Methodology

Scope 

Throughout our analysis we use the National Pupil Database (NPD)1 which captures pupils and students in state education in England.  

For our analysis of pupils up to age 16, we include all state-funded schools except for those whose sole, or main, registration was in alternative provision, a pupil referral unit or a hospital school. Independent schools are not included.  

For 16-19 education, we include all students and qualifications entered in this phase in all state-funded 16-19 provision. We exclude students on an apprenticeship as flagged in the key stage 5 National Pupil Database.  

In 2021 the Department for Education made changes to which students were included in that year’s results for the 16-19 phase. The impact of this is that data from 2020 and 2021 are not directly comparable to 2023 or earlier years. We have therefore not reported on 2020 and 2021 16-19 data in this report. Further detail about the impact of this change can be found in the 16-19 methodology chapter of our Annual Report in 2023. 

Attainment 

Here we set out which attainment measures we use for each phase, as well as a little detail on the changes in overall attainment since 2019. For all phases, we focus on comparisons with 2019 as the most recent year when assessments and exams were unaffected by the pandemic. 

Early years foundation stage 

To measure children’s development at age 5, we use the statutory teacher-led assessments against the early years foundation stage (EYFS) profile. Specifically, we use a pupil’s total point score across the twelve (out of seventeen) early learning goals which correspond to the Department for Education’s ‘good level of development’ measure.2 Although the EYFS was reformed in 2021 affecting both the goals themselves and how these are assessed, we have adjusted our time series to allow us to compare post-2021 attainment data to earlier years. Our time series begins in 2013, as the EYFS scoring system changed in this year, significantly affecting the score distributions in earlier years. 

For each goal, children are assessed as either meeting the expected level of development at the end of reception year (score = 2) or not yet reaching this level (score = 1). The total points score aggregates scores across the twelve goals, giving a maximum possible score of 24 for pupils meeting the expected level in each. 

Primary school 

At the end of primary school pupil attainment is measured by statutory key stage 2 assessments. Specifically, we base our attainment measure on pupils’ average scaled score in reading and maths to provide the most consistency since the start of our series in 2011. The spelling, punctuation and grammar assessment was only introduced in 2013, whilst the writing assessment has been teacher-assessed since 2012. Where pupils are missing either result for reading or maths, the average takes the value of the subject they do have a score for. Scaled scores for these domains are derived from national test results and can take values between 80 and 120. We also include teacher-assessed attainment scores for pupils who do not reach the lowest measurable score in the test, whose scores range from 59 to 79, to enable them to be included in the point distribution. 

2016 was the first year when pupils were assessed against a new national curriculum, in tests that were designed to be more challenging, and with a new scoring system. Whilst these reforms make direct comparisons of attainment difficult over time, our focus is on relative differences in attainment between specific groups of pupils. As with EYFS assessments, there were no key stage 2 assessments during the pandemic years of 2020 and 2021.  

Secondary school 

To assess overall attainment at the end of secondary school – known as key stage 4 – we measure pupils’ average GCSE grade across English and maths. This provides a measure that, while relatively narrow, is not affected by changes in GCSE subject entry patterns over time so gives maximum stability in our time series which starts in 2011. Achieving good grades in these core subjects is often a prerequisite for progressing to further study. Pupils failing, or not entering, a relevant English or maths qualification by the end of key stage 4 receive a score of zero for that component, to reflect the expectation that all pupils should study these core subjects. 

In 2017, GCSE English and maths were reformed, with a new grading scale from 9 (the highest grade) to 1 (the lowest grade) which replaced the old A* to G grading scale. To account for changes in the grading structure for pupils sitting GCSEs prior to 2017, we adjust average GCSE scores in the earlier years by mapping across the old score boundaries to the new, to produce a new adjusted point score for these pupils. Whilst the shift from unreformed to reformed GCSEs makes (absolute) comparisons of attainment difficult over time, our focus is on attainment gaps between specific groups of pupils.  

During the pandemic, in 2020 and in 2021, exams were cancelled and replaced with centre assessed grades (CAGs) and teacher assessed grades (TAGs) respectively. Grades were more generous during these years. 

Pre-16 attainment gaps 

For education phases up to age 16, we calculate attainment gaps using a ‘months of learning’ measure based on the following steps: 

1. We rank all pupils by their attainment score, as per the measures described above. 
   
2. We identify the relevant group of interest (e.g. disadvantaged pupils) and calculate the mean attainment rank of pupils in this group. 
3. We subtract the mean rank of the group of interest from that of the reference group (e.g. non-disadvantaged pupils). 
4. We convert this mean rank difference to a months of learning measure, by applying a scalar of 43 for the Early Years Foundation Stage, 64 for key stage 2, and 99 for key stage 4. 

16-19 Education 

At the end of 16-19 study, we measure attainment as a total point score over students’ best three qualifications (based on size rather than number of qualifications). The 16-19 disadvantage gap is then calculated as the mean average of the total point score for disadvantaged students, subtracted from the mean for non-disadvantaged students, expressed in equivalent A level grades. 

For our total point score measure, we map all level 1-3 qualifications onto the same scale. Full details of our approach can be found in the 16-19 methodology chapter of our 2023 Annual Report and Annex C of our initial ‘measuring the 16-19 disadvantage gap’ report3. 

When looking at gaps within specific qualification types such as A levels, we use an average point score per qualification measure, and only students that have entered one or more relevant level 3 qualification are included. 

Our 16-19 non-participation measure includes all students at the end of key stage 4 in each year. We then measure the proportion of these students that were not working towards a substantial qualification, apprenticeship or internship as of the 31st of October (i.e. the beginning of year 12), as recorded in the ILR, or Post-16 Learning Aims data. We define a substantial qualification as those equivalent in size to half an A level if at level 3, of equivalent in size to a GCSE if at level 2 or below. 

Identifying disadvantaged children and young people 

To identify disadvantaged pupils and young people we use eligibility for Free School Meals (FSM) as the best available proxy for measuring trends in attainment over time. However, it is important to distinguish FSM eligibility from poverty. FSM eligibility is an administrative measure based on entitlement to certain benefits, rather than a direct measure of household income, living standards or material hardship. As a result, it does not capture all pupils living in poverty: some may not meet the eligibility criteria, while others may be eligible but not registered. Conversely, some pupils recorded as FSM-eligible may no longer be in the same household circumstances. We therefore treat FSM-based measures as proxies for economic disadvantage, rather than direct measures of poverty. 

Changes to how we identify disadvantaged children and young people 

In this year’s Annual Report, we have changed how we use FSM eligibility to define economic disadvantage, moving to an ‘EverFSM’ measure. Specifically, a pupil or student is classified as disadvantaged if they have been recorded as eligible for free school meals (FSM) at any point during their time in state-funded schooling. This replaces our previous headline definition of disadvantage based on FSM eligibility at any point in the previous six years (FSM6). While this change increases the number of pupils classified as disadvantaged, it has relatively little effect on the measured attainment gap. Instead, it provides a more stable basis for tracking trends over time. 

The new definitions for disadvantage for each phase are as follows:

Definition by phase

Phase	Previous definition (FSM6)	New definition (EverFSM)
Age 5 (EYFS)	FSM eligible in reception year	FSM eligible in reception year (unchanged)
Age 11 (Key stage 2)	FSM eligible at any point between year 1 and year 6 (6 years)	FSM eligible at any point between reception and year 6 (7 years)
Age 16 (Key stage 4)	FSM eligible at any point between year 6 and year 11 (6 years)	FSM eligible at any point between reception and year 11 (12 years)
16–19 phase	FSM6 derived from the Key Stage 4 record, capturing FSM eligible students at any point between year 6 and year 11 (6 years)	EverFSM derived from the Key Stage 4 record, capturing FSM eligible students at any point between reception and year 11 (12 years)

Why are we changing our disadvantage measures? 

We have made this change because the rollout of Universal Credit, and the transitional protections introduced alongside it, have made FSM6 increasingly difficult to interpret as a consistent measure of economic disadvantage over time. FSM6 was designed to capture pupils who had experienced FSM eligibility relatively recently. However, between 2018 and 2026, some pupils have remained recorded as FSM-eligible because of transitional protections, even where household circumstances may subsequently have improved. As these protected households accumulate, year-on-year changes in FSM6 increasingly reflect historic rather than current family circumstances, as well as changes in underlying economic disadvantage.

It is important to distinguish FSM eligibility from poverty. FSM eligibility is an administrative measure based on entitlement to certain benefits, rather than a direct measure of household income, living standards or material hardship. As a result, it does not capture all pupils living in poverty: some may not meet the eligibility criteria, while others may be eligible but not registered. Conversely, some pupils recorded as FSM-eligible may no longer be in the same household circumstances, particularly during the period covered by Universal Credit transitional protections. We therefore treat FSM-based measures as proxies for economic disadvantage, rather than direct measures of poverty. EverFSM does not measure current low income either.

However, it provides a consistent historical record of whether a pupil or student has ever been recorded as FSM-eligible during their time in state-funded schooling. Because it is based on whether eligibility has ever been recorded, rather than whether it falls within a recent six-year window, it is less distorted by Universal Credit transitional protections. This makes EverFSM a more stable basis for tracking attainment gaps over time, and the most suitable measure for EPI’s Annual Report against the backdrop of the government’s target to halve the disadvantage gap. KS4 is the phase most directly affected by Universal Credit protections. However, it also has an impact in trends at KS2 and in the 16–19 phase. The measure for Early Years is unaffected as it only considers eligibility within reception year, and is unaffected by UC protections.

Secondary School  

Evidence from our sensitivity analysis suggests that, although extending the FSM window changes who is classified as disadvantaged, it makes relatively little difference to the size of the attainment gap itself. For example, at KS4 in 2023/24, an EverFSM definition covers 33.4% of pupils compared with 27.4% under FSM6, but the estimated national gap remains very similar – 19.0 months under EverFSM compared with 19.1 months under FSM6.

M1: KS4 national gaps under FSM6 and EverFSM: 2025 cohort

 

The chart below explains why moving from FSM6 to EverFSM has only a small effect on the measured gap. The wider EverFSM definition adds pupils who were eligible for FSM more than six years ago, but not more recently. We refer to this group as FSM>6, shown in orange on the chart. These pupils have lower average attainment than pupils who were never recorded as FSM eligible, but higher average attainment than pupils identified as disadvantaged under FSM6. As shown by the dotted arrows in the chart , moving the FSM>6 group from the non-disadvantaged group into the disadvantaged group  changes both group averages, but only slightly. The disadvantaged average rises because it now includes some pupils with relatively stronger outcomes. The non-disadvantaged average also rises because some lower-attaining pupils have moved out of that group. Overall attainment is unchanged, and the gap between the two groups remains broadly the same. 

M2: KS4 national mean attainment under FSM6 and EverFSM: 2025 cohort

 

The similarity between EverFSM and FSM6 can also be seen in the trend analysis below. Both series follow a very similar pattern, with only slight divergence in more recent years, including before the effects of Universal Credit transitional protections begin to affect the measure from 2024. This suggests that the deterioration in attainment since the pandemic was somewhat less pronounced among pupils with low family incomes during primary school only, as opposed more recently. However, the divergence is small, and in both cases the gap increased by between 5 and 6 per cent between 2018 and 2024. 

M3: KS4 Disadvantage Gap by FSM Measure

 

The chart below shows how the proportion of the KS4 (year 11) cohort identified as disadvantaged differs under FSM6 and EverFSM over time. EverFSM identifies a consistently larger share of pupils as disadvantaged than FSM6 throughout the series. The gap between the two measures widened up to around 2022 (to 10 percentage points), largely because the FSM6 share fell while the EverFSM share remained comparatively stable. Since the impact of UC protections in 2024 the coverage gap has begun to narrow rapidly, and in 2025 was down to 6 percentage points. 

M4: KS4 Population Share by FSM Measure 

 

 

16-19 Education 

The chart below explains why moving from FSM6 to EverFSM has a larger effect on the measured gap in the 16-19 phase than at KS4. As at KS4, the wider EverFSM definition adds students who were eligible for FSM more than six years ago, but not more recently. In the 16-19 phase, however, this FSM>6 group has attainment that is much closer to the FSM6 group than to the non-disadvantaged group. As a result, moving the FSM>6 group from the non-disadvantaged group into the disadvantaged group changes the two group averages unevenly, as shown by the dotted arrows. The disadvantaged average rises only slightly, because the additional students have outcomes closer to those already in that group. The non-disadvantaged average rises by more, because some lower-attaining students have moved out of that group. Overall attainment is unchanged, but the gap between the two groups is larger under EverFSM. 

M5: 16-19 national mean attainment under FSM6 and EverFSM: 2025 cohort 

 

 

The similarity in disadvantage gap trends between EverFSM and FSM6 is evident in the 16-19 phase, as shown in the chart below. Although the gap is around 0.15 grades larger under EverFSM, both series follow a very similar pattern, with a reduction of just 0.1 grades between 2018 and 2025. This reflects the fact that FSM6 in the 16-19 phase is based on pre-16 eligibility, meaning the effects of Universal Credit protections are delayed and remain minimal by 2025. 

M6: 16-19 Disadvantage Gap by FSM Measure

 

 

 

The chart below shows how the proportion of the 16-19 cohort identified as disadvantaged differs under FSM6 and EverFSM over time. EverFSM identifies a consistently larger share of students as disadvantaged than FSM6 throughout the series. As at KS4, the EverFSM trend is relatively stable across the 2018 to 2025 period, remaining at around 30 per cent of the cohort. By contrast, the proportion identified under FSM6 fell from 24 per cent in 2018 to 21 per cent in 2025. Correspondingly, the gap between the two series widened from 6 percentage points in 2018 to 9 percentage points in 2025. 

In the 16-19 phase, however, the composition of the disadvantaged group is shaped not only by pre-16 entitlement rules, registration take-up and administrative practice, but also by patterns of participation in post-16 education. As discussed elsewhere in this section, disadvantaged students are less likely to participate post-16, and this difference appears to have grown in recent years. This may help explain why the downward trend in FSM6 coverage is more pronounced for the 16-19 cohort than at KS4. 

M7: 16-19 Population Share by FSM Measure 

 

 

 

Primary School  

At KS2, moving from FSM6 to EverFSM has even less effect than at later phases. This is because the difference between the two measures is smaller: EverFSM extends the eligibility window by one year only, from years 1 to 6 under FSM6 to reception to year 6 under EverFSM. By 2025, the two measures are effectively identical. The 2025 cohort was in reception in 2018/19, after Universal Credit transitional protections had been introduced, meaning pupils recorded as FSM-eligible in reception would generally have retained that status into later primary years even if their household circumstances subsequently changed. As a result, almost all pupils captured by the additional reception year are already included within FSM6, leaving very little scope for the move to EverFSM to affect the measured KS2 gap. 

The chart below further explains why moving from FSM6 to EverFSM has only a very small effect on the measured KS2 gap. The additional group captured by EverFSM but not FSM6 is very small, at around 400 pupils. This group consists of pupils who would, in the absence of Universal Credit protections, have been recorded as FSM-eligible only in reception year. Their average attainment sits roughly midway between the FSM6 group and the non-disadvantaged group. Moving this small FSM>6 group from the non-disadvantaged group into the disadvantaged group therefore changes both group averages only slightly. Overall attainment is unchanged, and the measured gap remains virtually identical. 

M8: KS2 national mean attainment under FSM6 and EverFSM: 2025 cohort 

 

This similarity can also be seen in the chart below. Prior to 2025, although the gap is typically slightly smaller under EverFSM, both series follow a very similar pattern over time. By 2025, however, the FSM6 and EverFSM gaps are effectively identical. 

M9: KS2 Disadvantage Gap by FSM Measure

The chart below shows how the proportion of the KS2 cohort identified as disadvantaged differs under FSM6 and EverFSM over time. FSM7 identifies a slightly larger share of pupils as disadvantaged than FSM6 in earlier years, reflecting its longer eligibility window. However, by 2025, FSM6 and EverFSM identify the same pupils, and therefore the same share of the cohort as disadvantaged.  

M10: KS2 Population Share by FSM Measure 

 

Which other measures are affected? 

In addition to our headline gaps, a number of our other measures have moved to use EverFSM rather than FSM6. This includes: 

• Geographic breakdowns, including Regional, Local Authority, and Parliamentary Constituencies 

• Disadvantage gaps by ethnicity 

In both cases the change will have affected the results compared to previous years. In areas where there are a large proportion of young people who were eligible for FSM more than 6 years ago, but not since, or where the attainment of this group is atypical, there may be more notable changes. However, we consider the EverFSM measure to now be more robust than the FSM6 measure for measuring changes over time.  

We have discontinued our persistent disadvantage measure because Universal Credit transitional protections have substantially changed the composition of pupils recorded as eligible for FSM for most of their school lives. As a result, changes in the size or attainment of this group increasingly reflect the effects of entitlement rules and administrative protections, rather than changes in the prevalence or consequences of long-term economic disadvantage. 

Future reporting 

Overall, the move from FSM6 to EverFSM changes who is counted as disadvantaged, but not the substantive trends in attainment gaps. EverFSM’s main benefit is that it provides a clearer and more stable basis for reporting. As FSM entitlement rules change again from 2026/27, we will continue to review how best to measure economic disadvantage in future reports. 

Data access 

Our analysis was carried out in the Secure Research Service, part of the Office for National Statistics (ONS). It contains statistical data from ONS which is Crown Copyright. The use of the ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. This work uses research datasets which may not exactly reproduce National Statistics aggregates.  

Geographic classifications were derived from Office for National Statistics geography products, licensed under the Open Government Licence v3.0. This contains OS data © Crown copyright and database right 2026.

The Department for Education is responsible for the collation and management of the NPD and is the Data Controller of NPD data. Any inferences or conclusions derived from the NPD in this publication are the responsibility of the Education Policy Institute and not the Department for Education. 

 

Explore the report