You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I've been experimenting with the multimp option in cmest, which is great to have built-in; however, I've noticed that the imputation is carried out using only the variables included in the mediation models, even if the data frame includes additional variables.
It would be great if the package allowed the imputation models to differ (i.e., include a larger set of potential variables) to the mediation models. In cases where the exposure is missing, it may be that variables the are non-confounders within the mediation models are still useful for imputation.
I tried passing my own predictorMatrix as an additional argument to mice within cmest but this returned an error.
Thanks again.
The text was updated successfully, but these errors were encountered:
Hello,
Thanks again for the fantastic package!
I've been experimenting with the
multimp
option incmest
, which is great to have built-in; however, I've noticed that the imputation is carried out using only the variables included in the mediation models, even if the data frame includes additional variables.It would be great if the package allowed the imputation models to differ (i.e., include a larger set of potential variables) to the mediation models. In cases where the exposure is missing, it may be that variables the are non-confounders within the mediation models are still useful for imputation.
I tried passing my own
predictorMatrix
as an additional argument tomice
withincmest
but this returned an error.Thanks again.
The text was updated successfully, but these errors were encountered: