Generalized Frequentist Methods for Calculating P-Values and Confidence Intervals

May 16, 2006

Generalized frequentism addresses problems that are not exactly solvable using conventional frequentism. Such problems include the calculation of p-values and confidence intervals when nuisance parameters are present, or when interest is focused on a complicated function of the parameters of the model under consideration. Although generalized frequentist methods are based on exact probability statements, they do not necessarily yield coverage in the conventional sense. However, simulation studies indicate that these methods tend to overcover, and often surpass other available methods in terms of test power or interval length.

This paper was published in “Proceedings of PHYSTAT05: Statistical Problems in Particle Physics, Astrophysics and Cosmology, Oxford, UK, 12-15 September 2005,” edited by Louis Lyons and Müge Karagöz Ünel, World Scientific, Singapore (2006), pp. 7-10.

Download it here.