Wednesday 16 Oct 2013: Bayesian multivariate meta-analysis
Yingui Wei - Plymouth University
Plymouth University, Fitzroy 208 14:50-16:00
Multivariate meta-analysis is an alternative to standard univariate meta-analysis. It allows synthesizing effect sizes from multiple endpoints and multiple studies simultaneously. Recently there are increasing developments in statistical methods and packages for multivariate meta-analysis. A Bayesian approach, which is flexible and implementable, has been investigated less extensively than frequentist methods. I will briefly review the existing application areas of the Bayesian multivariate meta-analysis and discuss what might be the challenges. I will present the methods and application of Bayesian multivariate meta-analysis to situations when there are multiple correlated outcomes, using a dataset from a systematic review.