Hello guys?
When analysis my second order factor for the following factor, I came across with the following error message' "The model is probably unidentified. In order to achieve identifiability, it will probably be necessary to impose 1 additional constraint? '. When shall I do to remove the error? Thank you and stay safe!
If CFA is ok then go for initial model. If needed add or remove the links to modify maintaining reasonable value of GOF indicators.
It sounds like you are trying to estimate too many parameters for the number of variables. This can happen with a CFA if you have too few variables for the number of factors, for example. The additional constraint means fixing one of the parameters you are currently estimating, usually to 0 or 1, or to the value of another parameter.
mind it never miss the most important indicaor or dimension to modify the model. May be it's good for GOF but it has no theoretical value then.
Tool development is very important issue in SEM. 5 or 7 point Likert scale usually used for latent variables or you can take a continuous one.
Hello Paul,
In your AMOS model, the construct IT has only one single indicator (BPMS42) but 2 parameters to be estimated. Thus, the model is unidentified as indicated in the error message.
To fix this, you should add one more constraint, normally the variance of the error term (i.e., variance of e18 in your model). The value for this variance should be set as follows: "a conservative value for θδ, such as 0.1sx2, has been chosen, and its associated λ has been set at 0.95sx. Another conservative alternative to consider is to set θδ, for the single indicator at the smallest value found for the other estimated error variances (θδ)" (Anderson & Gerbing, 1988, p.415).
For more detail, please see:
Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin, 103(3), 411-423.
Best,
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India