STP103: Use of Statistics to Determine Precision of Test Methods

    Youden, W. J.
    National Bureau of Standards, Washington, D. C.

    Cameron, J. M.
    National Bureau of Standards, Washington, D. C.

    Pages: 8    Published: Jan 1950


    Abstract

    Some recent studies of A.S.T.M. Method D 877 - 46 T2 for the Determination of the Dielectric Streng h of Insulating Oils provide a good illustration of the types of statistical problems encountered in the study of a test method.

    These measurements are an example of a case of non-homogeneous measurements. An exposition is given of a method of taking repeated differences in a set of observations. By this statistical technique the errors of measurement may be isolated from the over-all dispersion which includes sampling and possibly other components. It is pointed out tha this may also be done by the analysis of variance technique. This latter technique is very useful in ascertaining whether data can be considered homogeneous.

    Several proposed methods for making multiple measurements on a sample of oil are compared.


    Paper ID: STP44979S

    Committee/Subcommittee: E11.03

    DOI: 10.1520/STP44979S


    CrossRef ASTM International is a member of CrossRef.