Alternative to -mi estimate-

[paulvonhippel]

When you work with multiply imputed data, the -mi estimate- command is nice, but it has some requirements and doesn't support every estimation command available in Stata.

But the formulas for combining estimates across multiply imputed datasets are quite simple and only require a point estimate and standard error for each imputed copy of the data. Has anyone written a command that implements the formulas from point estimates and standard errors? I'm thinking such a command could have been written before -mi estimate- or afterward to use with commands that -mi estimate- won't accept?

[Rich Goldstein]

what exactly you want to do is not clear; however, you might want to take a look at the user-written -mim- command which is meant to go with -ice-; while mi has an "mi import ice" command, I don't think there is an equivalent export command (but I could be wrong) so you might have to work that out; since I haven't used mim in many years, and I'm not sure what you want to do that mi estimate does not do, I don't know whether mim will do what you want - but, it is the only command I can think of that might do what you want

[Leonardo Guizzetti]

I am not sure what it is you really want, but if you problem is trying to use estimation commands that are not officially supported, take a look at the advanced options under -help mi estimate-, and in particular, -cmdok-. This option tells Stata to proceed even if the estimation is not one of the supported types. The caveat is that the estimation command must return at least b, V and N to e() per [P] Programming -- program properties.

[Felix Bittmann]

Maybe this one can be helpful? I have never tested it myself, tho:

help miparallel