How Can I Ethically Adjust Grading Without Violating My 'No Grade Curving' Policy?

Question

I always tell my students that I don't curve their grades, and that their final exam and quiz scores will directly determine their letter grades. I do this to avoid giving them false hope that they might still pass if the majority of students perform poorly.

However, in practice, I often adjust the grading rubrics to ensure that the average grades aren’t too high or too low, especially when the assessments turn out to be easier or harder than intended. This adjustment process makes me feel uneasy, as it feels like I'm effectively curving the grades despite my claim that I won’t.

How can I handle this dilemma ethically? Should I revise my policy, or is there a better way to approach grading adjustments without feeling like I'm compromising academic integrity? Or is there better way to gauge how difficult or easy an exam/quiz would be?

BTW, I do let students retake a portion of the quizzes and I drop the lowest grade of the quizzes.

---

My aim for not curving the grade is to discourage students to assume they will pass even if they get only something like 36% of the points, which some of my colleagues do. And I also want to avoid student come to beg for more points.

My aim for adjust rubrics is that some times I think a problem is very easy, which turn out not to be the case for most students.

Answer 1

Fundamentally grading needs to be fair to every individual. If a good grade by X results in a lower grade for Y then it is fundamentally unfair as the grade no longer depends on the efforts and skills of Y alone, but on something they can't control.

No individual class has a predictable distribution of skills and none will closely match the distribution of the population from which it is drawn. So, you can't make assumptions.

Suppose, hypothetically, that you have a class with 30 students and they happen to be the "best" 30 students on the planet. Assuming, hypothetically, that they are properly engaged, then they will all get full marks in any fair system. Any system that guarantees failure (or low marks) for some of those students is fundamentally unfair.

So, a grading rubric should be constructed when the exam is constructed and "points" per question should be determined at that time. You then live with those decisions (caveat below). If they all deserve top marks based on those decisions then they should get them.

What you can do is, in future delivery of that course, make the rubric a bit lower or make the questions harder. I used to put one very hard (near impossible) question on every exam and warn the students that it would be there. I didn't give it especially large weight, but it would tell me who was excelling in the study of the material.

Caveat. You can advantage a student slightly as long as it doesn't disadvantage other students. So, if 90 was required for A, then 89 would probably also get an A. This only recognizes that my questions aren't perfect and might be misunderstood and my grading isn't perfect either, requiring some small grained decision making on answers. If I'm not perfect, then my grading scheme isn't perfect, so some sloppiness is included so that no one is disadvantaged.

Another practice of mine was, at the end of the course, look at the grade distribution (actually the numeric total, not the letter grades). I'd make a determination informally whether I thought that distribution matched what I considered the actual learning of the students generally (I had reasonable scale for this: 30 or so students.). If I thought the distribution was too low, I'd up it a bit, giving everyone a bonus, raising the distribution. No one was disadvantaged. Students considered me "tough but fair". People didn't pass without effort, but if the effort and results were there, students got what they considered a fair grade. Aim for that.

I've had classes where every student is excellent and so the distribution of grades is naturally at the high end. There is nothing wrong with that. I do my job. They do theirs. Fairness over all.

Changing the rubric after seeing the grades can hardly be fair. It makes grading something other than skill-building into a competition where the students don't actually have any idea of the rules until after the exam.

If department heads and deans complain that your grades are too high, explain the above and defend the students. Toughen the course for the future, if you like. And if you see grades higher than your judgement of the overall skills early enough, toughen the course for future exams. But after the fact downward adjustments are just wrong.

Answer 2

Making good exams and quizes is more of an art than a science.

When I design a test, then I do that with the goal that anyone who knows everything I taught during my lectures should get 100%. But people actually acing a test is very rare. When a student answers a question wrong, then there are three possible reasons:

• It's the student's fault, because they didn't pay attention or didn't study.
• It's my fault as a teacher, because I didn't teach that thing properly or didn't properly communicate its importance.
• It's my fault as a test author, because I phrased the question in an awkward way and the student had no idea what I actually wanted from them.

If it's the students fault, that's on them and they deserve the bad grade. If it's my fault, then I might want to consider being lenient on grading. But how can I tell?

Usually by looking at the results for an individual question in isolation. If everyone got a question wrong, including people who answered most other questions correctly, then it was probably my fault and the question wasn't possible to solve. I might want to reduce the number of points for that question or remove it from grading altogether. But if there are some students who answered a question correctly, then it seems like it was indeed possible to answer it, and those who didn't should lose their points for it.

Answer 3

The rationale of "curving" as you call it is (at least in many places) to produce the same average or distribution of grades in every course. The idea behind that is that this makes grades comparable across courses (and particularly when different people teach the same course at different times), but obviously it will bias things if one group of students in a course as a whole is stronger than another, which can easily happen.

