Correlation-Based, Parameter-Based, or Fixed-Effect? A Comparison of One-Stage Meta-Analytic Structural Equation Modeling Approaches

Meta-analytic structural equation modeling (MASEM) is increasingly used in psychological research to evaluate multivariate models, such as path or factor models, across multiple primary studies. Several approaches to MASEM have been proposed that differ in whether and where they represent between-study heterogeneity, most notably the fixed-effect, correlation-based, and parameter-based approaches. This study investigates under which conditions these approaches target equivalent or divergent structural parameter estimates. We derive analytical conditions for equivalence and divergence and examine them in three simulation studies comparing one-stage implementations under parameter-based data-generating mechanisms. The results show that the approaches produce approximately equivalent estimates when the model-implied correlations depend linearly on the heterogeneous parameters. However, multidimensional, correlated parameter heterogeneity produces target discrepancies between the parameter-based and the other two approaches. These discrepancies increase with the magnitude of between-study heterogeneity. Nonlinear parameter transformations introduce additional discrepancies, even under unidimensional heterogeneity. Under the standard implementations used, almost no differences emerged between the fixed-effect and the correlation-based population targets. Two applications to psychological research further illustrate the model comparison. Overall, the findings show that the location of random effects can affect the target of structural parameter estimation in MASEM even when the study-specific structural models are correctly specified. It follows that the choice among the approaches constitutes a substantive decision for researchers between distinct statistical estimands: the mean of study-specific structural parameters (parameter-based MASEM) and a common parameter vector that best represents the mean correlations across studies (fixed-effect and correlation-based MASEM).

Authors

Publication Details

Journal
PsyArXiv (OSF Preprints)
Published
2026-11-15
DOI
https://doi.org/10.31234/osf.io/5evt3
Primary Topic
Meta-analysis and systematic reviews
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Correlation-Based, Parameter-Based, or Fixed-Effect? A Comparison of One-Stage Meta-Analytic Structural Equation Modeling Approaches

Julian F. Lohmann, Alexander Robitzsch, Oliver Lüdtke
PsyArXiv (OSF Preprints)
Meta-analysis and systematic reviews
preprint

Correlation-Based, Parameter-Based, or Fixed-Effect? A Comparison of One-Stage Meta-Analytic Structural Equation Modeling Approaches

Julian F. Lohmann, Alexander Robitzsch, Oliver Lüdtke
preprint en

Abstract

Meta-analytic structural equation modeling (MASEM) is increasingly used in psychological research to evaluate multivariate models, such as path or factor models, across multiple primary studies. Several approaches to MASEM have been proposed that differ in whether and where they represent between-study heterogeneity, most notably the fixed-effect, correlation-based, and parameter-based approaches. This study investigates under which conditions these approaches target equivalent or divergent structural parameter estimates. We derive analytical conditions for equivalence and divergence and examine them in three simulation studies comparing one-stage implementations under parameter-based data-generating mechanisms. The results show that the approaches produce approximately equivalent estimates when the model-implied correlations depend linearly on the heterogeneous parameters. However, multidimensional, correlated parameter heterogeneity produces target discrepancies between the parameter-based and the other two approaches. These discrepancies increase with the magnitude of between-study heterogeneity. Nonlinear parameter transformations introduce additional discrepancies, even under unidimensional heterogeneity. Under the standard implementations used, almost no differences emerged between the fixed-effect and the correlation-based population targets. Two applications to psychological research further illustrate the model comparison. Overall, the findings show that the location of random effects can affect the target of structural parameter estimation in MASEM even when the study-specific structural models are correctly specified. It follows that the choice among the approaches constitutes a substantive decision for researchers between distinct statistical estimands: the mean of study-specific structural parameters (parameter-based MASEM) and a common parameter vector that best represents the mean correlations across studies (fixed-effect and correlation-based MASEM).

PsyArXiv (OSF Preprints)
Meta-analysis and systematic reviews
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon