Method one: students choose their own groups

This is the default in most courses because it is free and students prefer it.

What it does well. Self selected groups form quickly and start working sooner, because the members already have some trust and usually some shared expectations about effort. Conflict is lower. Students report higher satisfaction, and for short, low stakes tasks that is often the whole story.

What it does badly. Self selection reproduces the existing social structure of the cohort, with predictable consequences. International students, mature students, commuters and anyone who missed the first two weeks end up in leftover groups or alone, which is a documented equity problem rather than an occasional accident. Groups of friends also tend to be homogeneous in skills and in viewpoint, which reduces the quality of the work and removes exactly the difficulty that group work is supposed to teach students to handle.

There is a further problem that only shows up at grading time. Friendship groups are the least likely to report honestly on each other's contributions, which means the peer evaluation data is least reliable in precisely the groups where free riding is most socially protected.

When to choose it. Short tasks, low stakes, early in a course when the priority is getting students talking to anyone at all.

Method two: random assignment

Assign students to groups by an arbitrary rule such as where they are sitting or a randomised list.

What it does well. It is fast, it is visibly fair, and nobody can argue that the allocation was biased. It also mixes the cohort, which self selection does not.

What it does badly. Randomness distributes by chance, which reliably produces a few very unbalanced groups. In any cohort of two hundred there will be one group where three of the five students are struggling and one where four are strong, and those two groups will have completely different semesters. Random assignment also ignores practical constraints that determine whether a group can meet at all, such as timetables, placements, campus and time zone, which is the most common reason a group fails for reasons that have nothing to do with its members.

When to choose it. Short in class activities and rotating teams, where the cost of an unbalanced group is one session rather than one semester.

Method three: criteria based assignment

Students complete a short survey and are assigned by an algorithm that balances the groups against criteria the instructor sets.

What it does well. It is the only method that lets you control the composition deliberately. You can distribute prior knowledge so that no group is left without anyone who can start. You can avoid isolating a single student of any particular background, which matters because a lone member of a minority in a group of five participates measurably less than when there are two. You can respect practical constraints such as availability and location, so that the group can actually meet. And you can weight the criteria by importance, so that availability might matter more than skill mix, or the other way round, depending on the project.

What it does badly. It requires preparation. You have to write the survey, get students to complete it, and decide what you are optimising for before the allocation runs. That is perhaps an hour of work the first time, and it has to happen early enough that the survey closes before groups are needed.

When to choose it. Any project that runs for more than two weeks, carries meaningful credit, or matters for the students' experience of the course.

Deciding what to balance on

Criteria based grouping only helps if you choose the right criteria, and there are four categories worth considering.

Practical constraints come first, because they are binary. If the group cannot meet, nothing else matters. Availability, campus, time zone and mode of study belong here, and in most courses availability should carry the highest weight.

Prior knowledge or skill is the criterion people reach for first. The usual choice is to spread ability across groups rather than to cluster it, so that no group is without a starting point. The research here is more mixed than people assume, with some evidence that mixed ability groups benefit weaker students without harming stronger ones, and some evidence that the strongest students disengage if they end up doing the work. Weight it moderately and pair it with contribution tracking rather than relying on it alone.

Diversity of background and perspective is where the algorithm can do something a human allocator cannot do at scale. The principle worth applying is to avoid the group of one. A single international student, or a single student from any underrepresented group, in a team of five is in a measurably worse position than two would be. Setting the criterion to avoid isolated members is a small change with a real effect.

Student preference can be partially honoured. Allowing each student to name one person they would like to work with, and then treating that as one weighted input among others, captures most of the comfort benefit of self selection while keeping the balancing.

Writing the survey

Keep it to six or eight questions and make every one of them something you will actually use in the allocation. Students notice when they are asked for information that clearly did nothing.

Ask about availability in concrete blocks rather than in general terms. Ask about relevant prior experience in specific terms, such as whether the student has used a particular method or tool before, rather than asking them to rate their own ability, because self ratings are unreliable and vary systematically by gender and background. Ask about the working style question that matters for your project, which is usually something like whether the student prefers to plan everything up front or to start and adjust.

If you want a fuller set of examples, we have a post on survey questions for group selection.

Tell students how the groups were made

This is the step that is almost always skipped and it costs nothing.

When groups are released, explain in two sentences what the allocation optimised for. "These groups were formed to balance prior experience with the software and to make sure everyone in a team has overlapping free time on at least two weekdays" changes how students read their allocation. It moves the group from something that happened to them to something that was designed, and it markedly reduces the number of requests to move.

Keep the ability to adjust

No allocation survives the first week intact. Students drop the course, someone has a placement that was not in the survey, and occasionally two people genuinely cannot work together for reasons you did not know about.

Being able to move a student with a simple adjustment in the dashboard, and have the change sync back to the LMS so the groups students see stay correct, is what makes criteria based grouping practical rather than theoretical. Plan for roughly one adjustment per twenty students in the first fortnight and the process stays calm. If you want to know more about how changes sync back to your LMS, our Group Formation FAQ explains what happens on both sides.

In short

For anything short and low stakes, let students choose or assign randomly and spend your energy elsewhere. For any project that runs for weeks and carries real credit, spend the hour on a survey and balance the groups deliberately, weighting practical availability highest, avoiding isolated members, and spreading relevant experience. Then tell students what you did and why.

Where to go next

If you would like to see how this works inside the tool itself, the Group Formation solution page shows how criteria, weighting and LMS syncing work in practice. The Group Formation FAQ is a good next stop if you have the practical questions that usually come up before a first run, such as how the activity sits in your LMS, what students see, and what happens when someone misses a deadline.

It is also worth looking at the tools that sit naturally alongside this one, including Group Member Evaluation, Team Based Learning and Peer Review. You can see how they all fit together on the FeedbackFruits tool suite page, or read more about what a feedback and assessment solution actually is if you are building the case for your institution.

And if you want to keep reading on this topic, we have Mixed or matched? How to decide what to balance in student groups, Create effective surveys for better group selection and Make group work work with effective group selection.

‍