Year
1972
Abstract
The determination of random error components in a measurement process is frequently accomplished by a statistical testing procedure known as a nested design. Where analytical costs are a significant consideration, the standard nested design may require excessive sampling at the lowest levels and too little sampling at the highest levels. One method of solving this problem is by selective \"pruning\" of the lower levels of the design to retain the symmetrical balance in the design, but to re-allocate the samples in such a way that improved estimates of measurement errors for the higher levels may be obtained by sacrificing some of the precision available at the lower levels. A four-level balanced staggered nested design is described. The test was used to evaluate the measurement error of plutonium determinations in a mixed oxide.