The article sort of dances around the core problem: a few broad "ethnicity" categories lacks precision and fails to account for a ton of diversity within those

The article sort of dances around the core problem: a few broad "ethnicity" categories lacks precision and fails to account for a ton of diversity within those categories. If you're doing a serious study, I'd keep that question, then add an additional question (probably: "where were your grandparents born?"), and see what the data looks like.