Details of the data requested and how it is collected is available at the Coronavirus (COVID-19): attendance recording for educational settings webpage. The following education settings were asked to complete the form:
- academies (including free schools and studio schools)
- local authority maintained schools
- local authority nursery schools (no longer required to submit data from 12 October)
- independent schools
- non-maintained special schools
- alternative provision
- university technical colleges
- FE colleges and sixth form colleges
- special post-16 institutions or specialist colleges
Validation of submitted data
There are some automatic validations on the data submitted by settings, some of which prevent submission and others that advise the respondent to check their submission.
Checks are carried out to ensure that no setting is double counted - if they submit data more than once per day the latest submission is used.
Manual checks are applied to the data during data processing and comparisons are made to other reported data. Outcomes from these checks are used to develop the validation rules within the data collection.
Data sources
The education settings survey asks schools to provide the number of pupils on roll for use in attendance calculations. For non-responding settings, the total number of pupils on roll on the spring census in the 2019/20 academic year is used.
Data is linked to Get Information About Schools (GIAS) to match in further information about phase and school type.
The following school types from GIAS are included in figures for state-funded schools:
- Local authority maintained schools (Community schools, Foundation schools, Voluntary controlled schools, Voluntary aided school)
- Academies and free schools and sixth form colleges (Converter academies, Sponsor led academies, Free schools, Academy 16 to 19 sponsor led, Free schools 16 to 19, Academy 16-19 converter, Studio schools, University technical colleges)
- Special schools (Community special schools, Foundation special schools, Non-maintained special schools, Special converter academies, Special sponsor led academies, Special free schools)
- Alternative provision (Pupil referral units, Converter academy alternative provision, Sponsor led academy alternative provision, Free school alternative provision)
- City technology colleges
Figures for special post-16 institutions, further education colleges and independent schools are included separately in the ‘Open status and attendance by school or college type’ section of the publication. Figures for local authority nursery schools are not included, as these settings are included in estimates from the local authority early years survey.
We regularly update our GIAS cuts to reflect the live nature of the data. A breakdown of the cuts in relation to the reporting dates can be seen below:
GIAS cut | Dates reporting |
25 August 2020 | 1 September - 10 September |
11 September 2020 | 11 September - 23 September |
24 September 2020 | 24 September - 11 October |
12 October 2020 | 12 October - 1 November |
02 November 2020 | 2 November – to date |
Response rate
The response rate has typically been between 70-80% among state-funded schools.
Since the response rate is not 100%, we estimate the situation in settings who have not responded in order to get a full national picture (a ‘grossing methodology’) for most national-level measures. This methodology is set out below. Note that local authority level data is not adjusted for non-response.
Figures prior to 12 October relate to data as at 4pm on each day. From Monday 12 October, figures are given as of midnight for completeness. This does not impact comparability, negligible changes to the estimates were seen between 4pm and midnight.
Non-response adjustment methodology in the 2020/21 autumn term
To establish the number and proportion of state-funded schools open to pupils, data collected via the education setting survey was triangulated with information from regular calls between DfE and all local authorities. The calls aimed to identify all schools that were closed or not fully open due to COVID-19. This information was cross-referenced with data collected via the survey to inform the final estimate.
A sample of 200 state-funded schools that had never responded to the survey since 1 September was taken, as of midnight 8 September. DfE attempted to contact schools by phone to find out if they were fully open. Schools were also encouraged to complete the survey in future.
- Of the 200 schools, contact was made with 166. For the remaining schools, searches were carried out for information that indicated they were not fully open - including media reports, school websites and school social media pages.
- All 200 schools were established to be open to pupils. Of these, 18 (9%) were considered not fully open, which is a similar proportion to responding schools. Schools were not fully open for both COVID and non-COVID related reasons.
- 104 of the 200 schools responded the following day (Thursday 10 September). Their reported absence was slightly higher than responding schools (16%). However, secondary schools were over-represented in this group of schools and this higher rate was accounted for by two large secondary schools who were staggering entry of year groups until week commencing 14 September.
