Methodology

National Tutoring Programme Statistics Methodology

• Published: 15 December 2022
• Last updated: 15 December 2022
  See all notes (3)

  1. 15 December 2022

     Rewriting of methodology to accompany new statistics released in the '2021/22' and '2022/23' publications.

  2. 19 July 2022

     Update to School Led Tutoring, Tuition Partners (2021/22) and Academic Mentors (2021/22) sections.

  3. 26 May 2022

     Addition of methodology for calculating Edset response rate, number of responders, and flags. Clarification of methodology for calculating school participation for NTP and SLT.

National Tutoring Programme (2022/23)

NTP (2022/23) is covered in the ‘2022/23 Academic Year’ release in this publication series.

Overview and Data 

In the 2022/23 academic year, all National Tutoring Programme (NTP) funding was given directly to schools to spend on any of the three NTP routes: School-Led Tutoring, Tuition Partners or Academic Mentors. Funding was allocated based on the number of pupil premium (PP) students in the school and covered 102.5% of the PP numbers. 

This year, data for NTP Statistics is collected through the School Census. State-funded schools are required to provide the Department for Education (DfE) with a school census return (using a DfE-designed survey instrument) covering a wide range of information on individual pupils once every school term. 

Data on NTP delivery is collected through two pupil level data items: 

• whether the pupil has received tutoring through any of the three NTP routes and 
• how many cumulative hours of tutoring (to the closest 0.5 hours) they have received up to the day of the census. 

This statistical release is based on data from autumn 2022, which covers the school year from 1st September 2022 up to and including 6th October 2022. 

Starts on Courses by Pupils 

To estimate the number of starts on courses by pupils we total the number of hours of tutoring each pupil who has been tutored has received. If a pupil has received up to 16 hours of tutoring, we consider them to have started one course, above this we consider them to have started two courses. This is in line with NTP guidance to schools suggesting a course should be between 12 and 15 hours, with an additional margin of one hour above 15 hours to avoid counting marginal cases.

Not all courses delivered to pupils were 15 hours long. Individual pupils may undertake more than one course. Where a tutor delivered a course to multiple pupils, that course was counted once per pupil.

We have removed a small number of pupils as they had been given less than half an hour of tutoring, which we did not consider to be a ‘start’. 

A small number of schools who had reported delivering more than three times their allocated courses or hours were also identified in the data before statistics were produced. The figures in these schools were instead replaced using the remaining national average relative to allocation. 

School Participation 

School participation percentages are estimated by dividing the total number of schools that reported delivering NTP by the number of schools (as defined in ‘Other Notes’) with pupils between the ages of 5 and 16. 

It is important to note that a small number of these schools (0.7%) have been given no funding for NTP, however we have kept these schools in the denominator to allow for easy comparison to NTP figures from previous years when these schools would have been eligible for Tuition Partners funding.  
 

National Tutoring Programme (2021/22)

There are four statistical releases which predominately cover NTP in the 21/22 Academic Year: March 2022, May 2022, June 2022, and the 2021/22 Academic Year publication. The last one of these is the latest publication and the figures from it will feature in future publications. 

Overview and Data

In the 2021/22 academic year, there were 3 NTP routes: Tuition Partners (TP), Academic Mentors (AM), and School-Led Tutoring (SLT). The first two were delivered by DfE’s delivery partner, Randstad. School-Led Tutoring involved allocating money directly to schools. 

You can find details about how data was collected on each NTP route in the sections below.

It is possible for pupils to receive tutoring from more than one route of NTP. For example, a pupil could potentially receive tutoring through TP and SLT. Where a pupil receives tutoring from more than one NTP route, they are counted in the figures for total pupils of both routes. Where a school participates in multiple routes of NTP, they are counted in the figures for total schools participating in different routes. This is why overall school participation in NTP is generally less than the sum of the participation rates across the routes. 

Overall Courses Delivered to Pupils (All Releases)

To estimate the total number of tutoring courses delivered in the academic year 2021/22, we total up the estimates made across tuition partners, academic mentors and school-led tutoring, as outlined below. 

Not all courses delivered to pupils were 15 hours long. Individual pupils may undertake more than one course. Where a tutor delivered a course to multiple pupils, that course was counted once per pupil. 

Overall Hours Delivered to Pupils (2021/22 Academic Year Release Only)

To estimate the total number of tutoring hours delivered in the academic year 2021/22, we total up the estimates made across tuition partners, academic mentors and school-led tutoring.

An hour delivered to multiple pupils is counted multiple times. For example, an hour delivered to 3 pupils is counted as 3 hours of tutoring delivered in these statistics. This mirrors the approach taken in NTP policy and funding.

Overall School Participation

For the March 2022, May 2022 and June 2022 releases: We estimated overall NTP school participation in 2021/22 using the sample of schools that responded to the Educational Settings Status form. We counted a school participating in NTP where that school met any of the definitions of participation across SLT, TP or AM, as set out below. School participation estimates in these releases should be treated only as an estimate of school participation.

For the 2021/22 Academic Year release: We count a school participating in NTP where that school meets any of the definitions of participation across SLT, TP or AM, as set out below. For this release, ‘School-Led Tutoring: Year-End Statement’ data was used for the SLT component. We estimate overall NTP school participation in 2021/22 by comparing the total number of schools participating in NTP to the total number of schools (as defined in ‘Other Notes’) with pupils between the ages of 5 and 16 from Get Information About Schools (GIAS).  Please find more information about GIAS here: Get Information about Schools - GOV.UK (get-information-schools.service.gov.uk)

National Tutoring Programme (2020/21)

NTP in the 2020/21 Academic Year has not been the primary focus of a publication in this series. However, details are provided here because the number of courses delivered in 2020/21 is included in statistics regarding the number of starts on courses by pupils since NTP began in all publications. 

Data on Starts on Courses by Pupils

In the 2020/21 academic year there were two NTP routes, Tuition Partners and Academic Mentors. The tuition partners route of the National Tutoring Programme was co-ordinated by the Education Endowment Foundation. Data was collected from tuition partners on how many pupils had been committed to, and of those how many had started, tutoring programmes. Pupils who were reported to have received at least one tuition session were counted as starting a course. 

The academic mentors route of the National Tutoring Programme was co-ordinated by TeachFirst. Data on the number of pupils tutored by each mentor was collected from management information submitted by mentors in each quarter. Levels of completion were low, so data on this route was incomplete.  Therefore, an estimate of the total number of pupils tutored in each quarter was calculated, using the assumption that the average number of pupils tutored by non-responding mentors was the same as the average number tutored by responding mentors. 

Individual pupils may undertake more than one course. Where a tutor delivered a course to multiple pupils, that course was counted once per pupil. Where ‘Starts on Courses’ are described as ‘Courses Delivered’ (as in ‘Academic Year 2021/22’ statistics), this does not mean all courses in 2020/21 were 15 hours long.

We have not produced statistics on school participation for the 2020/21 academic year.

School Led Tutoring (2021/22) - School-Led Tutoring: Year-End Statement

The Year-End Statement data was used to create the statistics on courses delivered to pupils, school participation, hours of tutoring delivered and number of pupils receiving tutoring for the ‘2021/22 Academic Year’ statistics release for school-led tutoring.

For the March 2022, May 2022 and June 2022 releases, please see the section “School-Led Tutoring (2021/22) – Educational Settings Status Form”. 

Data

The Department for Education (DfE) collected data on the delivery of School-Led Tutoring (SLT) in the 21/22 academic year through the Year-End Statement form which was completed by schools. Data submitted up to 21st November 2022 was used for the statistics and DfE plans to revise the statistics in April 2023 to include data submitted since then. 

The form asked schools to report their total delivery of SLT for the last academic year and was used to calculate the recovery of SLT funding. Schools must sign a declaration when completing the form to confirm their responses are correct. All schools who received funding were asked to complete the form. 

On the form, schools are asked whether they accessed the grant, how many hours of school-led tutoring they delivered, how much they spent on school-led tutoring, and how many pupils they delivered that tutoring too. The form did not ask schools how many courses they have delivered.

Before the statistics are produced, around 5% of schools were identified as outliers. In the case where the school’s response was judged to be substantially greater than their allocation, we treat them as having delivered tutoring in line with the national average relative to each school’s allocation. These schools are still counted as participating in SLT. In the case where the school’s average hours per pupil is less than one, these schools are excluded and are not counted as participating in SLT. 

School-Led Tutoring (2021/22) – Courses Delivered to Pupils

Because the number of courses was not collected by the Year-End Statement form, DfE analysts have used data on the number of pupils receiving tutoring at each school and the number of hours of SLT delivered to estimate the number of courses delivered. 

