Inquiry About Estimating a Mediation SEM With Dichotomous Variables in lavaan

[heba megahed]

hello everyon,

I am currently working on a structural equation modeling study in which the model includes one independent variable, one mediator, and one dependent variable, and all observed variables are dichotomous (binary).

I would like to ask whether it is methodologically appropriate to estimate such a mediation model using the lavaan package in R.

More specifically, I would be grateful for your guidance on the following points:

1. Can lavaan adequately handle a structural equation model where the independent, mediator, and dependent variables are all dichotomous?

2. If this type of model is feasible, what is the most appropriate estimator for obtaining reliable parameter estimates?

3. Would estimators such as WLSMV or DWLS be recommended in this case, or is there another preferable estimation approach?

4. Are there any specific model specification considerations or limitations that should be taken into account when all variables in the mediation model are binary?

My goal is to estimate both the direct and indirect effects accurately under this categorical data structure.

Any suggestions, references, or practical examples would be highly appreciated.

Thank you very much for your time and support.

Best regards,
[Your Name]