CFA in Type Complex vs Type twolevel PreviousNext
Mplus Discussion > Confirmatory Factor Analysis >
 Kathrin Klipker posted on Sunday, October 04, 2009 - 4:32 am
I am writing my diploma thesis on nested data (within individuals) and need to do a CFA.

I ran the same model for type= complex and type=twolevel, but there ist an error message (ill conditioned fisher information matrix – change model) when running it with Type= twolevel.

How can that be? Any suggestions (literature recommendations) would be really helpful
Thank you.
 Linda K. Muthen posted on Sunday, October 04, 2009 - 9:20 am
It sounds like you use the same CFA on both between and within in the TWOLEVEL model as you use for COMPLEX. There is no reason to think that because it fits the overall data used by COMPLEX that it would also fit both the within and between data using TWOLEVEL. I would suspect zero variances on between. Please send your full output and license number to for further information.
 Michael Mitchell posted on Monday, November 21, 2016 - 8:34 am
I have a CFA I'm running that has dichotomous response data, and I want to account for non-independence of individuals within teams so I used type=complex, which provides acceptable fit indices. A reviewer for the mansuscript wanted to have ICCs reported to get a feel for differences in team vs individual responses. Is there a way of getting ICCs in the type=complex output? I ran type=twolevel, but ICCs are not printed for categorical data. Here is the code I used for type=complex:

Names are ResID TeamID Period BTI01-BTI15
BTI19-BTI25 BTI29-BTI36 BTI40-BTI41 BTI43-BTI49;
Missing are BTI01-BTI15
BTI19-BTI25 BTI29-BTI36 BTI40-BTI41 BTI43-BTI49 (99);

usevariables are BTI02-BTI07 BTI09 BTI11-BTI15
BTI23-BTI25 BTI29-BTI32 BTI34-BTI36 BTI40-BTI41 BTI45-BTI46;


Categorical are BTI02-BTI07 BTI09 BTI11-BTI15
BTI23-BTI25 BTI29-BTI32 BTI34-BTI36 BTI40-BTI41 BTI45-BTI46;

Team by BTI02-BTI07;
Motiv by BTI09 BTI11-BTI15;
Theory by BTI29-BTI32 BTI34-BTI36;
Admin by BTI23-BTI25 BTI40 BTI41 BTI45-BTI46;


output: sampstat stand modindices
Thanks in advance for your help
 Bengt O. Muthen posted on Monday, November 21, 2016 - 4:41 pm
Try Estimator = WLSMV;
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message