STP103

    Precision and Accuracy of Test Methods

    Published: Jan 1950


      Format Pages Price  
    PDF Version (264K) 14 $25   ADD TO CART
    Complete Source PDF (836K) 14 $55   ADD TO CART


    Abstract

    All A.S.T.M. committees make use of data which have been obtained by the use of physical or chemical test methods. It is important to know something about the reliability, the precision, and the accuracy of these methods.

    If a test method gives results which are erratic and unpredictable, it cannot be considered to be reliable. The reliability of a test method can be studied by planning a series of determinations so that the individual results can be grouped in some logical way. The variability within these small groups can then be used as a yardstick to determine whether all assignable causes of variation have been properly controlled. It is a waste of time to try to evaluate the precision and accuracy of a test method until it has been shown that the method is reliable.

    The precision of a test method is a measure of how variable the method is when used under well-controlled conditions. It is convenient to express precision by means of the standard deviation of individual measurements. The average range within small groups of test measurements can be used to estimate the standard deviation.

    A test method can be considered accurate when it will give measurements on a sample of test material which do not differ from the known value by more than can be explained by the random variations of the method.

    The results of properly designed experiments to establish the reliability of a test method can be profitably studied by the use of simple statistical control charts. Test method precision and accuracy can also be evaluated with the help of this technique.


    Author Information:

    Wernimont, G.
    Eastman Kodak Co., Rochester, N. Y.


    Paper ID: STP44978S

    Committee/Subcommittee: E11.20

    DOI: 10.1520/STP44978S


    CrossRef ASTM International is a member of CrossRef.