Developing new models and methods for meta-analysis

Meta-analysis is a statistical technique to synthesize the research findings from multiple studies for decision making. It has been increasingly popular in recent years, mainly due to its wide application in evidence-based practice. Also in the past several years, the PI and collaborators have made some significant progress in advancing statistical methods for meta-analysis, in particular for the newly developed methods for data transformation that are capable to serve as “rules of thumb” in meta-analysis and evidence-based practice.

The main purpose of this proposal is to further advance the literature by developing new models and methods for meta-analysis. Specifically, there are three major projects in the proposal. The first project is to develop a unified framework for simultaneous synthesis of the odds ratio (OR) and the risk ratio (RR) in meta-analysis with binary outcomes. The second project is to propose an Intrinsic measure for Quantifying the heterogeneity in meta-analysis, referred to as the IQ statistic, to overcome the limitations in the I^2 statistic. The third project is to develop new estimation and model selection methods for the fixed-effects model (FEM) with application to meta-analysis with few studies. We plan to accomplish, but not limited to, the proposed three projects in a three-year research plan. All three projects are of particular importance, and upon a successful completion, we expect that the new models and methods from our proposed studies will have potential to be widely applied in meta-analysis and evidence-based practice