As Holger said, you should use the r-package lavaan. R and its packages are free to download (http://www.r-project.org/).

The code is pretty simple, i.e.:

yourmodelname

Hi Behzad,

to use the correction you have to calculate the asymptotic covariance matrix of all model variables (with Prelis) and add it to the normal model syntax in the following way:

DA NO=1442 NI=9 MA=CM

CM=CENT.cov

AC=CENT.acm

Here, CM is the normal covariance matrix (called CENT.cov in my case) and AC is the asymptotic matrix.

As you see: Very complicated. I recommend using the lavaan package of R instead which is free and easy to implement. In lavaan you simply type "estimator=mlm" in the estimation-syntax and the job is done :) 

See: www.lavaan.org

Best,

Holger

Dear Behzad,

AMOS does not provide Satorra-Bentler Corrected Chi-square. However, Bootstrapping is the method available in AMOS for the purpose of correcting the Chi-square (Bollen-Stine Bootstrap) and S.E.s (Bias-corrected confidence intervals). However, in the case of Chi-square correction, AMOS and bootstrap method only provides the Chi-square P-VALUE after correcting for bias and DOES NOT give you the actual value of corrected chi-square.

If you are interested in using Satorra-Bentler method, here is what you need to do in EQS. I personally recommend Mplus as the software of choice.

------------

In EQS-->

Once you import the data and designed your model,

Go to "Build EQS"

Click on "Title/Specifications"

In "EQS Model Specifications" window, select "Robust Methods" under "Non-normal estimators corrections" tab. (Test S.E. is automatically selected).

Click "OK"

-------------

Again, Go to "Build EQS"

This time, click on "Run EQS" (you may need to expand the EQS array under the "EQS Working Array" tab.

Once analysis finished, you will be presented by "EQS output file".

Scroll down to "GOODNESS OF FIT SUMMARY FOR METHOD = ROBUST"

Here you can find "SATORRA-BENTLER SCALED CHI-SQUARE" as well as corrected NFI, NNFI, CFI, IFI, and RMSEA.

--------------

To access corrected S.E. and Test Statistics scroll down to "CONSTRUCT EQUATIONS WITH STANDARD ERRORS AND TEST STATISTICS"

You can find Robust statistics in (PARENTHESES).

--------------

Best of luck,

Kia

As Holger said, you should use the r-package lavaan. R and its packages are free to download (http://www.r-project.org/).

The code is pretty simple, i.e.:

yourmodelname

Hi Behzad

I agree with Timo, in R its easier that in Lisrel. 

I just add a little piece of code to analyse robust estimators with Satorra 

factor1=~item1.1+item1.2+item1.3

factor2=~item2.1+item2.2+item2.3

factor3=~item3.1+item3.2+item3.3

g-factor=~factor1+factor2+factor3

'

fit