Structural Equation Modeling Calculator

Here you can create and estimate a structural equation model online. The calculator combines a visual path diagram with lavaan model syntax and supports models containing measured variables, latent factors, regressions, covariances, indirect effects, and observed-variable interactions. The model is calculated directly in your browser.

How to create a structural equation model

1. Enter or paste your data into the table above the calculator.
2. Select the measured variables that belong in your model.
3. Use + Factor to add latent variables where needed.
4. Drag between variables to create connections. Select a connection to define it as a loading, regression, or covariance.
5. Choose the estimator, missing-value treatment, and factor-scaling method.
6. Select Run model to estimate the model and display the results.

You can move variables freely on the canvas or use Tidy up to arrange the model. Open Model syntax when you want to inspect or edit the corresponding lavaan syntax. The same model builder is available for CFA, path analysis, and SEM mediation.

Results provided by the SEM calculator

After estimation, the calculator reports the number of observations, convergence and model diagnostics, chi-square, degrees of freedom, CFI, TLI, RMSEA with its 90% confidence interval, SRMR, and, where available, AIC and BIC. Parameter tables contain estimates, standard errors, p-values, confidence intervals, and standardized estimates. Explained variance is reported as R-squared for outcomes and indicators.

The calculator also lists modification indices for connections that could be drawn on the diagram. Add a suggested connection only when it is supported by the theory behind the model; a large modification index alone is not a sufficient reason to change a model.

Which type of model do you want to calculate?

• Use the confirmatory factor analysis calculator to test a predefined measurement model.
• Use the path analysis calculator for directional relationships between measured variables.
• Use the SEM mediation analysis calculator to estimate indirect effects involving measured or latent variables.

Estimators and ordinal variables

For continuous data, choose maximum likelihood (ML) or robust maximum likelihood (MLR). MLR estimates the same model with robust standard errors and scaled fit statistics and is useful when continuous data depart from multivariate normality. Variables marked as ordinal in the data table are modeled as ordinal when they are indicators or outcomes; the calculator then automatically uses WLSMV and reports scaled fit statistics.

Missing continuous values can be handled by excluding incomplete rows or by full-information maximum likelihood (FIML). With ordinal variables, the available-pairs option is used instead of FIML.