Why is it that university grades do not predict very-long-term retention?

Bahrick (1992) argued that Conway, Cohen, and Stanhope's (1991) failure to find a strong relation between grade and very-long-term retention may, in part, be due to the learning schedules underlying knowledge acquisition in different types of knowledge domains. As an alternative it is proposed that grade is a multicomponent measure that is sensitive to both the amount of knowledge acquired and a learner's ability to use that knowledge-the learner's understanding of the acquired knowledge. A new analysis shows that the course-work component of grade reliably predicts very-long-term retention of knowledge, whereas the final examination component does not