Message/Author |
|
|
Hi, I am trying to fit a 2nd order CFA using 6 observed categorical variables, as follows: Fa by cogquin fevquin1 quingrip; Fb by srh adlbin chronicbin; health by Fa Fb; The output gives the following message THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED. CHECK YOUR MODEL. PROBLEM INVOLVING PARAMETER 32. THE CONDITION NUMBER IS 0.655D-17. On checking parameter 32 is the PSI between health and health ( which I think must be related to the variance of latent variable health). I have also tried fixing the variance of the latent variables to 1, which did not help either. I can't understand why this model is not identified, and don't know how to solve it to fit the model. Can you suggest anything? Any help would be much appreciated. Thank you. |
|
|
Three first-order factors are needed to identify a second-order factor. You would need to fix both factor loadings to one for the model with two first-order factors to be identified. |
|
|
Thanks for your answer, I have done so using this code Fa by cogquin fevquin1 quingrip; Fb by srh adlbin chronicbin; health by Fa Fb; Fa@1 Fb@1; and the model is now identified. Excuse my not understanding, but how can I now interpret the standardized coefficients for the factors Fa and Fb to health, because haven't I constrained them to be 1? Many thanks again. |
|
|
The same way that you would interpret any standardized coefficient. |
|
|
Ok, thanks. I realised now I was getting confused between having fixed me variances and fixing my factor loadings. Thanks for you time. |
|
Back to top |