Relationship between path coefficient... PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 daniel posted on Monday, February 24, 2003 - 10:21 am
In Output we always have correalation matrix for factors if TECH4 is used no matter what structure we assume among the factors. So for some pairs of factors there are two kinds of parameters: path coefficent and correlation coefficient.
What relationship is there between these two coefficients? How to intepret them simutaneously?
 Linda K. Muthen posted on Tuesday, February 25, 2003 - 9:24 am
TECH4 is the model implied correlation/covariance matrix. The model estimates are printed in the Model Results section of the output. For the following model:

MODEL: f1 BY y1-y5;
f2 BY y6-y10;
f2 ON f1;

the STDYX estimate will be the same as the correlation between f1 and f2 in TECH4.

For model complex models, the relationships are more complicated.
 liesbeth mercken posted on Thursday, November 16, 2006 - 7:38 am
Hi,
In my outputfile I receive estimates of the paths in my hypothesized model (using MLR method).Is it somehow possible to get correlations between those resulted estimates?
I want to be able to say that estimate of 'Y ON x1' is significantly higher/different from estimate 'Y ON x2'

Or do you now another way to be able to say that,

thank you very much,
kind regards,

Liesbeth Mercken
 Linda K. Muthen posted on Thursday, November 16, 2006 - 7:43 am
You can use the MODEL TEST command. See the user's guide for details. Or you can test two nested models -- one where the two regression coefficients are not held equal against a model where they are held equal.
 Renate Houts posted on Thursday, January 15, 2009 - 9:21 am
I get different results when using MODEL TEST and testing nested models. Is there a preference for one test over the other?
 Linda K. Muthen posted on Thursday, January 15, 2009 - 9:56 am
These should not differ much. If you want me to look at it, please send the two outputs and your license number to support@statmodel.com. Please use Version 5.2.
 ZHANG Liang posted on Tuesday, January 08, 2013 - 6:18 am
Hi. Today I conducted a series of simple analyses in Mplus, and encountered some questions.

There are two continuous variables, namely A and B. I standardized them and calculated their correlation coefficient in SPSS, the result is 0.607. Then in MPLUS, I estimated the paths A ON B, and B ON A£¬respectively, and not surprisingly, they were all 0.607. Then I ran A WITH B, I thought the result would also be 0.607, but A WITH B was estimated as 0.611. Does the WITH command follow a different formula from common correlation analysis?

Besides, I found that in the A WITH B result, both A and B's residual are estimated. I got confused here, can you offer some explanation on it?
 Linda K. Muthen posted on Tuesday, January 08, 2013 - 2:30 pm
Please send the output with standardized results and your license number to support@statmodel.com.
 Paula Yuma posted on Tuesday, November 22, 2016 - 9:17 am
Hello. My correlation matrix shows a positive correlation between two latent factors. The path coefficient between those two factors is negative. Can you help me explain what might contribute to the difference in the direction of the relationship? Specifically, I'm trying to understand why the correlation is positive.
I do expect the relationship to be negative, and in SPSS when I run correlations between the observed indicators of the latent factors, all correlations are indeed negative.
Many thanks.
 Bengt O. Muthen posted on Tuesday, November 22, 2016 - 3:29 pm
One needs to know the full model to answer that - perhaps the two factors have other variables influencing them. This general question is more suited to SEMNET.
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: