The Volokh Conspiracy

He got the C in a class called "Problems in Social Thought," which explored the works of theorists such as Thomas Hobbes, John Locke, and Karl Marx.

Maybe he would have faired better if he'd based his complaint on the labor theory of value!

...particularly when the school refused to do anything about it.

From what I can remember of college, "refus[ing] to do anything about it" is in the highest tradition of the academy.

I wonder how different the college experience would be if they were called 'customers' instead of 'students'.

I like the idea that law schools have mandatory curves, and generally support the idea of curves, believing they should be mandatory, at least for classes of moderate to large size. I prefer being a student under these conditions.

When I was in high school, some teachers were known to be tough graders and some lenient. When you had a teacher that gave only one 90 per class 1/3 of the school's grades were 90 or more, this was unfair to those students.

It is not much of an imposition to ask a teacher to rank his students, and then to use grades as a proxy for rank.

Yes-- some classes are more talented than others, but overall, as a student and as a teacher, I have encountered less variation among groups of students than among faculty grading practices.

Knowing that a student's grades reflect his standing at the school offering the grade also is helpful to both the student and to employers who are looking at the student.

I've taught at the college and university level as a TA for 5 years and as an instructor of record for 10 years: I cannot recall an instructor grading 'on a curve' without disclosing this in the syllabus or orally in the first week of classes.

I find the curving process itself to be stupid: if you have scored tests or graded essays, then students should be graded for the course based on their scores or grades, e.g., if you get 80 out of 100, then you have a B-.

There is no scientifically valid reason to believe that non-randomly assigned students should perform according to a normal distribution, the usual curve method. To assume that 68.2% of the students should perform with one standard deviation of the mean score is senseless.

Moreover, grading on curves does not provide accurate ranking information -- only a ranking provides that! Telling me that 68.2% of the students got Cs does not tell me who was at the lowest end and who was at the highest end of the C spectrum.

Absent positive evidence that the students in the class are a random sample of the population of all students, grading with a normal distribution in academic malpractice.

"if you have scored tests or graded essays, then students should be graded for the course based on their scores or grades, e.g., if you get 80 out of 100, then you have a B-."

That doesn't seem right. What if the next highest score in the class was a 40 because the instructor made the test very difficult. Of course, if the test was designed so that an 80 was reflective of B- mastery, then you could be right.

"Curving" only makes sense when there is a wide distrubution of scores and the curve is done to elevate. If the average score is 95, then the tests are too easy and a curve is silly. "Curving" seems like a lazy approach to proper instruction.

TerrencePhilip:

1981 &1982 are acts specifically directed towards racial discrimination. A complaint which doesn't allege racial discrimination doesn't fall under 1981 or 1982. Also, since there's no property transaction involved, it doesn't fall under 1982.

I can barely see a breach of contract claim here, but the rest of it is plainly frivolous. Filing such an idiotic lawsuit (particularly the Constitutional and civil rights claims) should keep him out of law school, not that one C.

NYU 3L,

I understand a 1981/1982 complaint can't survive if it doesn't allege "racial discrimination." But that is not the reason given by the defendants for bouncing it-- they say it should be dismissed because the plaintiff isn't a "member of a racial minority," adding for good measure in a footnote that he is a "white male."

It is improbable but possible that every student in your class performs identically and all get perfect scores. Do they all get Cs for the class?

I think this gets to the heart of one of the two issues with curving.

1. Curving makes marginally more sense when assignments are subjectively graded rather than objectively graded. If you have a test that can actually have a "perfect score." That is you can have a 100/100, or you know exactly that you got 82/100 correct or some such number. A bell curve becomes fundamentally unfair.

But where tests are graded on a "A" "B" or "C" scale anyway, there's always going to be variation. A bell curve still has problems, but it's not as intrinsically unfair.


2. Where rankings do matter, Curves are important.

It would be nice if everyone were unranked, but speaking as a Law School Student, that's just not the way it works. We might have a circular situation going on. Law Firms want top ranked students because law school ranks students because law firms want ranks.

A really good professor can design a test so that the grades will naturally fall into a proper distribution.

Students despise these tests, because quite simply, they're hard. But a test that it's almost impossible to get a perfect score on is going to be a better measure of learning ability than one where half the class gets a perfect score.

But even good professors miss shoot every once in a while. (the test where the average falls below 50% for example) and other professors simply arent' very good at writing tests.

Curving allows these different settings to be equalized for the purpose of ranking.

I admit curves are scary. My biggest concern when starting Law School was that as someone who'd consistently been in the top 10% of the class, that I'd be getting worse grades because the professor gave out 1 A, and made a C the average. But they do serve a valid purpose.

Did anyone else catch the plaintiff's mistaken and self-serving grade calculation in paragraph 16? He first totals up his response paper losses (lost 2.5% out of 20%), then his exam grade losses (10% out of 75%), then subtracts both out of 100%, and not 20+75=95%, to get his 87.5%. This percentage already has a 5% participation grade added in! Afterwards, he sneakily self-awards an additional 5% for his participation grade to claim a final score of 92.5%.
But in fact, his best possible grade is 87.5% assuming he got the full 5% class participation grade.

I wondered why he was making such a complicated matter out of calculating his semester grade.

One of the most onerous tasks that I face as an occasional university professor is developing criteria for assessing student performance and applying these criteria to calculate grades. I've perfected the following process: (1) assign pre-determined points to the perfect performance of each assigned activity (including a point or two for each class attended); (2) provide a table of what range of total points is associated with each final grade; (3) provide immediate feedback on the points given for each assignment so students knoiw exactly where they stand at all times. I still receive bitter complaints at the end of every semester from persons who did not perform as well as they had expected. Coincidentaly, much of my teaching experience has been at UMass.

In the ideal world, teachers would strive to teach, students would strive to learn, and independent testing and assessment systems would determine how well the members of each group performed their allotted tasks. In such an ideal system students would probably be suing professors for inadequate teaching if they didn't pass their independent assessment exams.

The papers were subjectively graded. Would he have felt any better if the professor simply reassigned the grades on a wider distribution, causing him to receive lower grades throughout the semester and then received his C? If so, what's the difference?

This reminds me of my CivPro final. The professor said the highest score was a 54/100. Does that mean we should have all failed? Of course not.

If this were a math class and he answered all the questions right in the same time it took all other students, he may have an argument. Now he just looks like a crybaby.

Grading on curve makes no sense to me. Never has.

From my experience writing tests and test questions I find that you get a much better measurement of students relative abilities with a test where the average grade is 50% than with one where the average is 80%. The 80% test typically has a lot of easy filler material in order to guarantee that even the bad students get 60-70% of the test right. The 50% test is more likely to challenge the students throughout and allow the instructor to gauge the student's knowledge and thought process more accurately.

With a 50% mean score test you can't use the standard score groupings (91-100 = A, 81-90 = B, etc.). You have to either arbitrarily choose different groupings, which will be suspect unless you;ve taught the course and given similar exams for many years, or use a curve. Unless your class is quite large the grades probably won't fall in a normal distribution (and maybe not then) but statistical information like means and standard deviations are still very useful for setting the appropriate grade groupings.

Of course, my justification for curved grading doesn't apply when the mean score is high. If the test turns out to be relatively easy or your students exceed your expectations they should be given credit for that. Fortunately I never had a professor curve grades down.

Note: my experiences (both as a student and instructor) are mostly in math, the physical sciences and engineering.

The TA seems to have handled the situation poorly.

But did any of you read the brief (which apparently was written by the student himself)? After reading that brief, I have grave doubts that he is capable of "A" work in anything.

Perhaps I simply don't get the liberal arts mindset, but how is an arbitrary 80-90=B scale more fair in any sense than a grade assigned by fit to distribution? Is this not just an expectation to grade inflation?

Given that all distributions converge to a Gaussian by the central limit theorem, you can make an argument that grade by fit to Gaussian distribution is the only fundamentally fair way to rank anything.

Regardless of the merits, the plaintiff's attorney misspelled "Parties" in the second caption of the complaint. Where is the proofreading?

Curving makes marginally more sense when assignments are subjectively graded rather than objectively graded.

Well, you've never taken a test where the average grade is a 57, the top grade is a 81, and the lowest grade is a 4. ;)

Most of my undergrad professors curved to a 2.5 GPA, so half the class got As and Bs, half Cs or below. Some were really generous and curved to a 3.0.

The only time problems arose was when the professors scaled the grades at the end of the class, not per test. Students had no idea what their grades were, because it changed every time someone dropped out. You could theoretically get B+s on every test and end up with a C.

Lesson for professors/TAs: don't curve at the end of the class. Curve every test or paper and let the final grades fall where they may.

From what I can remember of college, "refus[ing] to do anything about it" is in the highest tradition of the academy.

My own little amusing story on curving:

There was a professor at my alma mater, and I had to take the course when he taught it. He said that he never curved. He would start with seventy students, end up with twenty, and flunk half.

It was rumored that he had had his grades handed back to him with instructions that he could not flunk that many people.

He also handed out the grading formula, and I went into the final knowing that if I got a zero, I got a D (very reassuring); a twenty-five, a C; a seventy-five, a B; and since there weren't twenty-five extra credit points, I couldn't get an A.

I got an A. I suspect that semester fed into the rumors about getting his grades handed back.

Just skimmed the Complaint. Anybody else see the rather obvious mathematical errors of this Plaintiff in Paragraph 16 of the Complaint? (1) 20% - 17.5% = 2.5%, NOT 3.5%, as Plaintiff claims; and (2) Plaintiff subtracts 13.5% from 100%, rather than the 95% which the tests (75%) and the papers (20%) comprised. In other words, the entire basis of his claim, that he got 92.5% of all possible points, which he claimed was necessarily worth an A, was based on miserably bath math skills. Not having reviewed his Papers or Tests, I can express no opinion regarding his grasp of the philosophical constructs coveredd during the Course, but if I were the hiring partner at a law firm, I'd bounce the guy just on his poor grasp of quantitative concepts, as evidenced by the poor math skills demonstrated by that Paragraph of his Complaint.