I'm not exactly sure what distinguishes an art from a science, but my personal belief is that a large part of teaching is simply communication. While it is certainly a help to feel excited about my material, this will only help if I can communicate this excitement to my students, and also communicate the substance of the material which so excites me. And the problem with communication as a skill is that you can't assess the communication act without taking account of the context within which the act is performed. Delivering a theorem with a smile may be motivating for one student, but for another the smile may mean that the theorem is trivial. Presenting through diagrams may be good for one student, but presenting through role-playing may be better for another.
I think one problem with some educational research is that it seeks universals in abstraction from the communicational context in which they exist. Instead, I think it is more useful to investigate the nature of the teaching _relationship_, and what kinds of communication _interactions_ (rather than communications _acts_) are helpful in building and utilising this relationship.