Evaluation of the Efficacy of Meta-Analytic Structural Equation Modeling with Missing Correlations. Issue 3 (3rd May 2020)

• Record Type: Journal Article
• Authors: Lv, Jing; Maeda, Yukiko
• Is Part Of: Structural equation modeling. Volume 27:Issue 3(2020)
• Page Start: 414
• Page End: 437
• Publication Date: 2020-05-03
• Subjects: W-COV GLS with PD -- W-COV GLS with MI -- TSSEM -- missing correlations; Multivariate analysis -- Periodicals; Social sciences -- Statistical methods -- Periodicals; 519.535
• DOI: 10.1080/10705511.2019.1646651
• Languages: English
• ISSNs: 1070-5511

Abstract

Abstract : Researchers often face problems with missing correlations when applying meta-analytic structural equation modeling (MASEM). However, methodological investigation regarding the performance of MASEM paired with a large proportion of missingness is limited. This study investigated the impacts of missing conditions on the performance of correlation-based MASEM, utilizing weighted-covariance generalized least squares (W-COV GLS) with pairwise deletion (PD) and/or multiple imputation (MI), and two-stage structural equation modeling (TSSEM), with two factorial models. The results show the superiority of W-COV GLS with MI and the necessity of including studies with complete correlation matrices for TSSEM and W-COV GLS with PD. We identified the boundary conditions in which each approach performs satisfactorily and illustrated how the variation of factor loadings impacts the performance of correlation-based MASEM.