Incorporating Effect Size Variation into Meta-Analytic Structural Equation Modeling

Scholars are increasingly recognizing the potential of meta-analytic structural equation modeling (MASEM) as a way to build and test theory (Bergh et al., 2014). Yet, as MASEM is currently applied, it disregards variability in estimates of true score correlations. Specifically, evidence of moderation identified when establishing the bivariate correlations in random effect meta-analysis is disregarded when later inputted into multivariate MASEM. This practice means that MASEM results may not be as stable or generalizable as MASEM scholars often conclude. We first examine the potential problems of existing point-estimate MASEM techniques in a simulation study. Then, we introduce a bootstrap MASEM technique that retains both the estimates of true score relationships and the variability surrounding those relationships. Through a simulation and re-analysis of published MASEM studies, we demonstrate how this procedure assists researchers in establishing overall model fit and path estimates as well as establishi...