Skip directly to search Skip directly to A to Z list Skip directly to page options Skip directly to site content

NIOSHTIC-2 Publications Search

Search Results

Reconciling traditional accuracy assessment with the ISO Guide to the Expression of Uncertainty in Measurement (ISO/GUM).

Authors
Ashley-K; Bartley-DL
Source
J Occup Environ Hyg 2004 Apr; 1(4):D37-D41
NIOSHTIC No.
20031630
Abstract
The past five years have seen a move towards standardizing the documentation of measurement uncertainty through nearly worldwide adoption of the International Organization for Standardization Guide to the Expression of Uncertainty in Measurement (ISO/GUM).(1-5) The question then naturally arises as to how to interpret traditional accuracy assessments of measurement methods for industrial hygiene applications.(6-8) Secondly, how are uncertainty problems specific to industrial hygiene to be handled if not covered in detail within the ISO/GUM? These problems within workplace measurements may be summarized as follows: (1) The expanded uncertainty U is defined in the ISO/GUM so as to bracket measurand values. Within the ISO/GUM, U depends generally on the student-t distribution. A student-t variable is a ratio of a normally distributed variable to a chi-square variable (such as an estimate s2 of standard deviation squared). Both numerator and denominator in the ratio must vary to define the student-t variable. With workplace measurements, however, the denominator (s2) is not remeasured (i.e., does not vary) at each practical method application. This is because workplace air concentrations vary spatially and over time to such a degree(9) that cost precludes method evaluation (i.e., by taking replicate measurements). Variability in terms of s2 is generally measured only once in an extensive method validation, relying on adequate quality control to maintain stability of a given method. Specifics in terms of prediction in such a practice are required. (2) Several useful methods have unknown, yet controlled or minimized, residual bias or systematic error. That is, the measured value is off from measurand values by a constant amount in all applications of the methods. Details, again in terms of prediction, are needed.
Keywords
Statistical-analysis; Epidemiology; Analytical-processes
Publication Date
20040401
Document Type
Journal Article
Fiscal Year
2004
NTIS Accession No.
NTIS Price
Issue of Publication
4
ISSN
1545-9624
NIOSH Division
DART
Source Name
Journal of Occupational and Environmental Hygiene
State
OH
TOP