This general idea of removing bias crossed my mind. How should we, if at all, deal with section to section discrepancies? I like the idea of mixing up the exams and having the person's name on the very back page of the last problem. That way, you can keep the scoring page on the front, and when you hand back the exams, you just turn the pile over and they can look for their names without seeing each other's scores. It also reduces section bias. But undoubtedly there is differences for the homework portion and emphasis of material from section to section. Maybe one of the TA emphasizes a certain type of problem and grades with respect to that emphasis. So even unbiased grading on midterm and final exams will be lopsided. I definitely would either scale each section so that all sections are comparable, or assign grades based on individual sections. TAs might expect a certain student to do well or do poorly.
See also next post.
No comments:
Post a Comment