As a result of this exercise, it is assumed that non-responding schools are no more likely to be closed, partially open or have lower attendance than responding schools. To calculate an aggregate figure for ‘state-funded schools’, responses are weighted by school type (primary, special schools and alternative provision) to account for differences in response rates. Assumption checking was carried out. If attendance in non-responding schools was 5% below that in responding schools, the national attendance estimate for 10 September would be 87%, 1 percentage point lower than reported.
To calculate grossed estimates for each setting type, the attendance rate is calculated in responding settings using the submitted number on roll and then applied to the number on roll in non-responding settings using the 2019/20 spring census. Attendance estimates for children with EHCPs and with a social worker are calculated by assuming that the same proportion of pupils on roll in non-responding settings have an EHCP or a social worker as responding settings. The attendance rate in responding settings for these groups of pupils are then applied to this estimated number on roll.
Non-response adjustment in the 2019/20 spring and summer terms
Since the response rate was not 100%, attendance figures were adjusted to account for schools and colleges that did not respond to the survey to get a full national picture (a ‘grossing methodology’). This methodology was adjusted when more information about non-responding settings became available or the survey changed. The methodology was revised four times between 23 March and the end of the summer term:
- after the first full week of reporting
- from 1 June to reflect wider opening of primary schools
- from 15 June to reflect wider opening of secondary schools
- from 22 June to reflect the increase in open primary and secondary schools
The methodology used between 23 and 27 March 2020 assumed that schools and colleges that did not respond were as likely to be open as those that did.
Following review of the first week’s response patterns, analysts developed a new methodology that no longer made this assumption. Therefore, non-response is accounted for differently from 30 March onwards.
From 1 June, and again from 15 June and 22 June, the methodology was updated to account for wider opening of some schools. Further detail is set out below.
Non-response methodology 23 March to 27 March
On each day we calculated the proportion of responding settings which were open and closed, and the number of children and staff in attendance. These same proportions were then applied to non-responding settings.
Non-response methodology 29 March to 29 May
Following a full week of returns, we analysed return patterns and developed a new methodology as follows:
- For settings that responded that day, it was recorded whether they were open or closed.
- For settings that did not respond that day, but had previously:
- If they reported that they planned to close on the next working day, they were assumed to be closed.
- If their most recent return was the previous day, and they reported that they planned to open, it was assumed that they remained open.
- If their most recent return was prior to the previous day and they reported that they planned to open, it was assumed that 50% remained open.
- For settings that had never responded, it was assumed 50% remained open.
- Attendance rates were calculated for each individual setting type and used 2019/20 census returns to calculate proportions of all pupils on roll. This is likely to over-estimate attendance in cases where settings were caring for pupils from other settings. This practice may have increased during the Easter break. Analysis of returns suggested the likely effect of this was less than 0.1%.
- When calculating attendance rates for non-responding settings that are assumed open, similar rates were assumed as responding open settings.
The assumption of 50% having remained open was arrived at by choosing a random sample of 50 settings and determining that approximately half had closed via school websites and local authority announcements. Adjusting this assumption by ±10% changed the estimated open settings by around 1%, whilst it has a negligible effect on the low attendance rate. Additional assumption checking was carried out to better understand the impact of varying the 50% assumption for each setting type. Applying a 10% change to each setting type changed the estimated open rates by between 0 and 3 percentage points, with independent schools most impacted. The impact on estimated attendance rates by setting type was negligible.
The main difference this methodology change made was to better understand the proportion of settings that were open. We also improved the denominator used for further education colleges, this was the driver of the majority of the decrease in attendance rate. More granular information, such as attendance rates of pupils and teachers saw only minor changes (see table below).
