Combining Estimates of Measurement Precisions from Different Experiments

Commonly, estimates of measurement precisions are derived from several sources. These may include data routinely generated by an ongoing measurement control program, data from experiments conducted to investigate specifically the effects of certain variables on total errors of measurement, shipper-receiver differences data, data from participation in inter-laboratory experiments, data derived from inspections by audit groups, etc. In the calculation of limits of error for material unaccounted for and other accountability quantities, given values of the measurement precisions are used based on all available pertinent data sources. The problem of how to combine the information from all these data sources to arrive at overall estimates of the precisions of interest is considered in this paper.