Fitting Fixed- and Random-effects Meta-analysis Models Using Structural Equation Modeling with the Sem and Gsem Commands. (October 2015)

• Record Type: Journal Article
• Authors: Palmer, Tom M.; Sterne, Jonathan A. C.
• Is Part Of: Stata journal. Volume 15:Number 3(2015)
• Page Start: 645
• Page End: 671
• Publication Date: 2015-10
• Subjects: st0398 -- gsem -- meta-analysis -- meta-regression -- metan -- metareg -- mvmeta -- sem -- structural equation modeling; Statistics -- Periodicals; Statistics -- Computer programs -- Periodicals; 001.422
• DOI: 10.1177/1536867X1501500303
• Languages: English
• ISSNs: 1536-867X

Abstract

In this article, we demonstrate how to fit fixed- and random-effects meta-analysis, meta-regression, and multivariate outcome meta-analysis models under the structural equation modeling framework using thesem andgsem commands. While all of these models can be fit using existing user-written commands, formulating the models in the structural equation modeling framework provides deeper insight into how they work. Further, the heterogeneity test for meta-regression and multivariate meta-analysis is readily available from this output.