Mediation PreviousNext
Mplus Discussion > Categorical Data Modeling >
 jan mod posted on Thursday, February 09, 2017 - 1:09 am
I am testing a mediation with a binary mediator, a continuous exposure variable (X) and a continuous outcome (Y). I'm testing it like in section 8.4.1 in your latest book (using those equations).

1) Is this correct because that section deals with a binary X, binary M and continuous Y?

2) In my case, should I interpret the indirect and direct effects as odds ratio's or linear coefficients?
 jan mod posted on Thursday, February 09, 2017 - 1:18 am
This is the syntax:


file = buurt.dat;
type = imputation;

names = a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13
a14 a15 a16 a17 a18 a19;
usevariables = a11 a3 a4 a5 a6 a7 a8 a10 a14 a18 a101-a133 ;
categorical = a14;
estimator = mlr;
a11 on a10 a8 a3 a18 a6 a5 a7 a4 a14 a101-a133 ;
a14 on a10;
model indirect:
a11 ind a14 a10;
 Bengt O. Muthen posted on Thursday, February 09, 2017 - 6:25 pm
1) For a continuous X, the two X values (exposure levels) that the effects com-
pare are given in parentheses after X in the IND statement:

.... X(x1 x0);

where you can choose x0 as say the mean of X and x1 say 1 SD above the mean.

2) Linear effects because Y is continuous.

We ask that no more than one window is used per posting. If longer descriptions are needed, please send to Support.
 jan mod posted on Friday, February 10, 2017 - 12:39 pm
Thank you very much and sorry for the double post. I have three more questions.

1) X(x1 x0): does Mplus then use the exposure variable as binary for the indirect effect? So section 8.4.1 (on a binary mediator is then equivalent for continuous X with that statement.
2) PNDE is here then interpreted as the linear effect of x1 on y with the mediator that is allowed to vary like it would if those respondents would have been exposed to x0.
3) my TNIE and PNIE are the same. Is this possible?

 Bengt O. Muthen posted on Friday, February 10, 2017 - 4:24 pm
1) No, the exposure variable is kept as continuous. It is just that your effect is evaluated as a function of a change in X from x0 to x1. See Section 2.6.1 of our book for an explanation of the case of a continuous exposure (although in the case of a continuous mediator and moderation but that doesn't matter).

2) Correct.

3) In the general case they are not the same but perhaps without an interaction (moderation) they could be - I haven't worked through the formulas for this but see the formulas of 8.4.1.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message