Estimates of *specific* indirect effects in AMOS (not "how to test mediation in AMOS)

#1
Estimates of *specific* indirect effects in AMOS (not how to test mediation in AMOS)

Hi all,

I have a very general AMOS question. I am wondering if there is any (simple) way to get AMOS to output *specific* indirect effects, along with associated confidence intervals, etc.

That is, in the simplest case, if I have a 2 mediator model, where X -> M1, X->M2, M1->Y, M2->Y (i.e., X predicts Y through M1 and M2), all I can get in output is the *total* indirect effect of X on Y (through M1 and M2). This is the sum of the two indirect effects, though, and does not tell me each individual path (X->M1->Y & X->M2->Y). This is often a very important question, with multiple mediators of an effect in a model, and a desire to test individual mediated paths.

In order to come up with the total indirect effect, AMOS must be calculating each path in some way. The question is "where do I find output attesting to this??" So far, I've seen one other person with the question on the interwebs, unanswered, and other searches simply turn up how to test indirect effects in AMOS.

Any help from anyone out there would be appreciated.
 
Last edited:
#2
Re: Estimates of *specific* indirect effects in AMOS (not "how to test mediation in A

I should have mentioned that the option "buy MPlus" won't be super helpful - I have it, and while it does what I want, I want to know if this can be done in AMOS. And I know I should be using R. Sigh.
 
Last edited:

spunky

Can't make spagetti
#3
Re: Estimates of *specific* indirect effects in AMOS (not "how to test mediation in A

I should have mentioned that the option "buy MPlus" won't be super helpful - I have it, and while it does what I want, I want to know if this can be done in AMOS.
given the widely-known shortcomings that AMOS has (constant subect of discussion on SEMNet), i wouldn't be surprised if this was impossible to do in AMOS.
 
#4
Re: Estimates of *specific* indirect effects in AMOS (not "how to test mediation in A

given the widely-known shortcomings that AMOS has (constant subect of discussion on SEMNet), i wouldn't be surprised if this was impossible to do in AMOS.
Indeed - AMOS has some shortcomings. I've now realized that it cal all be done through user-defined estimands (coefficients, SE, CIs, ps) - in some ways less difficult than the full model specification that must happen in MP...but something I hoped would be easier to get. For other people searching the web who come across this post, these videos (estimating one and two estimands) is actually useful. Tried it: It worked fine. AMOS 22 is a little easier as well, but not as flexible.

http://www.amosdevelopment.com/video/index.htm

Thanks to those who have replied. :D
 
#5
Re: Estimates of *specific* indirect effects in AMOS (not "how to test mediation in A

Hi Laurent
Could you kindly explain how you managed to calculate specific indirect effects using AMOS? The links to videos you provide seem to examine a model that consists of a mediation chain including two mediators in a row, but not a multiple mediation model like the one you originally had (which is what I have). So I would like instructions for specific indirect effects when one variable affects another through two simultaneous mediators. Any help would be much appreciated.
Thanks! :)