I have a multidimensional scale with 4 subfactors (which are rather highly correlated) which I modelled through a second order factor.
Now I want to look closer at the subfactor structure and want to follow a procedure described by Reise, Bonifay, Haviland (2013) which allows to calculate Omega, OmegaH, OmegaS. This is the reason I want to model a bifactor model (see syntax below). Obviously, I take out the second order factor.
However, that is also where the problem arises, where I receive the following error message stating that something is not right in the PSI matrix: THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED. CHECK YOUR MODEL.
So my question is if somebody sees any obvious errors in the syntax that I have missed?
MODEL: G by X01 X02 X03 X04 X06 X07 X08 X09 Y01 Y03 Y04 Y05 Y06; F1 BY X04 X03 X07 X08 Y04 Y05 ; F2 BY X09 Y06 ; F3 BY Y01 Y03 X06 ; F4 BY X01 X02 ; G with F1-F4@0; f1 with f2-f4@0; f2 WITH F1 F3-F4@0; f3 with f1-f2 F4@0; f4 with f1-f3@0;
The non-identification message is because you have 2 factors with only 2 indicators. You need to fix both of those loadings at 1 because with 2 indicators there is only 1 residual correlation that is fitted (it can be picked up in the factor variance or in one loading if the factor variance is fixed at 1).