Turning SPSS PROCESS Macro coding into Stata (sequential moderated mediation)

[Mojtaba Salem]

Dear Stata Users,

I've tried to compute a sequential moderated mediation model (IV -> Mediator 1 -> Mediator 2 -> DV conditional that the path"IV -> Mediator 1" is moderated by W).

First, the coding for a normal moderated mediation is available at http://www.ats.ucla.edu/stat/stata/faq/modmed.htm. However, I am looking for the coding for sequential moderated mediation; does anyone have any hint how I can extend the normal coding to compute a sequential moderated mediation model?

Second, I've obtained the SPSS coding for sequential moderated mediation. Is there any sort of guide or special command so I can use or translate the SPSS syntax in the STATA?

Thanks,
Mojtaba

[Daniel Ruiz-Palomo]

Hello. Is the mediation sequential or not? Is W categorical or continuous?
If mediation is sequential and W is not categorical, you can modelize it as follows:

gen wxiv = iv*w
sureg (m1 iv w wxiv cv) (m2 m1 x cv) (dv m2 m1 x cv)

Then you can get significance of indirect effects as follows:

nlcom _b[m1:iv]*_b[m2:m1] (here you test if m1 mediates iv->m2)
nlcom _b[m1:iv]*_b[dv:m1] (here you test if m1 mediates iv->dv)
nlcom _b[m2:iv]*_b[dv:m2] (here you test if m2 mediates iv->dv)
nlcom _b[m1:iv]*_b[m2:m1]*_b[dv:m2] (here you test mediation is sequential)

If mediation is not sequential and W is not categorical, you can modelize it as follows:

gen wxiv = iv*w
sureg (m1 iv w wxiv cv) (m2 x cv) (dv m2 m1 x cv)

Then you can get significance of indirect effects as follows:

nlcom _b[m1:iv]*_b[dv:m1] (here you test if m1 mediates iv->dv)
nlcom _b[m2:iv]*_b[dv:m2] (here you test if m2 mediates iv->dv)

[Bruce Weaver]

The UCLA link in #1 no longer works. I suspect this is the new version of that same page:

• https://stats.idre.ucla.edu/stata/fa...tion-in-stata/

HTH.

[Jon Hoefer]

Hello. Is the mediation sequential or not? Is W categorical or continuous?
If mediation is sequential and W is not categorical, you can modelize it as follows:

gen wxiv = iv*w
sureg (m1 iv w wxiv cv) (m2 m1 x cv) (dv m2 m1 x cv)

Then you can get significance of indirect effects as follows:

nlcom _b[m1:iv]*_b[m2:m1] (here you test if m1 mediates iv->m2)
nlcom _b[m1:iv]*_b[dv:m1] (here you test if m1 mediates iv->dv)
nlcom _b[m2:iv]*_b[dv:m2] (here you test if m2 mediates iv->dv)
nlcom _b[m1:iv]*_b[m2:m1]*_b[dv:m2] (here you test mediation is sequential)

If mediation is not sequential and W is not categorical, you can modelize it as follows:

gen wxiv = iv*w
sureg (m1 iv w wxiv cv) (m2 x cv) (dv m2 m1 x cv)

Then you can get significance of indirect effects as follows:

nlcom _b[m1:iv]*_b[dv:m1] (here you test if m1 mediates iv->dv)
nlcom _b[m2:iv]*_b[dv:m2] (here you test if m2 mediates iv->dv)

Hey could someone please elaborate what the code would look like for just one mediator and how I could include the bootstrapping procedure in it? In other word do what SPSS does with its PROCESS tool in Stata?