I am running an SEM model and have tested for several moderators. When I apply the formulat for robust chi square diff for MLR based on the link below, I get negative values for the 'cd' in the formula and the chi square diff is also negative. http://www.statmodel.com/chidiff.shtml
I researched on the website and found Web note 12, however, I have a hard time understanding it and applying the examples to my own data.
In my data I have depression levels before treatment and depression levels after treatment. I want to test if type of treatment, sex, and other categorical variables moderate changes from before and after treatment.
Could u please advise how to overcome the negative values on 'cd' and the scaled chi square diff test? Could you please perhaps explain Web note 12? I'm hoping I'll be able to apply it with some guidance. Thanks very much.