How do college curves work? Can they bring down the top students' grades?

I always thought that if I get an A on every assignment, there’d be no way I can get anything but an A for a final grade. This thought has not been disproved by my transcript yet (thank goodness), but something interesting was brought up by one of my teachers.

He said that his classes tend to have a double dip distribution, which he explained that the implication of having such a distribution meant that to normalize the curve, or to “shove the two humps into one,” he’d have to bring the students located on the right side of the right hump to the left. (His visual explanation was a little more gracious than this written one).

He eventually said that’s the reason for his exclusion of a curving mechanism from his grading policy, but it got me thinking:

If a threshold for an A is a 94 and I score 96 or better on every assignment (or on essays, an A, which I’m guessing can be no lower than a 94 if converted into numbers), and barring a personal vendetta against me, is it possible for me not to get an A?

On the other hand, how is it possible for someone to get B-'s to B+'s on every assignment and yet still land an A?

Let’s just say in all my years of college instruction that has never happened – getting A’s on everything during the course and not getting an A for the course. Ditto for all B’s etc. But you’re not in my class and an instructor essentially has carte blanche to grade however he or she wants. But if the grading is fair then whatever curving is used a student’s final grade should reflect his or her individual achievements during the course. And yes, I do use a curve, but it almost always improves a student’s grade vs. the other way around.

The curve is, theoretically, designed so that the top whatever percent of the class get an A. There is no way for a curve to reduce the grade of the top scorers in the class. On the other hand, if you got 96 on every exercise, but every other person in the class got 100, if the grades were curved you would get a lower grade than if they were awarded on absolute points.

That is the theory, anyway. Actually, I agree with nivlac in that it just wouldn’t be done that way.

Whether things like this are possible depend on what method the instructor is using to determine course grades. This should be clearly spelled out in the syllabus. If the specified system is “Final grades shall be assigned by spinning the Wheel O’ Grades located in the instructor’s office,” or even something like “A’s will be awarded to the 10% of the class’s members who have the highest overall average, B’s to the next highest 25%…” then sure, it’s possible. I’ve never seen anyone actually grading like this, though, but it’s one thing that “grading on a curve” could mean.

You may find more information in these old threads:
Define: Grading On The Curve?
Grading on a curve
GPAs, grading on a curve, and grade inflation

When I was a TA in engineering grad school I worked on the curve with the professors for the class I helped teach. Never once did we curve people’s grades down - roughly speaking, 90% was always an A, 80% was always at least a B and so on.

But as noted, your professor could do things differently.

Back in the dark ages when I was in college, I had a teacher use a really nasty “negative” curve for a test, but not for final grades. For the first exam the teacher underestimated the classes abilities and made the test too easy. I don’t remember the exact nubers, but I think you needed 100% for an A and 98% for a B, I think I got something like 96% which was a C. I ended up getting a B in the course after the teacher figured out how to make up exams.

We had true bell curve grades* in some of my larger undergraduate graduate classes especially in the psychology department and lower levels of the other departments. In those classes, you didn’t get grades, you just got a score and a had to see where you were tracking against other students by looking at your current z-score. It was pretty brutal and you were in true competition with your classmates. The top 10% got A’s, the next 20% got B’s or something like that. The only reason I put an asterisk was that it was possible for no one to fail as long as everyone made a minimum score but there was a minimum number of D’s as well.

Organic chemistry classes and other pre-med classes are notorious for this in some schools. The tests are so difficult that the top score in the class may be less than 50% for the entire semester let alone just one test. On some quizzes, 30% was the top score and you don’t know what grade you are getting until all of the students scores are compared against a normal curve. You most certainly could fail those classes easily.

College prof here… who cannot for any rational reason figure out why you’d want to screw with students’ grades just for the sake of the “curve”.

What if everyone did well on a test… because they worked hard?

Or what if most students got A’s and B’s, because… you did a good job teaching the material?
Are you going to take the B students and say “Sorry, the D students did better, so now you’re getting a C.”

Do teachers care about students? And what they’re learning? I sometimes wonder.

I’m reminded of some wisdom from my dad:

“The day you get your first job is the day your college grades don’t mean shit anymore.”

Approaching 30 now, he is absolutely correct.

I always thought the curve was to make up for the test being too hard, or the teacher being too bad. That’s why, in one of my psychology classes, I got an A, despite getting a C on all the tests (and I know it wasn’t for attendance. The guy was a horrible teacher, so I started skipping classes.)

I didn’t get A B C D F grades on tests and assignments in college chemistry classes, just number scores. I never really knew what my grade was going to be until I got it at the end of the semester. As long as I always scored well above average I figured I was ok.

In 40 years of teaching at universities, the scenario described never happened. What did happen was this. Write the marks on a piece of paper and look at them. There are always natural breaks. Because people tend to either be able to answer a question or not. Yes, there are part scores for partial answers, but you almost always see the breaks. And if they aren’t there you manufacture then. The, assuming there were good students, you draw a line at the first break and everyone above that line gets an A. Then you find another break and draw another line and that determines the Bs. And so on. Of course, you have to temper this with what you know of the students. But this was never used to lower marks. You did it because there were so few students that could answer all the questions even approximately well.

These were math classes where there was always a correct answer (and a plethora of wrong answers). Marking was reasonably objective. I still recall with pleasure a student who came up on an exam with a completely novel way of doing something. It was much easier than the traditional solution. It blew my mind. Sadly, she went off to law school when she graduated. She’s probably making ten times what I ever did.

I suppose this assumes you don’t go back to school (to pursue postgraduate degrees, get another degree altogether, etc.) after you get your first job.

(It also assumes your real first job you likely took in High School is a mulligan.)

Assuming also that your first job is not a tenure-track academic job where you fail to get tenure, since your next academic tenure-track job application will involve your college grades again.

Not that that’s happened to me.


I was in grad school and helped grade in one place which primarily graded on a curve and another which primarily used absolute breakpoints. If you mix these, you might have to give someone who got all As a B. But if you used a pure system, this would not actually happen. Say you used a curve, and gave tests do easy that everyone got over a 90. You’d then set the curve to go from say a C and 90 to an A at 98 and above. Thus no one getting a 90 would actually get an A. If you defined 90 and above as an A, and then curved the grades, then you’d have problems.

And I agree that it won’t happen in practice since any good instructor who knows how to construct good tests won’t make them this easy - and there is always one or two losers in each class.

Example of the class of cultures: when I taught a data structures class, I made the first quiz reasonably hard, and class average was about 70. The class totally freaked, since though I never gave an absolute grading scale they just assumed that 90 was A, etc. I had to reassure them that they were not all getting Cs.

I’ve only ever seen curves work in favor of the students. I remember one particularly brutal Evolutionary Biology test where I scored a 65%, but ended up with a B+.