We all use it in our classrooms at some point. Group work. Sometimes, because it is expected, sometimes because it is practical, and sometimes because collaboration itself can be an important learning goal in education. And it can help with critical thinking, too. Yet it remains an approach where you never seem completely sure of the outcome. Which students can work together? Which team will not work? This post will be about group composition.

One time, you see students genuinely engaging with each other, exchanging ideas, helping one another forward. Another time, one or two students do most of the work, two follow along, and one mainly stares at the ceiling. I once heard students describe the others as parasites. It does not always come together. I’ve noticed a lot of my students actually hate group work. I do understand them, but abolishing group work would be wrong. And usually, we look for the explanation in what happens during the group work itself: task division, clear instructions, roles, and guidance.

A recent study by Sun and colleagues turns that perspective around and asks a simpler, more fundamental question: what if the problem is not in how students collaborate, but in who we put together?

In their study, the researchers first mapped who students actually want to work with when they get stuck. Not based on friendship, but through concrete questions, such as who they would turn to for help with a difficult problem. This leads to a network of preferred collaborators within the classroom. Some students are mentioned frequently, others rarely.

They then used this information to form groups in different ways. In one condition, groups were composed randomly. In another, they were based on these collaboration preferences.

The difference is clear. Groups that aligned with students’ preferences functioned better. Students were more engaged, interactions were more balanced, and the overall quality of collaboration was higher. Not spectacularly so, but consistently.

That sounds almost obvious, but it challenges a common assumption. We often think that the success of group work mainly depends on how we organise it. This study suggests that composition plays a key role as well.

At the same time, there is an important nuance. This is not an argument for simply letting students choose their own groups. Friendship and effective collaboration are not the same thing. What matters here is not who students like, but who they see as helpful when learning becomes difficult.

In other words, these networks capture something closer to perceived competence and trust in learning situations. And that makes them interesting, but also complex. Because we can influence such perceptions by many factors, including prior achievement, status within the group, or earlier experiences.

The study itself is careful. The effects are not exaggerated, and the limitations are clearly acknowledged. We do not fully see what happens inside these groups. We mainly observe the outcomes, not the detailed processes. But the pattern is robust enough to take seriously.

For classroom practice, the message is not to replace one simple idea with another. Group work is not a single intervention that you can apply in the same way every time. It is a combination of choices. Task design, guidance, and group composition all interact.

We often focus on what students do during collaboration. This study reminds us that who sits together may matter just as much.

And perhaps that is the uncomfortable part. Because it is easier to adjust instructions than to rethink how we form groups. Yet if we want group work to work, that is exactly where part of the answer seems to lie.