Statistical Procedures for Determination and Verification of Minimum Reporting Levels for Drinking Water Methods
详细信息 查看全文
- 作者：Stephen D. Winslow ; Barry V. Pepich ; John J. Martin ; George R. Hallberg ; David J. Munch ; Christopher P. Frebis ; Elizabeth J. Hedrick ; and Richard A. Krop
- 刊名：Environmental Science & Technology
- 出版年：2006
- 出版时间：January 1, 2006
- 年：2006
- 卷：40
- 期：1
- 页码：281 - 288
- 全文大小：214K
- 年卷期：v.40,no.1(January 1, 2006)
- ISSN：1520-5851
文摘
The United States Environmental Protection Agency'sOffice of Ground Water and Drinking Water has developeda single-laboratory quantitation procedure: the lowestconcentration minimum reporting level (LCMRL). The LCMRLis the lowest true concentration for which future recoveryis predicted to fall, with high confidence (99%), between50% and 150%. The procedure takes into account precisionand accuracy. Multiple concentration replicates areprocessed through the entire analytical method and thedata are plotted as measured sample concentration (y-axis) versus true concentration (x-axis). If the data supportan assumption of constant variance over the concentrationrange, an ordinary least-squares regression line isdrawn; otherwise, a variance-weighted least-squaresregression is used. Prediction interval lines of 99% confidenceare drawn about the regression. At the points where theprediction interval lines intersect with data quality objectivelines of 50% and 150% recovery, lines are dropped tothe x-axis. The higher of the two values is the LCMRL. TheLCMRL procedure is flexible because the data qualityobjectives (50-150%) and the prediction interval confidence(99%) can be varied to suit program needs. The LCMRLdetermination is performed during method development only.A simpler procedure for verification of data qualityobjectives at a given minimum reporting level (MRL) isalso presented. The verification procedure requires a singleset of seven samples taken through the entire methodprocedure. If the calculated prediction interval is containedwithin data quality recovery limits (50-150%), thelaboratory performance at the MRL is verified.