Teaching style / technique came up at the MPIA mess hall lunch table last week. I said that I use the highly interactive style. Some at the table opined that it probably only helps the best students and is bad for the worst. I differed, but realized that I have essentially no objective evidence. There are certainly studies on the subject, many of which do limited forms of controlled experiments, and they tend to support my view. But it is truly impossible to do a proper differential experiment.
You can't teach two classes at identical quality in two different styles to two identical sets of incoming students. It gets worse the more you think about it. For just one example of the biases: Even if you get two teachers to agree, the one that agrees to do the newer style will tend to be younger; that is, very different when it comes to interacting with the students. And for another: There is no agreed-upon evaluation of student knowledge or aptitude, before or after, and the ones that exist (FCI, for example) favor certain kinds of knowledge.
This relates to a bigger issue I sometimes expound upon from my soap box: When it comes to human or important things, like teaching, hiring, promotion, and supervision, scientists tend to become non-scientists, and throw repeatable empiricism out the window. They are convinced that what they are doing is right despite being able to muster no piece of objective evidence (indeed, evidence is often concealed behind confidentiality or human subjects rules).
For all these reasons, I am a big fan of those who are trying to actually make measurements of the effectiveness of various educational strategies. More power to them, and I will continue to take their advice.