Multi-category outcome models need to have an argument that specifies which category should be labelled as fitted. The discrete class/category prediction should be called something else (probably category, level, or fitted.category etc.).
In Stata it is outcome() (as in margins xvar, predict(outcome(3))), which is ambiguous.
This will simply be margins() calls because margins.polr(), etc. methods can cascade that argument (or cycle through all levels - but that's not what Stata does).
Multi-category outcome models need to have an argument that specifies which category should be labelled as
fitted. The discrete class/category prediction should be called something else (probablycategory,level, orfitted.categoryetc.).In Stata it is
outcome()(as inmargins xvar, predict(outcome(3))), which is ambiguous.This will simply be
margins()calls becausemargins.polr(), etc. methods can cascade that argument (or cycle through all levels - but that's not what Stata does).