Hi
I am running a model in AMOS but keep getting error " The model is probably unidentified. In order to achieve identifiability, it will probably be necessary to impose 1 additional constraint."
i have tried every possible way, when I run EFA, my pattern matrix factor loading is good >.7 and no cross loading. Can anyone suggest another way to run this, may be SmartPLS?
Hello Reema,
Model identification depends on: (a) the number of variables included in your model; and (b) the number of parameters that must be estimated for the specified model. Under-identification is caused by a model having too many parameters relative to the number of variables. The fix is to: (a) add variables; (b) reduce the number of parameters to be estimated; or both. You can reduce the number of parameters by eliminating paths or by fixing multiple paths to have the same value.
The maximum number of parameters cannot exceed mv * (mv+1)/2, where mv is the number of manifest (measured or observed) variables in a model.
Here is a very good resource on understanding model df and identification:
Rigdon, E. E. (1994). Calculating degrees of freedom for a structural equation model. Structural Equation Modeling: A Multidisciplinary Journal , 1, 274-278. doi:10.1080/01705519409539979
Good luck with your work!
Hi Reema,
your model contains two second-order variables (Mooring and PULL) which have only two primary factors. These are only identified if their effects on "Switching" are significant (and I'd guess, substantial).
The rules about identification of second order variables are the same as the rules for identifying primary factors (see David's post): you must have 3 "outcomes" to get an just-identified structure (with no testable implications) or at least 4 to get an over-identified structure with testable implications).
Please be aware that an overall model may indeed have nonzero df but sub-portions of the model do not.
Best,
Holger
Thanks David Morse and Holger Steinmetz , appreciate your responses!
Reema
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running