Comparison of figures on Friday 27 March using the previous and revised methodologies
Previous methodology | Revised methodology | |
---|---|---|
Number of open establishments | 23,700 | 20,000 |
Proportion of open establishments | 96% | 81% |
Number of pupils attending | 140,000 | 124,000 |
Proportion of pupils attending | 1.6% | 1.3% |
Teaching staff | 115,000 | 101,000 |
Support staff | 92,000 | 80,000 |
Non-response methodology 1 June onwards
From 1 June, settings could be open to children of critical workers, vulnerable children and those from wider years (nursery, reception, year 1 and/or year 6) or open only to children of critical workers and vulnerable children. From 15 June, settings could be open to children of critical workers, vulnerable children and those from wider years (nursery, reception, year 1, year 6, year 10 and/or year 12) or open only to children of critical workers and vulnerable children.
A new methodology was needed to account for this among the non-responders, and to distinguish between settings not open to wider years who normally have pupils from those years and those who do not, such as secondary schools and further education colleges.
Following a full week of returns, we analysed return patterns and developed a new methodology as follows:
- For settings that responded that day, we recorded i) whether they were open or closed to vulnerable children and critical workers, and ii) whether they were open or closed to children from wider years, or where this was not applicable.
- For settings that did not respond that day, but had since 1 June:
- If they reported that they planned to be closed on the next working day, they were assumed to be closed. Settings were asked this question separately for i) children of critical workers and vulnerable children and ii) children from wider years. Therefore settings were categorised separately accordingly.
- If their most recent return was the previous day and they reported that they planned to open, it was assumed they had remained open.
- If the setting is normally open to wider years, their most recent return was prior to the previous day and they reported that they planned to open for one or both groups – 40% were assumed to be open to children of critical workers and vulnerable children only and 30% were assumed to be open to these children and those from wider years. From 15 June, these proportions continued to be used for settings with children in nursery, reception, year 1 or year 6. For settings with students in years 10 and 12 - 20% were assumed to be open to children of critical workers and vulnerable children only and 30% were assumed to be open to these children and those from wider years.
- If the setting was not normally open to wider years, their most recent return was prior to the previous day and they reported that they planned to open then 50% were assumed to be open.
- For settings normally open to wider years that had never responded:
- From 1 June, it was assumed 40% were open to children of critical workers and vulnerable children only and 30% were assumed to be open to these children and those from wider years.
- From 15 June, these proportions continue to be used for settings with children in nursery, reception, year 1 or year 6. For settings with students in years 10 or 12 - 20% are assumed to be open to children of critical workers and vulnerable children only and 30% are assumed to be open to these children and those from wider years.
- From 22 June, these proportions were adjusted to reflect the increase in open primary and secondary schools. For settings with students in nursery, reception, year 1 or year 6 – 80% were assumed to be open to children of critical workers and vulnerable children only and 70% are assumed to be open to these children and those from wider years. For settings with students in years 10 or 12 - 60% were assumed to be open to children of critical workers and vulnerable children only and 50% are assumed to be open to these children and those from wider years.
- For settings not normally open to wider years that have never responded, it is assumed 50% have remained open.
- Attendance rates are calculated for each individual setting type and use 2019/20 census returns to calculate proportions of all pupils on roll. This is likely to over-estimate attendance in cases where settings are caring for pupils from other settings. This practice may have increased during the Easter break. Analysis of returns suggested the likely effect of this was less than 0.1%.
- When calculating attendance rates for non-responding settings that are assumed open, similar rates are assumed as responding open settings.
To arrive at these assumptions, a random sample of 125 non-responding settings was chosen, stratified by school type and region. Information about the open status of 51 of these schools was captured by contacting them by phone on 4 June. Information about a further 50 settings was obtained via school websites. From 15 June, proportions for settings with students in year 10 or 12 were based on information about the open status of 60 settings obtained via school websites. From 22 June, adjusted proportions were based on information about the open status of 100 settings obtained via school websites.
Additional assumption checking was carried out to better understand the impact of varying the assumptions on each setting type. Adjusting all assumptions by ±10% changes the estimated open settings by around 3%, whilst it has a negligible effect on the attendance rate (<0.5%).
Changes to education settings survey from 12 October
Between 1 September and 9 October schools and colleges were asked to to report as ‘fully open’ or ‘not fully open’ to identify those affected by COVID-19. Schools were asked to report as ‘not fully open’ if:
- a group of pupils without COVID-19 symptoms themselves or in their households had been asked to self-isolate
- they were unable to provide face-to-face teaching for all pupils for the whole of the normal school day
- they were not open to all year groups
- they were not open to pupils because of a training or INSET day
To reduce burden of data collection on settings, only those that self-identified as ‘not fully open’ due to ‘suspected or confirmed cases of COVID-19’ were asked to report how many pupils had a confirmed case, suspected case or had been requested to remain at home due to potential contact with a case.
From 12 October, all responding schools and colleges are asked to report pupils in the following categories:
- Suspected case of coronavirus (COVID-19)
- Confirmed case of coronavirus (COVID-19)
- Requested to remain at home due to potential contact with a case of coronavirus (COVID-19) inside your educational setting
- Those remaining at home due to potential contact with a case of coronavirus (COVID-19) from outside your educational setting including self-isolation
We now report the number and proportion of schools that have requested 1 or more pupils to remain at home due to potential contact with a case of coronavirus (COVID-19) inside their educational setting. This is not comparable to previous data on schools ‘not fully open’, which intended to capture schools with 'a group' of such pupils.
The data published today is not comparable to previous weeks because the question we are asking has fundamentally changed. Previously a subset of schools and colleges were asked about groups of pupils isolating. Now all schools are asked about whether one individual pupil or more is isolating.
The changes we have made mean that the online portal now asks all responding schools and colleges that are open the same simpler and more reliable set of questions on the number of confirmed and suspected Covid cases among pupils and on pupils isolating due to contact both in the setting and outside it. The data is reported as having ‘at least one pupil’ self-isolating due to potential contact inside the setting, with no scope for interpretation.
Comparability of attendance estimates and typical absence rates
The overall absence rate in state-funded primary, secondary and special schools in England in the 2018/19 academic year was 4.7%. There are some differences in the calculation of this figure and our attendance estimates that affect comparability. These differences all depress our attendance estimates when compared to the 2018/19 absence rate.
2018/19 overall absence rates include pupils aged 5-15 in state-funded primary, secondary and special schools.
The following groups are included in our attendance estimates from 9 September, but excluded from the 2018/19 absence rate:
- Children in school-based nurseries, for whom attendance is not mandatory and many attend part-time (excluded from 12 October)
- Children in state-funded alternative provision, for whom the absence rate is typically much higher than average
- 4 year-olds on roll in state-funded schools, for whom the absence rate is typically slightly higher than average
- 16 to 18 year-olds on roll in state-funded schools, for whom the absence rate is likely to be slightly higher than average
We compared attendance in primary schools with and without nurseries, which suggested excluding children in school-based nurseries could increase our attendance estimate for state-funded schools by up to 1.5 percentage points. From 12 October, schools were asked to exclude nursery children from their submission. A corresponding decrease was observed in submitted number on roll in schools with nurseries. We estimate the other remaining differences depress our estimates by less than 0.5 percentage points.
COVID-19 related pupil absence
From 12 October, the education settings survey changed and therefore some metrics that used to be presented have been discontinued. The changes were made to give more comprehensive data about how many pupils are self-isolating for different reasons. There are no comparable figures to these from previous weeks.
This data is as reported directly by schools via DfE's daily education settings survey. It is not the primary source of data on infection, incidence and COVID-19 cases overall. Further detail can be found within the coronavirus in the UK dashboard, national COVID-19 surveillance reports and coronavirus infection survey pilot statistics.
Where COVID-19 related pupil absence categories are combined - for example to report the total proportion of pupils absent due to COVID-19 or the proportion self-isolating that are not a confirmed/suspected case - the proportion is reported as a range to account for possible double counting. Numbers of pupils are presented as ‘up to’ the upper bound of the range. Settings are asked to not count pupils in multiple categories, however analysis of responses found evidence of double counting.
The range is calculated as such:
- Minimum assumes maximum double counting by taking the largest figure from across each of the categories
- Maximum assumes no double counting by taking the sum of all categories
Some examples are provided below.
Confirmed case of COVID-19 | Suspected case of COVID-19 | Self-isolating - potential contact inside | Self-isolating - potential contact outside | Minimum total COVID-19 related absence | Maximum total COVID-19 related absence | |
School 1 | 1 | 3 | 26 | 0 | 26 | 30 |
School 2 | 0 | 2 | 0 | 5 | 5 | 7 |
School 3 | 0 | 0 | 0 | 4 | 4 | 4 |
Attendance of vulnerable children during the COVID-19 outbreak
Attendance of vulnerable children and young people was prioritised between March 2020 and the end of the summer term, when attendance was limited for other pupils to reduce transmission of coronavirus (COVID-19).
Before 1 June, settings were asked the number of vulnerable children attending. Settings were also asked to provide separate figures for the number of pupils with an EHCP and those with a social worker attending.
Some schools provided a count of critical workers with either EHCPs or a social worker but did not include these in their respective counts for vulnerable children. Therefore, the estimate of the number of vulnerable children may be an undercount. The data suggests the effect of this is c5%.
The proportion of vulnerable children in attendance was based on an estimate of the total number of pupils who either have an EHCP or are classed as Children in Need or have an EHCP, derived from the Children in Need and School Censuses. Settings would have included children classified as ‘otherwise vulnerable’ in their count of vulnerable children. As these children were not represented in the denominator, the proportion of vulnerable children would have been an undercount.
From 1 June, a total number of vulnerable children attending is no longer collected. Settings are asked to provide a count of the number of children of critical workers, with an EHCP, with a social worker and otherwise vulnerable attending.
From 16 June, the proportion of children in attendance with an EHCP or a social worker was reported alongside the number of children classified as ‘otherwise vulnerable’ in attendance. This calculation was backdated to 23 March and published in the accompanying data tables.
While all children were expected to attend from the start of the 2020/2021 autumn term, provision for vulnerable children and their attendance continues to be monitored. Schools are asked to report the number of pupils with an Education, Health and Care Plan (EHCP) and the number with a social worker on roll and in attendance each day.
Data for children of critical workers
In the 2019/2020 spring and summer term, the attendance of children of critical workers was prioritised. Schools were asked to provide a count of the number of children of critical workers attending.
The proportion of children of critical workers in attendance is based on an estimate of the number of pupils who have a critical worker parent derived from the Labour Force Survey.
The number of children of critical workers in attendance may have been under-reported from 1 June. This is because some settings may not be counting all children of critical workers who were attending in nursery, reception or years 1, 6, 10 and 12.
Autumn 2020/21 half term
Data between 19 October to 23 October, and data for 2 November are included in a separate underlying data file called :
“Table 1A- Daily attendance in education settings during the COVID-19 outbreak (data for half term dates only - 19th October - 23rd October and 2 November).”
This file can be downloaded from the associated files section of EES. The data for these dates are in a separate file as some local authorities are on half term between these dates, so data is not comparable with previous dates.
Adjustments to methodology for figures between 19 - 23 October and 2 November.
Between 19 and 23 October, schools in the following local authorities were expected to be on half term for some or all of that week: Leicester, Leicestershire, Lincolnshire, North Lincolnshire, Nottingham, Nottinghamshire, Rutland, Warrington, West Berkshire, Swindon, Wiltshire and Norfolk. On 2 November, Bedford and Isles of Scilly were expected to still be on half term.
Some schools in other local authorities were also on half term or had inset days - academies are not required to follow local authority term dates and schools can set their own inset days.
In previous weeks, our estimates are adjusted for non-response to provide estimates for 100% of schools. To provide the clearest picture of attendance in schools that would normally be open, figures between 19 and 23 October, and also 2 November are adjusted for non-response to provide estimates for all schools that are not on half term. This means our estimates are not directly comparable to previous weeks, which are nationally representative.
Table 1 sets out the proportion of schools included in our estimates each day. Term dates are not held for individual schools, so this is estimated based on local authority term dates.
