3

Our University follows a letter grading system with discrete A through D ( including half grades like AB, BC etc) I have a relatively large class ( around 100 students).

No matter where I select my grade transitions ( eg A vs AB), there's always some student(s) who "just" lost an A and thay creates a barrage of emails requesting me to consider a better grade. That's the inherent problem of converting a continuous distribution of marks to a discrete set of grades. In other words, if I just had to report a continuous score I don't think the 79 student would agonize about not having recieved an 80. On the other hand, when 79 leads to a BC but 80 leads to a BB, the student does have a greater incentive to complain.

The situation is worse for large classes since the distribution is so dense that there are no natural breaks where say there are no students in a 5 mark interval and that can serve as a grade break. A student does not mind losing an A by 5 marks than "just" 1 mark.

Of course, I kind of see the students point of view as well. On one hand we do tell them not to fight for each small mark but on the other hand a 79 could mean a BC vs a 80 a BB. So I guess the subjectiveness gets amplified.

Is there any workaround that anyone has? How do I make the grading "fair". Or at least give it some semblance of fairness so that I avoid this annual headache.

Would pre annuancing thresholds make a 79-vs-80 penalty seem more fair? I don't see a way out.

Ps. Personally I've started thinking this mapping of continuous to discrete ( at individual course level) to then continuous again ( at a degree level) while calculating a CPI ( which can be a 9.1 and not just an A) does not make much sense. I would rather the university just had us report a continuous score and summed it up over all the courses. But that's beyond me to change!

2
  • 1
    If you need to make a distinction, well, there really is no way around it. Like many other times in life…
    – Jon Custer
    Commented Apr 19 at 17:59
  • @JonCuster so basically go with the scores, ignore the complaints and not try to second guess myself? Commented Apr 19 at 19:01

2 Answers 2

2

Common practice where I've worked: recognise that no-one has perfect intra-marker or inter-marker reliability at the one-percentage-point level. If a student gets 79 (i.e. one mark short of a grade boundary), remark the work and/or get a colleague to remark it blind as many times as necessary (will probably only need to be once or twice) to get a clear majority vote as to whether it's closer to 78 and clearly in the lower grade band or closer to 80 and clearly in the higher grade band.

(If you're getting a colleague to do it, you can use the same remarks to satisfy your institution's requirements for moderation of a sample of marks, so overall little or no extra effort is required.)

3
  • Thanks! But my 79 or 80 is a composite score. Ie it's the weighted and normalised sum of a couple of homeworks, quizzes, a mid and end sem exam etc. So basically to regarde a 79 would be logistically impossible. Commented Apr 19 at 19:00
  • @curious_cat It's a bit less elegant, but you can still do it by picking the ones with an aggregate 79 for a remark on the chronologically last assessment (presumably the end of semester exam). It doesn't introduce any unfairness, because the mark is still (roughly) equally likely to go up or down - it's just applying the highest precision in marking where precision matters most. Commented Apr 19 at 19:34
  • (Of course, if you're doing the remark yourself, that last suggestion might have to be done after you've de-anonymized the work, assuming your usual practice is anonymous marking. But having an independent colleague do it is probably the best practice anyway.) Commented Apr 19 at 19:44
2

I use linear interpolation, picking what I think is an A and what I think is a C such that I get the median I want, and then interpolate in between. It's easy and fair and completely avoids the question of what to do about students who are right on the border: They get what the formula says and that's that. It tends to discourage students begging to be bumped because they were "so close".

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .