Add trtAssignment as a "distribution" to defData to ease flow? about simstudy HOT 5 CLOSED

I see the appeal of that, though I do have to say I like the current flow of keeping the missing data process different from the underlying (true) data generation process. They are two different processes, so I think I would like to keep them separate. As you know, though, I am very keen on being able to define the randomized treatment assignment in the data definition - that to me is a key part of the underlying data generation process. And the truncation obviously.

Maybe by excluding the missing data from this will simplify things so that it is not as big a lift once the new dataDef arguments are in place.

from simstudy.

I see what you mean, makes sense. I will likely have time to look it over at the end of the week.

from simstudy.

Should be able to use trtAssign code, or some of it. I guess we would keep trtAssign as well.

from simstudy.

This just sparked an idea... we could possibly rework the data definitions / "dist" column to contain not only distributions but rather "modifications" that are applied to the data so: dists, trtAssign, addMissing, user defined functions(#71) ....
That way the complete data workflow would be contained in the definition and clearly readable, which, as i understood, is a high priority for you.

This would defintely be quite some work but could be of value and as we are considering breaking changes in several different places, it might be a good time to implement such sweeping changes (maybe as simstudy 1.0.0 ?)

from simstudy.

#71 and #75 both seem relevant!

from simstudy.