Value-added scores can be confounded because a great teacher may be assigned challenging students and show only modest improvement, analogous to hospitals with low survival rates because they take the hardest surgical cases—so good evaluation must separate the teacher's effect from the composition of the class.