2nd order multilevel CFA with formati... PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Matthew Wong posted on Wednesday, October 16, 2013 - 3:09 pm
Hi Drs. Muthen, I am interested in examining a second-order formative construct that is formed by 5 sub-dimensions. The sub-dimensions are measured with 3-reflective items each. The formative construct itself has 3 reflective measures for identification purposes. The data is multilevel with individuals nested in companies. From various examples on the Internet and the Topic 1 formative indicator specification I have assembled this syntax snippet and wanted to know if it appears correct (assuming this model can be tested).

At the end of the %BETWEEN% syntax in a TWOLEVEL model I have:

ecb BY;
ecb ON oeb@1 sab cb ldsb odcb;
ecb@0;
EC1 EC2 EC3 ON ecb;

Where oeb, sab, cb, ldsb, odcb are latent variables and EC1-3 are the reflective observables of ecb.

Thank you!
 Bengt O. Muthen posted on Thursday, October 17, 2013 - 9:13 am
If oeb, sab, cb, ldsb, odcb are latent variables measured by reflective indicators, this looks correct and that it would be identified.
 Matthew Wong posted on Thursday, October 17, 2013 - 11:13 am
Hi Dr. Muthen, thanks for the reply. Those variables are latent variables measured by reflective indicators. They are specified in a 5-within and 5-between factor model.
 Tunde Ogunfowora posted on Tuesday, March 25, 2014 - 12:27 pm
Hi Dr. Muthen,

I am relatively new to mplus. I have a formative construct measured with 5 items. I am wondering how to specify this construct at both the between and within levels. My reading of previous posts is that one needs to specify the following in a non multilevel context:

EINJ by;
EINJ@0;
EINJ on EINJ@1 EINJ2-EINJ5;

In my sample, respondents self-reported on these five questions but they are nested within organizations.

I have tried to specify the following to account for the multilevel nature but the model is not running (I have two factors, f2 is not formative).

%within%
f2_W by x1-x3;
EINJ_W by;
EINJ_W@0;
EINJ_W on EINJ1@1 EINJ2-EINJ5;

%between%
f2_B by x1-x3;
EINJ_B by;
EINJ_B@0;
EINJ_B on EINJ1@1 EINJ2-EINJ5;

I get the following error messages:

THE MODEL ESTIMATION DID NOT TERMINATE NORMALLY DUE TO AN ILL-CONDITIONED
FISHER INFORMATION MATRIX. CHANGE YOUR MODEL AND/OR STARTING VALUES.

THE MODEL ESTIMATION DID NOT TERMINATE NORMALLY DUE TO A NON-POSITIVE
DEFINITE FISHER INFORMATION MATRIX...........


Any help would be greatly appreciated.
 Linda K. Muthen posted on Tuesday, March 25, 2014 - 1:36 pm
Please send the output and your license number to support@statmodel.com.
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: