Every spring (including the spring about to start), I teach an honors course that normally has an enrollment of about 25, with 15 to 20 of those coming from the top few percent of our students and the rest coming from the top 20%. I give a series of quite challenging assignments that require both a lot of technical skill and a lot of creativity, and students are not permitted to work together. Usually about 2/3 of the students earn grades of A. Grades below B are rare but not unheard of.

I have always felt like the students are missing out on an excellent experience by not being allowed to work together, but I have also always worried that if they work in groups, then it will become impossible for me to tell the B and C students from the A students. This matters for a lot more than just their grades, because a large fraction of these students end up asking me for letters of recommendation and I like to single out particularly brilliant solutions they've come up with. With groups, I never know who had the ideas.

This term, for complicated reasons, I will have less TA support than usual, which creates an argument for dividing the class into say, groups of three, with each group submitting one paper --- thereby cutting the grading by 2/3.

Possible policies are:

  1. Stick with individual submissions and figure out a way to get them graded (maybe by enlisting some star undergraduates from recent years).

  2. Let students work in groups to swap ideas, but then submit their own individual papers crafted in their own words. This still means figuring out a way to get them graded.

  3. Accept group submissions. This has advantages and disadvantages alluded to above.

What should I do?

Edited to add: Guided in part by the many thoughtful responses below, I've cobbled together a grading system I think will work. Of course I don't know whether it will work until I've tried it. I'm therefore not sure whether it's appropriate to post the relevant portion of my syllabus as a self-answer, or as an edit to the question, or to do nothing at all (at least until the semester is over). I'm sorry that I'm new to this site and not at all sure of the culture.

Edited further to add: In response to a comment below, I've posted my plans for the semester as an answer below. If this sort of thing is on-topic here, I'll come back at the end of the semester with a report on how it worked.

  • Figuring out how to pay undergraduates to be your TAs seems harder than just assigning a few group apers, no? Jan 6, 2022 at 21:16
  • @AzorAhai-him- No, I think I have a substantial stable of undergrads I could draw on for this. I have always hesitated to ask undergrads to grade the papers of other undergrads who might be their friends, but aside from that, recruiting should be no problem.
    – WillO
    Jan 6, 2022 at 21:20
  • 1
    Sorry, I didn't mean recruitment, I meant the overhead of hiring them as TAs, teaching them how to be a TA, etc. Jan 6, 2022 at 21:21
  • I would just assign a few group assignments (change groups) and a few individual ones, you should have more than enough information to pull out the high-performers Jan 6, 2022 at 21:22
  • @AzorAhai-him- : Without going into a lot of detail that would be irrelevant to the question, I'm sure I've got that covered.
    – WillO
    Jan 6, 2022 at 21:22

7 Answers 7


Here is what I (the OP) am going to do. It's more complicated than I would like, but I think it serves my purposes.

  1. Students are warned that this is an experimental system and might have to be tweaked mid-semester, but I expect it to stay mostly in place.

  2. Each student must choose (once, for the entire semester) among four options: A) The "loner" option --- you do all your homework yourself. B) The "partner" option --- you team up with a partner and the two of you do all your homework together. C) The "rover" option --- I will assign you a different one or two teammates for each assignment. D) The "roving partners" option --- you have a partner you work with all semester, and I will assign you an additional one or two teammates for each assignment. (Therefore team sizes will range from one to four.)

  3. Teams work together, and cannot get any help from anyone not on the team.

  4. Each team submits one paper, together with a small spreadsheet indicating which students believe they have completely understood which solutions.

  5. Each student separately submits an individual assessment of his/her teammates' contributions.

  6. Most of the time, the whole team gets the same grade, with three exceptions:

Exception 1: If Carol claims (on the spreadsheet from point 4)) to understand considerably less than Alice and Bob, she might get a slightly lower grade than they do.
Exception 2: I might randomly call on you to explain a solution you claimed to understand. If you can't explain it, you will get a much lower (and possibly failing) grade for the entire assignment. (This is explicitly presented to the students as a disincentive to exaggerate on the spreadsheet.)
Exception 3: In cases where the applicability of Exceptions 1 and 2 is ambiguous, I will consult the evaluations submitted per point 4).

