Volume 50, Issue 1 p. 414-437
ORIGINAL ARTICLE
Open Access

Failing children with Special Educational Needs and Disabilities in England: New evidence of poor outcomes and a postcode lottery at the Local Authority level at Key Stage 1

Francisco Azpitarte

Corresponding Author

Francisco Azpitarte

Loughborough University, Loughborough, UK

Correspondence

Francisco Azpitarte, Loughborough University, Loughborough LE11 3TU, UK.

Email: [email protected]

Search for more papers by this author
Louise Holt

Louise Holt

Loughborough University, Loughborough, UK

Search for more papers by this author
First published: 27 November 2023
Citations: 2

Abstract

This paper sets out original findings from analyses of the English National Pupil Database of Key Stage 1 (KS1) attainment, to examine educational outcomes of children with Special Educational Needs and Disabilities (SEND). The schooling of these children has been entirely within the context of the current SEND system, defined by the 2014–2015 policy of the Children and Families Act and Code of Practice. With a strong focus on children's needs and outcomes, the policy intends to achieve high educational outcomes for children with SEND. Our new results show, however, that children with SEND are one of the most disadvantaged groups in education, and they are far less likely to meet expected learning standards than their peers at KS1. For instance, about 44%, 31% and 23% of children with SEND met the standards in phonics, reading and writing, respectively, compared to 88%, 83% and 78% of children with no SEND. Further, our spatial analysis shows for the first time that this disadvantage displays large spatial variability across Local Authorities: there is a postcode lottery in the education of children with SEND. The new findings provide strong evidence that the new SEND policy is failing many children with SEND, and that this performance varies markedly across space. This adds further weight and evidence to a growing recognition, even from government, that the SEND system needs to change, and that the ambitious aims of the transformation of education and care for children with SEND in 2014 and 2015 are not being realised.

Key insights

What is the main issue that the paper addresses?

The aim of this paper is to investigate early educational outcomes of children with Special Educational Needs and Disabilities (SEND), whose education has taken place entirely within the context of the current SEND framework, defined by the 2014–2015 policy of the Children and Families Act and Code of Practice.

What are the main insights that the paper provides?

Despite the focus on promoting children's educational outcomes, our new results show that children with SEND continue to be one of the most disadvantaged groups in education, and they are far less likely to meet expected learning standards than their peers. Further, our spatial analysis shows for the first time that this disadvantage displays large spatial variability across Local Authorities, providing strong evidence of a postcode lottery in the education of children with SEND.

INTRODUCTION

The 2014 Children and Families Act and the statutory guidance in the accompanying Code of Practice of 2015 represented a major overhaul of the education system of children with Special Educational Needs and Disabilities (SEND) in England. Regarded by many as the most radical reform in decades (Lindsay et al., 2020), the new policies proposed a holistic approach to SEND provision and education, a greater focus on early and school-based intervention, and placed new responsibilities on Local Authorities (LAs) to create a framework where children with labels of SEND ‘reach their potential’ and ‘get the best possible outcomes’ (Department for Education, 2014, p. 14). Despite laudable aims, recent new evidence suggests that the new SEND policy has been failing in its intentions, as exemplified by the most recent government review, the publication of a new Green Paper and SEND and Alternative Provision Improvement Plan (Department for Education, 2023a; Department for Education & Department of Health and Social Care, 2022), and growing evidence that many LAs are failing to provide adequate support for children with SEND in their communities (Ofsted, 2021).

The current policy framework defined by the 2014 Children and Families Act has early identification of children's needs and developmental outcomes as the keystone of an adequate SEND provision. However, the significant weakness identified by Ofsted, the Care Quality Commission and national government bodies, such as the Department for Education (DfE) and the Education Select Committee, demonstrate that the aims of the new policies are not being met. The aim of this paper is to empirically investigate that question by evaluating recent trends in early academic outcomes of children with SEND, whose entire education career has been overseen by the 2014–2015 policy framework of the Children and Families Act and Code of Practice. This will also provide a benchmark against which to evaluate the impact of future changes in policy and practice. The paper uses publicly available data on the academic attainment of children at Key Stage 1 (KS1), the Phonics Screening Check and KS1 teacher assessments in reading, writing, mathematics and science in England from the 2019 National Pupil Database, a near 100% sample of children drawn from the School Census.

The paper contributes to the existing literature by investigating the spatial variability in children's academic attainments using harmonised data that allows the comparison of attainments of children with SEND across LAs. The analysis of spatial inequalities is motivated by recent policy assessments, which indicate that the new SEND system is failing many children and that the implementation of the new SEND policies is spatially differentiated (House of Commons, 2020). The spatial variation is corroborated by the outcomes of inspections of local SEND systems by Ofsted and the Care Quality Commission since 2016, which found that many LAs are failing to meet their statutory duties regarding SEND provision (Ofsted, 2021). The aim of this paper is to explore the consequences of these local disparities by evaluating spatial inequalities in the early attainments of children with SEND, to provide further insights into the spatial disparities of the current SEND system.

We present two key original findings about children who have been exclusively educated during the new framework for SEND. First, we emphasise that the outcomes at KS1 of children with SEND labels are much lower than their non-disabled peers, suggesting that schools and LAs are systematically failing to achieve the key aim of the legislative changes embedded in the SEND Code of Practice 2015, to ‘reach their potential’ and ‘get the best possible outcomes’ (DfE, 2015). Second, we demonstrate significant spatial variability in outcomes at KS1 across LAs in England, which dwarfs the spatial variability in outcomes observed for children with no SEND. In sum, we argue that a label of SEND has a significant impact on the outcomes of children, mediating known patterns of inequality and education, and that there is a postcode lottery in terms of outcomes for young children with SEND. Some LAs significantly outperform others in relation to the numbers of children with labels reaching expected levels in the phonics screening check and the national curriculum teacher assessments for reading, writing, mathematics and science at KS1.

The paper proceeds through five key sections. In the next section, we situate our argument in debates about disadvantage, education, inequality and SEND. The third section discusses our data and methods and the key statistical analyses we use to analyse the educational attainment of pupils at KS1 of education, and spatial and non-spatial inequalities in these attainments. The fourth section presents the key findings from these analyses. The final section discusses the findings and their implications and presents our conclusions.

SITUATING DISADVANTAGE, EDUCATION, INEQUALITY AND SEND

There is an interesting paradox that education simultaneously enhances quality of life and opportunities, and (re)produces existing patterns of advantage and disadvantage within societies. For instance, Raudenbush and Eschmann (2015) emphasise that expanding education reduces class-based inequalities and provides general benefits to societies (see also Downey & Condron, 2016). Conversely, a substantial literature documents how education reproduces enduring advantages and disadvantages, particularly in education systems which are characterised by neoliberalism, incorporating competition, ‘choice’, marketisation and fiscal responsibility (e.g., Ball, 2018; Calarco, 2014; Haelermans et al., 2022).

