19

It's my first time teaching a college class as an instructor. My grade breakdown is as follows:

  • Worksheets 15%
  • Lab attendance and participation 15%
  • Homework 40%
  • Final Exam 30%
  • Discussion forum 3% extra credit

I have two TAs who have been doing the homework and worksheet grading. I provided the rubric for the homework to my TAs and they told me that there was a "distribution" in the grades. But after the initial check-in, I never checked the grade distribution again.

Now it is the end of the term and after receiving the (published) homework grades from my TAs, I see that they have a 98% average with virtually all students having a grade above 95%. That is: 55% of the students' grades are nearly perfect.

The attendance and participation grades have not been published yet and the final exam has not been written. Since this is my first course as an instructor, I fear the department's response because the grades might be too high.

But irrespective of the department's response, I'd like to have a grade distribution that's a bit more spread out. I've been writing and re-writing reasons for wanting a grade distribution but it turns out finding these reasons is harder than I thought. I guess it's just an intuition.

Obviously, I'm to blame big time for not having more grade components, to begin with. I'm also partially to blame for not keeping up with my TA's grading but I also told them repeatedly that I was looking for a "distribution" on the homework and feel like they did not warn me that the grades were very high. It's too late for all of that now.

Be all of that as it may, my students will take their final exam soon and they have not received their participation grades yet. I would like to see a grade distribution that roughly ranks the students according to their skills and that makes the testing harder than it has been so far. What are my options, and which one is best? Make use of the participation grade? Make the final exam harder?

Update:

Thanks for the amazing support, your sympathy with my situation, and your helpful responses!

The class I'm teaching is an intro STEM class at a highly selective school. This particular STEM major is very popular, so I imagine that the department wants to use the grading scheme to reduce the number of students.

I based my homework assignments on those from a previous instructor who had a more stereotypical grade distribution. If anything I made them a bit harder. So it's probably the grading instructions to my TAs that led to the differences.

Few of you have said anything about the role of participation grades. Most students were not particularly engaged during lectures (despite me trying to interact with them and using tools like jamboards and surveys etc). So I've been considering not giving every student a 100% for participation but more like something centered around 90% and differentiating from there. I said nothing about the way participation would be graded in the syllabus.

I also didn't specify the grade cut-offs in the syllabus. AFAIK, I can determine the cut-offs myself, within reason. Some students will understandably be annoyed if the cut-off for an A is set at a 93%.

Improve this question
14
  • 2
    Much of this would depend o things like where you are teaching, what department are you with, what level are the students, and so forth. I was in a very challenging US EE dept, but for example, it was considered normal for grad students to get very high grades and also for students in lab courses to get very high grades.
    – JosephDoggie
    Sep 7, 2022 at 12:45
  • 2
    How endemic is cheating on homework at your institution? I tend to give higher weight to exams and individual projects than to homework, so that I’m relatively sure the grade reflects the individual’s work.
    – pjs
    Sep 7, 2022 at 15:55
  • 4
    Hmmm. Then it isn't cheating.
    – Buffy
    Sep 7, 2022 at 16:11
  • 10
    Wouldn't it be expected that an "intro" class in a "highly selective school" would end up with mostly/all high grades?
    – Michael Richardson
    Sep 7, 2022 at 18:09
  • 3
    I'm concerned about your statement of "so I imagine that the department wants to use the grading scheme to reduce the number of students"... don't imagine it is. Find out from your department leadership if it is. Then you will know if it is, or is not a fact.
    – CGCampbell
    Sep 9, 2022 at 13:44

6 Answers 6

Reset to default
61

Let me add a few points to supplement the excellent answers of Juan and cag51.

There are some reasons why you might have such a narrow distribution, including an anomalous group of students. This happens, and statistically is guaranteed to happen (very) occasionally. I've had such a group in the past. I had to work especially hard to keep up with them, producing challenging work. They did well, but they earned it. There is no need to "spread" the distribution for a group of hard working excellent students. Doing so artificially is actually unethical.

