J. Padilla Fernández-Vega, B. González Pérez, L. Sanz San Miguel
We develop a Bayesian procedure to test the independence of two factors in r × s contingency tables. The approximation is based on calculating the posterior probability of the null hypothesis using a mixed prior distribution. Using simulated and real data, we compare the decisions reached by the Bayesian and frequentist approaches using the posterior probability and the usual p −value. This comparison has been developed through heatmap between the different values of the posterior probability and the p- value. To do this, data analysis is performed over r × s tables with a fixed sample size.
Palabras clave: Independence, contingency table, posterior probability, p-value, heatmap
Programado
L06.2 Grupo de Inferencia Bayesiana: Bayesian model selection in Classification and Clustering
5 de septiembre de 2016 12:55
0.09 - Aula de proyectos 2
