Structural Equation Modelling (SEM) and PLS-SEM Analysis Service
Test complex theories with several constructs, paths and mediators in one model. We choose between covariance-based SEM and PLS-SEM, build the model, and report it to the standard reviewers expect.
CB-SEM or PLS-SEM: which one do you need?
Covariance-based SEM (in AMOS, R lavaan or STATA) is the usual choice when you want to test how well a theory fits the data and your sample is large enough. PLS-SEM (in SmartPLS) is often preferred for prediction-focused or exploratory models, complex models with many indicators, smaller samples and formative constructs. We explain the trade-offs for your study and justify the choice in the methods section.
What we deliver
- Measurement model assessment: indicator loadings, composite reliability, AVE, Fornell-Larcker and HTMT
- Structural model assessment: path coefficients, R-squared, effect sizes and predictive relevance
- Bootstrapping for significance and confidence intervals
- Mediation, moderation and moderated-mediation models
- Multi-group analysis and measurement invariance
- For CB-SEM: global fit (CFI, TLI, RMSEA, SRMR) and respecification with theoretical justification
- Path diagrams and tables formatted for publication
Sample size and model planning
SEM results depend heavily on having enough data. Rules such as the 10-times rule, power analysis and the inverse square root method give different answers; our free regression and PLS-SEM sample size guide compares them. We can advise on this before you collect data, then help you structure the questionnaire so the constructs are properly measured.
Building on a clean measurement model
A structural model is only as good as its measures. We usually begin with factor analysis and reliability and validity testing to confirm that each construct is sound, then add the structural paths. For simpler models without latent variables, regression analysis may be enough.
How the work runs
- You share your conceptual model, questionnaire and data.
- We agree the SEM approach and specify the model with you.
- We assess the measurement model, then the structural model, then robustness checks.
- You receive tables, diagrams and a written results section, plus the project files.
Frequently asked questions
What sample size do I need for SEM?
It depends on model complexity. Many studies aim for at least 200 cases for CB-SEM, while PLS-SEM can work with smaller samples, but power analysis is better than a single rule of thumb.
Can you use SmartPLS and AMOS?
Yes. We work in SmartPLS for PLS-SEM and in AMOS, R (lavaan) or STATA for covariance-based SEM.
My model fit is poor. What can be done?
We examine the measurement model first, look at loadings and modification indices, and respecify only where theory supports it. We never tweak a model just to improve fit.
Can you test mediation in SEM?
Yes, with bootstrapped indirect effects and confidence intervals.
Do you help write the results?
We provide results tables, figures and draft text for your own report or thesis.
Related statistical analysis services
Factor Analysis (EFA & CFA)
EFA and CFA to find and confirm the structure behind your items.
Regression Analysis
Linear, logistic, moderation and mediation models with diagnostics.
Reliability & Validity Testing
Alpha, omega, test-retest and validity evidence for your scales.
Statistical Analysis Services
See every statistical method we support.
Also related: Questionnaire design and sampling