Difftest with 'negative' chi-square d... PreviousNext
Mplus Discussion > Categorical Data Modeling >
 Thomas Klausch posted on Thursday, July 12, 2012 - 5:51 am
Dear Linda, Bengt,

Sometimes I encounter the finding that the more constrained model has lower Chi-square than a less constrained model. (How) Is this possible?

For example:
Model 0: df=28, Chi=142
Model 1: df=26, Chi=144

How should I interpret the adj. Chi-square difference (difftest) test in this case? It gives a positive chi-square (about 8, df=2).

Thanks - Thomas
 Linda K. Muthen posted on Thursday, July 12, 2012 - 9:59 am
What estimator are you using?
 Thomas Klausch posted on Friday, July 13, 2012 - 6:44 am
I am using WLSMV in a categorical factor analysis model.
 Linda K. Muthen posted on Friday, July 13, 2012 - 4:43 pm
The chi-square values cannot be compared with WLSMV. This is why DIFFTEST is required for difference testing. You should be looking at only the p-values.
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