www.cmm.bristol.ac.uk/forum

Hi,

I'm using Realcom Impute for a research project I'm working on (using the realcomImpute command from Stata), and was wondering about the reproducibility of the imputations.

If I run Realcom on the same dataset twice, will the imputations be the same for both? If not, is there a way to set a seed so that the imputations are the same regardless of when/where Realcom is run?

Many thanks,
Brennan

Realcom-Impute does not have a way of setting a random number seed however according to http://www.mathworks.co.uk/help/matlab/ ... artup.html Matlab, which Realcom-Impute is written in, will always have the same random number state after it is started up. This means that if you do exactly the same sequence of operations in two instances of Realcom-Impute you should get the same results.

Thanks Chris, much appreciated.