Rules Comm posted on Wednesday, June 29, 2011 - 2:46 pm
I am doing an one-factor CFA: I got the following numbers as part of the results:
MODEL RESULTS Latent variable by 1st observed variable ESTIMATE 1.000 S.E. 0.000 EST./S.E. 999.000 Two-Tailed P-value 999.000
The model fit numbers are not very good. Is it because the model fit is not good or any other reason? (I will not use the number in the section of "model results", but I will have to use the model fit. This is why I ask).
Worthy of notice is that the estimates of the same observed variable in STDYX, STDY, STD sections are all normal.
On the other side, the estimates for Variances of the latent variable in the section of MODEL RESULTS are normal, but those in STDYX, STDY, STD sections are abnormal, such as the latent variable: 1.000 0.000 999.000 999.000
Meanwhile, the estimates for other observed variables are also normal. Why is the above one so abnormal?
The result you show is for the first factor indicator that is fixed at one as the default to set the metric of the factor. The value 999 is given when a value cannot be computed.
If the model does not fit the data, the results should not be interpreted.
Shirley posted on Sunday, February 19, 2017 - 8:47 pm
Dear Dr. Muthen, Following the question above, I observed that the "*********" is printed for the "Est./S.E." statistic for the variance of the first factor in the STDYX Standardization section of a 5-factor CFA model. I understand that "Est/S.E." cannot be calculated in this case, as the estimate of the standardized factor variance is 1 and S.E. is 0. But is there a particular reason why *********, instead of 999 like the case for the "Est./S.E." statistics of the other 4 standardized factor variances, is printed for the first standardized factor variance?