Three way interaction marginsplot After SEM?

[Mariame Shaw]

Hi Everyone,

I was wondering if anyone had experience or could help me create a graph for a three-way interaction between two continuous variables (sleep and depression scores) and one categorical variable (sex) after an SEM command.

Here is what I have tried so far and the error I ran into:

sem (c_sat_composite <- c_cesd_wthsleep c_avg_sleepdis c_depXsleep c_depXsex c_sleepXsex c_depXsleepXsex ce_rage rasex rarace ///
                 ramari ra_edu emp_status smoking_now alcohol_ever meds_use diff_func gn_health rate_memory), allmissing method(mlmv)     
            
                *Graph Significant 3-way interaction
                margins rasex, at(c_cesd_wthsleep==(0(1)7) c_avg_sleepdis=(0(1)2))
                marginsplot, by(rasex) noci xdimension(c_cesd_wthsleep)

However, I get the error that

factor 'rasex' not found in list of covariates

Any guidance or help would be greatly appreciated!

[Mariame Shaw]

I realized that I was getting an error as I could not specify that my sex variable (rasex) was a factor with the SEM command, which I had to use to implement the full maximum likelihood method. Does anyone have any insight on how I might go about plotting a three-way interaction after the SEM command, given that I have a continous*continous*categorical interaction term? Any help would be greatly appreciated

[Erik Ruzek]

This might work, but I'm not 100% sure without having some data to play with.

margins , at(c_cesd_wthsleep==(0(1)7) c_avg_sleepdis=(0(1)2) rasex=(0 1))
marginsplot, by(rasex) noci xdimension(c_cesd_wthsleep)