Asymptotic normality for the K-phi-divergence goodness-of-fit tests

Abstract

In this paper for a wide class of goodness-of-fit statistics based Ko-divergences, the asymptotic normality is established under the assumption n/m(n) --> a is an element of (0, infinity), where m(n) denotes sample size and Mn the number of cells. This result is extended to contiguous alternatives to study asymptotic efficiency.