This is a case where it might prove valuable to examine a larger sample. You can always find one example of something working out or something failing. Instead, we should see what the real probablily is. Perhaps we'd find out that 90% of student/teacher romances end in disaster for the student, and an additional 5% end in disaster for the teacher (pure speculation). I'd be inclined to think that the vast majority of these types of relationships aren't healthy.