Stephen Brand
www.statisticsdoc.com
Dear Dr. Joshi,
Your model is underidentified: i.e., there are an infinite number of
possible parameter estimates. The solution this problem depends on the
source of the problem. Without seeing the model, I am speculating a little,
but here are the two broad types of underidentification.
One type of problem is structural underidentification. This condition
occurs when there are not enough known correlations or covariances to
estimate the parameters in the model. Have you constrained at least one of
the loadings on each factor in order to scale the variance of the
underlying factor?
Another problem is empirical underidentification. Do you have a parameter
estimate that is close to zero? Perhaps the correlations between two of the
scales are close to zero, and AMOS is treating them as if they were zero.
You may need to specify a fixed non-zero value for the parameter in order to
achieve identification.
The modification indices are a different matter. These indices are useful
for improving the fit of the model once you have an identified (or
overidentified) model. You must request them on the output tab if you are
using AMOS.
HTH,
Stephen Brand
For personalized and professional consultation in statistics and research
design, visit
www.statisticsdoc.com
-----Original Message-----
From: SPSSX(r) Discussion [mailto:
[hidden email]]On Behalf Of
DR VEENA Joshi
Sent: Saturday, September 23, 2006 11:08 AM
To:
[hidden email]
Subject: AMOS question
Hi,
I am very new to AMOS and trying to confim my model which I have got
from Exploratory analysis.
After getting the diagram, when i asked for estimates, Amos Text notes
mentioned I need two additional constraints to get model.identified.
What does that mean?
Where can i see modification indices?
Thanks
Veena
Locked
