In education, we often talk about being evidence-informed. In England, we are rightly proud of being one of the most evidence-informed education systems in the world. But that only really matters if we are willing to let our views evolve as new evidence emerges, and to see that evidence not as delivering final answers but as helping us to build a clearer, more detailed picture over time.

At its best, engaging with evidence helps us to refine our understanding, challenge assumptions and sometimes reconsider positions. Single studies rarely settle debates - but knowledge accumulates incrementally, with new research sharpening, qualifying or reframing what we thought we knew.

A new study from the Education Endowment Foundation on how we group pupils for learning is a good example of this process in action. It adds important new insights to a longstanding and often polarised debate. And in doing so, it allows our thinking and professional knowledge to evolve.

The key question is familiar: when deciding how to organise classes, should pupils be grouped by prior attainment or should classes include a broad mix of attainment?

Pupil grouping is an area where we often hold strong views, shaped both by experience and by diverse research literature. A simplified version of the debate is this.

On one side: grouping pupils by prior attainment risks entrenching disadvantage. Lower-attaining pupils could be less likely to access high-quality teaching or opportunities to progress, and pupils may internalise expectations about their ability.

On the other: mixed-attainment classes risk insufficient differentiation to meet pupils’ diverse attainment needs. Higher-attaining pupils could be insufficiently stretched, while lower-attaining pupils may not receive the targeted support they need.

One of the challenges in interpreting the research literature in this area is that much of the international evidence covers forms of “tracking” (for example, placing students on different types of course or in different types of school) that differ from the way grouping by attainment typically operates in England.

And schools often operate more than one approach to grouping by attainment (for example, they may operate combinations of setting and streaming), which are hard to disaggregate in research. As a result, it has been difficult to draw clear conclusions about the specific practices used in English schools.

At the same time, there has been relatively little direct comparison between attainment grouping and mixed-attainment practices within this context.

Is setting better than mixed-attainment grouping?

This is where our new EEF-funded study, conducted with UCL, makes an important contribution. It focuses specifically on grouping practices in English secondary schools, and on mathematics in particular, providing a clearer lens on how these practices play out in a familiar context.

The new study is large in scale and uses a robust design, including a clear comparison group. It is the first to directly compare the impacts of mixed-attainment practice and setting (the most common form of attainment grouping in England, across subjects). It is therefore an exciting contribution to the research in this field, and its findings merit careful attention.

What did we find? Pupils with higher prior attainment made slower progress in mixed-attainment maths classes than in maths classes set by attainment, equivalent to around two months’ less progress.

Lesson observations provide some possible insight into why. We found that lessons from mixed-attainment schools tended to be more similar to the bottom set than to top set lessons in schools using setting, especially in terms of pace.

At the same time, the group of pupils with lower prior attainment, which also includes pupils from socioeconomically disadvantaged backgrounds, appear to make similar progress in both types of grouping. In other words, this study does not find significant evidence that setting in maths negatively affected the progress of these pupils.

The study also found that, while the self-confidence of low-attaining pupils was generally lower than their higher-achieving peers, the self-confidence of low-attaining pupils in mixed-attainment maths classes was lower compared with those low attainers who were set by attainment.

Taken together, these findings add nuance to the picture. They suggest that some of the potential risks associated with attainment grouping may be less pronounced, at least in the case of maths, than is sometimes assumed. At the same time, they highlight possible trade-offs in mixed-attainment settings, particularly in relation to challenge for higher-attaining pupils.

What does this mean for schools?

It is important not to over-interpret what is, ultimately, one study (though a robust one). It focuses specifically on mathematics, and on the particular attainment grouping approach of setting, within English secondary schools. Other subjects, phases or models of implementation may look different. As with all research, it forms part of a wider and still-developing evidence base.

What this study does do is help us to better understand the consequences of different grouping practices in this specific context.

Perhaps the most important implication is that, whichever approach schools adopt, careful attention is needed to ensure that all pupils are well served.

For schools using mixed-attainment grouping, the study findings point to the need to think carefully about how to maintain high levels of challenge and stretch, particularly for higher-attaining pupils.

For schools using attainment grouping, the wider research shows it remains essential to focus on the experience of lower-attaining pupils (among whom socioeconomically disadvantaged learners are disproportionately represented). This means high expectations, strong teaching (including considerations on how to allocate staff) and opportunities to progress. For school leaders, this reinforces the idea that decisions about grouping are live, context-dependent questions that should be revisited as new evidence emerges and as our understanding evolves.

More broadly, this speaks to what it really means to be evidence-informed. It is not about finding a single study that confirms what we already believe, nor about searching for definitive answers to complex questions. It is about engaging with an evolving body of evidence and being open to updating our views, while recognising uncertainty and the limits of what any one study can tell us.

New evidence rarely closes debates. More often, it helps us to ask better questions, see more clearly the trade-offs involved and make more informed, thoughtful decisions in practice.

Becky Francis is chief executive officer of the Education Endowment Foundation