Set max. number of iterations for imputation (mi impute chained)

[Nicole Kapelle]

Dear all,

Is there a way to restrict the maximum number of iterations for an imputation model? My model does (currently) not reach convergence, but I don’t want to wait for 16,000 iterations to see what is going on. If it doesn’t reach convergence after a couple of hundred iterations, it probably won’t do so at around 16,000 iterations.

I am using the mi impute chained command in Stata 16.

Thanks!

[Nicole Kapelle]

In case anyone else is also looking for a solution: I figured out that one can simply restrict the number of iterations through the option iterate(#). It doesn't seem to be explained in the Stata documentation for mi impute (or at least I couldn't find it).

Here an example:

mi impute chained  (logit, augment iterate(200)) pedu = i.female i.mig i.cohort i.psample c.age, chaindots replace  rseed(123456789)  burnin(20) noisily