While some institutions frown on it, it should be possible for every student in a class to win full marks and possible for every student to fail. If a group is motivated, and if you do your work well, the former is possible. If they are disengaged, don't care about the subject, the latter can happen. I've usually told my students this as a motivator. I've rarely had to fail anyone who actually worked on the material. I wasn't as "enlightened" early in my career, but grew into the role of instructor and mentor.

You are new at this, and novices (even novice teachers) aren't perfect. You may have made some mistakes in targeting the course to your audience. The solution isn't to change the expectations late in the game ("Oh, by the way, this isn't a hundred-yard dash, it is really a marathon. Heh heh heh"), but to rethink the process for the next time you deliver this (or another) course.

Your assignments set your expectations. If the students meet those expectations, then they should get the rewards. You can set higher expectations, but not in mid-stride.

As a learning experience, you might ask a colleague or two to look at your assignments and exams, etc. to give you advice. Perhaps they will have something to say that will help you set more appropriate expectations for the next course. And perhaps they will just say that you did fine.

I'll also suggest that if there was an error on your part, it may not have been grading too leniently, but not offering enough work that was sufficiently challenging.

Teaching is also a learning experience.

Improve this answer
5
  • 27
    "it may not have been grading too leniently, but not offering enough work that was sufficiently challenging" ... my thought exactly. TAs do sometimes tend to grade too leniently, but if everybody did (almost) perfectly the most likely explanation is that the graded homework just wasn't difficult.
    – xLeitix
    Sep 7, 2022 at 14:20
  • 17
    This is an excellent answer, and I want to highlight this portion: 'There is no need to "spread" the distribution for a group of hard working excellent students. Doing so artificially is actually unethical.' More teachers need to keep this in mind.
    – SirTechSpec
    Sep 7, 2022 at 21:33
  • 1
    This is a valid point, particularly because students may learn and work together and for sure they communicate, so high motivation of some socially influential students may spread through the whole class. There is a temptation to think of students as independent and therefore to expect that roughly the same "quality distribution" will appear in every class, but in fact there are good reasons to expect that extraordinarily good and bad classes occur from time to time.
    – Christian Hennig
    Sep 9, 2022 at 9:23
  • But, if 98% of the students are achieving above a 95% in the class, they're obviously cheating. That's what my current university professor brother-in-law told me when I read this question to him....
    – CGCampbell
    Sep 9, 2022 at 13:50
  • 1
    @CGCampbell, your brother-in-law is making some assumptions that may not apply. Simple work and lenient grading can lead to such a result even with "ordinary" students.
    – Buffy
    Sep 9, 2022 at 13:53
46

The goal of a test is not to deliberately establish a grade distribution. There is a standard to you need them to meet and the test should reflect that. If they all pass the standard comfortably then they should all have high grades.

I understand most courses have a different distribution from what you describe here, but I think it is best that you produce the test without being influenced by the current grades, and just honestly set a bar for the standard you need the students to cover on the subject, nothing more, nothing less.

I think the best way to shield yourself from criticism is to produce tests and assignments that follow the standard required of the course. That will be your best defense even if someone criticizes the test scores later.

Improve this answer
13
  • 6
    If all the students merit high grades, there may be an admissions issue (or students being advised to take modules below their level of ability).
    – Dikran Marsupial
    Sep 7, 2022 at 12:56
  • 5
    @DikranMarsupial Before a course, the required outcome or what is required to master can be easily accessed. Either you master these requisites, or you don't. If my peer and I both have the same expectations on ourselves after the course, but my peer performs higher than me, this should not be a reason to fail me, but both should pass. The idea that grades should be a distribution is stupid in my opinion. Either yuou know what you shoiuld after the course, or you don't.
    – DakkVader
    Sep 8, 2022 at 7:39
  • @DakkVader students benefit from taking courses that push them hard to maximise their potential. Where there is no failure and everything is easy for you, there are no grounds for a sense of achievement or satisfaction. I want my students to have a sense of achievement. This is not a criticism of the course or the teaching, just a suggestion that the problem may lie elsewhere (and thus don't make changes to fix a problem that doesn't exist).
    – Dikran Marsupial
    Sep 8, 2022 at 10:17
  • 1
    @DikranMarsupial so you'd fail students or give them lower grades based on each others performances, and not on their performance in the topic? What if one year the average performance is lower? or higher? do you just shift the bell curve?
    – DakkVader
    Sep 8, 2022 at 15:20
  • 1
    @DikranMarsupial "But if all students score highly on a module, that means there is either a problem with the module, with the program or with admissions." Why does that mean a problem exists? Why is it not possible for all of them to have earned that high score?
    – trenthaynes
    Sep 8, 2022 at 22:15
28

Obviously, I'm to blame big time for not having more grade components, to begin with.

Actually, your grading system seems reasonable. You have a test, homework, a lab component, and a participation component. Having more than one exam might have been a good idea, but your system is totally defensible.

I'm also partially to blame for not keeping up with my TA's grading but I also told them repeatedly that I was looking for a "distribution" on the homework and feel like they did not warn me that the grades were very high.

Actually, I think you should consider this more carefully. You have multiple TAs: did they both grade the same way? If so, it sounds like you were not as clear as you thought you were. If not, then you may have a bigger problem: students who had one TA got lower grades than students who had the other. Either way, you should have caught this before now.

It's too late for all of that now.

Yes, it is. Try to understand what happened, but it would be unfair to regrade the homework or change the grading scheme at this point (unless one TA was way harder than the other, in which case, you may need to "normalize"...which will reduce your distribution even further).

What are my options, and which one is best? Make use of the participation grade? Make the final exam harder?

It would be unfair to give an insanely-difficult test just to force the required distribution. Instead, give a reasonably test with a reasonable grading scheme. "Reasonable" is a broad term...you can choose something on the more difficult side of reasonable, but don't overdo it.

As for the participation, it is likely too late there. You should have been clear from the beginning what the participation requirements were and how the grading would be done. And if you were not, you will have to use some reasonable, commonly-accepted standard.

Finally, consider the conversion between numeric grades and letter grades. If you have already defined how percentages translate to grades, you should not change them now. But if you have not, you may have some flexibility here. As before, don't overdo it -- it is not fair to say that 99.5% is an A and 99.3% is a B; if you end up in a situation like this, you will have to award lots of good grades and take your lumps from the department. But on the other hand, a curve like 93%=A, 85%=B would be fine, even if a curve like 85=A, 70=B would be more typical for certain types of courses.

One last suggestion: warn the students (without blaming the TAs) that many students have very high grades going into the exam, so the exam scores will be very important. Students are rightfully upset when they go from an A+ to a C after an exam, especially when the exam is only 30% of their grade. Telling them that this may happen in advance will soften the blow.

Improve this answer
14
  • 26
    I'll add that the most likely outcome of an "insanely difficult" final is a similar distribution, but with lower numbers, punishing everyone. Just. Say. No.
    – Buffy
    Sep 7, 2022 at 12:47
  • 8
    Indeed. Worse, too-difficult exams tend to be arbitrary: the students who perform best tend to be those with stronger backgrounds or more experience working under pressure, rather than those who have the best overall mastery of the course material.
    – cag51 ♦
    Sep 7, 2022 at 14:41
  • 3
    "students who had one TA got lower grades than students who had the other" The easiest way to prevent this problem is to have one TA grade (for example) the even numbered problems and the other grade the odd numbered ones, so that any variation between the TA's is distributed evenly over all students.
    – David E Speyer
    Sep 7, 2022 at 14:54
  • 3
    Even if not actually disallowed, I would not recommend using a very aggressive conversion range. The range I suggested (93=A, 85=B) is likely fine though -- this is the standard scale for many humanities classes in the US (and students are familiar with it from high school), though it's more unusual in STEM. FWIW, I usually had the opposite problem -- students would be very upset about getting a 75% on an exam, and I had to assure them that a 75% was already a high B and would probably become A-range once the homework and everything was factored in. But students don't like low numbers.
    – cag51 ♦
    Sep 7, 2022 at 15:23
  • 6
    @cag51 I would discourage changing the grading scheme to be harder to get high marks. I feel this is only ethical if done at the start of the semester and outlined in the syllabus. Otherwise you're moving the goalposts in a way that harms the student. Suppose someone needed an A to meet some GPA cutoff so was shooting for a 90%, and then last minute you tell them 90% is now a B. They should do this adjustment starting next semester.
    – Drake P
    Sep 7, 2022 at 16:53
8

