Unwanted correlations PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Rosemarie Huver posted on Tuesday, March 01, 2005 - 1:05 am
I find that Mplus adds correlations unspecified by my model. In the User's Guide it reads that correlations between exogenous variables are included automatically. Although I do not fully understand the rationale behind this and I wouldn't mind adding these correlations myself, I understand what is going on.

I find, however, that also correlations between exogenous and (latent) endogenous variables are added.

I wonder why this is and what I can do to prevent this, other than setting the loadings @0.
 Linda K. Muthen posted on Tuesday, March 01, 2005 - 6:09 am
Defaults are chosen to fit what is most often seen in practice. If a parameter is free by default and you don't want it to be free, you can fix it at zero.
 rosemarie huver posted on Thursday, March 03, 2005 - 3:54 am
In this part of the input I have fixed all correlations to 0 that aren't between variables/factors in a 'column'. I realize that variables/factors that aren't in the same column can be related, but I prefer thinking of these as causal relationships, indicated by 'on' statements that I plan to add if indicated by the Modification Indices.

x1-x5 with f1-f9@0;
f1-f4 with f5-f9@0;
f5-f7 with f8-f9@0;
f8 with f9@0;

Does this make sense?

Will the input above lead to difficulty in estimating beta-weights?
 Linda K. Muthen posted on Thursday, March 03, 2005 - 7:35 am
Please show your full MODEL command.
 Rosemarie Huver posted on Thursday, March 03, 2005 - 11:39 am
model: f1 by y1 y2;
f2 by y3 y4;
f3 by y5 y6;
f4 by y7 y8;
f5 by y9 y10;
f6 by y11 y12;
f7 by y13 y14;
f8 by y15 y16;
f9 by y17 y18;
f1 on x1 x2 x3 x4 x5;
f2 on x1 x2 x3 x4 x5;
f3 on x1 x2 x3 x4 x5;
f4 on x1 x2 x3 x4 x5;
f5 on f1 f2 f3 f4;
f6 on f1 f2 f3 f4;
f7 on f1 f2 f3 f4;
f8 on f5 f6 f7;
f9 on f8;
x1 with x2 x3 x4 x5;
x2 with x3 x4 x5;
x3 with x4 x5;
x4 with x5;
f1 with f2 f3 f4;
f2 with f3 f4;
f3 with f4;
f5 with f6 f7;
f6 with f7;
x1-x5 with f1-f9@0;
f1-f4 with f5-f9@0;
f5-f7 with f8-f9@0;
f8 with f9@0;
 Linda K. Muthen posted on Thursday, March 03, 2005 - 5:47 pm
It is unnecessary to include both ON and WITH statements for the same variables. For example:

f9 on f8;

and

f8 with f9@0;

Because f9 is a dependent variable, the WITH statement refers to the residual of f9. Such residuals are always uncorrelated with indepdent variables such as f8 as a default in Mplus.

I would remove all of the WITH statements and run the model with the Mplus defaults. See what you get and go from there.
 swamp_billy posted on Monday, June 12, 2006 - 10:36 am
Hello,

When I add one WITH statement between just two variables MPLUS sometimes automatically outputs many WITH statements among other variables in the model. When I remove the WITH statement all the others are no longer included in the output. Is this a bug or something I haven't picked up on? I have version 3.12. Thanks for your help.
 Linda K. Muthen posted on Monday, June 12, 2006 - 10:49 am
This sounds strange. I would have to see it to commnent. Please send your inputs, data, outputs, and 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: