When Joe Kirby and I presented data on whether Teach First is working at ResearchED2013 earlier this month, the best audience question came from Arthur Baker who wanted to know whether the improvements in average school attainment following Teach First participation meant that the most disadvantaged students were benefiting, or not.
This question struck a particular chord with me because I had recently read Machin and Silva’s chapter in The Tail, an excellent collection of essays about underachieving pupils in English schools. They showed that, although the early sponsored academies had indeed improved average GCSE outcomes (see Machin and Vernoit for evidence on this), there was no improvement in GCSE attainment for those starting secondary school at the bottom of the Key Stage two distribution. So, a policy designed to improve life chances for those at risk of being left behind in under-performing schools seemingly did not impact on those within the schools who were most educationally disadvantaged.
To return to Arthur’s question: what sort of pupils benefit most from a school’s participation in Teach First? We didn’t answer this in my original paper with Jay Allnutt, but I’ve re-opened up our datasets to have a look. In the paper we compare the performance of a particular pupil across their core subjects of GCSE English, maths and science to see how well the pupil performs in the department(s) who chose to take Teach First participants, compared to those who did not. We showed that departments taking on Teach First participants were more ineffective than others within the same school before joining Teach First (scoring about 10% of a GCSE grade worse), but that by years two and three of participation they were outperforming their neighbouring departments who did not join Teach First (by about 15% of a GCSE grade).
To mirror the analysis of Machin and Silva on academies, I looked at how lower and higher ability children responded to the presence of Teach First participants in the department. The chart below shows that the overall pattern of impact of Teach First is similar for children across the ability distribution. Those scoring in either the top 10% or bottom 10% at Key Stage two do not appear to be particularly suffering in their attainment within the department who will soon join Teach First. (We can only speculate as to why – e.g. very high achieving pupils have strategies and other support systems for overcoming any teacher quality issues; very low prior attainment pupils have specific circumstances that dominate GCSE grades across all subjects.) In years two and three after joining Teach First, it appears that improvements in attainment are shared across the ability spectrum. It is hard to make claims as to why this should be so, not least because Teach First themselves have not systematically collected information on what year groups and which ability sets within year groups their participants are teaching.
Finally, to answer the question that Arthur Baker was most interested in, this chart shows that the responsiveness of free school meals pupils to the placement of Teach First participants in departments is very similar to the response of other pupils. If anything, the free school meals pupils appear to benefit a little more (and this difference is statistically significant in year two).