This paper focuses on the outcomes of young people with SEND and we argue that they are one of the most disadvantaged groups in education. Further, we emphasise that their experience is socio-spatially shifting, with both generalised poor outcomes and a geographically variable postcode lottery. Following Domina et al. (2017), who assert that schools create ‘categorical inequality’, we argue that SEND is a specific, powerful category which exists within school spaces, although it has connections to other labels and experiences (e.g., specific impairments). It has significant categorical power and is an important element in (re)producing educational disadvantage, with impacts across a young person's life course. This is not to suggest that SEND does not exist in a real sense—in common with critical accounts of (dis)ability (Hall & Wilton, 2017), we argue that SEND is both a factor of different and sometimes painful and frustrating mind–body emotions and how they intersect with social and spatial processes, such as norms and expectations of ability and curricula, which are politically constituted. Such arguments have great purchase and visibility in relation to people with recognised impairments, particularly those of the body; however, they are less generally accepted when theorising learning, neurodiversity or social–emotional and mental health differences. Under this framing, it should not be accepted that young people with labels/experiences of SEND should under-achieve educationally. Instead, the gaze should be turned upon curricula, ideas of achievement, norms of appropriate behaviour, the way school environments are socio-spatially organised and associated choices about which groups of students fall outside the general educational institution. These are powerful acts, as the labelling of young people often intersects with other axes of power, class, race and gender (Shaw et al., 2016; Youdell, 2010; Youdell & Armstrong, 2011). In addition, as Chatzitheochari and Butler-Rees (2022) emphasise, the experience of young people with labels of SEND is intersectional, with those from poor and socially excluded backgrounds experiencing enhanced stigma, which influences their experiences of school and potentially their outcomes.

In 2014 and 2015, SEND education was subject to a major overhaul as a new Children and Families Act and Code of Practice were implemented to address the failings of the existing framework (Department for Education, 2014, 2015). There were some new (and more familiar) strands to the policy, including a focus on early intervention, a move away from an adversarial system, a more integrated system for meeting education, health and social care needs through statutory requirements for joint commissioning, an increase in the age range of support to 25 years of age and a greater requirement to work in partnership with families at both an individual and strategic level. The Code of Practice also enshrines an associated statutory duty on LAs to provide information via a Local Offer, providing information about all the services available to children with SEND in the area, including those within the LA and nearby, which local children and their families might use.

The Code of Practice enshrined a shift of provision and resources to schools so that the assistance for most young people with SEND was provided through SEND support in schools. This would provide support early and without the need for labelling, diagnosis or the bureaucratic lag tied to seeking the legal status of the new, more holistic Education, Health and Care Plan (EHCP), which replaced the ‘Statement of Special Educational Needs’. For most young people, support should be coordinated by schools in liaison with external agencies as appropriate. To facilitate this shift, the Code of Practice set out that each state school will have funds allocated within their budget for SEND. The intent of the new Code of Practice was that new statutory EHCP would be reserved for children and young people with higher levels of need, who required more intensive and specialist support. Despite this laudable aim, the numbers of young people gaining EHCPs has increased by 118.9% since 2014 (calculation by authors, using data provided by Department for Education, 20142023b). This rise suggests that SEND support in schools is not working. This failure is perhaps connected to the fact that the school-level budget is ‘notional’ rather than ‘ring fenced’, and therefore there is no specific requirement for schools to spend the funds on SEND. Within the context of austerity and reducing per capita budgets on education, there is no guarantee that schools will use their nominal SEND budget for the purpose it is intended. In this context, it is probable that the gaining of EHCPs is also influenced by the social, cultural and economic capital of families, as much as by the ‘needs’ of children. Poor and socially excluded children are more likely to experience SEND (Shaw et al., 2016) and in common with its predecessor, the Statement of Special Educational Needs, it is probable that families with higher levels of capitals who have children with SEND are more likely to successfully gain both an EHCP and a specific label for their child (Holt et al., 2019).

In this paper, we examine the attainment at KS1 of children with labels of SEND in 2019, thus capturing the attainment of children whose entire education has taken place under the new policy framework. We find that the attainment of these young children with SEND is both considerably poorer than those without such labels, and geographically variable at the LA level. Thus, the new empirical findings provide evidence that the current policy framework so far has not delivered its promises. However, the descriptive nature of the analysis does not allow the establishment of any causal evaluation of SEND policies, which would require further research over time to identify the causal links between policies and educational outcomes.

DATA AND METHODS

We analyse publicly available data from the DfE on the outcomes of statutory assessments conducted in state-funded primary schools in England in 2019, which captures children whose entire education journey has taken place within the new policy framework. Specifically, we use data on the Year 1 phonics screening check and the national curriculum teacher assessments for reading, writing, mathematics and science conducted at the end of KS1.1 The phonics screening check is designed to assess pupils' understanding of phonics and is administered to pupils in Year 1, typically aged 6.2 The assessment consists of 40 words (20 real words and 20 nonsense words) and pupils are considered to meet the expected standard when they read at least 32 words correctly.

The teacher assessments of reading, writing, mathematics and science are carried out in Year 2, involving students typically aged 7, and are designed to evaluate whether pupils meet the expected standard in these subjects by the end of KS1. These evaluations are based on a broad range of evidence on pupils' knowledge and performance in the subjects, including mathematics and reading tests performed in Year 2. When conducting the assessments, schools must follow the teaching assessment frameworks produced by the Standard and Testing Agency for each subject. Furthermore, to ensure the assessments are appropriate and consistent with national standards, LAs must moderate a quarter of their schools every year and ensure every school is moderated at least once every 4 years (Department for Education, 2019). Thus, teacher assessments are designed to be comparable across regions and therefore suitable to examine spatial inequalities.

Our dataset comprises information on the proportion of pupils who met the expected standard on the phonics screening check and the KS1 assessments in each LA in 2019. Importantly, the data produced by the DfE contains disaggregated data for pupils with and without SEND, which allows derivation of the proportion of pupils meeting expectations in each of these groups. Data on the phonics screening check were collected from 16,487 schools, whereas 16,521 schools contributed their data on KS1 outcomes. Both samples represent more than 99% of KS1 state-funded schools across England, and can therefore be regarded as representative of that population of schools. A detailed list of the variables and data sources used to construct the dataset is provided in Appendix A.

The final dataset contains data on the outcomes of all pupils and pupils with and without SEND in 149 LAs.3 The dataset also has a range of variables known to influence outcomes and which might help to explain local variation in the outcomes of pupils with SEND. These variables were constructed using public data from the DfE on the characteristics of the local pupil population enrolled in Year 1 and Year 2, including the number of pupils in the LA with SEND support and the number on an EHCP, the number of pupils eligible for free school meals (FSM) and the number of pupils for whom English is the first language. Furthermore, the dataset contains information on specific characteristics of the pupils with SEND in Year 1 and Year 2, including their gender, primary type of special need and the type of school they were attending in 2019.4 These variables have been used extensively in the academic literature aimed at investigating the academic progress of children with SEND and its drivers (e.g., Crawford & Vignoles, 2010; Parsons & Platt, 2017), and there is ample recent evidence documenting differences in early educational attainments of groups defined by these characteristics that justify their inclusion as covariates in the statistical analyses, as they could contribute to explaining spatial differences in early educational attainments across LAs. Thus, evidence from the DfE shows important variations in attainment by levels of SEND support, language, gender and primary needs of children with SEND (Department for Education, 2019, 2022). This is also true for FSM eligibility, which is widely used as a proxy for socioeconomic status and disadvantage in academic research and government statistics in England and Wales (Department for Education, 2022; Taylor, 2018).

Data produced by the DfE defines 12 categories of primary need: autistic spectrum disorder; visual impairment; multi-sensory impairment; physical disability; other difficulty/disability; moderate learning difficulty; profound and multiple learning difficulty; severe learning difficulty; specific learning difficulty; social, emotional and mental health; speech, language and communication needs; and SEN support but no specialist assessment of type of need. Data on the type of school released by the DfE allows us to quantify the number of pupils with SEND in each LA attending state-funded mainstream schools, maintained special schools, pupil referral units and non-maintained special schools. We used these variables to examine the attainment of children and assess possible explanations for any spatial disparities tied to population differences (e.g., primary need, eligibility for FSM, English as a second language, different type of education provision).

Children's early educational attainments are analysed using descriptive statistics and a range of spatial and non-spatial measures, which are presented in the next section. To account for spatial inequalities in children's attainments and evaluate the contribution of local characteristics to explaining these inequalities, we use regression analysis and decomposition techniques, which are also discussed in detail in the next section.

The data and analyses provide new evidence of children's outcomes who have been educated exclusively under the new SEND framework, providing novel insights into children's educational outcomes under the new framework and the extent and drivers of spatial inequalities in these outcomes. There are some limitations of the data. For instance, many children with SEND have intersecting needs, and consequently which ‘primary need’ is identified will differ according to local, and even school, context. In addition, the use of aggregate data produced at the LA level does not allow the analysis of outcomes and the influence of socioeconomic factors at the individual level. In particular, there is a known interconnection between poverty and SEND, and we can explore these connections using aggregate data on FSM available at the LA level, but we cannot examine the influence and interconnections at an individual level here.

FINDINGS

Early inequalities in the academic outcomes of children with SEND

This section presents evidence on the inequalities and spatial patterns in academic outcomes of children with SEND in KS1. Table 1 shows the number and percentage of pupils in that education stage by SEND category in England in 2019. Just over 13% of pupils enrolled in Year 1 had special educational needs, whereas the percentage in Year 2 was about 15%. SEND support was the most common form of support provided to pupils with SEND, with this group representing about 11% and nearly 13% of pupils in Year 1 and Year 2, respectively. Pupils with EHCPs were a small minority of the school population, accounting for about 2% of pupils in both years.

TABLE 1. Number and percentage of pupils by year and SEND category in 2019, England.
Year 1 Year 2
N % N %
All 650,479 100.00 667,987 100.00
Not SEND 564,023 86.71 565,833 84.71
SEND 86,456 13.29 102,154 15.29
SEND support 72,400 11.13 85,818 12.85
EHCP 14,056 2.16 16,336 2.45
  • Source: Authors' own calculations.

Table 2 presents the outcomes of the phonics check and KS1 assessments by pupils' SEND status across all LAs. The table includes descriptive statistics on the average, maximum, minimum and standard deviation, and the Gini index of the percentage of pupils meeting the expected standard in each assessment across LAs in 2019.5

TABLE 2. Percentage of pupils meeting expected standards in the phonics check and KS1 assessments across LAs in 2019, England.
Mean (%) Max (%) Min (%) SD (%) Gini
Phonics
All 82.11 87.13 77.08 2.18 0.02
Not SEND 88.38 92.12 83.56 1.87 0.01
SEND 44.14 61.63 31.06 6.24 0.08
SEND support 48.88 69.33 35.66 7.08 0.08
EHCP 19.49 50.00 0.00 8.72 0.24
Reading
All 75.05 83.30 66.92 3.00 0.02
Not SEND 83.47 89.81 74.34 2.72 0.02
SEND 30.67 44.52 18.79 5.53 0.10
SEND support 34.18 53.52 20.43 6.46 0.11
EHCP 12.56 29.63 0.00 5.64 0.24
Writing
All 69.45 76.40 60.81 3.31 0.03
Not SEND 78.31 86.26 69.20 3.18 0.02
SEND 22.58 37.87 12.85 5.22 0.13
SEND support 25.32 45.84 13.77 6.17 0.14
EHCP 8.35 22.58 0.00 4.18 0.25
Mathematics
All 75.80 82.02 69.84 2.75 0.02
Not SEND 83.80 88.87 76.90 2.45 0.02
SEND 33.57 49.67 20.81 5.60 0.09
SEND support 37.40 56.50 23.41 6.50 0.10
EHCP 13.56 31.00 0.00 6.12 0.24
Science
All 82.27 90.69 75.08 3.17 0.02
Not SEND 89.97 95.97 83.10 2.51 0.02
SEND 41.90 63.82 26.17 6.22 0.08
SEND support 47.05 72.44 30.16 7.19 0.09
EHCP 15.32 37.00 0.00 6.96 0.24
  • Note: SD denotes the standard deviation of the distribution.
  • Source: Authors' own calculations.

Overall, about 82% of Year 1 pupils eligible for the phonics screening check met the expected standard in 2019. The level of performance of Year 2 pupils eligible for KS1 assessments varied across subjects, ranging from 82% in science to 69% in writing. Outcomes in writing were the lowest on average, and also the ones that showed the largest variation across LAs, as indicated by our estimates of the standard deviation (3.31%) and the Gini index (0.03).

There were marked differences in outcomes between children with and without SEND in the data. The gap between the two groups ranges between about 44 percentage points in the phonics check (88% vs 44%) and 55 percentage points in writing; only 23% of children with SEND met the expected standard in writing, compared to 78% of children without SEND. Within the SEND population, children with an EHCP tend to perform, on average, worse than those on SEND support across all the assessments.6 Thus, for instance, in writing (the assessment with the lowest outcomes for both SEND groups), only about 8% of children on EHCPs met the national standard, with that percentage increasing only up to 25% among those with SEND support. Our specific analyses show that known patterns of lower attainment for young people with SEND labels (e.g., Chatzitheochari & Platt, 2019; Department for Education, 2019; Department for Education & Department of Health and Social Care, 2022) are evident even for children whose school education journeys and most of their education have been entirely within the context of the new SEND framework.

There are stark inequalities between different LAs. The summary measures reported in Table 2 suggest large spatial differences in the outcomes of these assessments. The Gini index for children with SEND is more than five times larger than that of children with no SEND for the phonics check and the KS1 assessments. These findings demonstrate that there are large spatial inequalities in attainment affecting children with SEND, more than any other group of young people, providing compelling evidence of a postcode lottery in attainment for children with SEND at KS1.

The comparison of the Gini coefficients and standard deviations of attainment levels of different groups of SEND provision reveals larger levels of spatial inequalities for children with SEND with an EHCP than for those with the lesser degree of intervention, SEND support. The Gini index for children with EHCPs is the largest of all groups, at 0.24 or higher for all assessments. Comparatively, the outcomes of children with SEND support are more evenly distributed, although still substantially unequal across space, with Gini indices ranging between 0.8 in the phonics check and 0.14 in writing, which are still substantially more unequal than the outcomes of children with no SEND. These findings are further supported by results from the kernel density analysis of the spatial distributions, which are included in Appendix A.

The geography of inequalities in the outcomes of children with SEND

In this section we present graphical evidence and estimates of spatial measures of inequality, to show the geographical dimension of these patterns across different LAs. Specifically, we characterise spatial distributions using Moran's index of spatial autocorrelation (Moran, 1950), which quantifies the degree of concentration and clustering of areas with similar attributes—in this case, the percentage of students meeting a certain standard. For any assessment and group of children, the index is defined as
I = l k w l , k z l z k / W l z l 2 / N ,
where the superscripts for the assessment and group of children have been omitted for the sake of clarity, and z l and z k are the percentage of pupils meeting the standard in local areas l and z , respectively; w l , k is the element of the spatial weight matrix corresponding to locations l and k ; W = l k w l , k is the sum of all spatial weights; and N is the number of locations. Widely used in spatial analysis to describe spatial patterns, positive (negative) values of Moran's index indicate positive (negative) spatial autocorrelation and the propensity of areas to be surrounded by areas with similar (dissimilar) levels of academic achievement.

Table 3 reports estimates of Moran's index for the percentage of pupils meeting the expected standard in the phonics check and each of the KS1 assessments. In addition, the right-hand column shows the indices for the average of the KS1 assessments, which was derived by assigning each LA the average of the four percentages of pupils meeting the standard in each of the four KS1 assessments. All estimates of Moran's index were computed using the GeoDa open-access software (Rey et al., 2015) and queen-based contiguity weights of order one. Results from robustness checks (available upon request) showed that the main qualitative conclusions presented here are robust to the use of alternative spatial weights.

TABLE 3. Moran's spatial correlations in 2019, England.
Phonics Reading Writing Mathematics Science Average KS1 subjects
All 0.34 0.36 0.37 0.37 0.28 0.34
Not SEND 0.36 0.33 0.34 0.37 0.24 0.33
SEND 0.53 0.46 0.42 0.36 0.39 0.44
SEND support 0.57 0.53 0.48 0.43 0.46 0.51
EHCP 0.35 0.20 0.21 0.21 0.19 0.23
  • Source: Authors' calculations.

Our estimates of Moran's index provide strong evidence of positive spatial autocorrelation in the level of attainment across LAs in England. This means that the spatial distribution of attainments at the LA level is not random and there is evidence of clustering of LAs by levels of attainment: LAs with low (high) levels of attainment tend to be clustered near LAs with similar low (high) levels of attainment, leading to a non-random distribution of attainment across space. This is true for all groups of children and all types of assessment, as indicated by the positive values of our estimates, consistent with the concentration and clustering of areas in terms of academic attainments. The identification of these clusters is something we will address below using graphical evidence on the spatial distributions of academic outcomes. The value of Moran's index for children with SEND is higher than that for children with no SEND for most of the assessments considered. Thus, for instance, Moran's index for the outcomes of the phonics screening test for pupils with SEND is 0.53, which is above the 0.36 found for children with no SEND. Similar results were found for all the KS1 assessments except mathematics, where the degree of spatial correlation for the outcomes of that test were similar for both groups of children. Overall, these results highlight the greater importance of space for the inequalities in the academic attainment of children with SEND relative to children with no SEND, as the spatial distribution of attainments of children with SEND shows more spatial clustering than that of children with no SEND.

Inspection of the results for the different SEND groups shows large differences between these groups regarding spatial autocorrelation. In particular, our estimates suggest a larger spatial clustering in the outcomes of children with SEND with support than in the outcomes of those with EHCPs. Thus, Moran's index for the outcomes of children on SEND support is well above that of any other group of children for all assessments considered. For instance, the estimated indices for the outcomes of that group in the phonics check and the reading assessments are, respectively, 0.57 and 0.53, the largest observed for any group and type of assessment. This suggests an important role of location for the outcomes of children on SEND support, far more important than for any other group of children. In contrast, levels of spatial autocorrelation for the outcomes of children on an EHCP are the lowest of all groups, including those with no SEND. This is true for all the assessments, which implies a lower spatial autocorrelation and clustering in the outcomes of children with the most severe and complex needs.

The maps in Figure 1 provide further insights into the spatial patterns of the academic outcomes of children with SEND. In particular, the maps present the spatial distribution of the percentage of pupils with SEND meeting the standard in the phonics check (left panel) and the average percentage across the four KS1 assessments (right panel). For both maps, LAs were grouped into quartiles where the bottom (top) quartile comprises those areas with lowest (highest) levels of attainment. Maps for the individual KS1 assessments revealed very similar patterns and are available upon request.

Details are in the caption following the image
Percentage of pupils meeting the standard in the phonics screening test and KS1 assessments in 2019, England.

The maps provide evidence of large spatial inequalities and clustering in the attainment of students with SEND. Both maps show a cluster of higher performing LAs in London and its surrounding areas. Thus, most London boroughs were in the top two quartiles of the outcomes in the phonics screening check and KS1 assessments in 2019. However, not all the top-performing LAs were in London. Thus, for instance, Northumberland and Tyneside in the north, Southampton and Dorset in the south, and Herefordshire in the west were all among the best performing LAs in terms of the outcomes of children with SEND. Some of these areas, including Northumberland, Southampton and Dorset, ranked in the top quartiles of both the phonics check and KS1 assessments in 2019.

At the opposite end, the maps show a low-performing cluster of LAs comprising areas in the East Midlands and east of England. This cluster includes areas such as Staffordshire, Nottinghamshire, Lincolnshire, Leicestershire, Peterborough and Suffolk, which are all in the bottom quartile of both the phonics and KS1 assessments. In all these areas, the percentage of pupils with SEND meeting the expected standard in the phonics check was below 40% and the average performance in the KS1 assessments was below 28%, well below the (low) national average across LAs. Many local areas in the northwest region of England also performed poorly in terms of the outcomes of children with SEND. Thus, coastal areas including Cumbria, Lancashire and Cheshire West, as well as inland places such as Bolton and Wigan, were all in the bottom quartile of performance in the phonics check, with most of these areas also being in the bottom quartile of the KS1 assessments.

Accounting for inequalities in the outcomes of children with SEND

The evidence on spatial and spatial measures of inequality discussed in the previous section shows large spatial inequalities in the attainment of children with SEND in KS1 education across areas in England. Importantly, these inequalities are substantially larger than those found for children with no SEND, highlighting the important role of place for the educational opportunities and attainment of children with SEND in the context of England.

To shed light on the reasons behind the large spatial inequalities in the outcomes of children with SEND, this section aims to identify potential drivers and their contribution to these inequalities. For instance, the large spatial inequalities could be driven by differences in the demographic and socioeconomic characteristics of local populations and children with SEND. To evaluate these contributions, we make use of regression models of the form
y i = X i β + u i ,
where y i denotes the outcome variable whose variation one is interested in explaining—in our case, the percentage of pupils with SEND meeting the national standard in each assessment in any LA i ; X i is a vector of factors that could contribute to explaining the variation in outcomes across geographies; β is the vector of parameters summarising the relationship between each of these factors and the outcome variable; and u i is an error term including unobservable factors influencing the outcomes of children with SEND.

To account for the inequalities in the outcomes of children with SEND, we consider a set of factors that existing research has identified as likely to influence children's outcomes, and whose variation across areas may explain the inequalities in attainment. Specifically, we consider characteristics of the overall pupil population (gender, FSM eligibility) and also specific characteristics of children with SEND attending state-funded schools in each LA in 2019. Eligibility for FSM or pupil premium as a proxy for socioeconomic hardship is associated with lower outcomes at all key stages (Ashraf et al., 2021; Department for Education, 2019). A relatively high proportion of young people diagnosed with SEND experience socioeconomic hardship or poverty, as measured by FSM eligibility (Keslair & McNally, 2009; Shaw et al., 2016). Further, recent analysis demonstrates that poverty and/or social class and a label of SEND intersect in specific ways in schools, such that young people with special educational needs from poor or socially excluded backgrounds are particularly disadvantaged (Chatzitheochari & Butler-Rees, 2022). To match the timing of assessments, the analysis for the phonics screening check will consider the characteristics of the cohort enrolled in Year 1, whereas the analysis for the KS1 assessments will use the characteristics of children in Year 2.7

To account for differences in the prevalence of children with SEND across LAs, the models include as covariates the proportion of students in Year 1 and Year 2 with an EHCP and with SEND support. Furthermore, differences in the characteristics of the SEND population across areas are captured by a set of covariates including information on the gender composition of pupils with SEND, the type of school provision and the type of primary need of those children. Thus, the models include a variable with the percentage of boys in the local SEND population, the proportion of children with SEND attending state-funded mainstream schools, maintained special schools, non-maintained special schools and pupil referral units (PRUs). Information on the type of primary need is coded into four variables capturing the four most prevalent types of need among KS1: children with SEND including speech, language and communication needs; social, emotional and mental health; moderate learning difficulty; and autistic spectrum disorder. A fifth variable, denoted other primary need, was created to capture the proportion of children with less prevalent forms of need, including visual impairment, multi-sensory impairment, physical disability, other difficulty/disability, profound and multiple learning difficulty, severe learning difficulty, specific learning difficulty and SEN support but no specialist assessment of type of need.

To evaluate the contribution of each individual covariate to the goodness-of-fit of the model and the spatial variation in SEND outcomes, we apply a Shapley-type decomposition method (Sastre & Trannoy, 2002; Shorrocks, 2013). Based on the concept of the Shapley value, this type of decomposition quantifies the contribution of each covariate with the expected marginal impact on the R 2 measure of including the covariate in the model. The expected marginal impact on a given covariate j is derived considering all possible subsets of covariates that covariate j could be added to and it is given by the following expression:
Sh j = S K , j S s 1 ! K 1 ! K ! e S e S \ { j ] e K with j = 1 K Sh j = 1 ,
where Sh j denotes the Shapley contribution of factor j ; S K is any subset of the K covariates including covariate j ; s is the number of covariates in the subset S ; and e is the statistic one is interested in explaining, which in our case is the R 2 of our regression models.

Before presenting the regression and decomposition results, Table 4 shows key descriptive statistics of the covariates used in the regression analysis. In particular, the table reports the characteristics of Year 2 children used to account for the spatial variation in the outcomes of KS1 assessments. The characteristics of Year 1 children used for the analysis of the phonics screening test are very similar and are not reported here but are available upon request.

TABLE 4. Characteristics of Year 2 pupils across LAs, England.
Characteristic Mean Max Min SD Gini
All children
% of children eligible for FSM 16.94 34.00 5.68 6.29 0.21
% of children with English as first language 77.10 98.38 24.63 18.56 0.13
Children with SEND
% of children with SEND: boys 68.45 75.51 61.70 2.44 0.02
% of children with SEND: mainstream schools 93.80 100.00 87.31 2.49 0.02
% of children with SEND: special schools (maintained) 6.01 12.47 0.00 2.46 0.22
% of children with SEND: special schools (not maintained) 0.07 1.32 0.00 0.22 0.41
% of children with SEND: pupil referral unit 0.11 1.26 0.00 0.21 0.37
% of children with SEND: speech, language and communication needs 35.19 56.99 18.99 7.58 0.12
% of children with SEND: moderate learning difficulty 17.27 43.51 2.60 8.19 0.27
% of children with SEND: social, emotional and mental health 14.64 25.79 7.44 3.32 0.13
% of children with SEND: autistic spectrum disorder 9.45 23.27 1.53 3.94 0.23
% of children with SEND: other primary need 23.46 45.27 7.81 5.94 0.14
  • Source: Authors' calculations.

The evidence reported in Table 4 suggests a large variation in the characteristics of the local pupil population and children with SEND across LAs. The proportion of Year 2 children eligible for FSM (a standard indicator of socioeconomic disadvantage at the local level) ranged from 6% to 34%, with an average of 17% across LAs. Similarly, there are substantial disparities in the proportion of children with English as first language, ranging between 25% in London boroughs with a relatively small White British population (e.g., Newham and Tower Hamlets) to 98% in areas where being White British is the overwhelming majority in the population (e.g., Redcar and Cleveland, Northumberland).

Most children with SEND in Year 2 were attending a mainstream school in 2019, with the proportion ranging between 87 and 100 across LAs. Less than 1% of children with SEND were attending non-maintained special schools in 2019 but there are large inequalities in that type of provision, as indicated by the value of the Gini index (0.41; the largest of all covariates). Similarly, a small proportion of children with SEND were attending PRUs but, as with non-maintained special schools, that provision is very unequal, as indicated by the Gini index of 0.37, which suggests that the role of these educational settings in the local offer for children with SEND differs markedly across areas.

Speech, language and communication needs are the most prevalent form of need, representing the primary need of more than 35% of children with SEND. Moderate learning difficulties is the second most frequent need, at 17%. However, its prevalence varies largely across areas, ranging between 3% and 44%, which makes it the most unequal of all primary needs, with a Gini index of 0.27, followed by autism with an index of 0.23. These differences are important, as children's educational attainments are likely to be influenced by their type of primary needs, which may contribute to accounting for the inequalities in attainment in KS1.

Spatial differences in the socioeconomic characteristics of the local population and the population of children with SEND could in principle contribute to explaining the inequalities in the attainment of children with SEND across LAs. To evaluate this contribution, Table 5 presents the estimates of the regression models for the percentage of children with SEND meeting the national standards across LAs in England in 2019.

TABLE 5. Regression for the percentage of students meeting standards across LAs, England.
Phonics Reading Writing Mathematics Science Average KS1 subjects
% eligible for FSM −0.21*** −0.24*** −0.14** −0.21*** −0.49*** −0.27***
(0.07) (0.07) (0.06) (0.07) (0.09) (0.07)
% with English as first language −0.14*** −0.08*** −0.08*** −0.09*** −0.01 −0.07***
(0.02) (0.02) (0.02) (0.02) (0.03) (0.02)
% with EHCP 1.18** 0.94* 0.36 0.73 1.77*** 0.95**
(0.54) (0.50) (0.46) (0.50) (0.62) (0.47)
% with SEND support 0.81*** 0.91*** 0.91*** 1.07*** 1.00*** 0.97***
(0.20) (0.19) (0.18) (0.20) (0.24) (0.18)
% SEND: boys 0.05 −0.05 0.03 0.17 −0.35** −0.05
(0.13) (0.13) (0.12) (0.13) (0.16) (0.12)
Type of SEND provision
% SEND: special school (maintained) −0.44*** −0.25* −0.16 −0.34** −0.40** −0.29**
(0.14) (0.14) (0.13) (0.14) (0.17) (0.13)
% SEND: special school (not maintained) −0.23 2.24 1.71 3.18** 4.93*** 3.02**
(1.91) (1.39) (1.28) (1.41) (1.72) (1.32)
% SEND: pupil referral unit −0.67 −3.03** −1.70 −2.62* −2.68 −2.51*
(2.20) (1.48) (1.37) (1.50) (1.84) (1.41)
Type of primary need
% SEND: Social, emotional, mental health −0.12 0.10 0.05 0.00 0.00 0.04
(0.12) (0.11) (0.10) (0.11) (0.13) (0.10)
% SEND: moderate learning difficulties 0.02 −0.34*** −0.33*** −0.30*** −0.31*** −0.32***
(0.09) (0.05) (0.05) (0.05) (0.06) (0.05)
% SEND: autistic spectrum disorder −0.29*** 0.10 0.07 0.09 0.10 0.09
(0.06) (0.10) (0.09) (0.10) (0.12) (0.09)
% SEND: other primary need −0.15* −0.11 −0.15** −0.08 −0.09 −0.11*
(0.08) (0.07) (0.06) (0.07) (0.08) (0.06)
Constant 54.01*** 38.04*** 26.00*** 25.50** 66.52*** 39.02***
(10.59) (10.38) (9.59) (10.50) (12.88) (9.87)
Observations 149 149 149 149 149 149
R2 (%) 61.35 60.20 61.73 60.24 51.44 60.85
  • Note: Standard errors in parentheses. ***p < 0.01; **p < 0.05; *p < 0.1.
  • Source: Authors' calculations.

Our regression results provide evidence of strong statistical associations between local characteristics and the outcomes of children with SEND. Eligibility for FSM and the prevalence of English as first language are both negatively associated with the attainment of children with SEND. Specifically, our research suggests that areas with large proportions of children eligible for FSM and children for whom English is the first language have, on average, lower attainments than other areas, and this is true for all assessments as indicated by the negative, statistically significant value of the coefficients. For instance, the estimated coefficients imply that a one percentage point increase in the number of children eligible for FSM is associated with a decline in the proportion of children with SEND meeting the standard, ranging between 0.14 points in writing and 0.49 points in science. These findings give further evidence of the intersectionality of SEND with other characteristics at the LA level (e.g., Chatzitheochari and Butler-Rees, 2022; Keslair & McNally, 2009; Youdell, 2010). Note, however, that these and the other estimated coefficients cannot be given a causal interpretation as the regression design is purely descriptive, consistent with our aim of accounting for spatial inequalities in SEND outcomes.

Our evidence suggests the existence of economics of scale in the provision of educational opportunities for children with SEND. Specifically, our research suggests that LAs with larger populations of children (with SEND) tend to perform better, on average, than other areas, as indicated by the positive and statistically significant estimates of the percentage of children on EHCPs and SEND support in the LA. Although the significance of the effect varies across assessments, all the estimated coefficients indicate a positive effect, with a larger presence of children with SEND being associated with better attainments of those children.

The type of educational provision and the primary needs of children also influence the outcomes of children with SEND at the local level. Our results suggest that having a larger proportion of children with SEND enrolled in non-maintained special schools is associated with better outcomes in most assessments, and this association is particularly strong and significant for mathematics and science. Thus, based on our estimates, increasing the proportion of children with SEND in those settings by 1% would lead to an increase of 3 and 4 percentage points in the proportion of children meeting the standard in mathematics and science. On the contrary, the use of PRUs for SEND provision is associated with lower attainment. LAs with larger proportions of children with SEND in PRUs have, on average, worse outcomes than other areas in most assessments, and this effect is particularly large for reading and mathematics.

A larger proportion of children with SEND with moderate learning difficulties is related to lower levels of attainment in all assessments except phonics. The estimated coefficients for these assessments are all negative and statistically significant, and suggest that a one percentage point increase in the proportion of children with moderate learning difficulties in the LA is associated with a decline of about 0.3% in the proportion meeting the standards, and this effect is very similar across assessments.

Overall, the regression models can explain a sizeable part of the spatial variation in SEND children's attainments across LAs. The R 2 measure of all models is about 50%, which suggests that differences in the characteristics of LAs and children with SEND account for more than half of the variation in children's outcomes across areas. Table 6 presents the Shapley contribution of each covariate to the goodness-of-fit of the model, which can be interpreted as the covariate's contribution to the part of the variation in outcomes that is explained by the model.

TABLE 6. Shapley contributions to the explained variation in outcomes across LAs, England (all variables expressed as a percentage).
Phonics Reading Writing Mathematics Science Average KS1 subjects
% eligible for FSM 2.09 2.23 1.17 1.67 10.05 3.15
% with English as second language 20.93 11.06 13.64 12.80 2.20 9.79
% with EHCP 4.15 3.20 2.23 2.47 3.08 3.00
% with SEND support 5.72 5.19 6.94 7.63 4.07 6.33
% SEND: boys 0.24 0.06 0.12 0.66 1.23 0.06
Type of SEND provision
% SEND: special school (maintained) 3.97 1.72 1.19 2.95 2.32 2.27
% SEND: special school (not maintained) 0.18 2.45 1.93 3.55 5.80 3.79
% SEND: pupil referral unit 0.20 1.01 0.32 0.84 0.80 0.81
Type of primary need
% SEND: Social, emotional, mental health 1.94 0.89 0.76 0.60 0.38 0.62
% SEND: moderate learning difficulties 14.77 26.05 25.31 21.34 17.96 25.21
% SEND: autistic spectrum disorder 4.13 5.29 5.71 4.76 2.37 4.86
% SEND: other primary need 3.04 1.04 2.39 0.96 1.17 0.96
Total (R2) 61.35 60.20 61.73 60.24 51.44 60.85
  • Source: Authors' calculations.

Differences in the primary needs of children with SEND are the biggest explanatory factor, accounting for at least a fifth of the variation explained by each model. This contribution varies across assessments, and it is particularly important for reading and writing, where differences in the primary needs of the local SEND population account for more than a third of the inequalities in attainment levels across LAs. Within the categories of primary needs, the moderate learning difficulties category is the single most important factor, with a contribution ranging from 14% in phonics to more than 25% in reading and writing.

Differences in the socioeconomic characteristics of the overall child population also contribute to explaining the inequalities in early outcomes of children with SEND across LAs. Taken together, differences in the proportion of children eligible for FSM and the proportion of children for whom English is not the first language account for more than 12% of the variation in each of the assessments included in the analysis. This result is likely to reflect the association of these two variables with structural socioeconomic inequalities across space in England. For most assessments, the effect of these characteristics is mostly due to differences in the concentration of children for whom English is not the first language, whose contribution to the spatial differences ranges from 2% in science to nearly 21% in the phonics check.

DISCUSSION AND CONCLUSIONS

Some reflections

The analyses and results above identify large spatial inequalities in the early outcomes of children with SEND labels across LAs in England in 2019. These inequalities are much more significant than those of children with no SEND. Our evidence shows that differences in the characteristics of the child population across LAs can go some way to explaining the inequalities in outcomes, with population effects accounting for approximately 60% of the spatial inequalities. Important intrinsic population effects are ‘English as a second language’ and ‘eligibility for FSM’.

Other variables are tied to labels of SEND specifically. Notable here is the type of provision; LAs with a higher proportion of children in non-maintained schools have better outcomes for children with SEND. Of particular interest is the categorisation of primary need; LAs with a higher proportion of children with labels of moderate learning difficulties (MLD) perform particularly poorly. Given inequalities in the effectiveness of identifying primary need across LAs (Department for Education, 2021), high proportions of young people categorised within the amorphous label of MLD might be expressive of LAs which are not identifying and supporting the needs of children with SEND sufficiently well (a potential supported by the research of Hutchinson, 2021; Keslair & McNally, 2009; Riddell & Weedon, 2016).

Notably, 40% of the spatial variation cannot be attributed to characteristics of the overall population of children within the LA. So, the question is: what factors can explain the 40% spatial difference found between LAs which cannot be explained by the population characteristics of the children? Given the stark variations in outcomes across LAs, we argue that differences in the interpretation and implementation of the Code of Practice are likely to be a key factor in explaining this spatial variation. The latest evidence from Ofsted (2021) demonstrates that the UK education inspectorate is aware of significant variation in the local offer and provision for children with SEND across LAs, with the provision in many authorities being very poor quality. Just over half (51%) of all LAs were found to have significant weaknesses in their duty to provide better outcomes for young people with SEND. In this analysis, we have examined inequalities between LAs. This is a critical and often overlooked spatial scale (Hutchinson, 2021), given the pivotal role of the LA in managing the support of young people with a label of SEND.

Despite some excellent aims, our findings suggest that the new policy framework is not working to deliver high-quality support and ambitious outcomes for children with SEND. A 2018 Education Select Committee Review concluded that the reforms are being hampered by poor implementation, a lack of sufficient funding and a lack of accountability. Ultimately, they state: ‘This generation is being let down—the reforms have not done enough to join the dots, to bring people together and to create opportunities for all young people to thrive in adulthood’ (p. 4). A new SEND and Alternative Provision Improvement Plan (Department for Education, 2023a) has resulted from a broad acknowledgement of the failings of the SEND system and a wide-ranging review (Department for Education & Department of Health and Social Care, 2022). Whether this new plan can address the current poor educational outcomes of young people can be assessed via future research; however, the responses from a variety of interested stakeholders suggest it does not go far enough to address the systemic failures of SEND policy.

CONCLUSION

This paper has presented data from the National Pupil Database, which is a near-universal sample of school-aged children in England, to demonstrate that the outcomes of young people at KS1, at age 6 and 7, with SEND labels are much lower than their non-disabled peers. It is evident that experiencing SEND has a significant impact on the outcomes of children. The paper provides new empirical evidence to support a growing awareness that current educational provision is largely failing to ‘provide better outcomes’ for young people with SEND, which was a key aim of the 2015 Code of Practice (see Education Select Committee, 2018). The gap in attainment for young children with SEND compared to their counterparts without a label is stark; having a label of SEND is the most significant variable in not achieving the expected levels of attainment at KS1. Since the children in this study have been educated entirely since the new Code of Practice was enacted, it is evident that the Code of Practice has, so far, failed to achieve its aims. Even in the highest performing LAs, only 61.63% of children with SEND achieve the expected levels, and Ofsted has identified that 51% of LAs are failing to meet in full their statutory obligations under the Code of Practice. SEND intersects with other axes of power relations, and those from more affluent and educated backgrounds with SEND labels are more likely to achieve age-expected outcomes at KS1. This finding adds empirical weight to the need to urgently enhance the SEND system.