To do this, we review evidence on the numbers of hours of tutoring reported by schools. In some cases, pupils have received more than 15 hours of tutoring on average, meaning that at least some of these pupils have started more than one 15-hour course of tutoring. This could be because they are receiving tutoring in multiple subjects, or because they are receiving intensive tutoring in one subject which has totalled more than 15 hours. We estimate that pupils in the school have received one course of tutoring if there are less than 16 hours per pupil. If the average number of hours per pupil is between 16 and 31 hours, we estimate pupils in that school started two courses, and so on. We include an extra hour in here to allow for cases where a school has marginally gone over the 15-hour average. 

Not all courses delivered to pupils were 15 hours long. Individual pupils may undertake more than one course. Where a tutor delivered a course to multiple pupils, that course was counted once per pupil. 

School-Led Tutoring (2021/22) – School Participation 

Using the Year-End Statement data, we estimated the total number of schools who were doing SLT and then calculate this as a percentage of all schools (as defined in ‘Other Notes’).  Not all schools receive funding for school-led tutoring, but all schools are included in the calculation of the percentage for consistency across the publication.

School-Led Tutoring (2021/22) – Hours of Tutoring Delivered and Hours per Course

These metrics are calculated by taking the sum of hours of school-led tutoring delivered and dividing this by the total number of courses.

School-Led Tutoring (2021/22) – Number of Pupils Receiving Tutoring

The number of unique pupils receiving school-led tutoring is taken from the Year-End Statement form data. We assume the same pupil is not included on more than one school’s form. 

School-Led Tutoring (2021/22) - School Census

The School Census data was used to create the statistics on pupil characteristics for the SLT route in the ‘2021/22 Academic Year’ release. 

Data

Schools are required to provide the Department for Education (DfE) with a school census return (using a DfE-designed survey instrument) covering a wide range of information on individual pupils once every school term. In the 2021/22 academic year the spring census date was 20th January 2022, and the summer census was on 19th May 2022. 

Data on SLT delivery is collected at each census through two data items: whether the pupil has had tutoring through the SLT route and how many cumulative hours of tutoring (to the closest 0.5 hours) they have received up to the day of the census.

School-Led Tutoring (2021/22) – Pupil Characteristics 

Some pupil characteristics variables are only updated once a year in the spring census while others are collected every term. In these statistics we have therefore matched pupil characteristics information from the spring 2022 census to the most recent pupil level information on SLT from the summer 2022 census. This data was used to calculate the breakdown of pupils tutored on SLT by each characteristic. 

Please note: The number of pupils in this section will not match the number of pupils in the SLT volumes section of the release. This is because the census covers a shorter time period (up to May) than the Year-End Statement data used to estimate pupil numbers (up to August). 

At pupil level, pupils with less than one hour of tutoring recorded were removed from the analysis as outliers. At school level, schools are treated in line with the ‘School-Led Tutoring: Year-End Statement’ data (see above). The percentage of tutoring provided to each pupil group was calculated after removing these outliers.

Pupil characteristic definitions:

• Eligibility for free school meals (FSM) during the previous six years: at each census the school records whether a pupil is in receipt of free school meals on the day of the census. Pupil records across census are matched after the spring census each year to create a cumulative record of FSM eligibility over six years.
• Gender: is self-declared by a parent/guardian or the pupil and is recorded as either female or male.
• Special Educational Needs Provision: this is recorded by the school and includes pupils with special educational needs support or those with Education, Health and Care (EHC) plans. 
• Ethnicity: is self-declared by a parent/guardian or the pupil. The specific ethnicities collected in the census are grouped into six major groups: Any Other Ethnic Group, Asian, Black, Mixed, Unclassified and White. The unclassified group includes pupils where the information was not returned to the school or where the information was refused.

School-Led Tutoring (2021/22) - Educational Settings Status Form

For the March, May and June 2022 statistics releases, the Educational Setting Status form was used to estimate course starts by pupils. This data source was the best way of producing estimates of school-led tutoring course starts at the time, following the introduction of school-led tutoring. For better estimates of school-led tutoring courses delivered throughout the academic year 2021/22 please use the more recent ‘2021/22 Academic Year’ release.

