search

amices / mice

Multivariate Imputation by Chained Equations
https://amices.org/mice/
GNU General Public License v2.0
444 stars 107 forks source link

Reference Based Imputation #476

Closed LeaSchuurmans closed 2 years ago

LeaSchuurmans commented 2 years ago

Hallo,

I was wondering if there is a way to do reference-based imputation like J2R according to Carpenter (2013) with mice. Are there any possibilities?

Thank you, Lea

gerkovink commented 2 years ago

Hi Lea,

You're in luck: As of yesterday there is a whole package devoted to reference based imputation on CRAN called rbmi.

All the best,

Gerko

stefvanbuuren commented 2 years ago

My understanding is that reference-based imputation imputes missing outcomes under the active treatment as if these subjects were allocated to the control group. There is no ready-made procedure in mice that does this, but it should be relatively straightforward to adapt an existing mice.impute.xxx() imputation method to do this for you.

LeaSchuurmans commented 2 years ago

@gerkovink thank you very much, this seems to do exactly what I was looking for! Even though it would be more convenient to implement the J2R in my mice routine as @stefvanbuuren has mentioned.

Thank you for your replies!

gerkovink commented 2 years ago

Cool. FWIW: if you could give us a reprex of your problem and what you would like to get out of an imputation routine, we can see if it fits or can be implemented in the mice workflow. I am not too familiar with reference based imputation to be honest, so the question can we do it remains a bit vague if it is not directly apparent what it is.