If you think about it, Jonathan's answer makes total sense. The right thing for a common name like Smith is very different than for an uncommon name like mine. I define members of the study as anyone who had the name "Garrigues" or one of the common variants even if they aren't descended, so spouses are included. Female children are also included, but their children aren't (unless I find one who took their mother's name, which I haven't yet).
This does not mean that I'm not interested in finding all descendants and I add them to wikitree when I identify them, but I don't consider them part of the study.