Meta-Analysis: A Structural Equation Modeling Approach

Chapter 7Meta-analytic structural equation modeling

This chapter covers meta-analytic structural equation modeling (MASEM), a technique that combines meta-analysis and structural equation modeling (SEM) to synthesize correlation or covariance matrices and to fit structural equation models on the pooled correlation (covariance) matrix. We begin this chapter with a discussion on the need to synthesize existing research findings when SEM is applied as the methodology in the primary studies. MASEM is then proposed as a statistical method to synthesize these research findings. Conventional methods based on the univariate approaches and the generalized least squares (GLS) approach are briefly reviewed. The fixed- and the random-effects two-stage structural equation modeling (TSSEM) are introduced in details. Issues related to conducting MASEM are discussed. Several examples are used to illustrate the procedures in the R statistical environment.

7.1 Introduction

As introduced in Chapter 2, SEM is a popular statistical technique to test hypothesized models in the social, educational, and behavioral sciences. What makes SEM so popular in applied research is that theoretical models can be translated into a set of interrelated equations involving latent and observed variables. The proposed models can be path models, confirmatory factor analytic (CFA) models, or general structural equation models. The proposed models can be empirically tested by the use of a likelihood ratio (LR) statistic ...