In addition, we have demonstrated significant spatial variability in outcomes at KS1 across LAs in England; there is a postcode lottery in terms of outcomes for young children with SEND, with some LAs significantly outperforming others in relation to the numbers of children with labels reaching expected levels in the phonics screening check and the national curriculum teacher assessments for reading, writing, mathematics and science conducted at KS1 of education, typically at age 6 and 7. There is a significant spatial variation at LA level. Some of this (60%) can be accounted for by population characteristics, however, 40% is independent of these population factors and suggests some other factor at work, which we argue is the variable interpretation of the SEND legislative and policy framework. We suggest that the poor and variable implementation of the SEND Code of Practice at the LA level is inbuilt to a Code of Practice which is high on aspiration and low on detailed mechanisms for how to enact these aspirations—what we have labelled a ‘Christmas wish list’. No child expects to receive everything on their Christmas wish list, yet the Code of Practice sets up as a right for children with SEND an ambitious list of resources, provisions and outcomes without any real scaffolding mechanism for how education institutions at a variety of scales, from the classroom to the LA, can achieve them. Further, the new legislative framework has been implemented by LAs during a period of extreme fiscal retrenchment; according to the Institute for Government, LAs' spending power decreased by 16% between 2010 and 2019 (Atkins & Hoddinott, 2021).

Despite this general picture of poor provision and poor outcomes for children with SEND, it is evident that some LAs are achieving much better early outcomes for children with SEND; further research is needed to interrogate how these LAs are achieving such better outcomes and the ways in which this best practice could be shared and disseminated. LAs which have a smaller proportion of children identified with MLD and LAs with larger numbers of young people with SEND have the best outcomes for young people with SEND. These findings point to some possible interventions that could improve outcomes for young people. First, having fewer children labelled with MLD suggests more effective interventions for young people with SEND to identify and address their underlying needs. Second, having more children with SEND provides more opportunities for economies of scale for specialist services, support and information to schools and children. We propose that smaller LAs with smaller populations of children (with SEND) could collaborate to provide facilities and services similar to the larger LAs. We acknowledge that further research is required to unpick the mechanisms behind the patterns we have presented in this paper.

The failure of the current policy framework, suggested in our analyses, has recently been acknowledged by government following a 3-year review (Department for Education & Department of Health and Social Care, 2022). Proposed changes to the system have been outlined in a Special Educational Needs and Disabilities (SEND) and Alternative Provision (AP) Improvement Plan: Right support, right place, right time. The Improvement Plan makes an effort to address some of the issues raised in this paper. Our analysis provides a mechanism to review the effectiveness of these changes, empirically, in future research. Evaluation and adaptation must be an ongoing real-time process for education policy, particularly when parents and other advocates (e.g., the Children's Commissioner) are highlighting systematic failures. The delay in review and revision of the policy represents a failure to achieve an appropriate level of attainment in education for cohorts of children with Special Educational Needs, which will have a negative and sometimes devastating impact on their future trajectories throughout their life course, representing a very real loss of human capital to society.

FUNDING INFORMATION

Francisco acknowledges support from Grant No. PID2019-104619RB-C41, funded by MCIN/AEI/10.13039/501100011033.

CONFLICT OF INTEREST STATEMENT

Not applicable.

ETHICS STATEMENT

The data is publicly available. The research was conducted in line with Loughborough University's Ethical Policy Framework.

Endnotes

  • 1 Department for Education (2019) provides a detailed description of the methodology and validity of these assessments.
  • 2 The phonics screening check is also administered to pupils in Year 2 who missed or did not pass the test in Year 1. The DfE data do not include a breakdown of these pupils' outcomes by SEND status and are therefore excluded from the analysis.
  • 3 The original dataset includes 151 LAs, but following the practice of the DfE, the City of London and the Isles of Scilly were excluded from the analysis as they only have one school each.
  • 4 Unfortunately, the data produced by the DfE does not contain fine-grained data at the LA level on eligibility for free school meals and the use of English as a first language by SEND status, so we cannot characterise the SEND population in terms of these attributes.
  • 5 Widely used in inequality analysis (Atkinson, 1970; Cowell, 2000), the Gini index provides a summary measure of the level of inequalities in academic outcomes across LAs for the different groups of children. The Gini index ranges between 0 (absolute equality in the distribution of outcomes) and 1 (the highest level of inequality in such a distribution).
  • 6 This might be expected, given that children with EHCPs should have a higher level of need.
  • 7 As discussed in the data section, our analysis for the phonics screening check is focused on Year 1, excluding students who missed out or did not pass the test in Year 1 and took the test in Year 2, as data on the outcomes of that group does not allow disaggregation by SEND status.
  • APPENDIX A

    The dataset used in the analysis draws on data produced by the DfE, which is part of the releases Phonics screening check and key stage 1 assessments: England 2019 (www.gov.uk/government/statistics/phonics-screening-check-and-key-stage-1-assessments-england-2019) and Special educational needs in England (https://explore-education-statistics.service.gov.uk/find-statistics/special-educational-needs-in-england/2019-20#dataDownloads-1 and www.gov.uk/government/statistics/special-educational-needs-in-england-january-2019). In what follows, we discuss these sources and their contribution to the final dataset.

    Data on the outcomes and characteristics of LAs

    Data on the outcomes of the phonics check come from the 2019_PHONICS_LA_CHAR_UD_1 and 2019_PHONICS_LA_CHAR_UD_2 files, whereas information on the KS1 assessments is from 2019_KS1_LA_CHAR_UD_1 and 2019_KS2_LA_CHAR_UD_2 files (https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/854578/Underlying_data__1_.zip). These files contain data on the number of pupils eligible for the assessments and the number of those pupils who met the national standard in each LA in England. They also contain disaggregated data on the outcomes of pupils with and without special educational needs, which allows derivation of the proportion of pupils meeting the expectations in each of these groups. The files also include information on the demographic and socioeconomic characteristics of the local pupil population in Year 1 and Year 2, including gender, the number of pupils on SEND support and Education, Health and Care Plans, the number eligible for free school meals and the number of pupils with English as their first language. Unfortunately, data on free school meals and first language is not disaggregated by SEND status in the DfE files, thus not allowing analysis of the intersection of SEND with these socioeconomic characteristics.

    Data on the characteristics of SEND pupils

    Information on SEND provision and the type of primary needs was taken from the file SEN_NCYEAR, which contains count data from the School Census on a range of characteristics of SEND pupils in different academic years at the LA level in 2019 (https://content.explore-education-statistics.service.gov.uk/api/releases/521fba4e-dfb7-4e1e-9ca7-d0c7b9880732/files). In contrast to the outcomes file discussed above that provides data for the LA of Bournemouth, Christchurch and Poole, this file provides data for the LAs of Bournemouth (Pre LGR) and Poole (Pre LGR), which together with the non-metropolitan district of Christchurch were merged in 2019 to create the LA of Bournemouth, Christchurch and Poole. The characteristics of the SEND population in that LA were derived by combining the information from Bournemouth (Pre LGR) and Poole (Pre LGR) (Figure A1).

    Details are in the caption following the image
    Kernel density estimates for the percentage of pupils meeting expected standars across local authorities in 2019, England.

    Source: Authors' own calculations.

    DATA AVAILABILITY STATEMENT

    The data that support the findings of this study are produced by the Department for Education and are available in the public domain. Below are listed the datasets used in the paper and the links from which they were retrieved: