The contribution of individual correlated parameters to the uncertainty of integral quantities

Modern libraries of multigroup cross-sections often include covariances of the multigroup data. Data covariances are especially useful in calculating the variance of an important applied quantity z that can be calculated as a function of the data. Frequently, it is also of interest to know the magnitude of the contributions to this variance arising from various subsets of the data. This knowledge can be used, for example, in judging the benefits of performing specific new measurements, calculations, or evaluations on a given data subset. Here we describe a method for determining the contribution of an identified set of parameters to the variance of z that can be applied without difficulty even when the data have general correlations.