Table 1: Proportion of schools not on half term on each day included in estimates. Proportions are given to the nearest 1%, apart from when the proportion is >99%,
Date | Proportion of schools included in estimates |
19 October | 95% |
20 October | 95% |
21 October | 95% |
22 October | 92% |
23 October | 90% |
2 November | 99.7% |
Grossing methodology for estimates 19-23 October and 2 November
To reduce burden of data collection, schools were not required to respond to the survey during half term. Schools that did not respond in local authorities that were expected to be on half term each day are not included in our grossed estimates for 19 to 23 October and 2 November.
Some schools on half term did respond and reported as ‘Closed for non-COVID related reasons’. All schools in this category have been excluded from our estimates. This means a small number of schools (typically less than 10) that were closed for non-COVID related reasons, but were not on half term, will also have been excluded.
All schools that reported inset days are also excluded from these estimates and considered on half term. This is a change from our usual reporting – schools on inset days are normally included in our open status estimates and counted as ‘Open’ and included in our attendance estimates and counted as having no pupils attending. Typically, a very small number of schools (less than 10) report inset days between Mondays and Thursdays.
Ensuring estimates are comparable on Thursday 22 October to previous weeks
While figures between 19 to 23 October and 2 November are not directly comparable to previous weeks, we have analysed returns and report whether changes in headline measures on Thursday 22 October compared with Thursday 15 October broadly reflect a national-level change or are due to schools on half term in certain areas being excluded.
We looked at attendance in state-funded schools that responded on both 15 and 22 October and did not report an inset day or closure for non-COVID-19 related reasons on either day. This analysis captures 64% of all state-funded schools. The table below shows that grossed estimates (excluding inset days) of attendance, absence and open status rates for 15 October are very similar to the rates for this group of responding schools on the same day. Similar changes between 15/10 and 22/10 are seen both nationally and among this group of schools, suggesting that changes in these rates broadly reflect a national-level increase unrelated to half term. The increase in COVID-related absence (approximately 2 percentage points) accounts for approximately two-thirds of the overall increase in absence between 15 and 22 October (approximately 3 percentage points).
15/10 - grossed all schools excluding inset days | 22/10 - grossed all schools without inset days or half term | 15/10 - responded both days excluding inset days and non-COVID closures | 22/10 - responded both days excluding inset days and non-COVID closures | |
Attendance in state-funded schools (%) | 89.4 | 86.2 | 89.5 | 86.5 |
Attendance in state-funded primaries (%) | 92.0 | 89.6 | 92.0 | 89.7 |
Attendance in state-funded secondaries (%) | 86.5 | 82.4 | 86.7 | 82.7 |
Confirmed COVID cases (%) | 0.1 | 0.1 | 0.1 | 0.1 |
Suspected COVID cases (%) | 0.5 | 0.4 | 0.5 | 0.4 |
Self-isolating (%) | 4.3 | 6 | 4.3 | 5.9 |
All COVID absence (%) | 4.9 | 6.5 | 4.9 | 6.4 |
Open state-funded schools (%) | 99.7 | 99.3 | 99.7 | 99.3 |
Closed state-funded schools (%) | 0.3 | 0.7 | 0.3 | 0.7 |
Schools sent 1 or more pupils to self-isolate (%) | 21 | 26 | 21 | 25 |
Monday 26 October to Friday 30 October
Between 26 October to 30 October 137 local authorities were expected to be on half term. To reduce burden of data collection, schools were not required to respond to the survey during half term. Data for this period is therefore not included in the underlying data files since the majority of schools were on half term.
Ensuring estimates are comparable on Thursday 5 November to previous weeks
To ensure estimates on Thursday 5 November are comparable to previous weeks, figures reported are compared to Thursday 15 October.
Data sharing
Data from the Education Settings Survey are shared across national and local government on a restricted basis for operational purposes. This enables government to monitor the impact of coronavirus (COVID-19) and effectively target support. Occasionally unpublished data shared in this way has entered the public domain.