A few days ago we got the long-expected and dreaded grades for the catastrophic programming exam. To nearly everyone’s complete surprise, they were quite good. In fact, they were so good that they appeared almost fictional, like they had nothing to do with the exam at all. Consider that the best programmers in the semester group had barely managed to complete half of the exam, and most had accomplished a lot less than that. According to the original grading scheme, under standard conditions (i.e., with 50% of the total available points needed for a passing grade), just the best of us should have gotten a D, and about 90% should have failed. Yet almost everyone passed, people who had completed just a fraction of the assignments got reasonably good grades, and the best ones of us had the A’s and A+’s merited by their programming abilities.
In short, our professor must have adjusted the grading ceiling so dramatically that he could give us the grades we deserved almost without any reference to what happened in the exam itself. I haven’t seen the scale (and I’m not certain I will), but the grading ceiling must have been at 50% to at best 60% of the total points. I personally, with two of the smaller assignments fully done, two more half done, and the two huge assignments barely begun, got an A+; certainly the most unexpected ever. Mind you, I am among the top programmers in our semester group, so one might say this grade is a reasonable reflection of my abilities. Still, it feels fishy, because it’s like the exam never happened. Almost as if our professor had decided to grade us according to what he had seen of our programming skills in the practical course during the term, rather than based on the exam results. Which is of course the system I would prefer, but not the one prescribed by our study regulations.
I talked about shifting grading ceilings before, but so far I hadn’t realized the full extent of the leeway our professors have in adjusting it after the fact–i. e. when they already know how the students did in the exam. The outcome of the automata theory exam this term was already quite surprising for me–the “buffer” of points available above the grading ceiling eventually was a lot larger than our professor had hinted before–but in the programming exam the adjustment was so out of proportion that the result appears almost arbitrary. Meanwhile I am convinced a professor can get almost any result s/he wants. It seems like the professors compute the point totals for all students taking the exam, then either determine how many of them they want to pass, pushing everyone else upwards in proportion–or conversely, decide who should get an A+ and then project the scale downwards from there. Or even both. Our math professor, for one, has published the tally of students getting each individual grade in each exam. Logic had a very large number of people getting a D, and very few with a C or B–or an F. That looks a lot like he set the minimum requirements so that a large number of people would just meet them, maybe including many who otherwise wouldn’t have.
There is something to be said for this flexibility. We have certainly benefited from it in that we got good grades in an exam that went catastrophically wrong. But I can see quite a number of drawbacks in this system as well. For instance, if it’s always possible, by shifting the grading scale, to extract the desired result from almost any exam it’s not essential to really consider how many assignments the students can reasonably complete in the given time. In other words, professors might feel justified to throw everything they can at us, knowing that they can figure out the grading scale later. I am fairly sure that’s what happened, at least unconsciously, with the automata theory and the programming exams. But of course for the students that’s scarce comfort during the exam; even with a reasonable hope that the grading will be kind, the time pressure caused by more assignments than can possibly be completed is real.
There is also the risk, alluded to before, that expectations are made to match the performance of the students, rather than the other way round. And there is certainly a danger of injustice. With 30 to 40 students, semester groups in computer science at UAS are way too small for irregularities to be cancelled out by large numbers. A small group of good people can pull the whole average upwards–or the other way round. So if a student who theoretically would exactly meet the average requirements ends up in a better-than-average semester group, and grading scales are adjusted on the fly, he may get worse grades than another student of the same abilities in a worse-than-average group. There are also minor injustices at either end of the grading scale. For instance, let’s say our programming professor might have reasoned that in his exam the best programmers in the group should get an A or A+ (and we all did). Now that includes both me and my programming partner, whose performance and productivity is certainly 1.5 times to twice mine. If we get the same grade, that’s certainly somewhat unfair. In machine-oriented programming this discrepancy was even measurable. 13 out of 33 people got an A because the grading ceiling was so low. The A+ started at 82 out of 108 possible points, and there were people who made more than 100 points. So there is certainly a lot of different achievement lumped together in that A+.
But then again, it’s hard to argue that exceptional performance should be the benchmark for the best grade, because the effect on the requirements for the average grade would be absurd. In programming, if my partner’s abilities would define an A+, I should merit at best a C-, and many people would fail. And such a grading scale would certainly be entirely accidental. One semester group will contain such a genius, the next will not.
There are some professors who do not accept the philosophy of the flexible grade scale. Our first-term business administration professor sets the grading ceiling at 100% of the available points, meaning the A+ starts at 95% (I missed it by just 3 points out of 100, but only because I made the same error consistently three times and got points subtracted every time; I argued in an email that this is unfair but so far the professor has not honored me with a reply). And our database professor this term said A+ means you know basically everything that is asked and explain it well. So both offer no buffer at all, and apparently accept the results as they come. In a way, though unwelcome to the students, one might argue this approach is fairer and more consistent.