[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Multiple imputation
Reza Daniels" <email@example.com>:
> Dear Statalisters,
> I was wondering whether an identical multiple imputation algorithm for
> IveWare's -%impute- has been written for Stata8. IveWare is the imputation
> and variance estimation add-on program for SAS written by Raghunathan,
> Solenberger & Van Hoewyk (2002).
Patrick Royston has developed a similar procedure called "multivariate
imputation sampling" -mvis- that also uses regression techniques for
multiple imputation. -mvis- is based on a technique originally developed by
van Buuren (1999).
I'm not sure it is _identical_ to the procedure of Raghunathan et al.
However, van Buuren states on his website that IVEWARE is similar to his own
program called MICE so it might be just what you are looking for:
Patrick Royston's software is still in a pre-release version however, it has
been submitted to the Stata Journal for review. He has also been kind to let
me try the preview version and to pass it on to interested members on the
list. Up till now, I have passed it on to just over 20 inquirers. And now, I
will add your name to the list. Check your mail.
I have tried it myself and it seems to work fine. However, for various
reasons, in my current project I have chosen another approach so my own
practical experience in using the program is limited.
van Buuren S., H. C. Boshuizen and D. L. Knook. 1999. Multiple imputation of
missing blood pressure covariates in survival analysis. Statistics in
Medicine 18:681-694. Also see http://www.multiple-imputation.com.
The reason for the software not being available from SSC is to keep the
numbers down until after the review process. However, Patrick Royston is
happy with distributing the software on a personal basis until then. I'm
quite sure it will soon be available (possible in revised form) for download
from SSC. If you can't wait however, feel free to contact me and I will send
you a copy of the preview version.
* For searches and help try: