Smaller teaching groups backed by learning support assistants may not be the best way of helping special needs pupils, says Government-backed research.
There is no convincing evidence that this approach works for children with moderate learning difficulties, according to a report commissioned by the Department for Education and Employment.
It says teachers and other education professionals have no evidence of what works best. Instead, they rely on practice and their experience of what works on the ground.
The result of this "resource and hope" policy is a wide variation in what schools do, and in how much they spend. But there is no link between what is spent and how well pupils perform, either academically or in their personal development.
The report, produced by researchers from Newcastle University, recommends sustained research to establish what works best for pupils with moderate learning difficulties, national benchmarks to indicate what they might be expected to attain, and local and national frameworks to ensure some consistency between education authorities and schools.
Individual schools would then no longer have to "invent" their own targets and forms of support, and ultimately such a system - based on greater delegation of money to schools - would get rid of the need for statementing most MLD students.
This would mean a radical departure from the current model of SEN funding - introduced following the Warnock report in 1981 - and based on the assessment of individuals' needs. Instead, say the authors, there should be a focus on outcomes.
"We see it as being in line with current government education policy . . . which has emphasised the centrality of outcomes and has accordingly begun to identify national outcomes targets and national strategies for provision," says the report.
"It is possible on the Warnock model - as was realised in the statementing procedure - to allocate resources to children on the basis of their needs without any indication of what particular outcomes those resources would lead to for the child.
"In the longer term, our proposed system may prove to be applicable to other groups of pupils with SEN and may offer a way forward from a system which has been in place for some two decades and which was designed for a very different education system from the one we have now."
Based on an international research survey and their own analysis of MLD support in 33 schools in eight education authorities, the Newcastle team found a lack of evidence about the effectiveness - or otherwise - of the most common MLD interventions. These included smaller class sizes, the use of learning support assistants, setting by attainment, interventions by education authorities, and differentiation in mainstream classrooms.
Education authorities should review their costs and the efficiency of their strategies, given the variations in costs associated with different patterns of inclusion, the team concludes.