Standare error of estimated mean of t... PreviousNext
Mplus Discussion > Categorical Data Modeling >
Message/Author
 Lianjiang Li posted on Tuesday, July 23, 2002 - 5:53 pm
I try to compare the means of a latent construct across two groups. All observed variables are binary. I use WLS as estimator. The results make sense as the mean of the reference group is 0 and the mean of another group is .34. But the standard error is computed and is listed as ******. Did I do something wrong?

Curiously, when I change the estimator to either WLSM or WLSMV, I was told that my model may be misspeified. But with WLS no such warning is issued.

Lianjiang Li
Hong Kong Baptist University
 bmuthen posted on Wednesday, July 24, 2002 - 7:38 am
Please send input, output, and data to support@statmodel.com.
 Lianjiang Li posted on Wednesday, July 24, 2002 - 7:43 pm
I'm sure other users won't make the mistake that I made, but I'd like to post Professor Linda Muthen's response to my inquiry for the record. I, of course, am very grateful for her kindness.

Lianjiang Li
Hong Kong Baptist University

Here is Professor Muthen's message:

Lianjiang,

The problem with you model is that you have not set the metric of the factor. To do this, you need to allow one factor loading to be fixed at one or fix the variance of the factor to one. I have fixed the variance of the factor at one and attach the output.

Please let me know if you have any other questions.

Linda
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: