There's a large and growing body of research on best practices of teaching. Most of it seems premised on the idea that we know exactly how students learn and therefore we know exactly how to teach them. It is largely based on the behavioral model of psychology. Push a button, get a response. Nevermind that behaviorism is largely discredited.
So, I think there's an inherent dishonesty at work in this research on teaching and learning. Dishonest because it denies an inherent complexity and subjectivity which lies at the heart of teaching.
A more honest approach to the question of how best to teach is offered by an educator universally regarded as one of the best in his field. When he taught at Caltech he was not only very popular but also very effective. Give this a listen:
Does this answer bother you? Not measurable? How do you assess? What are the learning outcomes? Yeah, those were not considerations for Feynman. Great teaching was the focus. An interest in engaging students with passion, curiosity, and knowledge. The interest was to create a unique subjective experience for students. As a result of this, learning occurred.
There was no teaching to the test, no asking whether this or that learning outcome was measurable. Feynman was interested in getting students to understand physics.
The question we should be asking is whether we are encouraging such teaching today or discouraging it. It seems that the focus is on doing things that discourage the very sort of teaching that is most effective. It's time to stop discouraging great teaching. It's time to take the risk of letting teachers do great things and sometimes fail.
Do we want to avoid errors at all cost thus assuring mediocrity or do we want to aspire to something more for our students? And ourselves.