AMI Parameterization PreviousNext
Mplus Discussion > Confirmatory Factor Analysis >
Message/Author
 Qixiang Fang posted on Friday, June 08, 2018 - 4:58 am
Dear All,

I am trying to test approximate measurement invariance (AMI) of a three-item scale across 22 countries, and while doing so, I have encountered three different ways of parameterization.


1. Free all loadings and intercepts; fix factor means @ 0 for the reference group; fix factor variance @ 1 for all the groups (http://daob.nl/wp-content/papercite-data/pdf/lek2017approximate.pdf.)

2. Free all loadings and intercepts; fix factor means of the reference group @ 0, and factor variance of that group @ 1. (https://www.statmodel.com/download/usersguide/Chapter5.pdf)

3. I also tried freeing all loadings, intercepts, and factor means, while fixing only the factor variance of the reference group @ 1. The model still converged and provided reasonable estimates.


I wonder what is the theoretical difference between these three approaches, which one is preferred, and which one is perhaps wrong?

Many thanks!
Kind regards,
Qixiang
 Bengt O. Muthen posted on Friday, June 08, 2018 - 1:42 pm
1. Are you referring to the input in Figure 4 of that document?

2. Which example are you referring to here?

3. I would have to see the full output - send your output to Support along with your license number.
 Qixiang Fang posted on Monday, June 11, 2018 - 8:23 am
Dear Prof. Muthen,

1. Yes.

2. Example 5.33

3. I will send it right away!

Thanks a lot!

Best,
Qixiang
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: