Can robust statistics aid in the analysis of NAA results?

In this work, which is part of a larger effort to develop a software to automate instrumental neutron activation analysis calculations, the elemental concentration in a sample was calculated using either a set containing only the gamma-ray peaks recommended in the literature or a set containing all peaks identified. The results for each element were reduced using five tools: the usual unweighted and weighted means, plus the limitation of statistical weight, Normalized Residuals and Rajeval. The results were compared to the certified value for each element, allowing for discussion on the performance of each statistical tool and on the choices of peaks. (author)