You could do an encouragement design to try and get at a causal effect here. Randomly assign some classes to be encouraged to take notes, not take notes, or do nothing. It'd take a lot more work to get your statistical power, but if you see that over a large aggregate the kids who don't take notes tend to perform better over the long run (you'd definitely want to look at performance in subsequent classes that aren't actively being experimentally manipulated when measuring outcomes), that'd certainly be indicative.

An encouragement design might be more appropriate to the question, anyway. The policy decision from an institution's perspective isn't really, "Should students take notes or not?" It's, "Should we be advising students to take notes or not?"