When I first computed the grades, the average grades were terrible. I had felt that people getting < 40, say, out of 150, ought to be failing; it seemed to me unreasonable that one couldn’t manage 27% on this not-so-hard exam. But such a policy would have meant failing an unusually large fraction of this class. After studying the data for a long time, I sill felt uncertain. I don’t see any reason why the final exam numbers would be untrustworthy—there were no messed-up questions, students had plenty of time, I rechecked the grading on many of the exams—and yet the data seemed to show signs of unreliability, like a worse-than-usual correlation between midterm and final scores. So I decided to raise many grades.
I hope that many of you liked and gained from this course (even those whose final exams didn't make this so very clear). It will grown on you with time, I hope. Good luck in all your future studies. Do good work.
Regards,
Phil Rogaway