Ryan Hong posted on Thursday, September 15, 2016 - 1:25 am
I am trying to conduct a multiple group CFA (groups = S & P) with 10 binary items loading on 3 factors. I started with the configural model here.
***** USEVARIABLES ARE m1-m10; GROUPING IS Group (0=S 1=P); CATEGORICAL ARE m1-m10; Parameterization = theta; MODEL: affect BY m4@1 (L4) m5* (L5); impul BY m1@1 (L1) m2* (L2) m3* (L3) m6* (L6) m10* (L10); self BY m7@1 (L7) m8* (L8) m9* (L9); [m1$1*] (I1); [m2$1*] (I2); ... [m10$1*] (I10); affect* impul* self*; affect WITH impul*; affect WITH self*; impul WITH self*;
In Model S, the commands are repeated but without labels.
I got an error message: THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED. CHECK YOUR MODEL. PROBLEM INVOLVING THE FOLLOWING PARAMETER: Parameter 42, Group P: M2
I am using MGCFA to examine measurement invariance with categorical indicators.
One of the scales (4 items) I am examining has passed the configural invariance step but did not show adequate fit for scalar invariance. I am proceeding to measure for partial scalar invariance (DIFFTEST compared with configural model) by constraining factor loadings and thresholds one item at a time while relaxing other items (factor loadings and thresholds).
However, I am receiving the following error:
THE CHI-SQUARE COMPUTATION COULD NOT BE COMPLETED BECAUSE OF A SINGULAR MATRIX.
I wonder why I am receiving this error since I have used this same analysis for the other scales I am examining and was able to compute all calculations with the same data set, except with the specific scale mentioned above.
I changed one of the items on the scale just to see if that made a difference and it was able to compute but the DIFFTEST could not be computed and gave this error:
THE CHI-SQUARE DIFFERENCE TEST COULD NOT BE COMPUTED BECAUSE THE H0 MODEL MAY NOT BE NESTED IN THE H1 MODEL. DECREASING THE CONVERGENCE OPTION MAY RESOLVE THIS PROBLEM.