Closed BernhardClemm closed 2 years ago
@BernhardClemm thanks for the excellent, reproducible example!
I updated the version on GitHub to call model.frame
with just the relevant variables, so you can use the instructions on the main page to install that. Your example should run once you do that.
Also, please be sure to confirm I spelled your name correctly in the NEWS as well.
As expected, the
mix
call applies an omission for missing values. However, it seems to apply this omission to observations that are missing on any variable in the provided data set, not just the variables relevant to the model.For example, if these are the data:
The function will fail:
However, after dropping the one column that is NA for all obs, it works:
I suspect this is caused by line 124 in the file
adaptiveQuad.R
. I don't know whether it is intended, but in other similar packages, i.e. lme4, only relevant variables are checked for NA.