力验证z比分数评价结果可能偏离参加者真实能力水平的问题,建立了"z值评价-风险评估-技术核验"统计评价模型,以"一带一路"共建国家"水中铁和氟化物的检测"国际能力验证项目为例进行了验证。结果显示:样本量为14家(低样本量)、指定值与样本真值相对误差的绝对值大于4%时,铁检测项目获得"满意"评价结果的14家参加者中,5家存在数据"不准确"中/高风险;氟化物检测项目获得"满意"评价结果的11家参加者中,7家存在数据"不准确"中/高风险。技术核验反映出的突出问题是参加者未按照检测方法的要求实施质量控制措施,标准曲线绘制、试剂配制、关键仪器参数设置记录缺失,原始记录可追溯性较差。由此证实了数据风险点的存在,同时也印证了该统计评价模型的可靠性和必要性。 |
| 英文摘要: |
| In this paper,a statistical evaluation model of "z-score evaluation-risk assessment-technology verification" was established to reduce the deviation caused by low sample size when using z-score for the evaluation of proficiency testing from the true level of participants’ results.Taking an example of the international proficiency testing project of "Iron and Fluoride Detection in Water" for the countries along the Belt and Road,the developed model was further verified.The results showed that when the number of testing results was only 14 (low sample size) and the absolute value of the relative error between the assigned value and the standard value of the testing sample was over 4%,five participants within 14 participants who had obtained satisfactory evaluations on the iron item were at medium/high risk of inaccurate.Among 11 participants who had received satisfactory evaluation on the fluoride item,7 participants were at medium/high risk of inaccurate.The typical problems revealed by the technical verification were that the quality control process was implemented without following the requirements of detection methods,and the records of standard curves,reagent preparation,and key parameters of instrument were lack.It was hard for the traceability of testing records.Both the existence of risk when z-score evaluation used for proficiency testing with low sample size and the reliability and necessity of the statistical evaluation model was further verified. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
