You could try forcing the random number generator in front of -ice- by
-set seed # -. If there are any sorting operations in -ice- before the
point where -seed- is actually used for the first time (which is not
how it should be done though), then the results will differ because
-sort- breaks ties in sorting variables in random order, and hence is
affected by the settings of pseudoRNG.
On Thu, Aug 21, 2008 at 9:43 PM, Medina, Tait Runnfeldt
<[email protected]> wrote:
> I am unable to reproduce regression results using imputed data generated via Royston's -ice- command, although I am using the -seed- option. A comparison of a mock original and an attempted replication, illustrating this problem, can be found in step #4 below.
>
> I am using Stata version 10.1. The version of -ice- and -micombine- being used can be found in step #5 below.
--
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
