Skip to content Skip to navigation

Bootstrapping P-Values of F-Divergence Tests

Volume 8, Number 1 (2009), 1 - 15

Bootstrapping P-Values of F-Divergence Tests

Price: $20.00

Abstract

In this paper, we review some of the basic ideas of bootstrap inference. The smallsample behavior of the bootstrap is investigated as a method for estimating P-values in parametric model, with a new approach based on divergence types measures in testing statistical hypotheses. Under the relatively weak conditions needed for a test statistic to be asymptotically pivotal, bootstrap tests should perform better than asymptotic tests. Monte Carlo methods are used to examine the bootstrap test of f¡divergence statistics and the results suggest that the bootstrap can accurately estimate the rejection probabilities. In this case, bootstrap tests work well, and very few steps are needed.