Page 1 of 1

missing values and multinomial models

Posted: Mon Apr 19, 2010 2:49 pm
by venimore
I am trying to fit an ordered multinomial repsonse model. It is working fine for covaraaites without missing data but will not allow addition of covariates with missing values. Is this correct? I have not come across this for other models (binary and continuous). I assume this means for univariable analysis creating different datasets for every covariate (if , as i have , there are different missing values for different covariates).