Possibly selection bias - is there evidence that a student good enough for Oxford would get worse outcomes when attending, say, Harvard, and missing out on tutoring?