Moderation in a Growth model for two ... PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
Message/Author
 Ana Margarida Graça posted on Monday, May 07, 2012 - 10:49 am
Hi,
I have two variables each one measured 5 times. I want to know if the growth in one variable is related to the growth of the second variable. I want also to test how is that relationship according to different levels of a third variable, i.e., test a moderation.
How can this be done in MPlus? Is there any reference with the sintax?
Thanks in advance!
 Linda K. Muthen posted on Monday, May 07, 2012 - 5:17 pm
See Example 6.13 which is a parallel process model. If the moderator is categorical, you can do a multiple group analysis.
 Ana Margarida Graça posted on Tuesday, May 08, 2012 - 10:10 am
Thank you for your response!
The moderator is continuous. What could be the best procedure?
 Linda K. Muthen posted on Tuesday, May 08, 2012 - 2:44 pm
Then you need to create an interaction term using the DEFINE command if both variables are observed or the XWITH option if one or both are latent.
 Daniel Lee posted on Friday, January 12, 2018 - 7:51 am
Hi Dr. Muthen, to follow up on this question, if I wanted to test the interaction between X (an observed variable and continuous) and an intercept factor, and the interaction between X and the slope factor, I would use the Define command and use the XWITH option?

Thank you so much for your guidance.
 Daniel Lee posted on Friday, January 12, 2018 - 8:05 am
Sorry, and I forgot to mention the interaction between X and the growth factor, would be predicting another growth factor.
 Bengt O. Muthen posted on Friday, January 12, 2018 - 5:11 pm
See page 759 of the V8 UG for a summary of ways to obtain interactions.
 In Seo La posted on Monday, April 13, 2020 - 9:24 pm
Thank you so much for your videos and support.
I would like to know how to conduct latent growth model with moderation.
I have three waves of a outcome (ment1-ment3), three waves of a predictor (rest1-rest3), and SPS1 as a moderator (continuous). In this command, I was also trying to control clustering issue using type = complex.

I could find error *** ERROR Undefined variable used in transformation: IR.

How could I add interaction terms to paths between intercept and slope?

USEVARIABLES = ment1-ment3 rest1-rest3 SPS1 w1 w2;
cluster = ptID;
Missing are all (-99);

Define:
w1 = ir*SPS1;
w2 = sr*SPS1;

Analysis:
Estimator is MLR;
Type = complex;

MODEL:
ir sr | rest1@0 rest2@3 rest3@6;
im sm | ment1@0 ment2@3 ment3@6;
im ON ir SPS1 w1;
sm ON ir SPS1 w1;
sm ON sr SPS1 w2;
 In Seo La posted on Tuesday, April 14, 2020 - 9:45 am
I am so sorry to bother you. Even though I tried to use define in previous question, ) .

1. If I cannot use "Define" for latent variable and observed continuous,
Could I use "type = complex random" to make interaction term (observed continuous with continuous latent (intercept and slope)) and controlling for clustering issue at the same time. SPS1 as a moderator (observed continuous).

2. Should I revise comments, starting from "im ON ir SPS1 irxSPS ....?

USEVARIABLES = ment1-ment3 rest1-rest3 SPS1;
cluster = ptID;
Missing are all (-99);

Analysis:
Estimator is MLR;
Type = complex random;

MODEL:
ir sr | rest1@0 rest2@3 rest3@6;
im sm | ment1@0 ment2@3 ment3@6;

irxSPS | ir XWITH SPS1;
srxSPS | sr XWITH SPS1;

im ON ir SPS1 irxSPS;
sm ON ir SPS1 irxSPS;
sm ON sr SPS1 srxSPS;


OUTPUT:
SAMPSTAT;

Thank you so much again.
 Bengt O. Muthen posted on Wednesday, April 15, 2020 - 3:47 pm
We need to see your full output - send to Support@statmodel.com along with your license number.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: