Performing a Backward Hierarchical Stepwise Selection on Multiple Imputed Data

[Maria Elizabeth Mercado]

Hi can someone help me figure out how to perform this in Stata 15? I looked for possible solutions and I just don't know how to do the LASSO solution. And "stepwise" prefix does not work with mi estimate:.

I also downloaded the mim command, and that does not allow stepwise as well.

Hope someone could point me in the right direction or let me know how they do this.

[Maria Elizabeth Mercado]

I installed a different mim command file and that seemed to work.

[daniel klein]

Just a critical comment if I may: Multiple imputation is all about getting the inference right when missing data are present; stepwise selection is all about well, I do not really know, it has been severely criticized (see, e.g., What are some of the problems with stepwise regression?) and is certainly not about getting inference right. This is to say, just because you technically found a way to do something does not mean that doing it is theoretically sound. Obviously, you need to make that judgment.