Abstract: | We compare two distribution-independent methods for determining confidence intervals for the characteristics of systematic and random error and for interquantile intervals of complete error, the methods compared are the bootstrap and L0 methods. We compare the properties of computational difficulty and reliability. Both methods have approximately the same reliability, but the bootstrap methods requires 4–5 times as much work. Translated from Izmiritel'naya Tekhnika, No. 10, pp. 9–12, October, 1996. |