Classes have not yet started, so I will be very receptive to criticisms, issues I might not have thought of, suggested tweaks to make this a better grading system, suggested tweaks to make this seem less complicated, and anything else you think I should be thinking of.

  • 2
    I like this. Two suggestions: (1) I would simplify #2; as is, it seems so confusing that it might get you off on the wrong foot. Personally, I would make option C the default, and tell students "if you prefer to work alone, or if you have someone that you want to team up with, let me know, I will consider exceptions." This is less confusing and lets you magnanimously grant exceptions. And (2) I would similarly simplify point #6 -- e.g., have a schedule for when students will be interviewed and a rubric for how these will be graded. "Schedule" and "Rubric" sound better than "3 exceptions"
    – cag51
    Jan 10, 2022 at 0:05

In your question, you mentioned "a series of assignments". Is there something preventing you from doing both types, i.e. some of the assignments done in groups and some individually? That would give you the best of both worlds.

  • I might be overthinking this, but: These are pretty serious students, who I think will mostly follow the rules. But I feel like once they get used to the idea that it's okay to work together on some assignments, the temptation to continue working together even when it's disallowed will be significantly increased.
    – WillO
    Jan 7, 2022 at 18:07

I had a lot of students work on small-team projects. Sometimes I would form the groups and sometimes I'd let them. More below on selection.

Usually there was more to the course than the project so there was an opportunity to get some additional ideas about performance. But not always. In the "senior capstone project course" one of the requirements was a presentation of their project at the end where everyone was required to participate. I graded that separately (preparation, presentation, etc).

But, I never assumed that all the students would contribute to the project in the same way and so it was impossible for me to decide who did "more". And they worked outside my view mostly. But there were two features that I generally included.

First was that everyone got the same grade for the project unless there were compelling reasons to do otherwise. And I'd tell them at the start that an individual taking on all the work and closing out their team mates wasn't a good way to get a good grade. Team work should be team work. If they haven't worked in teams before you may need to teach them how: meetings, discussions, small continuous improvements... I also kept an open forum where students could ask questions and get answers, either from me or from members of other teams.

The second feature was "peer evaluation", which is not the same as "peer grading". Each student was required at the end of the course to name two or three (depending on team size) team members who contributed the most and to say why. I didn't ask for slackers to be named. Additionally they were asked for their own primary contribution to the team.

I was once surprised when a team praised a teammate for their important contributions to making it all work when I'd thought that person was rather a poor performer. They actually got a boost in grade from peer evaluation.

I'd discourage teams of size greater than five unless there were special circumstances. You can permit individual projects or not, but in my field (CS) teamwork is valued so I didn't permit it, since they needed the experience.

You can also, perhaps, let everyone vote during the presentations on who had done the best job on the projects and boost that team a bit.

I never tried this, but my own doctoral advisor used to hold five or so minute "oral exams" with each student (in every course, actually). You can learn quite a lot in a few minutes if you ask the right questions.

More on selection.

One way to choose who is on a team that works in a small (~25) group of students who know one another is the "sandlot baseball" method. Choose a set of "captains" equal to the number of groups(randomly, by perceived skill, volunteers, voting...). Then in rotation the captain (perhaps in consultation with other members already chosen) chooses a person from those not yet on any team. Yes, somebody will be last, which is a disadvantage, but it seemed to work when I used it. This spreads the talent among the teams without being too cliquish.

Another way, was randomization. I once needed two teams, so I made up a set of index cards, half of them with the word "Fire" and half with "Ice". On one of each of those sets I also wrote "captain". Face down, I had each student draw a card. That told them their team and gave the team a name that they actually liked. I explained that the "captain" wasn't the manager, but only the one responsible to communicate team needs to me as required, including a list of the team member initially. This also worked. One of the "captains" was a bit isolated from the rest of the class but getting the "job" seemed to spark something in him that was positive for everyone.

