Some Statistical Aspects of Bias Corrections

An important practical question is whether bias corrections should always be made to a measurement process or should be made only when a test of the measurements of a standard indicates a statistically significant, non-zero bias. Some statistical aspects of this question are studied. The results are expressed in comparative terms, as ratios of mean-square deviations for the two procedures. It is assumed that the standard deviations involved with the measurement process and the standard are known.