Grouping children by attainment within classes is more effective than putting them in sets, according to new guidance from a government-backed body.

It comes as a major research project raised concerns about setting, including large numbers of pupils being put in the wrong maths sets (especially black pupils and girls), less qualified teachers being assigned to lower sets, and pupils in those groups having less confidence.

The Education Endowment Foundation (EEF), which today published its first guidance about the use of within-class attainment grouping, says the evidence “indicates that it is likely to be beneficial for all learners”.

In contrast, it says that “on average, pupils experiencing setting or streaming make slightly less progress than pupils taught in mixed attainment classes”.

The new section in its teaching and learning toolkit says that within-class attainment grouping provides an average benefit of three months’ additional progress, although it notes that “there appears to be less benefit for lower attaining pupils than others”.

On setting, it says: “The evidence suggests that setting and streaming has a very small negative impact for low and mid-range attaining learners, and a very small positive impact for higher attaining pupils.

"There are exceptions to this pattern, with some research studies demonstrating benefits for all learners across the attainment range.”

It adds that overall the effects are “small”, and “it appears that setting or streaming is not an effective way to raise attainment for most pupils”.

The EEF funded a two-year trial, developed by the UCL Institute of Education, which aimed to see what would happen if some of the poor practices associated with setting were stripped away.

Focusing on maths and English in Years 7 and 8 it had three principles:

• to randomly allocate teachers to sets to prevent the lower groups having less experienced teachers
• assign pupils to sets based on independent measures of attainment rather than subjective judgements
• provide opportunities to reassign pupils though the year based on up-to-date measures of their attainment.

Independent evaluators from the National Foundation for Educational Research (NFER) found that some teachers believed that what they were being asked to do was no different to their existing practice; others said that changing the way they put pupils in groups was onerous and hard to do.

In today’s update to its toolkit, the EEF says this attempt to introduce best practice in setting “had no overall impact on attainment”, but says this finding was “largely because it was a challenge for schools to change their setting practices”.

A second EEF-funded project sought to examine mixed attainment teaching, but the researchers found it hard to recruit schools willing to try this approach for the study.

The NFER found that staff who did take part had mixed views, with some enjoying it, and others struggling with differentiation of pupils, but the evaluators conclude that “most interviewees felt that the intervention had a positive effect on student outcomes; with most believing that the least able students had particularly benefitted”.

Steve Higgins, a professor at Durham University’s School of Education and lead author of the toolkit, said the updated analysis “shows us that the impact of attainment grouping is dependent on the type of grouping used”.

He added: “Based on the evidence we have, setting or streaming pupils in different classes by prior attainment appears, on average, to have had a small negative effect. In contrast, grouping pupils by attainment within classes had a positive overall impact in the studies we reviewed.”

Professor Becky Francis, director of the UCL Institute of Education, and her colleague Jeremy Hodgen will outline the findings of the best practice in grouping students project at the ResearchED conference in London tomorrow.

This is an edited article from the 7 September edition of Tes. Subscribers can read the full article here. To subscribe, click here. This week's Tes magazine is available in all good newsagents. To download the digital edition, Android users can click here and iOS users can click here.