  • 2
    Depending on the class, I can usually determine the grade level (A/B/C/D/F) within five minutes of an oral exam and the individual grade level (A+/A/A-) after another five minutes. I started doing oral exams at the onset of COVID, and I'm not going to go back to written midterms and finals if I can help it. Orals allow me to see each student individually, and to make the exams formative as well as summative because I can give immediate feedback on their answers. Students seem to be more open to such feedback in-the-moment. YMMV
    – Peter K.
    Jan 7, 2022 at 18:35
  • @PeterK. What kinds of courses/class sizes are you talking about?
    – Kimball
    Jan 8, 2022 at 22:27
  • @Kimball The largest cohort I’ve done is 75, but it’s generally smaller. My classes are electrical engineering and software engineering classes: circuit analysis, programming, data structures, requirements engineering.
    – Peter K.
    Jan 8, 2022 at 23:17

I suggest group assignment is a good choice because can improve our skill development: Working in a group allows students to develop a range of skills, including communication, leadership, negotiation, and conflict resolution. These skills are valuable not only in academic settings but also in professional and personal contexts.

Group assignments enable the division of tasks among team members. This helps in efficient time management and allows individuals to focus on specific aspects of the project where they excel, contributing to a higher quality final product.


In my opinion individual assignments also have many benefits. For example, each student’s success hinges completely on their own actions rather than on the contributions of a group of fellow students. Other than that, they don’t need to discuss a certain topic, just like in the group (within the assignment’s parameters), so they can write a paper on whichever topic they choose. Since all students work independently, there is no conflict between them. During individual projects, students will become responsible for their learning and finishing, which will lead to their learning more about their topic.


I suggest first and foremost that you are missing out on the significance that you can bring to your honors students by giving them exercises that teach them how to work effectively in teams (I prefer the word teams rather than groups, as the former implies a greater or more formalized degree in collaboration and sharing). By example, what balance do you expect your students will have after they graduate to the real world between working entirely by themselves on a project versus working in teams on a project?

I might also challenge your statement that "students are not permitted to work together". We could enter entirely new threads on what the students believe this means in principle or practice versus what you intend it to mean as well as how effectively you can truly monitor or grade against this restriction even after all sides (you and your students) fully understand its meaning in precisely the same ways.

To your problem. In a nutshell, you have less manpower to grade your assignments. You can find replacement manpower or restructure how your assignments are to be graded. If I review your proposals individually, I would comment as below.

  1. Employing undergraduates to grade their peers or peers from other classes is fraught with potentials for abuses. In the US, we would immediately recognize FERPA limitations have to be respected. You will probably have to clear this proposal through your department chair. My experience is that we employ senior undergraduate students as peer mentors to our lower level courses, but we never employ them as graders. The administrative overhead (and potential for abuse) is too high.

  2. This approach promotes exactly the wrong ideas about what is involved to work effectively in and be evaluated entirely at the end as a team. Indeed, it is this example that I give to distinguish "groups" versus teams. The former is a collection of students who swap ideas about the questions, all come up with the best answer, each submit it, and then each get graded individually for having one (unified) "best" answer. Or, back to my challenge above ... How will you explain and police this proposal to meet what you believe are the standards for working in "groups" versus working in teams?

  3. Are you building a false presumption ... allowing team work projects prevents you from having any ways to separate B and C students ... to avoid being creative in restructuring your assignments instead? Why not have both team-work and individual-based assignments? Examples for the latter (mentioned also in other responses) could include individual oral reviews on the team-work reports, quizzes or exams, and homework. Why not also restructure your grading in ways that will take less time for the TA's (and you) to grade. Examples include assigning ten problems on a homework but choosing randomly only three of the problems to grade (choosing after the assignment is submitted).

In summary, I have great respect for the hard work that is required of an instructor to administer a proper mix of individual-based and team-work-based assignments in a course. I propose that taking on this responsibility will be the best response to your immediate problem.


You could let the students work in groups. They submit a paper as a group and this is part of their grade. You can let the other part of the final grade be based on individual conversations with the students.

I understand that 25 individual conversations is also time consuming, but it could give you the opportunity to identify students who actually understand what they did from students who benefitted from the collaboration more, and with 10 minutes per student it should be doable in 25*10 / 60 = just over 4 hours, plus some additional time for planning and admin so let's say 8 hours.

You must log in to answer this question.

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