The EAL Academy


What do the data tell us about EAL pupils?

When the 2016 GCSE results were released on the DfE’s performance table web site a school in Yorkshire came bottom of the EAL league table. I have a little trepidation in saying so, though, because these are high-stakes examinations for schools and authorities as well as for students, and because what seems like a straightforward set of outcomes rests on a series of decisions about what to measure and how.
My example of the ‘worst performing’ school is a good example. I had to choose between using Progress 8 and Attainment 8 as the key measure, and then specify a minimum number for the cohort (50 in this case) that gives some measure of statistical validity. Doing so involves shaping the data as I analyse it; to understand what the analysis tells us, we have to understand the data sets themselves. There is, in fact, a huge amount of EAL data in the public domain that relates to English schools. This article seeks to explore what is there and how we might best use it. I will make passing reference to data sources to which there is restricted access so readers can pester their school, LA or MAT for access, but mostly I will stick to freely available data.

Key demographics

The school census collects information about ethnicity and language every January, and in late June of every year the Schools, pupils and their characteristics survey is published. It tells us that there are now a total of 1.3 million EAL pupils in our schools (equating to 20.6% of pupils in state-funded primary schools and 16.1% in state-funded secondaries). That’s one in every five primary school pupils and one in six secondary pupils.
When we look at numbers by local authority we can see that Birmingham has the largest number of EAL pupils (just under 72,000) and – if you exclude the Isles of Scilly, the City of London and Rutland – Redcar and Cleveland has the smallest number (272). In percentage terms Tower Hamlets tops the table (73% EAL), while Redcar and Cleveland has the lowest EAL density (1.5%).
These data become a bit more useful when we look at change over time. The same information can be found going back to 2005. So we can say, for example, that Herefordshire has the largest percentage increase in EAL pupils since 2005 at a staggering 922%. However, that’s not really very helpful because what it actually means is that the number from just under 150 to just over 1500 out of a total of nearly 25,00 pupils. Of more use is percentage point growth. Barking and Dagenham has an increase 34.1 percentage points since 2005. So it’s gone from one in six pupils having EAL (16.4%) to half of its pupils (50.5%) in just twelve years. Over the last five years the national increase is 3.4 percentage points, but seven LAs have increased at double that rate.
You can also find in Schools, pupils and their characteristics school level data for all English state schools showing EAL numbers and percentage along with ethnicity at the sub-code level. For example, Black is broken down into Black Caribbean and Black African, but you can’t find the number of Nigerian or Somali pupils. Here we find out that five schools in the country are 100% EAL and a further 200 are more than 90% EAL. Over 2000 schools are more than 50% EAL.

Performance data

We now turn to performance data. The earliest sight of anything at LA level is usually in Nexus, but access is limited, so most of us have to wait for DfE data from the main government web site.
National data for Key Stage 1 are available broken down by ethnicity and EAL. White British outcomes are lower than those of almost all minority ethnic groups (Gypsy/Roma and Irish Travellers are the exceptions), although EAL outcomes are a bit behind English first language outcomes. The LA break down at ethnicity level is not really worth the trouble as the categories are at too high a level (White, Mixed, Asian, Black and Chinese). However, the comparison of EAL pupils and English first language pupils is more helpful. For example, in Reading there is a five percentage point gap (75% of English first language pupils meet the expected standard while only 70% of EAL pupils do), but there is significant regional variation. The gap is three percentage points in London but eight percentage points in the North East, South West and East Midlands. And in Barnsley it is 23 percentage points.
For Key Stage 2 much more is available. National, regional and LA data for 2016. Here we can learn that just 13% of Gypsy Roma pupils achieved the expected standard in Reading, Writing and Mathematics and that 50% of EAL pupils reached the expected standard in all three subjects compared to 54% of pupils whose first language is English. However, when you turn the LA tables you find that the EAL figure has become 52%. I think the discrepancy is because the national tables include all and the aggregate of the LA results excludes pupils who started UK education after the beginning of Year 5. LA outcomes for EAL pupils range from Doncaster and Bedford at 34% to Bromley at 71%. You cannot get data on the performance of EAL pupils by school from this source, but you can get it from the performance tables web site, where you can search individual schools, MATs and LAs and filter the date to show outcomes for EAL pupils. It is also possible from this web site to download the results of all the state fund schools in the country and find out there are 35 primary schools (with more than five EAL pupils in the 2016 cohort) where no EAL pupils reached the expected standard in Reading, Writing and Mathematics. The performance tables also let you look at progress scores for EAL pupils by school and LA. They are usually much higher for EAL pupils.
The GCSE data sets are in a very similar format to Key Stage 2. You can find them at and on the performance table web site. Nationally EAL pupils do almost as well on the Attainment 8 measure as pupils whose first language is English (49.9 compared to 50.0- or 50.0 and 50.8 if you use the aggregated figures form the LA tables) and are far ahead on the Progress 8 measure (0.39 compared to -0.09). Regionally, the trend is very clear, London’s Attainment 8 for EAL pupils is the best at 52.8 and Yorkshire and the Humber’s the worst at 45.8 On the Progress 8 measure the North East comes out at 0.6 while Yorkshire and the Humber and the West Midlands are both only 0.25. The school in Bradford, by the way, is -0.55. Again, searching and filtering gives all the schools in an LA. Here’s an example of Islington secondary schools.

Performance data

What’s the use of it all? You can compare your school’s EAL outcomes to your LA and the national regional and national pictures but you need to be cautious, as you do with all data. For many colleagues, national and LA data are a valuable tool to help raise expectations of EAL pupils in their school. If you combine all of the publicly available data, you can also start to look at what factors might underlie the achievement of particular groups. A lot of my recent work has been with schools with significant Pakistani populations. That makes me interested in why Pakistani attainment nationally is so much lower than that of EAL pupils in general and Indian and Bangladeshi pupils in particular. So my task this autumn is to investigate the regional, local authority and school-level variations.
This is a slightly extended version of my article in the Autumn 2017 edition of EAL Journal.
EAL data

EAL data: finding it and using it

Online course

To bolster your data skills, consider this short online course. The course aims to develop data analysis and presentation skills in relation to EAL learners. It covers both what we can learn from school and wider data sources and how we can present that data to influence colleagues and improve EAL provision.

I attended a training session organised by Causeway Education in which you gave an incredibly thought-provoking presentation on understanding ethnicity and attainment data.
Andi Marsh
Schools and Universities Associate, Rare Recruitment