Reducing the attainment gap: good ways and bad

Figure 1: Percentage of pupils achieving 5 A*-C GCSEs including English and Maths (2012-2015)
Sample sizes: All schools: 2,778-2,814. Ofsted groups: 73-1,441. Regional groups: 143-409.
Sources: Department for Education; SchoolDash analysis.
Figure 2: Relative GCSE performance for high- and low-attaining pupils by school type (2015)
Notes: The size of each dot represents the number of pupils in that group of schools; hover over each one to view relevant data. School deprivation figures based on pupils' eligibility for free school meals, with bands defined by the DfE. Local deprivation figures based on the mean IDACI of postcodes within a 4km radius of each school, with schools then divided into three roughly equally sized groups. Coastal schools are within 5km of the coast. Small schools have fewer than 700 pupils, large ones have more than 1,200. A small proportion of low attainers means less than 12% and a high proportion means more than 18%. Urban, suburban and rural groups use ONS rural-urban categories applied to school postcodes.
Sources: Department for Education; Department for Communities and Local Government; Office for National Statistics; SchoolDash analysis.
Figure 3: Relative GCSE performance for high- and low-attaining pupils by local authority (2015)
Notes: Two small authority areas, the City of London and the Isles of Scilly, are not included because relevant data were unavailable.
Sources: Department for Education; Department for Communities and Local Government; Office for National Statistics; SchoolDash analysis.
Figure 4: Average capped GCSE point score (2012-2015)
Sample sizes: All schools: 2,778-2,814. Ofsted groups: 73-1,441. Regional groups: 143-409.
(Sample sizes vary because not all data fields are available for all schools.)
Sources: Department for Education; SchoolDash analysis.
Figure 5: Relative GCSE point score for high- and low-attaining pupils by school type (2015)
Notes: For details on how the data in this graph were derived see the notes accompanying Figure 2.
Sources: Department for Education; Department for Communities and Local Government; Office for National Statistics; SchoolDash analysis.
Figure 6: Relative GCSE point score for high- and low-attaining pupils by local authority (2015)
Notes: Two small authority areas, the City of London and the Isles of Scilly, are not included because relevant data were unavailable.
Sources: Department for Education; Department for Communities and Local Government; Office for National Statistics; SchoolDash analysis.
  1. Schools that do the most to close the attainment gap in good ways are:
    • Schools in many London boroughs
    • Ofsted 'Outstanding' schools
    • Boys' schools
  2. Schools that generally perform well but tend to increase the attainment gap include:
    • Schools with small proportions of low-attainers
    • Rural schools
    • Schools in the East of England as well as Herefordshire, Worcestershire and York (among other areas)
  3. Schools that reduce the attainment gap but do so in unhelpful ways (ie, by holding back high attainers) include:
    • Schools with high levels of deprivation
    • Schools with large proportions of low-attainers
    • Schools in Torbay, Kent, Salford and Blackpool (among other areas)
  4. Schools that combine low attainment with an increase in the attainment gap include:
    • Ofsted 'Inadequate' schools
    • Schools in Yorkshire and The Humber, the East Midlands and the North West
    • Schools in Knowsley, Barnsley, Stoke-on-Trent and Brighton (among other areas)
  1. Free schools have been omitted from this chart because they represent an outlier, with the data indicating extremely poor performance for high-attaining pupils (-19.0 percentage points) as well as rather bad performance for low-attaining pupils (-4.7 percentage points). However, this is based on a small number of schools (representing a total of 32,925 pupils), many of which have not been going long enough to issue any GCSE attainment data.
  2. The fact that this measure is limited to the 8 best GCSEs for each pupil is the reason that it is more properly referred to as the 'capped' point score. Each grade A* receives 58 points, each grade A 52 points, each grade B 46 points and so on (see this DfE document for further details). The theoretical maximum is therefore 8 x 58 points = 464 points.
  3. Note that standards vary from year to year, and in particular seem to have got tougher in 2014, disproportionately affecting the point scores of low-attaining pupils. This makes it look like the gap has increased, but in reality it has probably just become more visible.

Immigration and school capacity

Figure 1: Average primary school occupancy (2015) against change in percentage of non-British white pupils (2011-2015)
Sample size: 17,016 schools. Individual local authority areas are represented by between 17 and 482 schools.
Two small authorities, The City of London and the Isles of Scilly, each represented by one school, have been omitted.
Notes: Hover over each dot to see the name of the local authority and exact numbers.
Click on the legend to turn individual regions on or off. Double-click to view just one region on its own.
  1. The Department for Education does not capture data on the nationalities of pupils (or their parents), only their ethnicities. For this reason we have used the 'non-British, non-Irish white' ethnic category as a proxy for recent European immigration. It is certainly imperfect, but we believe it to be a reasonable one given the available data. For a further discussion and analysis, see our previous post on this topic.
  2. This analysis includes only mainstream state primary schools in England (though some of these schools also have a secondary phase). We excluded 10 very small schools (defined as as a capacity of less than 30 pupils) because these sometimes have unusually high or low occupancy rates that correspond to small absolute variations in pupil number but can nevertheless materially affect local averages. These schools are: Newton Blossomville School, , Milburn School, Combs Infant School, Earl Sterndale Primary School, Ormskirk Lathom Park Primary School, Foston Primary School, Nun Monkton Primary School, Netherton Northside First School and Holy Island First School. For similar reasons we also excluded four schools with very high reported occupancy rates (greater than 200%). Whether these extreme values are due to highly unusual circumstances or errors in the data is unclear, but none of them appear to be typical of either local or national trends. These schools are: Windrush Primary School (capacity 200, pupils 515), Littlemoor Children's Centre and School (capacity 112, pupils 227), Gearies Primary School (capacity 260, pupils 699) and West Thurrock Academy (capacity 203, pupils 553).

Immigration's impact

Map 1: Proportions of non-British white pupils by parliamentary constituency (2015)
Map 2: Changes in proportions of non-British white pupils by parliamentary constituency (2011-2015)
Map 3: Changes in proportions of non-British white pupils by parliamentary constituency (2011-2015)
Map 4: Changes in proportions of non-British white pupils by school (2011-2015)
Map 5: Proportions of non-British white pupils by parliamentary constituency (2015)
Figure 1: Change in percentage of non-British white pupils against starting percentage (2011-2015)
Notes: Hover over each dot to see the name of the local authority and exact numbers.
Click on the legend to turn individual regions on or off. Double-click to view just one region on its own.
Figure 2: Changes in percentages of non-British white pupils and pupils with English as an additional language (2011-2015)
Note: Hover over each dot to see the name of the local authority and exact numbers.
Click on the legend to turn individual regions on or off. Double-click to view just one region on its own.
Figure 3: Changes in percentages of non-British white pupils and pupils eligible for free school meals (2011-2015)
  1. 'All schools': All mainstream state primary schools in England (17,095 schools).
  2. 'Low white immigrant schools': All schools in group 1 that have 5% or less non-British white pupils and 5% or less EAL pupils (6,748 schools).
  3. 'High white immigrant schools': All schools in group 1 that have 25% or more non-British white pupils and 25% or more EAL pupils (273 schools).
  4. 'Similar low white immigrant schools': A subset of the schools in group 2 that are as similar as possible to those in group 3 except in terms of their proportions of non-British white and EAL pupils (253 schools).

Figure 4: Pupil characteristics (2015*)
 
Sample sizes: All schools: 14,563-17,047. Low WI schools: 6,327-6,748.
Similar low WI schools: 266-273. High WI schools: 266-273
(Sample sizes vary because not all data fields are available for all schools.)
Sources: Department for Education; SchoolDash analysis.
* 'Special educational needs' data are from 2014.
Figure 5: Key Stage 2 attainment (2015)
 
Sample sizes: All schools: 6,453-14,559. Low WI schools: 2,172-6,747.
Similar low WI schools: 88-272. High WI schools: 88-272
(Sample sizes vary because not all data fields are available for all schools.)
Sources: Department for Education; SchoolDash analysis.
Figure 6: Key Stage 2 results (2015)
 
Sample sizes: All schools: 14,560-14,563. Low WI schools: 6,747-6,748.
Similar low WI schools: 272-273. High WI schools: 272-273
(Sample sizes vary because not all data fields are available for all schools.)
Sources: Department for Education; SchoolDash analysis.
Figure 7: Relative value-added score (2015)
 
Sample sizes: All schools: 6,431-14,507. Low WI schools: 2,164-6,725.
Similar low WI schools: 88-271. High WI schools: 88-271
* Value-added scores provided by the DfE use a median value of 100 but have been rebased here to a median value of zero.
(Sample sizes vary because not all data fields are available for all schools.)
Sources: Department for Education; SchoolDash analysis.
Figure 8: Pupil characteristics (2015*)
 
Sample sizes: Similar low WI schools outside London: 89-96.
High WI schools outside London: 89-96. High WI schools in London: 177.
(Sample sizes vary because not all data fields are available for all schools.)
Sources: Department for Education; SchoolDash analysis.
* 'Special educational needs' data are from 2014.
Figure 9: Key Stage 2 attainment (2015)
 
Sample sizes: Similar low WI schools outside London: 20-96.
High WI schools outside London: 20-96. High WI schools in London: 72-176.
(Sample sizes vary because not all data fields are available for all schools.)
Sources: Department for Education; SchoolDash analysis.
Figure 10: Relative value-added score (2015)
 
Sample sizes: Similar low WI schools outside London: 20-94.
High WI schools outside London: 20-94. High WI schools in London: 72-175.
* Value-added scores provided by the DfE use a median value of 100 but have been rebased here to a median value of zero.
(Sample sizes vary because not all data fields are available for all schools.)
Sources: Department for Education; SchoolDash analysis.
  • The average national increase in the proportion of non-British white (NBW) pupils between 2011 and 2015 was just over 1 percentage point across all schools, but was disproportionately concentrated in particular areas, especially ones where such pupils were already relatively common. Certain parts of London and a few other specific areas, mainly in southern England, showed the largest increases.
  • There is little evidence that schools with high proportions of NBW and EAL pupils do much worse than those serving predominantly native British children, and some evidence that they can do much better, at least in London.
  • The effects of EU immigration in terms of both numbers and their effects on academic attainment therefore appear very local, with most areas largely unaffected, a few areas heavily affected and London positively thriving, at least in terms of academic outcomes.

  1. I'm the son of a Polish immigrant, the husband of a Japanese immigrant, and have been an immigrant myself, both in Japan and in the US. It will therefore not surprise you to learn that in general I'm in favour of people being able to move about the world freely. However, as usual, our purpose here is not make the case either way but rather to analyse the numbers as objectively as we can.
  2. Four schools (St Edmund's Catholic School in Dover, The Five Islands School in the Isles of Scilly, The Bay Church of England Primary School on the Isle of Wight and The Priory Ruskin Academy in Grantham) have been omitted from this analysis because their numbers show huge year-on-year swings that materially affect local and regional averages. These usually take the form of very large reductions between 2011 and 2012. Note also that the The Five Islands School is the only school in the Isles of Scilly, so that local authority area does not appear in any of these analyses.
  3. The City of London, a small authority area with about 2,350 school children, showed a very large drop of 5.5 percentage points in the proportion of EAL pupils (along with a modest fall of 0.4 percentage points in the proportion of non-British white pupils). It has therefore been omitted to make the graph more legible.
  4. The group of 'similar low WI schools' was created by taking each of the 273 'high WI schools' and matching it with the 'low WI school' that most resembled it in terms of deprivation (as indicated by the proportions of pupils who were eligible for free school meals or the pupil premium), prior attainment (as measured by the proportions of pupils who showed high or low attainment at the end of Key Stage 1) and special educational needs (as indicated by the proportion of pupils with a SEN statement or on School Action Plus). Note that any given low WI school can be matched with more than one high WI school. The vast majority of schools were matched only once, but 18 were matched twice and 2 were matched three times. As a result, the number of different schools in the similar low WI school group is 253. In addition, when calculating statistics for high WI schools, each school was only included if data for its corresponding similar low WI school was also available (and vice versa). This allowed us to make sure we were comparing like with like, though in practice it usually made little difference to the numbers.

 

 | Copyright © 2017 |