I was wondering for a CFA model, when variables are specified as either CATEGORICAL or NOMINAL, does Mplus recode such that the response option with the lowest category becomes 0, with other options assigned higher values in increment of 1?
The reason for asking is because a lot of sources identify weighted least square as estimator but do not highlight how a categorical CFA is then interpreted.
Xin Yuan posted on Thursday, August 09, 2018 - 6:07 am
I checked the Manual for the recoding of categorical variables in CFA. Mplus always automatically recodes the variables as lowest category becomes 0, with other options assigned higher values in increment of 1.
In my data set, I have categorical factor indicators, which have categories 0,1,2,3,4,7. First, I run the program without recoding. Then I recoded "7" to "5". So the categories become 0,1,2,3,4,5. After recoding, the result changed (model fit index, factor loading, etc.). Because Mplus always automatically recodes the variables. The result shouldn't change, right?
I don't understand why this happened. Do you know why?