Likelihood ratio when the difference in the log-likelihood is a large negative number
Using the Geyer-Thompson algorithm, you can compute the ratio of two likelihoods (needed to do a likelihood ratio test or to compute an information criterion like AIC or BIC). This is useful when comparing two hierarchical models using data cloning, which uses the Bayesian machinery to compute ML estimates. However, you need to compute the average of the ratio on the data rather than the log scale.
Likelihood ratio when the log-likelihood is a large negative number
Using the Geyer-Thompson algorithm, you can compute the ratio of two likelihoods (needed to do a likelihood ratio test or to compute an information criterion like AIC or BIC). This is useful when comparing two hierarchical models using data cloning, which uses the Bayesian machinery to compute ML estimates. However, you need to compute the average of the ratio on the data rather than the log scale.