Definition: cAIC

Category: Memobust Glossary

Definition 1: As model selection measure, cAIC is well -suited for small area estimation. It is relevant to inferences regarding the clusters, or areas, in the context of linear mixed models, inferences regarding the clusters, or areas, in the context of linear mixed models. The criterion is  based on the conditional likelihood for fixed and random effects vectors evaluated at their estimated values, and y is the data. The effective number of degrees of freedom is essentially given by the trace of the hat matrix H.

Definition 2: As model selection measure, cAIC is well-suited for small area estimation. It is relevant for inferences regarding clusters, or domains, in the context of linear mixed models. The criterion is cAIC = 2peff – 2log(lik), where lik is maximum values assumed by the conditional likelihood, that is the likelihood function when fixed and random effects vectors evaluated at their estimated values. The effective number of degrees of freedom peff is essentially given by the trace of the hat matrix H. https://ec.europa.eu/eurostat/cros/content/memobust-glossary-pdf-file_en