FSCORES not the same as factors PreviousNext
Mplus Discussion > Confirmatory Factor Analysis >
Message/Author
 lopisok posted on Monday, February 11, 2013 - 7:48 am
Hello,

I have a question about saving fscores after CFA.

In a previous thread http://www.statmodel.com/discussion/messages/23/4239.html?1336064215 Linda Muthen said:

"Factor scores are not the same as the factors in a model. How close they are to each other can be seen in the factor determinacy score. See the FSDETERMINACY option of the OUTPUT command."

1) I wonder why they are not the same? Why aren't the same scores saved so they can be used as dependent variables for regression?
2) In another thread Linda said that these fscores can't be used as dependent variables in regression. Why is this?
3) Is there any way I can save the factors from CFA (for example the STDYX outcomes) as they are in the model and use these for regression as dependent variables?

Thank you for the help,
kind regards,
Filip
 Linda K. Muthen posted on Monday, February 11, 2013 - 10:39 am
1-2. See the following paper:

Skrondal, A. and Laake, P. (2001). Regression among factor scores. Psychometrika 66, 563-575.

3. Factor scores are not used for model estimation in an SEM model. In SEM, the model parameters are directly estimated from the data.
 Jane Doe posted on Tuesday, March 24, 2015 - 2:42 pm
I have categorical measures for my latent variable and when I ask for FSDETERMINACY it gives me the following error:

*** WARNING in OUTPUT command
FSDETERMINACY is not available for analysis with at least one categorical
dependent variable. Request for FSDETERMINACY is ignored.


Is FSDETERMINACY not available if the measures for the latent variable are categorical? If this is the case then how do I know how close the factor scores are to the factors?

Thanks in advance.
 Linda K. Muthen posted on Tuesday, March 24, 2015 - 3:19 pm
In this case, look at the information function plots which tell you the precision with which the factor scores are estimated for different values of the factor. The information function value is 1/SE**2. The higher the value, the more precise the factor scores.
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: