Nominal as mediators PreviousNext
Mplus Discussion > Categorical Data Modeling >
Message/Author
 Marcus Valenzuela posted on Tuesday, January 12, 2016 - 5:22 pm
Dear Dr. Muthen,

I have a nominal (4 groups) mediator with both X and Y continuous. I just read you paper (Muthen, 2011) and I am just confused as to how parameters were fixed in Tables 48-49, and if a similar "fixing" approach is possible for a group of 4 instead of 3. Thank you for our time.

https://www.statmodel.com/download/CausalmediationPaper-Mplus%20input%20appendix.pdf

Mark Match
 Bengt O. Muthen posted on Wednesday, January 13, 2016 - 12:38 pm
Not sure what you mean - no parameters are fixed there.

The setup can be generalized to other number of mediator categories in a straightforward manner. But it does involve being able to handle the Model Constraint formulas.
 Marcus Valenzuela posted on Wednesday, January 13, 2016 - 1:06 pm
I apologize for the confusion. I meant to say freeing parameters (e.g., y*.75), or at least that was my understanding. I believe I have an understanding on the formulas and labels, but I ma not sure how to deal with the '*', or if I have too. I hope I am making some sense.

(Model section from Table 48)
[c#1*-1] (gamma01);
[c#2*-.5] (gamma02);
c#1 ON x*.7 (gamma11);
c#2 ON x*.3 (gamma12);
y ON x*0;
y*.75;
%c#1%
[y*-2] (beta01);
y ON x*-.5 (beta11);
%c#2%
[y*0] (beta02);
y ON x*-.3 (beta12);
%c#3%
[y*2] (beta03);
y ON x*-.2 (beta13);
 Bengt O. Muthen posted on Wednesday, January 13, 2016 - 6:39 pm
Ah, those * values are just starting values that I chose to match the simulated data - you don't need to give any such values.

Btw, what is your nominal mediator?
 Marcus Valenzuela posted on Friday, January 15, 2016 - 10:01 am
Thank you Dr. Muthen. I have extended the equation to include 4 groups, but I must be doing something wrong (possibly with the summation) since I have very big estimates and standard errors for some parameters (p21, p31, p41, term11, term01, tie, total, pie), which I did not obtain when I run my data with your code of 3 groups only. My nominal variable is acculturation group. Would it be possible to post here what I am doing for clarification purposes?
 Linda K. Muthen posted on Friday, January 15, 2016 - 10:47 am
Please send the relevant files and your license number to support@statmodel.com.
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: