Closed Generalized closed 1 year ago
No, this is not possible. My package is based on object classes and methods. While I suppose it's possible to define methods for class list
, it would be inappropriate to do so because a list
is a very general thing, whereas what you have here is specialized to multiple imputations.
I understand the reason. Indeed, it would be dangerous... I will try to "wrap" the list into a "mira" class or wait until the "mmrm" will support the "mice". Thank you for checking this.
Let's use this naive scenario (totally nonsensical, but just to illustrate)
While "mice" works with the nlme::gls()
But it doesn't work with mmrm. I reported this to the authors ( https://github.com/openpharma/mmrm/issues/247#issuecomment-1458476424 )
But still there's a workaround to make mmrm working with mice:
Let's compare both results pooled:
But problem is that while gls gives proper "mira" object, mmrm - gives an ordinary list, which won't work with emmeans.
Would it be possible, to allow the same that can be done with mira, only using a list, as above?