www.cmm.bristol.ac.uk/forum

I specified 10 imputations in a Realcom run, but only nine were generated. I discovered this after closing Realcom. Not sure how this occurred since impvals.txt shows 10 files. I'm wondering if it would be valid to run a single iteration and add that to the nine that I have. There is a huge time cost to running 10 imputations, so that would be beneficial. I just don't know how the random numbers generator is seeded.

The random number generators are seeded using Matlab's default method, which I believe means that it is always set to the same value when you start the application (see http://www.mathworks.co.uk/help/matlab/ ... artup.html). This would mean that if you used exactly the same settings for burnin, iterations and imputation iterations you should end up with exactly the same imputed data sets as before.

Bummer. I guess I'll have to start all over. Thanks for the info nonetheless.