Likelihood ratio statistics, that is, twice the difference of the log likelihoods, are computed between successive models. This type of analysis is sometimes called an analysis of deviance since, if ...