3

There seems to be a trend with science and math classes especially, where target distribution for the absolute grades are somewhere in mid Fs. But then the curve brings all the students with 30s and 40s into high Bs or even As.

What's the rationale here? Isn't it effectively the same as having assignment weights sum up to over 100% sans the stress aspect?

Layman
  • 1,190
  • 1
  • 9
  • 17
  • 5
    Sorry, this isn't very understandable. And where do your assumptions come from about "trend" and such? – Buffy Mar 02 '23 at 17:56
  • Is this simply a case of a flood of students into a "popular" major and a very difficult entry level course or two to push students into other majors since the place can't handle such a flood in later courses? This happens periodically in CS, for example, but less often in math. – Buffy Mar 02 '23 at 18:03
  • I find it impossible to believe that students scoring in the 30s or 40s can be bumped to an A except in very exceptional circumstances, such as a technical error on the exam, and exceptional circumstances don’t make a trend. I suppose the instructor could be entirely incompetent but again I doubt this is a trend. – ZeroTheHero Mar 02 '23 at 18:04
  • 8
    @ZeroTheHero - one advanced physics course I took was set up so that a grade in the 30's was in fact an A. The homework and tests were made so that any given problem was extremely difficult to actually finish fully and correctly. The point was for the professor to see just how far along you actually got. Different philosophy. An extremely good course all the same. – Jon Custer Mar 02 '23 at 19:22
  • 1
    @JonCuster I can only say I strongly disagree with this approach, and I know of no-one operating on this philosophy: it is disrespectful to students as it assumes they have nothing better to do than work on impossible assignments. it's easy to make arbitrarily hard assignments, and if indeed this was the case then I join the OP is question. – ZeroTheHero Mar 02 '23 at 19:27
  • Can you clarify your definition of "absolute grade"? Does your professor or institution have a well-defined grading scale that says 40% = F? Or do you just have a preconception that a 40% "should" be an F? – cag51 Mar 02 '23 at 19:32
  • 2
    @ZeroTheHero - and yet it was one of the best classes I ever took. I learned a great deal about how to wrestle with difficult problems, which has served me very well over the years. Learning how to approach hard problems in the real world needs practice and training, not a bunch of problems with a neat, simple. "right" answers. – Jon Custer Mar 02 '23 at 19:52
  • Are you having a certain region or even school in mind? – user111388 Mar 02 '23 at 19:55
  • 2
    @ZeroTheHero I don't see any disrespect towards students in asking them to attempt something hard. It would be far more disrespectful to assume they are only capable of something easy. I know I hated math classes where you were primarily graded on demonstrating all the intermediate steps: for some problems, the solution became obvious to me and it was silly to write out the rest of the basic algebra for a course in calculus. – Bryan Krause Mar 02 '23 at 20:06
  • In my math classes the numeric grade was always the percentage of correct questions weighted by difficulty points. Grades above 90 and A+ were pretty usual. Is that not the norm for math? – user253751 Mar 02 '23 at 20:46
  • @BryanKrause I disagree. Students have other things to do than work on arbitrarily hard assignments to the detriment of their other courses. Some may have hold jobs outside the university, some may have other obligations. I recognize the need to have challenging assignments (for which I am locally notorious) and exams, but if the class average is such that students who achieved 30% get a high B or an A in this course, something's wrong with the format. – ZeroTheHero Mar 02 '23 at 21:04
  • 2
    @ZeroTheHero I don't see how posing a difficult question in an exam causes the student to fail their other obligations. – Bryan Krause Mar 02 '23 at 22:38
  • @BryanKrause if every student is expected to study 100 hours per test to get a passing grade because exams or assignments are needlessly hard, the course load in the class is unnecessarily high. – ZeroTheHero Mar 02 '23 at 23:15
  • 4
    @ZeroTheHero No one has suggested an exam that is needlessly difficult with respect to passing. The situation described is one where the threshold for passing an exam is far lower than the maximum possible credit on the exam, rather than a more typical "90% for an 'A'", etc. – Bryan Krause Mar 03 '23 at 00:06
  • 3
    I don't think this is a duplicate of the associated question. That question is about whether curving is fair. This question is about why target a low average score on an exam. One can make 40 a B and 50 an A without curving. – Ian Sudbery Mar 03 '23 at 09:44

2 Answers2

7

The conversion of numbers to grades is entirely arbitrary. For example, in England the norm is something like:

  • 40-55 = C
  • 55-70 = B
  • 70-80 = A
  • 80-85 = A++
  • 90 = you should be teaching not me
  • 95 = you should get a Nobel Prize or the equivalent
  • 100 = I am repenting of all my sins because you must be the Second Coming

Learning flexibility as to conventions is an important life skill. Setting the scale as it is has some useful meaning to the instructor. (For me, 1 point on an exam = what a student who basically knows how to do this problem can do in 30 seconds.)

Also, doing different multiples can relinearize the scale in different ways. If you want to require that an A student gets twice as many points as a C student, you can't do it simply by multiplying all the points as extra credit.

Alexander Woo
  • 24,741
  • 3
  • 59
  • 91
  • 1
    As an outsider to the English system: is this much exaggerated or are exams really created in a way that, say, 1 person in 20 years ever hat 95%? – user111388 Mar 02 '23 at 19:58
  • @user111388 https://academia.stackexchange.com/q/191600/17254 – Anyon Mar 02 '23 at 20:16
  • 1
    I got a 96 for one of my UK MSc courses. I'm sure my Nobel Prize must be in the mail =) – Allure Mar 03 '23 at 01:17
  • 1
    @user111388 it is a little exaggerated in the 85-100 range. But in my 8 years as faculty, I've only given one 90, and I'm quite a generous grader. I'd say that there is probably one 90 a year in our department (on a single assessment, usually coursework, not averaged over the year). Also worth noting that we don't use letter grades. So what this answer calls A+++ we call "1st class with distinction", A - 1st class honors, B encompases 2nd class honors (upper and lower), C encompasses 3rd class honours and pass without honors. Curving is also less common in the UK I believe. – Ian Sudbery Mar 03 '23 at 09:22
  • @IanSudbery: Thank you, that's really interesting! So, is 100 different from "everything was answered correctly"? In other words, if you would immediately after you created an exam, solve it, would you obtain 100 points? – user111388 Mar 03 '23 at 21:26
  • @user111388: Let me point out that it is trivial for me to create an exam that I could not solve all of: just put the Riemann Hypothesis on it. (Note the Riemann Hypothesis is equivalent to a statement about the distribution of prime numbers that can be explained to a calculus class.) – Alexander Woo Mar 03 '23 at 21:58
  • @AlexanderWoo: Yes. But I mean not an artificially hard exam, but an exam you deem fair (say the last exam you gave). Would you score 100? – user111388 Mar 04 '23 at 07:45
  • @user111388 I don't know how it works in maths, but in biology we set questions where there isn't a 100% correct answer. There is always something more you can say, more points to be made. It's not exactly about being subjective, but about there being a while discipline of things you could say, and picking which to use. – Ian Sudbery Mar 04 '23 at 09:44
  • @user111388 - If I were teaching in England, I would definitely write exams where I would not score 100. – Alexander Woo Mar 04 '23 at 15:45
2

You seem to assume that percentage of correct answers is independent of the questions asked.

Any experienced exam writer can manipulate the median / mean in a wide range. Sometimes, it is more important to give difficult questions and expect a good student to solve a quarter of them then to give easy questions and expect a good student to solve over 90% correctly.

You seem to be complaining more about the stress level caused by not being able to answer questions. That shows probably a lack of communication by the instructor in charge.