Data

During the academic year 2021/22, the Department for Education (DfE) collected data on the uptake of school-led tutoring approximately every four weeks through the Educational Setting Status form. The questions asked were:

• Is your school currently using the School-Led Tutoring grant to provide tuition?
• Since the beginning of the Autumn term, how many pupils have received tuition through the School-Led Tutoring grant in total? 
• Since the beginning of the Autumn term, how many pupil hours of tuition have been delivered through the School-Led Tutoring grant in total? 

The date of the form used to inform our statistical publications is indicated in the release. This was a non-compulsory survey which meant not every school responded.

A small number of outliers were removed. For example, where schools had reported delivery of unfeasibly high average numbers of hours of tutoring per pupil or an unfeasibly high number of pupils given the number of pupils recorded in the School Census. Such schools are still treated as participating in SLT.

For the release on the 31st March 2022, we tightened up our criteria for outliers, compared to previous ad-hoc releases. This reduced the estimated starts on courses by pupils for school-led tutoring by around 7%. 

We did not find responding and non-responding schools were significantly different in characteristics (for example size, phase, location). As such we did not feel it was necessary or appropriate to make any adjustment for non-response bias. This is consistent with how other data from the Educational Settings Status form was reported, for example, absence rates. However, that does not mean that we can rule out the risk of non-response bias in these estimates.

Starts on Courses by Pupils

We reported estimates of the number of starts on courses made by pupils based on schools’ responses to the Educational Setting Status form. We assumed that the average number of pupils receiving tutoring at non-responding schools was equal to the average number of pupils receiving tutoring at responding schools.

For these statistics, we had to infer what non-responding schools were doing from the sample of responding schools, therefore estimates are uncertain. This uncertainty increased as the response rate fell and as the population size fell. As such, we have greater confidence in the national estimates, which are based on a relatively large group of schools, than in the local authority estimates, which are based on a much smaller group. Local Authority estimates should be treated with caution, particularly when comparing against previous estimates.

Starting with the May 2022 publication, to be transparent and give an indication of the uncertainty, we included the number of schools that responded to the Educational Setting Status form and the response rate next to every School-Led Tutoring estimate. In line with DfE Statistics on “Attendance in education and early years settings during the coronavirus (COVID-19) pandemic”, we also flagged School-Led Tutoring estimates for Regions and Local Authorities where the estimate was:

1. Based on 10 or less responding schools, or
2. Based on a response rate of 50 per cent or less.

The flags are provided in the downloadable data tables. They act as a guide to users as to where the uncertainty of the estimates may be higher, although they should not be interpreted as hard rules on data quality. There will always be some variability in estimates calculated from a sample of schools.

For school-led tutoring, we received data on the number of pupils undertaking tutoring at each school. To make these figures consistent with those we received for tuition partners and academic mentors (where pupils can undertake multiple courses of tutoring) we additionally estimated how many course starts had been undertaken by these pupils.

To do this, we reviewed evidence on the numbers of hours of tutoring reported by schools. In some cases, pupils received more than 15 hours of tutoring, meaning that they in effect started more than one 15-hour course of tutoring. For example, some schools reported an average of more than 15 hours per pupil being delivered since the start of the academic year, meaning that at least some pupils must have received more than 15 hours of tutoring. This could be because they were receiving tutoring in multiple subjects, or because they were receiving intensive tutoring in one subject which totalled more than 15 hours. 

We estimated how many pupils this affected by assuming that tutored pupils at each school received the average number of hours of tutoring for that school. We then found the percentage of schools where the average pupil was receiving more than 15 hours tutoring. We applied this proportion to the estimated number of pupils receiving school-led tutoring to estimate starts on courses by pupils. This ratio was applied nationally, rather than calculated separately for each Region and Local Authority. 

Not all courses delivered to pupils were 15 hours long. Individual pupils may have undertaken more than one course. Where a tutor delivered a course to multiple pupils, that course was counted once per pupil. 

School Participation

Using the Educational Setting Status data, we estimated the total number of schools doing SLT based on that data and then calculated this as a percentage of all schools. Not all schools received funding for school-led tutoring but all schools (as defined in ‘Other Notes below) are included in the calculation of the percentage. As with the estimate for starts on courses by pupils on School-Led Tutoring, we have included information about the Educational Settings Status responses and flags to indicate where Regional and Local Authority estimates may be less reliable. We started doing this in May 2022. 

Tuition Partners (2021/22)

Tuition Partners (2021/22) - Data

In the 2021 to 2022 academic year, the tuition partners route of the National Tutoring Programme was co-ordinated by Randstad. 

