Page 1 of 1

Imputation and perfect prediction

Posted: Wed Dec 10, 2014 11:47 am
by odukesucl
Hi,

I am developing a multi-level imputation model for several categorical variables and want to include many predictor variables in the model. How does REALCOM IMPUTE cope with perfect prediction? I am anticipating PP to be an issue, and read in the book 'Multiple Imputation and Its Application' (Carpenter & Kenward) that a method to cope with this issue in the maximum indicant model is to bound acceptable draws for the latent normal variables being imputed. Is this method included in the software, or does it need to be implemented manually?

Thanks

Re: Imputation and perfect prediction

Posted: Fri Dec 12, 2014 6:18 pm
by Harvey Goldstein
There is no feature in realcom to do this I'm afraid.
Harvey Goldstein

Re: Imputation and perfect prediction

Posted: Sat Feb 14, 2015 1:15 pm
by Harvey Goldstein
We haven't really explored this. There are some bounds that are placed around draws that may deal with this, but you would do best to avoid perfect predictions if you can. Interested to know if it works for you however.
Harvey Goldstein