Thomas Schwarz
  • 22,862
  • 2
  • 54
  • 92
  • That's not quite what I meant. My thinking is more along the lines of if you are expecting a very good student to say 5 out of 20 questions tops, why not weight each question at 20 points instead – Layman Mar 02 '23 at 19:35
  • 3
    @Layman - what's the difference? If everything is announced ahead of time, then you know what is going on. – Alexander Woo Mar 02 '23 at 19:36
  • @AlexanderWoo if there is no difference then what's the point of doing this? The instructor is simply finding what students don't know rather than what they know. – ZeroTheHero Mar 02 '23 at 19:39
  • 2
    @ZeroTheHero - Solving problems that are new to you (but based on what you should know) is an important life skill, and I want to test whether students have learned to do this. Problem solving in the context of a timed exam has a high failure rate. – Alexander Woo Mar 02 '23 at 19:46
  • 1
    @Layman - that doesn't work if you're expecting a C student to get 5 out of 20, a B student to get 10 out of 20, and an A student to get 15 out of 20. – Alexander Woo Mar 02 '23 at 19:52
  • @AlexanderWoo I don't have a problem with challenging problems which force students to reflect on the material and, with suitable guidance, can be solved but if the outcome is such that students with 30% get an A then what kind of grade will students with 60% get? And if the best students systematically get 30%, the problem is with the course material, not the students. – ZeroTheHero Mar 02 '23 at 21:09
  • 1
    @ZeroTheHero - I write an exam with 5 boring problems I expect most students to do, and I want the best students to also solve 1 interesting problem. I know 10 interesting problems and decide the best students might as well have a choice of which interesting problem to do, the choice of trying to make partial progress on 2 or 3 of them, and also the choice of skipping some boring problems and doing several interesting problems instead. Then the exam becomes one where the best reasonable grade is around 40% (the 5 boring problems plus 1 of the interesting problems, out of 15). – Alexander Woo Mar 02 '23 at 21:25
  • @AlexanderWoo so what’s your saying (if I understand correctly) is that in effect your exam is out of 50, not 100. I do a variation of your scheme but the students know ahead of time they can complete only (say) 5 of 7 questions to get a perfect score. This does not yield a final distribution somewhere in the mid-F, nor does it causes students to panic because they only completed a fraction of the exam. – ZeroTheHero Mar 02 '23 at 22:18
  • @ZeroTheHero: I think we are all talking about various schemes where the students know (at least approximately) what the conversion of numbers to grades is ahead of time. I don't see why you can't just tell students that 15-25=C, 25-35=B, 35-100=A on such an exam. – Alexander Woo Mar 02 '23 at 23:31
  • @AlexanderWoo Under your scheme it is not possible to have an average of mid-F (or any F for that matter). If this is your interpretation of mid-F, it is now the way I understood the OP. – ZeroTheHero Mar 03 '23 at 00:01
  • @ZeroTheHero That description is just OP's misunderstanding of this scheme. They are used to a grade around 50 being an "F", so they are describing the median grade as an F. The exam is curved so that 50 is not an F. See AlexanderWoo's 5/10/15 example above for why it may not be appropriate to just declare the exam "out of 50". – Bryan Krause Mar 03 '23 at 00:57
  • @BryanKrause well that’s just not how I understood things (at all). The OP should clarify this point as it would change pretty radically the perspective of the question. – ZeroTheHero Mar 03 '23 at 01:06
  • 1
    @ZeroTheHero How else are you interpreting "distribution in the mid Fs before curve", where people actually get As Bs and Cs? Clearly they are not actually Fs, they are actually the grades the students receive, whatever number is put on those letters. – Bryan Krause Mar 03 '23 at 01:31
  • @BryanKrause so you agree that - as written by the OP - curving a distribution from mid-F to Bs and As doesn’t make any sense. Granted your interpretation is possible but (unfortunately) I know of cases where your interpretation did not apply - and it didn’t make sense either. You may also be interested by this story (although the context is different). – ZeroTheHero Mar 03 '23 at 02:01
  • @ZeroTheHero no, I think that they are disagreeing with the notion that a grade of 30-40 is inherently associated to an F. That may usually be the custom in K-12, but there’s no inherent tie between the letter grade and the numerical grade. – fyrepenguin Mar 03 '23 at 03:26
  • @ZeroTheHero Not arguing for setting exams where students only get 40 on average, but "the instructor is simply finding what students don't know" is generally thought of as being the most pedagogically valid purpose for an exam. – Ian Sudbery Mar 03 '23 at 09:29
  • @IanSudbery yes but in an exam where the average is 30% are you achieving this purpose? The instructor basically finds almost all the class cannot solve hard really problems, something no doubt they knew before the class started. – ZeroTheHero Mar 03 '23 at 16:28