From #456, found that region variable still had some missingness after attempting MICE imputation, so clearly MICE doesn't like the variable or the default imputation method for it (presumably polyreg as region is unordered categorical), even after casting to factor.
Creating issue as might be relevant for other variables. Partly fixed for other (also unordered categorical) variables in e7972aa.
From #456, found that
regionvariable still had some missingness after attempting MICE imputation, so clearly MICE doesn't like the variable or the default imputation method for it (presumablypolyregasregionis unordered categorical), even after casting tofactor.Creating issue as might be relevant for other variables. Partly fixed for other (also unordered categorical) variables in e7972aa.