You don't want to do this kind of thing, and I agree with you (assuming that there is no general policy in your university that prescribes this).

There is however another issue, which is that we may sometimes assess the time required for an assessment wrong, or we may misjudge the difficulty of some questions.

In fact we'd be happy to grade so that grades over different courses and different runs of the same course are comparable and fair, but this is not achieved by "curving". If possible, this should be achieved by qualitative descriptors of what the different grades mean in terms of competences and understanding of the students.

A personal grading policy could be guided by such descriptors, which allow for the possibility that some groups of students are just overall better than others (so no "curving"), but they also allow for the possibility to adjust the grading scheme in case that, for example, questions turn out easier or more difficult or more time consuming than expected.

It should be possible to communicate this to the students, making clear that (a) "curving" will not happen but (b) adjustments may happen (for good reasons).

Answer 4

My experience as a (former) student: in my University, it was a general rule that a gaussian curve must be applied before providing the grades for a final exam. In practice, what most professors would do is to grade each test (which were always MCQs in our case) according to the agreed-upon grading rubrics, compute score, and students who scored under 50% of maximum score achievable would fail. After that, the final grade would be given by dividing the score of each student by the highest score achieved by a student. I found this system to be good, because:

• Passing/failing would not be influenced by what others did, only by how each student fared against the exam questions.
• There would be always a perfect grade, even though nobody got some questions right (if the questions were too hard or not correctly written).
• Worst case scenario, if there were a student that aced the exam, everyone would get their grade according to the grading scheme. (Although in practice this rarely happened, and virtually nobody got a barely passing grade because not even top students would manage to ace the exams, but rather students would either fail or get one grade upper than the passing grade.)
• Adjustments were done always in favor of the student.
• If all students would be top performers, all would get (near) perfect scores.

The main caveat of this method would be that students could theoretically ally and ask the top performers to underperform intentionally, in order for bottom of class to get a higher grade, but this never happened because 1. This was a competitive environment (there were incentives for top students) and 2. Top students wouldn't risk having a bad day and accidentally fail by over-underperforming.

Answer 5

Your second topic: "Should I curve?"

When I taught, I looked at the median average. I ordered the scores from high to low and looked at the one in the middle. If that one was below 75 then everyone in the class got the same amount of points added: Whatever it took to get the middle score up to 75.

Example:

Abby: 95 +12 = 107 (recorded as 100)

Bob: 63 +12 = 75

Cedrick: 36 +12 = 48

My reasoning was that if average wasn't at least C, something was wrong. I wasn't teaching well enough. The material was too hard. (Good reasons to curve). I got a bad batch of students. Too many of them arrived lacking prerequisite skills. (Bad reasons to curve). If you feel like you can 'smell' why students weren't doing well then you'd have a better idea as to whether you should curve. If you have taught your classes 3 times or less then I'd say curves are on the table. You'll continually build better curricula, methods, and ideas about what students can accomplish.

Your first topic: "Is changing my rubrics curving?"

Are you changing rubrics after students have received grades so that the next batch of students has different rubrics? That's improving your classes and not curving.

Are you changing rubrics so that students will get different grades on the current test/assignment?

Sub question: Did the students clearly receive the rubrics you are using to grade?

Yes > then yes, that is curving.

No > it's curving, but only you will know. You get to decide whether it's ethical.

Answer 6

If you ask questions of somewhat different difficulty year to year, then it makes sense that the exact percentage grade needed to obtain an A, B, ... would vary from year to year.

It's hard to tell exactly how hard a question is until the you see the students' performance, so it makes sense to set the grade cutoffs based on student performance to some extent.

This is kind of a curve, since you'd be lowering the standard needed to achieve a grade and, to an extent, lowering it more if students struggled more. But it's definitely a different type of curve than deciding in advance how many As, Bs, etc you will give out.

In some of my classes, I follow this kind of policy. At the end of the term, I rank students by percentage grade in the class, and look back at students' work. If a lot of students in the 87-90% range are doing what I'd call A-level work, I may drop the cutoff for an A down to 87%. And so on.

Also, if you don't worry about designing the rubric to give the right average number of points, you can focus better on designing it to assess knowledge.

Answer 7

Curving and adjusting mean the same thing, and I recall being shocked (as a teaching assistant) when a professor who professed no curve explicitly told us to calculate the median and other statistics and assure that about top 10% get A's and so on. Whether it actually works on students is hard to say - some children learn early on that parents and teachers often fail to make good on their promises/threats.

However, as a teaching assistant I also adopted a lenient policy that I would add a few points here and there in an appeal. My justification is that, if a student bothered to come to the appeal, and especially if they bothered to figure out or to ask for a correct solution of the problems, they deserve a few extra points. This is especially true for those who had been generally diligent students, but had a bad day on the exam. One could even stretch this philosophy by saying that the exam does not end when the papers are collected, but only when the final grades are submitted to the administration.