ABSTRACT
Most broadly defined, meta-analysis includes any methodology for combining information across sources. Of most interest to statisticians are quantitative approaches to summarize all relevant information pertaining to a research question. Introductory texts on the subject include Refs 1-8. Each of these books focuses on methods for deriving a common estimate of effect. In contrast, Ref. 9 emphasizes the need to move this focus to one of quantifying and reporting the heterogeneity between studies. This message reflects the methodological development seen in the statistics field during the last decade. Statistical methodology for meta-analysis has been moving away from approaches focused only on fitting a common estimate toward approaches that include estimating the extent and sources of heterogeneity among studies. This chapter will review this progression, highlight complexities encountered, summarize how the chapters in the present volume address heterogeneity between studies, and suggest future directions for methodological development.
