Working together in maths class works. At least, that seems to be the conclusion of a new systematic review published in the British Educational Research Journal. Patricia Hampson and colleagues examined research on cooperative learning in mathematics classes with students aged 11 to 16. Their conclusion is quite positive: in eight of the nine selected studies, student performance improved. The calculated average effect size was around +0.62. That is substantial for educational research. But wait, it is far more complex than that.
To begin with, “cooperative learning” turns out not to be one clearly defined method. The label covers a wide range of approaches. Some studies used peer tutoring, others relied on Kagan structures, Team Assisted Individualisation, or variants of flipped learning combined with collaboration. That makes it difficult to determine what exactly is working. It is less a single technique than a family of instructional principles.
The theoretical foundation will sound familiar to many teachers. The authors refer, among others, to Vygotsky and the idea of the zone of proximal development. The concrete idea here is that students sometimes learn better when supported by a more capable peer. Social interdependence also plays a role: students need each other in order to progress. That may sound abstract to some readers, but in practice it translates into something quite concrete. Good cooperative learning does not simply mean “work together”, but involves clear roles, mutual dependence, individual accountability, and structured interaction. This aligns with earlier meta-analyses as well.
And that immediately leads to an important nuance. What works in these studies is usually not just group work. The stronger interventions often included considerable structure. Students received explicit tasks. They had to explain their ideas to one another. Also, they were individually assessed. And they worked simultaneously toward a group goal. This fits closely with earlier insights from Johnson & Johnson and Slavin. When that structure is absent, what you often get is a few students doing the work while others simply follow along and learn very little. The review, therefore, does not confirm that simply “putting students into groups” is automatically effective. It rather suggests that well-designed collaboration can be powerful.
At the same time, there are reasons to remain cautious. The evidence base itself turns out to be surprisingly small. Of 115 studies, only 9 ultimately met the criteria. The researchers had to exclude many studies because they were too short. Some often lacked a proper control group. Or they did not report sufficient statistical information. In addition, most of the included studies were relatively small. The largest involved 254 participants. That is not trivial, but neither is it enormous.
The results also varied across groups. Girls appeared to benefit somewhat more from cooperative learning than boys. Findings for disadvantaged students were mixed. In some studies, the gap narrowed; in others, it did not. And among students with learning difficulties, cooperative approaches mainly improved computational skills rather than deeper conceptual understanding of mathematics.
That last point is particularly interesting. Collaboration alone does not automatically lead to deep understanding. That may partly depend on the nature of the interactions themselves. When students are mainly practising procedures, peer support can be effective. Conceptual understanding may require more explicit guidance.
The size of some of the reported effects also deserves nuance. One study reported effect sizes above 1.0, which is exceptionally high in educational research. The authors themselves acknowledge that some findings may artificially inflate the average effect size. That does not mean the overall conclusion is wrong, but it does mean we should be careful with spectacular numbers.
Still, the overall signal remains fairly consistent. Collaboration can work in mathematics, especially when it is carefully structured. And in some ways, that may simply make sense. Mathematics does not have to be purely individual puzzle-solving. Explaining concepts to others, comparing reasoning, discussing mistakes, and developing strategies together are all cognitively rich activities. Anyone who has ever truly explained something to a classmate knows that sometimes you only realise whether you understand it yourself while trying to explain it.