This is of course just an anecdote from universities I have experience with, but the best advice I can give is this: always prepare the class with the assumption that the grades handed to you by the TA will be so high as to make no difference to the student's final grades. As a result, I would try to make sure the big-ticket items like tests are graded by you, or via scantron. TAs are for labs, homework, etc, while the tests are for distribution enforcement.

At least in my experience, most TAs will universally grade too generously, as it's often what they're taught to do, and what the departments prefer as well. And furthermore, asking them to be the ones that make or break a student is not fair. They don't have the professional training to assume so much of the University's (legal) responsibilities, and it's just a heavy task.

It's little help now, but prepare your future classes expecting your TAs to keep handing you universal scores of 90+. Because among every TA I've ever known, that's pretty par for the course.

Improve this answer
1
  • 1
    For "too good" graded homweork: I don't think that really exists. If they all actually tried to do the homework, they should all get max grades on it. Homework grades IMHO are for effort and trying. Exam grades are whether you actually did learn it
    – Hobbamok
    Sep 9, 2022 at 17:06
2

Obviously this is only relevant for future occasions, but personally I'd assign a higher percentage to the exam, less percentage to homework (as it is relatively easy to submit good homework without actually having done good work themselves), and I'm reluctant to grade participation and attendance at all, because I don't like the students to participate or even attend because there are marks to be earned. I prefer them doing it out of genuine interest or the insight that this will likely improve their performance where it counts, and give them useful knowledge for their later studies and career. If grades are won through participation, this encourages annoying and meaningless contributions from students who think they should do this to improve their grades.

Note also that rather than grading some components, if your department allows it you can make it compulsory to achieve a minimum standard there for even qualifying for the exam, e.g., somebody needs 50% of homework marks to qualify, 75% attendance, and/or once present something in class, but these marks are then ultimately not counted.

Even though exams are not 100% reliable at assessing the achievements of a student, I think they're still much better than the other components, and you also give yourself more control of the final mark if you rate the exam higher.

Improve this answer
1

Why not answer this question with statistics? How many students are taking your class? If this is a large subject group, then there is bound to be a greater spread and less likely that all will maintain a higher grade. You never showed the actual class size. Second, how many have taken this class before. Pre STEM classes are absolutely everywhere right now. On a long enough timeline, with a large enough subject pool, you'll see regression to mean. Perhaps you haven't given it enough time or large enough pool?

OR perhaps you're not doing anything wrong... The last thing, and I honestly don't understand this, is why would a college want their students to fail? I spoke to a mathematics department about a computer program teaching mathematics. The end of the course resulted in a pass/fail final. From this I found that of the students who took the class, completed all of the homework and received 100% on participation and cognition, studied and tried... 17% would fail the class for no discernable reason. And I don't mean because of issues with mental health, stress, family, death, disability as those would allow the students to withdraw from the course. I mean failure for 'no discernable reason'. I realize schools need to make money, but you shouldn't continue to use programs which fail students intentionally. Perhaps you're doing nothing wrong. It's easy to forget you have an ethical obligation to your students, not just your college.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .