STP1075: Detection Limits: For Linear Calibration Curves with Increasing Variance and Multiple Future Detection Decisions

    Gibbons, RD
    Associate Professor of Biostatistics, University of Illinois at Chicago, Chicago, IL

    Jarke, FH
    Manager of Quality Programs and Director, Waste Management Environmental Monitoring Laboratory, Geneva, IL

    Stoub, KP
    Manager of Quality Programs and Director, Waste Management Environmental Monitoring Laboratory, Geneva, IL

    Pages: 14    Published: Jan 1991


    Abstract

    With increasing concern over chemicals that are potential health hazards at low levels, determination of detection limits has undergone considerable scrutiny. Most traditional detection limit estimators suffer from one or more statistical and/or conceptual limitations [1]. In this paper, a new detection limit estimator is proposed, that builds upon the previous work of Clayton and co-workers [1]. The proposed detection limit is computed directly from calibration data, and specifies both false positive and false negative rates (i.e., Type I and Type II error rates) at fixed levels. The new detection limit estimator extends previous work by permitting variability in the instrument response to be proportional to concentration, and by incorporating multiple future detection decisions, where the exact number is large and potentially unknown. Previous methods have only considered the problem of a single future detection decision, and have assumed that variability in the response signal is constant throughout the range of the calibration curve, or can be suitably transformed to bring about this condition. The new method is illustrated using calibration data for 10 volatile organic priority pollutant compounds.

    Keywords:

    method detection limit, weighted least squares, linear calibration, tolerance intervals, prediction intervals


    Paper ID: STP25492S

    Committee/Subcommittee: D18.95

    DOI: 10.1520/STP25492S


    CrossRef ASTM International is a member of CrossRef.