Relative bias is an indicator of how closely a reported measurement corresponds to the known tritium activity of a urine sample. It is calculated in the following manner. First, each observation for a spiked sample is corrected for the appropriate urine blank analyzed the same day. The result obtained is represented by yjkl where:
j=the laboratory; j=1, 2, ..., n
k=the spike; k=1, 2, ..., 3
and l=the aliquot; 1=1, 2, ..., 5
The relative bias of the result for each aliquot (bjkl) of a particular spike is calculated thus:
where ak=the known value for a particular spike.
The relative bias (bjk) for the mean of measurements on all the aliquots of a given spike is then obtained as follows:
where L=the number of replicate measurements (=5, in this intercomparison).
In this program, the standard deviation of the bias is taken as the measure of the reproducibility of an analysis, and is referred to as "relative precision" in this report. It is calculated as follows: