M. A. Negrín Hernández, E. Moreno, M. Martel Escobar, F. J. Vázquez Polo
In meta-analysis, the between-sample heterogeneity introduces model uncertainty that should be incorporated into the inference. We argue that the way of measuring the between-sample heterogeneity is by clustering the
samples and finding the posterior probability of the cluster models. The meta-inference is then obtained as a mixture of the meta-inferences for the cluster models, where the mixing distribution is the posterior model
probabilities. Although this topic has been ignored in the meta-analysis literature, the inference is sensitive to the cluster structure of the samples. We focus in very common practical situations where few studies and sparse data are present.
Illustrative examples with real data are analyzed and compared with previous meta-analyses.
Palabras clave / Keywords: meta-analysis, sparse data, clustering, model selection
Programado
Sesión GT08-1a Inferencia Bayesiana-1 (Parte 1): New Insights on the Role of the Bayesian Thinking in Model Selection Problems (BAYES-1a). Organizador: Gonzalo García-Donato
29 de mayo de 2018 10:50
Sala 3
