ESEM within CFA with categorical indi... PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Tamsyn Gilbertson posted on Thursday, October 16, 2014 - 10:46 pm
I'm attempting to conduct ESEM within CFA (EWC) with 3-level ordered categorical indicators.

I have example input for continuous data, which uses ESEM results as start values for factor loadings, means and residual variance, while fixing the factor variance to 1 and fixing one cross loading per factor to its ESEM value.

I've tried to apply this analysis to my own categorical data, omitting residual variance specifications and using thresholds in the place of means:

ANALYSIS:
ESTIMATOR=WLSMV;
PARAMETERIZATION=THETA;
MODEL:
F1 BY x1*1.147 x2@-0.029 x3*1.477 x4*0.707 x5*0.027 x6*0.166;
F2 BY x1*0.155 x2*0.800 x3@-0.055 x4*0.124 x5*0.700 x6*0.882;
F1-F2@1;
[x1$1*0.660 x1$2*1.493 x2$1*-0.090 x2$2*0.985 x3$1*0.371 x3$2*1.338 x4$1*0.484 x4$2*1.364 x5$1*0.140 x5$2*1.069 x6$1*-0.095 x6$2*0.591];

Output error message:
THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED...

I would be very grateful for any advice on how to conduct EWC with categorical data (e.g., Is it correct to use thresholds in place of means? What additional constraint(s) should be included for identification?).

Many thanks,
Tamsyn
 Bengt O. Muthen posted on Friday, October 17, 2014 - 11:43 am
It sounds like you are applying only 2*m restrictions to the EFA model with m factors. You need to specify m*m restrictions for the model to be identified. See our handout and video of Topic 1 on our website.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: