Advances and Applications in Statistics
Volume 7, Issue 3, Pages 341 - 356
(December 2007)
|
|
A META-ANALYSIS FRAMEWORK BASED ON HIERARCHICAL MIXTURE MODEL
A. Dimitri (Italy) and M. Talamo (Italy)
|
Abstract: Despite its widespread and growing use, meta-analysis continues to be a controversial set of not well-defined techniques to treat a set of well-known problems (publication bias, heterogeneous data and results, missing data, etc.). This study proposes a statistical framework that starts from the characterization of meta-analysis as the research of the maximum number of shared and large concepts. This research is made using hierarchical mixture models. In this general framework all the above well-known problems assume a new exact characterization. |
Keywords and phrases: meta-analysis, Bayesian meta-analysis, mixture models, statistical learning. |
|
Number of Downloads: 360 | Number of Views: 1342 |
|