Proc mixed SAS versus mplus on mixed ... PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Lisa Koonce posted on Wednesday, April 13, 2011 - 2:04 pm
Question: I'm having trouble replicating a 2 x 2 mixed model ANOVA within MPlus. I'm trying to replicate before i do an involved path model.

I have one within-subjects manipulated (two levels) variable and one between-subjects manipulated variable (two levels). One DV. Trying to model the two main effects and interaction, like one would do in a standard MIXED model ANOVA in SAS.

My estimates in MPLUS generally seem close enough for the interaction term, but are all over the place, not even close on the main effects.

I did also "back up" and try to replicate SAS regular PROC GLM for the between subjects variable only (it worked) and replicate the sas PROC MIXED results for the within subjects variable only. I did replicate those within MPLUS .

But when I go to the 2 x 2 world, something goes awry. Can you help. SEe code below.

THE MIXED MODEL that doesn't replicate in MPLUS:
VARIABLE: NAMES ARE subject GorL Bifurc GLxBi advant ;
USEVAR = GorL Bifurc advant subject bixGL ;
WITHIN = Bifurc bixGL;
BETWEEN = GorL;
CLUSTER = subject;

DEFINE: bixGL = bifurc * GorL;


ANALYSIS: TYPE=TWOLEVEL ;
ALGORITHM = INTEGRATION;
MODEL: %WITHIN%
advant ON bifurc bixGL ;
%BETWEEN%
advant ON GorL;
 Linda K. Muthen posted on Thursday, April 14, 2011 - 8:55 am
Please send the SAS and Mplus outputs as attachments and your license number to support@statmodel.com.
 Daniel Kopala-Sibley posted on Wednesday, September 07, 2016 - 1:02 pm
I am having a similar problem, only I"m looking at the effects of time-invariant (level 2) predictors versus the time varying component of those predictors. I get a similar pattern of p's in SAS vs MPlus, but very very different parameter estimates.

M = Mean, Dv=Deviation score. The dependent variable "Dysphoria" is raw on the Within path, and is a mean score on the between.

Do you have any suggestions on how to figure this out?

USEVARIABLES ARE id DvBFIN DvSC DvDependency DvRumination dysphoria
MBFIN MSC MRumination MDependency mdysphoria ;

MISSING = ALL (9999);
WITHIN = DvBFIN DvSC DvDependency DvRumination;
BETWEEN = MBFIN MSC MRumination MDependency mdysphoria ;
CLUSTER = id;
ANALYSIS:
TYPE = TWOLEVEL;
ESTIMATOR= MLR;

MODEL:
%WITHIN%
dysphoria ON DvBFIN DvSC DvDependency DvRumination;

%BETWEEN%
mdysphoria ON MBFIN MSC MRumination MDependency ;

OUTPUT:
STAND sampstat CINT ;


Thank you in advance!
 Linda K. Muthen posted on Wednesday, September 07, 2016 - 8:13 pm
Please send the SAS and Mplus outputs as attachments 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: