Assuming that the model specification has been saved in the "model" object, and the variables to be considered as categorical are called "var1" and "var2". The code would be as follows:

fit

You have to specify the model and then use sem function.

semplot package can be used to draw the diagram.

Please see the link:

http://web.missouri.edu/~huangf/data/mvnotes/Using_R_for_path_analysis.html

Regards

Alma

Hi Dorien,

in principle, lavaan allows the possibility to apply the DWLS estimator by labeling the relevant variables in the model as "ordered". I can only guess but this should work for binary mediators, too, as the estimator estimates the effects of a underlying continuous response variable. For more valid answers you may want to post to the lavaan group:

https://groups.google.com/forum/#!forum/lavaan

Testing moderation on the mediator-outcome links should be, however, tricky, if you have a continuous moderator. I don't know whether you can simply add a product of the binary mediator and the moderator. Perhaps the lavaan group can answer this.

Best,

Holger

Hi Holger Steinmetz ,

Thank you, this is very helpful. I will have a look at the link you sent me.

Dorien

Assuming that the model specification has been saved in the "model" object, and the variables to be considered as categorical are called "var1" and "var2". The code would be as follows:

fit

Hi Brian Norman Peña-Calero ,

Thank you, I will give this a try.

Dorien

Hi Alma Braimllari Spaho ,

Thank you for sending me that link, I will have a look at it.

Dorien

Any idea if the categorical variable is not ordered? I dummy coded the categorical IV (for example, political affiliation - republican, democrat, independent). But using Lavaan, I would have to order them?