Likelihood Ratio Tests

Abstract The likelihood ratio test is a widely used procedure for testing hypotheses. It rejects the null hypothesis when the maximum likelihood under the null hypothesis is significantly smaller than the maximum likelihood under the alternative hypothesis. In some situations, itsPvalue can be calculated exactly, but, in general, thePvalue must be approximated, usually by using a chi‐square table. Three examples are presented. Methods have been, and continue to be, developed for improving the approximation.