What I think in such cases is "unless I can convince him that this is a bad idea, then I should let him do it". Yes, I understand that sometimes we aren't entirely sure why a thing is bad, we just have a gut feeling that it would break something, or is a bad practice.

So first off I try to convince them with everything I got. That usually works if I have solid concrete reasons. If I don't and let them do it there way. Sometimes it turns out my instincts were right and I get to learn exactly what was right about them. Sometimes it turns out I was wrong and his way was much more efficient and I get to learn something new.

Either way, it doesn't make the junior dev feel I am imposing my will on him. People tend to work harder when they feel it was their idea.