Department of Statistics, UCLA

Statistical inference is an important feature of meta-analysis. Estimation is often a central goal, with hypothesis tests and confidence intervals used to address uncertainty. Expositions of meta-analysis make statistical inference a major theme. Indeed, a significant motivation for meta-analysis can be improving the precision of the estimates produced and increasing the power of any hypothesis tests. In the pages ahead, the use of statistical inference in meta-analysis will be examined. The intent is to consider the statistical models employed and the data with which they are used. Building on some previous work, a key issue will be whether the data were produced by mechanisms that the models require. The paper begins by describing some popular meta-analysis models. An assessment of their use follows.