Melanie posted on Thursday, March 28, 2019 - 5:03 am
I think there is probably a really simple explanation, hopefully, for this, but I'm a bit stuck.
I've run an LCA with 5 dichotomous variables and 1 nominal variable, which has 7 categories. I have declared the nominal variable as such on the NOMINAL ARE command and the output recognises it has 7 categories in the 'UNIVARIATE PROPORTIONS AND COUNTS FOR CATEGORICAL VARIABLES' section.
However, in the MODEL RESULTS section of the output, which I'm using the logits to manually turn into probabilities, there are only 6 categories listed for my nominal variable.
How can I work out which category isn't listed? and why isn't it listed?
This is how it should be. A nominal variable with 7 categories is described by 6 probabilities - the 7th is just the sum of the other 6. So there are only 6 different parameters. Same thing if you regress the nominal variable on x's. See the topic of "multinomial logistic regression (e.g. in our RMA book).
Melanie posted on Thursday, March 28, 2019 - 11:40 pm