The impact of specialist resource centres on autistic pupils' experience of mainstream school.

Pupils were not allocated to placements at random, and placements reflect need, advocacy and local availability. Differences between the groups may be differences between the pupils and between the schools, not effects of the placement itself. Only seven schools took part, too few to treat the school as its own level in the analysis, so school-level factors such as leadership, ethos, staffing and staff turnover could not be separated from placement type. Group sizes were uneven and the group in schools with no centre was small (23 pupils), so real differences may have been missed; several of the regression models were likely underpowered, meaning non-significant predictors may reflect too few pupils rather than no relationship. Some pupils had already been in their placement for one or two years when the first survey was done, so any early effect of that placement may already have occurred before data collection began. Parent-reported data were collected early on but lost too many respondents to analyse, so the picture rests on pupil self-report alone, which can be influenced by wanting to give the expected answer. Perceived support and the outcomes it was compared with were measured at the same time, so the study cannot show whether support improved wellbeing or wellbeing changed how supported pupils felt. Attendance and exclusion figures came from school-level summaries rather than individual pupil records and rest on few data points, so those comparisons are indicative rather than definitive. The placement groups differed in need: all 50 resource-centre pupils had an education, health and care plan compared with 30.4% of pupils in schools with no centre. This was controlled for statistically but other differences in severity or complexity may remain.

The study was funded by the Cullum Family Trust and carried out under contract for the National Autistic Society, which commissioned it. The Trust partnered with the National Autistic Society in establishing and supporting the resource centres being studied, and funded both the centres and this evaluation. The authors state that the Trust had no role in the analysis or reporting of the findings.

Do not read the resource centre findings as proof that resource centres improve outcomes. Pupils were not allocated to placements at random and only seven schools took part, so the links between resource centre placement and belonging, teacher support and attainment may reflect differences between the schools and the pupils who go to them rather than the centres themselves.