Using firthlogit after multiple imputation

[Edsel Ing]

I have a dataset of 32 subjects with 4 variables, and 4 missing data points for one variable.

I am performing logistic regression, and given the small dataset would like to do some sort of penalized logistic regression.

Is there anyway I can still use firthlogit after my data has undergone multiple imputation?

I also tried to use exact logistic regression but despite M=2048, I get a memory error.

Thanks

Edsel Ing

[Rich Goldstein]

have you tried with the "cmdok" option to "mi estimate"? see "h mi_estimate"

[Edsel Ing]

Thank you much Rich. Will try.