Multilevel model in dyadic analysis PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Donna Jingdan YAO posted on Thursday, July 19, 2012 - 10:10 pm
It's a moderated mediation multilevel model in a negotiation study.
IV: essentialism (TrEnt)
Mediator: Positive perception towards partner (PercPos)
DV: negotiation outcome (FinalPt)
Moderator: Cultural similarity (CultBg),
a categorical between-level variable moderating IV-> Med.
We are interested in the indirect effect of Moderator (1= dissimilar) on DV.
Does this syntax below looks correct?
Thanks!!

Define: Interact=CultBg*(TrEnt-3.37);
ANALYSIS: TYPE IS TWOLEVEL RANDOM;
MODEL:
%WITHIN%
TrEnt
PercPos
FinalPt;
PercPos ON TrEnt;
FinalPt ON PercPos;
FinalPt ON TrEnt;
%BETWEEN%
CultBg
TrEnt
PercPos
FinalPt
Interact;
PercPos ON TrEnt(a1);
PercPos ON CultBg;
PercPos ON Interact(a3);
FinalPt ON PercPos(b1);
FinalPt ON TrEnt;
FinalPt ON CultBg;
FinalPt ON Interact;
MODEL CONSTRAINT:
NEW(indb modval);
modval = 1;
indb=(a1+a3*modval)*b1;
OUTPUT: TECH1 TECH8 CINTERVAL;
 Linda K. Muthen posted on Friday, July 20, 2012 - 10:42 am
This looks correct.
 Donna Jingdan YAO posted on Saturday, July 21, 2012 - 9:22 pm
Linda, thanks for your reply.

One more question: the goodness-of-fit indices of this model are not good, with CFI below 0.40 and RMSEA around 0.25. Do these indices matter in a multilevel model? If yes, how can i improve the model fit?

Thanks!
 Linda K. Muthen posted on Sunday, July 22, 2012 - 10:49 am
I would say these matter. Look at modification indices using the MODINDICES option of the OUTPUT command. Perhaps a TWOLEVEL EFA will reveal a better model.
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: