Evidence that students cheating on tests graded on a curve is stoopid.

I’m looking for more evidence to back up what I see as a pretty clear thing: if one student cheats on a test that is graded on a curve, they are only hurting the other students. (just in case grading on a curve is not the term you furriners use, it means to simply take the best test score of the students and make it the top grade and take the lowest score and make it the lowest grade. Strict versions make the top score an ‘A’ or ‘1’, and the low score fails. This means that if the best score is 65%, it is an ‘A’.)

Seems like some students feel that cheating on tests is just something that affects only them and maybe the school or ??? Sorta like sticking it to the man, man. I want to develop a short paper on any number of examples where cheating is not as productive as one would think when the test is graded on a curve. Or at the very least that it hurts the person sitting next you in class much worse than it helps you. Sample size of 35 students in a class (so, yeah, I am looking for statistical and mathematical evidence).

Anyone want to help? Disagree? Give me examples where it doesn’t work like I think?


What kind of cheating are we talking about? Writing critical formulas on your hand, which would help you on a few questions but still not guarantee 100%? Stealing an answer key, which would?

Also, there are a zillion ways a test can be curved.

Of course cheating “helps” you more than it hurts any other individual: it moves your grade from being, say, a C, to being an A. That’s a big change in score! Whereas each other individual student is affected either a small amount (grade goes down a step, from a B to a B-, say), or not at all (received the same grade they would otherwise).

One unrigorous but helpful way to think about it might be by analogy to Newton’s third law of motion: every action has an equal and opposite reaction. In this case, the positive effect on the cheater’s grade is equal to the total negative effect on the 34 non-cheater’s grades. In a class of 35, individual noncheaters are hurt 1/34 as much as the cheater is helped (on average).

Clearly the cheater is not “only” hurting the other students, because they attain a much better grade than they would otherwise.

I can see one way where the cheating hurts the cheater immensely: cheater gets caught.

Not an answer key. But looking at the person next to you, answers written on your hand, etc. Actually, I suspect that

And there can obviously be more than one cheater.

Many ways to curve, yes, but it seems to me that it boils down to “For me to cheat and get a better grade means that someone else in class will be pushed down to a lower grade.”


Is this game theory?

My understanding of the concept pretty much stops at “The Prisoner’s Dilemma”.

I had a summer statistics class one year. The teacher was an abnoxious bastard, who mostly read to us directly from a poorly written text in a monotone for two hours at a time. His sole personal interjections were to abuse the reputation of our school and point out how it would never live up to the standards of his alma mater, a prestigious university in Massachusetts.

Everyone did equally poorly on his first exam, but because he was a statistics geek most everyone ended up with a B or a C (I don’t think he had a +/- thing going on, if my memory serves me correctly lo these many years). Now this particular class was a required course for several majors, pre-medical being one of these. The pre-med studnets, of whom there were about 20, got together and somehow, probably through a frat house, got their hands on some of this guys old tests. During class they didn’t grasp the coursework any better than anyone else. Being read statistics from 1 until 3 in the afternoon during a sultry Florida summer afternoon can have a bit of a soporific effect. But after the first test, all of this group that sat together during the class started to get perfect scores on the remaining tests. All of the tests would have 10 questions with no partial credit. The best anyone else in the class did on any of the four remaining tests was 70%. There were about 100 people taking the class.

So the teacher ended up giving the next best grade after “the cheating bastards” a “C”, and then down on from there. He was amazed this was the first time he’d seen some sort of double hump curve. Of course he was too dull to figure out that after the first test results some of these people had jumped up from a 1 or 2 out of 10 to 10 out of 10 on each test might indicate some sort of collusion.

Life is on a curve. It doesn’t matter if the individual class is curved or not. Everyone wants to get the top x% of a group, to get the elite. Whether that is represented by an A or a B+ doesn’t matter much.

That is, suppose the class ends up with 11 As in it instead of 8, due to 3 people cheating. The grades of those 8 fair-As may not have gone down, but they are facing more competition in anything they would want to do with those grades. Anyone picking students now has to pick from 11 instead of 8.

And that school is now rated better by whoever rates schools, and so on.

There’s a finite amount of quality out there, grades are a means to distinguish who has it. If you rate one person higher, everyone else is rated lower, whether directly or indirectly.

I agree with iwakura43 in that even if you can cheat yourself from the worst score in the class of 35 to the best, the worst outcome using most curving schemes would be that some, or possibly all, other students would receive a one step grade reduction. Guilt over this is unlikely to discourage anyone, if they are not put off by the possibility of being kicked out of school.

PS - ShibbOleth:
Is studying from old tests considered cheating? At my university it is considered one of the best ways to study in many courses. The Students’ Union even collects old exams and provides copies to students for a small fee.

Not necessarily. The OP specifies a curve where the highest score is an A and the lowest score is an F.

Take the worst student in the class, and have him cheat his way to an A. Everybody else slides down a notch on the curve. Assuming nothing else changes, the student with the lowest passing grade now becomes the student with the highest failing grade.

One student cheating has just caused another student to fail. That’s not a small effect!

Maybe I’m missing something, but it appears to me that a cheater in the curve system described in the OP would have zero effect on anyone else’s grade unless he got the highest or lowest score in the class.

Let’s say I’m in a class with Ed and Cecil. We all take a test and Ed gets a 15 and Cecil gets a 95. The class marks will be curved over a 80 point scale with everyone getting a mark proportionate to their place on that scale.

Now I didn’t study and would have gotten a 25 (by choosing A for every answer). But I secuced the professor’s wife and slipped into his den afterward to read the test. By doing so I got a 90. Obviously this moved my mark up. But it had no effect on anyone else’s mark - they were still graded on the same 15-95 curve.

What you described is more like a predefined grade distribution system. Grading on a curve involves using statistics to assign a grade based on your performance compared to the rest of the class. Often the grades are fit onto a gaussian curve so that, for example, 10% of students get an A, 20% get a B, 40% get a C, 20% get a D and 10% get an F. So even if the entire class scores above 80/100, some will get A’s and some will get F’s.

Wasn’t this an episode of The Wonder Years? IIRC, Kevin got the same percentage grade two weeks in a row-but the first week it was a C, the second it was a D.

What so few people in this thread seem to miss about a strictly-curved system is that this doesn’t happen.

Say there’s 80 people in the class and I set a curve with 10% As and 10% Fs (this isn’t accurate, but it makes the numbers work out right. That means that no matter how anyone does there will be 8 As.

So, say three F students cheat and get the top grades on the test. That doesn’t mean that there are 11 As. That means that there are still 8 As, and that the lowest three who should have gotten As now get Bs. The lowest 3 who should have gotten Bs now get Cs. And so on.

Make that “so many seem to miss” or “so few seem to get”. Rage makes me less coherent.

I teach statistics, and:

  1. I don’t curve my tests; and

  2. I publish all of my tests, keyed and all.

Problem more or less solved.

If you don’t structure your course like some sort of game, then people won’t game you. And if you publish your tests, then frat houses and student unions won’t have a leg up on the little guys. And the wide availability of said tests removes a large part of the pressure to cheat ( or at least on the rationale of the cheaters).

Then there’s the Honor Pledge and Sanctioned Toolsheet…

Curving is a stupid thing to do, made more problematic by the wide usage of curving by those who lack the statistical training required to do it properly. Testing to a standard of proficiency makes more sense. Curved scores don’t say anything beyond “how stupid you were relative to your peers.”

My $0.02’s worth. First, I think Mathochrist has hit the nail on the head insofar as the OP is concerned. IOW, yes, a cheater on a curved test is stealing from his/her fellow students, not from “the man.”

Second, I respect cerberus’ solution to the problem, but feel constrained to point out that (a) it requires a good deal of judgment and experience to craft a fair examination which measures achievement without a curve and (b) curved exams have the advantage of being able to be more difficult, and thus working the edges of the students’ understanding of the subject. I remember, in particular, with great affection, my high school physics teacher, who routinely gave very difficult exams. He graded on a curve. He also had a completely open book policy. You could bring in literally anything you desired: notes, study guides and of course the course textbook. I enjoyed those exams more than any others I took before or since. They were exhilirating, yet not intimidating, because I knew the curve would make it fair in the end. Though, yeah, I would have felt cheated if there had been cheaters. To my knowledge, there weren’t any and, given the broadness of his parameters, cheating would have been difficult. Assuming there to have been no cheaters, they seemed to me then (and now) some of the fairest exams I’ve ever taken.

This is true, if the purpose of grades is to evaluate whether a student has fulfilled a certain criteria to pass the course and eventually earn a degree. However, another major function of grades is to allow comparison between students - literally how (dare I say) smart you are in comparison to your peers. This is needed for things such as scholarships, acceptance into professional/graduate programs, and in some cases even jobs. Without curving, the top students in a program can all get A’s in everything, making the GPA useless.

I also agree with PBear42 that “way too hard” exams do a better job of testing understanding of the material. Getting my Electrical Engineering degree, I scraped out A’s from some mighty rough finals.

Of course I mean “fulfilled certain criteria.” :smack:

Pardon the triple post but I just had an idea for the OP. Rather than writing your paper on why a student shouldn’t cheat because it hurts his/her peers, maybe you should focus it on why honest students should report any cheating that they become aware of.

I wouldn’t expect the cheater to be altruistic enough to care about his peers. I would, however, expect the peers to act in their own self interest. In my experience, the prof/teacher rarely notices cheating, but the average student sees it all the time.