Managing directorMember of ASTM, Vortek Systems, Ltd., Edmonton, Alberta
The detection of bias depends on the precision of the sampling procedure that is employed to measure it. The significance of detected bias is therefore a relative matter; nonperformance of the bias test gives a one-sided answer only as smaller bias may exist. In fact, all measurements can be said generally to produce answers that are to some degree affected by systematic error. In this paper three different aspects are treated as starting points for formulating a standard method for bias: (1) the conditions under which bias occurs in the sampling of broken coal and factors that contribute to its control, (2) the means whereby bias can be reduced, and (3) the discussion of two procedures for the testing of bias. Attention is drawn to the terminology and the definitions that are related to the reduction and testing of bias, with special reference to the need for finding a common terminology in which the basic principles of probability are anchored down for public use regardless of the field of endeavor to which it is to be applied.
Paper ID: JTE11194J