Hi Statalist - Longtime reader, first-time poster. I am running an sem/gsem where the latent has 4 indicators, and interested in the interaction of three continuous predictors on the latent variable. So, the form of the command is below, using Stata v16 IC on MacOS.
gsem (LatentVar -> indicator1) (LatentVar -> indicator2) (LatentVar -> indicator3) (LatentVar -> indicator4) (c.ContinuousPredictor##c.ContinuousModerator1##c. ContinuousModerator2 -> LatentVar), latent(LatentVar)
The 3-way interaction is significant, so I want to plot the interaction, however the "margins" command that I typically use on regression (e.g., margins, dydx(ContinuousPredictor) at(ContinuousModerator1 = (-1 0 1) ContinuousModerator2 = (-1 0 1)) only provides the margins on the observed indicators, but not the margins on the latent variable LatentVar.
Is this possible to calculate and plot the marginal means for the LatentVar at selected points of the predictors using "margins" and "marginsplot"? Or is there another approach to generate a plot of the predicted values and CI's of the latent variable at different values of the predictor/moderator?
Thanks much for any help anyone can provide, John
0 Response to Plotting interaction on latent variable after sem/gsem
Post a Comment