Meta-Analytic SEM (MASEM) Services
Meta-analytic structural equation modeling tests an entire theoretical model on the accumulated evidence of a field—pooling correlations across many studies, then fitting a path or factor model to the synthesised matrix. It answers questions no single study, however large, was designed to answer.
Meta-analytic structural equation modeling (MASEM) combines meta-analysis with structural equation modeling. It pools correlation (or covariance) matrices across many studies into a synthesised matrix, then fits an SEM—a path or factor model—to that pooled matrix. This lets researchers test a full theoretical model using the entire body of evidence, not one sample.
What meta-analytic SEM does
A single study can test a theoretical model—a mediation chain, a set of structural paths, a factor structure—but only on its own sample, with its own limitations. Meta-analytic SEM asks a more ambitious question: does the model hold across an entire literature? It brings together two of the most powerful tools in quantitative research—meta-analysis and structural equation modeling—to test a full model on accumulated evidence.
It proceeds in two conceptual steps. First, it pools the correlations among the variables across all the studies that reported them, producing a single synthesised correlation matrix that represents the whole field. Second, it fits a structural equation model—paths, mediation, or a measurement model—to that pooled matrix, and evaluates how well the theoretical model fits the combined evidence. The result is a test of theory with the statistical power of many studies and generalisability that no single sample can offer.
When to use it
MASEM is the right tool when you have a theoretical model involving several variables—typically three or more—and a body of primary studies that report the correlations among them, even if no single study measured all of them together. It is especially valuable for testing mediation and competing models across a literature, for establishing whether a structural theory generalises, and for resolving debates where individual studies disagree.
It is widely used in management, applied psychology, and organisational research—fields rich in correlational studies of multi-variable theories. If your question concerns a single bivariate relationship, ordinary meta-analysis is enough; MASEM earns its complexity when the question is about a system of relationships and whether a model of that system holds across the evidence.
SEM in one study vs meta-analytic SEM
| SEM in a single study | Meta-analytic SEM | |
|---|---|---|
| Evidence base | One sample | Many studies pooled |
| Power | Limited by sample size | Large, from combined evidence |
| Generalisability | To that population | Across the literature |
| All variables measured together? | Required | Not required (pooled matrix) |
| Answers | Does the model fit here? | Does the model hold across the field? |
Two-stage MASEM—done properly
The current standard is two-stage SEM (TSSEM). Stage one pools the correlation matrices across studies—correctly, using a multivariate meta-analytic model that accounts for the dependence among correlations and tests whether they are homogeneous enough to pool. Stage two fits the structural model to the pooled matrix using an appropriate weight, so the SEM fit statistics properly reflect the evidence behind each correlation. This is more rigorous than older “naive” approaches that pooled correlations crudely and then treated the result like ordinary data—which produces incorrect standard errors and fit tests.
A central issue is heterogeneity: if the correlations differ substantially across studies, a single pooled matrix may not represent any real population, and fitting one model to it can mislead. A credible MASEM tests for this and, where heterogeneity is present, uses a random-effects pooling stage and considers moderators or subgroups rather than forcing one model on a heterogeneous field. Missing correlations—studies reporting only some variable pairs—are the norm and are handled within the multivariate pooling, but very sparse cells limit what can be estimated.
Pool the matrix correctly first, then fit the model. Two-stage SEM keeps the meta-analysis and the SEM statistically honest. Crude pooling followed by an ordinary SEM gives wrong standard errors and fit tests—and heterogeneity across studies must be tested, not assumed away.
What we deliver in
We fit MASEM in established, reproducible tools—principally R’s metaSEM package for two-stage SEM—and provide the pooled matrix, model fit, path estimates, and versioned code so the whole analysis can be checked and rerun.
How we deliver a meta-analytic SEM
MASEM draws on both our meta-analysis and SEM & psychometrics practices, run on a full systematic-review workflow—so the model is tested on a sound, reproducible synthesis.
We begin with a registered protocol and a specified theoretical model, a comprehensive search, and careful extraction of the correlation matrices (and sample sizes) each study reports. In stage one we pool the matrices with a multivariate model and test their homogeneity; in stage two we fit the structural model and evaluate its fit.
Reporting follows PRISMA standards alongside SEM reporting conventions, with pooling, heterogeneity, and model fit all reported transparently.
You receive the pooled correlation matrix, the fitted structural model with path estimates and fit indices, tests of heterogeneity (and any moderator or subgroup models), a comparison of competing models where relevant, and reproducible code and data. The result is a theory tested on the whole field—defensible in the most demanding review.
Meta-analytic SEM across Management & Allied Studies
MASEM is a natural fit for the correlational, theory-driven literatures at the heart of management and the social sciences—and we apply it across the disciplines we serve.
Management & Organisational Studies
Testing whether a theorised model—say, a driver’s effect on performance through engagement—holds across the accumulated literature.
Applied & Organisational Psychology
A core MASEM setting: pooling correlations to test multi-construct theories and competing structural models.
Marketing & Consumer Research
Testing mediation chains—attitude to intention to behaviour—across many correlational studies at once.
Economics & Public Policy
Synthesising the relationships among several variables into a coherent structural model across evaluations.
Information Systems
Testing technology-acceptance and adoption models on the full body of correlational evidence.
Education & Learning Sciences
Establishing whether a structural theory of learning or motivation generalises across studies.
Meta-analytic SEM: common questions
A theoretical model to test across a literature?
If your question is whether a multi-variable model holds across an entire body of correlational research, meta-analytic SEM can test it with the power of the whole field. We design and deliver it using rigorous two-stage methods.