Rules Comm posted on Thursday, May 06, 2010 - 3:33 pm
This is one factor CFA.
The results of the model are generally good. CFI>.96 and SRMR<.05.
Only the StdYX of one observed variable (MNDAAA9) is greater than 1, while the rest are smaller than 1. But I do not think this is a problem
But there is a message saying: THE RESIDUAL COVARIANCE MATRIX (THETA) IS NOT POSITIVE DEFINITE. I found the estimate of residual variance for the variable (MNDAAA9) is negative (-0.011), which is small and not statistically significant.
Then I follow the guide in some of the discussions in the forum and add a line MNDAAA9@0 in the model command to handle this problem. It does not work. All the model fit indices are weird and the StdYX of mndaaa9 equal 1.
What could be the reason that this strategy does not work? Is there any other strategy that I can follow? Thanks.