Extension of Grubbs' Method When Relative Biases Are Constant

The problem of estimating errors of measurement when two or more measurement "methods" are used to measure each of n items has been considered in previous INMM Journal articles [1], [2], [3], [4]. The term "methods" is used in a general sense and may apply to different analytical techniques, different laboratories, etc. In the example in [3], the methods referred to six different scales in a fuel fabrication plant, while in the example in [5], the different methods consisted of two laboratories with several individuals performing measurements of pellet densities using either the geometric or immersion technique. In [6], the example is concerned with the measurement of stress at 600% elongation of seven rubber samples as measured by 13 laboratories, using data reported by Mandel [7].