Favorite Sentence from Wu's paper

"It is an intriguing question how to judge students’ learning processes if they are fed extremely defective information."


He goes on for quite a bit about how in ed research, how we don't decouple the student from the teacher. We measure students' performance, but not don't control the input variables: teacher knowledge/methodology/pedagogy/etc/etc/etc.

Thinking of it this way, it's worth going back to even the most well known research on the effects of teaching on IQ and student performance and asking how much of those outcomes are simply transference: maybe they really are merely measuring teacher IQ, in a sense . Certainly, without decoupling, we can't tell which pieces matter. Catherine posted a few weeks ago about Miss Apple Daisy, who actually changed the outcome later in life for nearly every single student she had. Was she a necessary or a sufficient outcome for them?

Decoupling would require controlling the variables. That means controlling the teachers and the curriculum. It means Direct Instruction is a Necessary, but Not Sufficient condition. First, you have to create the means by which the teachers are controlled enough that we could even tell what script works and what doesn't--DI controls its teachers. But then we still have to test what makes all kids understand. Funny idea, that we would believe all kids could understand, isn't it?