Meta-Analysis and Hierarchical Models in Data Analysis and Prior Construction

Sala 6.4.30, FCUL, Lisboa

Por Prof. Robert Weiss (Dep. Biostatistics, Fielding School of Public Health, University of California at Los Angeles - UCLA)

Meta-Analysis is among the simplest of hierarchical models. Among the uses for meta-analysis and meta-analytic ideas is in prior construction. I detail some uses of meta-analysis for prior construction in three (time permitting) situations where (1) In a real sense there is no more data to be collected, (2) When you can't run the entire hierarchical model that you'd like to run, which might happen if computing resources are insufficient or different subsets of the data are stored in different locations and can not be released and (3) In a pharmaceutical example where proper priors give better frequentist behavior than the frequentist analysis.

CEAUL - Centro de Estatística e Aplicações da Universidade de Lisboa