# Grades: curve or no curve

Should you grade on a curve or not? If you are student, the answer is clear: go by whatever the instructor does. Otherwise, you have a choice. I don’t like to tell other instructors or faculty what to do because I respect their freedom. For my classes, there is no curve. Why? Well, the question really is: “why grade on a curve?” I don’t know the exact reason for particular instructors, but I can come up with some possible reasons.

### Curve for competition

This is a very common curving reason. The basic idea is that the class is a competition between the students. The strong survive. The weak fail. In this case, class is sort of like the olympics in that not everyone can win. But it is also a case that not everyone can fail. This is why many students ask to be graded on a curve (he can’t fail everyone, can he?).

I like to think that learning is not a competition. Competition in learning would discourage collaboration and students helping students (which I think is a great way for them to learn). Without a curve, it is possible for everyone to succeed.

### Curve to determine test difficulty

I like this reason better, although I still don’t agree. Some faculty say they curve because maybe the test was just too difficult. That can clearly happen. The question I think instructors should ask themselves “what are the key ideas I want to evaluate students?” Also: “what questions would be able to show that students know that stuff?” If you really know the answers to these two questions, then you should be able to make a test that is fair.

Example: After finishing the stuff on Newton’s second law, I would expect that students could:

• Correctly draw a free-body-diagram
• Solve for an unknown force on an object in equilibrium or with constant acceleration
• Determine the motion of an object with known forces on it
• etc….

So, a reasonable test question would be to perhaps find the acceleration of a block sliding down an inclined plane. One mistake I have seen many faculty make is that they are afraid. Afraid that everyone will get an A on the test. Trust me, that is not going to happen. That problem may seem trivially simple to you, but perhaps you forgot what it was like to to an introductory physics course. If that is a reasonable objective, it is a reasonable objective.

On a related note, I try not to make my tests too long. My general rule of thumb: I write the test, then I make out the key (with complete solution). If I can make the key in 10-15 minutes, then a student should be able to complete the whole test in 50 minutes (well, most students). I am not a big fan of time limits, but sometimes you have no choice.

Ok, back to curving. The idea of the curve in this case is to assume the following:

• Any given class is a good normal distributions of students from the whole population of students.
• The whole population of students has a normal distribution of abilities. Where by normal distribution I mean that most students would be in the C range. There would be fewer in the F and A ranges.

So, the instructor writes a test. Assuming the students are normal, you can essentially determine the universal difficulty of the test and adjust the grades accordingly. That seems like an ok idea, but when you have smaller classes it is possible that the class is not a good representation of the entire population of students. Also, doing this essentially makes class a competition (whether you want that or not).

### Curving just because

There are some instructors (not any that I know, but I am sure they exist) that curve just because. They curve because that is they way they were graded and that is what other instructors do. Perhaps they curve because the students want that. On a side note, students in introductory astronomy for non-science majors WANT multiple-choice tests even though they perform significantly poorer on them.

If an instructor (by instructor I mean anyone who teaches a course) falls into this category, he or she should think about why they do what they do.

1. #1 Uncle Al
January 27, 2009

Grade on a curve to objectively separate the better from the worse. The middle hump is mediocrity, the viscous basis of every society. The top slice is the Severely and Profoundly Gifted, the folks who create the future and must be cherished. The bottom slice is the incompetent who will become crooks, managers, and Congresscritters. In the real world the Severely and Profoundly Gifted are crushed while the Severely and Profoundly Stupid are cherished (for a preponderent majority cannot be wrong).

In the Fine Arts the competent are working drudges while the incompetent are avant-garde darlings (while they last). In the Liberal Arts everybody gets an A. How can anybody screw up “Hermeneutics of Brythonic Poetic Rhythms in Keith Ledger’s Dark Knight Monologues”?

2. #2 Paul
January 28, 2009

In my experience exam grades – and class grades as a whole – are generally not distributed normally. Instead they are some kind of bimodal distribution, with a clump of high grades, a clump of low grades, and some gap in the middle. My experiences come from teaching freshmen mathematics courses of 150-800 students (the wide range comes from varying target majors, as well as “off cycle vs. on cycle” course offerings). I have a respectable sample size. The underlying assumption(s) that students and their exam grades somehow fit a Normal distribution is thus faulty.

Curiously, the major exception to the bimodal phenomena comes when the exams are multiple choice in nature. Those types of exams seem to generate more normal grade distributions – almost as if the students guess randomly with a probability p of selecting the correct answer; the Central Limit Theorem really takes over to generate the normal distribution!

I have also found that grade distributions in upper level courses are not bimodal. In those courses the grades mostly clump in the high range, with a small number (less than 20% of the class) of stragglers.

In any case, I abandon curved grading schemes rather early on. The “curve” concept doesn’t seem to fit the data.

3. #3 Kimberly
May 15, 2009

Interesting article. I am a student who just completed an Economics class. I got a 95 percent in the class (which by our university standards is an A, worth 4 quality points) however, because of grade banding, which the School of Business mandates, I got downgraded to an A- (which is 3.7 quality points). I feel cheated and think it is unfair…but what can I do? Is there anything I can do to appeal this? Is this fair in your opinion?

4. #4 Rhett
May 15, 2009

@Kimberly,

I don’t think there is much you can do. Appealing will likely do little – especially if the curve was explicit in the syllabus. I don’t think the curve is a good idea, but if it is clear from the beginning that is the way it will be, I guess it is fair. It still stinks though.

5. #5 Ana
October 27, 2009

Hi, there!

I’m an Spanish-English translator and I happen to be translating an education programme, where the teacher says “there will be no curve”. I see you use the same expression “curve” in your blog. Could you please explain me what it means? I can’t find a translation or a definition and I’m on a deadline!! Thank you very much in advance.

6. #6 Rhett
October 27, 2009

@ana

Essentially a grading on “a curve” means that the grades will be adjusted to fit a normal distribution. Most people interpret the normal grade distribution as the average (and most common) grade being a “C”.

7. #7 Dave
February 28, 2012

Hello,

My question is the following: Does it “benefit,” and by “benefit” I mean does it actually grant more students a passing grade ( i.e ‘C’) if the standard deviation is smaller, say between 1-4, than a higher standard deviation? Let’s say that on the 1st exam the mean was 32 (out of 40 possible points) and I score a 20. Then on the 2nd exam the mean was 30 out of 40 possible points and I get 28. So, my question ultimately is does it benefit everyone if the mean is lower or closer to the perfect score?

8. #8 larry lurio
April 15, 2012

i dont really understand how you ca NOT grade on a curve. what is the alternative? do you just decide that 90% is an a and 80% a b etc. why? that seems totally arbitrary.