Statistical methods in sterile neutrino experiments

Abstract

We compare the current statistical methods applied to short-baseline neutrino oscillation experiments looking for light sterile neutrinos. The interpretation of the reconstructed confidence regions differs because of the different definition of the alternative hypothesis in the hypothesis test. When the alternative hypothesis extends over the whole physical allowed parameter space of the oscillation parameters, proper confidence regions are obtained both in case of a limit and a signal. Since the different definition of the alternative hypothesis prevents from a direct comparison of the results, a standardized analysis is proposed.