4.3.1.3 Relative standard deviation of replicate samples

The reproducibility of ISM replicates collected from a DU can be evaluated in terms of RSD, also known as the coefficient of variation (CV), which is the SD divided by the mean. Because both RSD and CV are commonly used, both are used interchangeably in this document. Although included as a performance metric in the simulation studies, the RSD does not provide an indication of the accuracy of the estimate of the mean or 95% UCL. Figure 4-6 illustrates the distinction between reproducibility (or precision) vs. bias, which, taken together, represent accuracy. For example, a low RSD indicates the estimates are precise. The values might be reproducible but still yield a biased estimate of the mean and corresponding UCL.

Four possible relationships between bias and precision.
Figure 4-6. Four possible relationships between bias and precision.