That feels like a poor statistical evaluation. Why not test along the way with progressive complexity/depth?

Using attendance is a carrot to get students to show up, which leads to better learning outcomes overall - which should be the goal.