Richard Hake is among those who have dedicated a significant portion of their careers to taking data about the learning that goes on in his classes and experimented to find ways to improve. He recently posted a follow-up to an earlier posting called "Re: Lecture Isn't Effective: More Evidence." If you teach through lecture, perhaps this is one entry point into a different way.
And I do encourage you to follow his advice and read Robert Morrison's The Lecture System in Teaching Science. It reminds me a bit of a graduate course I had from Dr. Joel Cyprus at Rice many years ago. On the first day of class, he gave a pop quiz right at the start (needless to say, we didn't do well, but we learned to be prepared!). Then he announced, among other things, that we had two weeks to read our textbook. After that, he'd consider it fair to assign homework or ask questions on quizzes or tests about any of the material in the text.
Dr. Cyprus didn't talk much about the textbook in class. He gave pop quizzes at least once a week, I recall, he showed us ways to analyze and design circuits, and he gave us plenty of opportunity to try out what we were learning in class. He also gave plenty of advice about being a professional.
While it's been some years since I have designed circuits professionally, I still remember most of what I learned in that class, and I use the ideas (understanding exponential growth and decay, for example) in other work I do today. I also try to emulate what I've learned from him and from those such as Richard Hake in my teaching.
By the way, that computer described in the article about Dr. Cyprus is one of the first two I ever programmed, the other being an IBM machine.