DfE received data on which schools participated and how many courses were started through Randstad, our delivery partner for the 2021 to 2022 academic year. Randstad provided management information to DfE on a continuous basis throughout the academic year 2021/22. 

Tuition Partners (2021/22) - Courses Delivered to Pupils

For tuition partners, data on each order of a tuition course was collected by Randstad. To qualify as a course delivered, there must have been an order record and at least one timesheet of tuition attributed to that course.

For some courses delivered to pupils, neither Randstad nor DfE hold data on the geographical location of those courses during the 2021/22 academic year. These courses were labelled as ‘Unassigned’ in the Regional and Local Authority tables in the relevant releases.

Individual pupils may undertake more than one course. Where a tutor delivered a course to multiple pupils, that course was counted once per pupil. 

Tuition Partners (2021/22) - School Participation

We counted a school as participating in the tuition partner route of the NTP if Randstad reported a course of tuition on the tuition partner route had started at that school. The March 2022, May 2022, and June 2022 NTP statistics publications may have slightly underestimated this rate when there were courses for which DfE nor Randstad had data. 

Tuition Partners (2021/22) - Hours of Tutoring Delivered and Hours per Course

Estimates of hours of tutoring delivered and hours per course come from timesheet data provided to the department by Randstad. The hours per course estimate was calculated as an average of the summed hours of tutoring delivered per course delivered.

Tuition Partners (2021/22) - Number of Pupils Receiving Tutoring

Tuition partners were required to submit information about the pupils they delivered tuition courses to. This information was provided to the department by Randstad and used to estimate the number of pupils receiving tuition via TPs and the characteristics of the pupils. 

Tuition Partners (2021/22) - Pupil Characteristics

We estimated the characteristics of pupils by calculating percentages for those pupils tutored by tuition partners. This information was sourced from schools.

Academic Mentors (2021/22)

The methodology described here is for the ‘2021/22 Academic Year’ release. The methodology used in previous releases is similar and the differences in methodologies are described within the ‘Academic Mentors (2021/22) – Starts on Courses by Pupils or Courses Delivered to Pupils in All publications’ section. 

Academic Mentors (2021/22) - Data

In the 2021 to 2022 academic year, the academic mentors route of the National Tutoring Programme was co-ordinated by Randstad. 

DfE received data on participating schools and how many courses had been started through Randstad, our delivery partner. Randstad provide management information to DfE on a continuous basis. The data on academic mentors is not complete, with many mentors supplying little or no data on their activity. 

Estimates on courses delivered to pupils, hours per course and pupil characteristics are therefore calculated using data from a sample of academic mentors who appeared to have supplied reliable data covering the courses they delivered, hours of tutoring they delivered and the number of pupils they delivered tutoring to. This group represents around 38% of all academic mentors and estimates have been produced assuming this sample represents the wider population. Data on the number of academic mentors employed by schools is complete, so we can estimate school participation with a high degree of confidence. 

Academic Mentors (2021/22) - Starts on Courses by Pupils or Courses Delivered to Pupils in All Publications

Randstad provided data collected from some academic mentors on courses of tuition. As with 2020 to 2021, the coverage of this information was not complete. We took the average number of recorded courses delivered to pupils per sampled mentor with a number of starts within reasonable delivery parameters, split by school phase. We assumed the average number of courses delivered to pupils made under non-sampled mentors who have been in post for over a week is the same as the average number for sampled mentors with a realistic number of starts, for each phase. 

To produce estimates of courses delivered to pupils in each Region and Local Authority, we used national averages of courses delivered per mentor. As such, these estimates should be used with caution. Individual pupils may undertake more than one course. Where a tutor delivered a course to multiple pupils, that course was counted once per pupil. 

The same data source for statistics on academic mentors was used in the March, May and June 2022 statistics releases. The sample used to create the ‘2021/22 Academic Year’ statistics on academic mentors was refined in order to produce a wider variety of statistics covering hours delivered, number of pupils and pupil characteristics as well as courses delivered. The sample of AMs used in the December 2022 statistics release is approximately 38% of all academic mentors, if the methodology had not been amended the sample of AMs would have been 47% of all academic mentors. 

We carried out statistical analysis to test bias in the sample and examined differences between the summary statistics of both methodologies’ samples. This analysis concluded that there was no significant bias in the sample and the difference in total courses delivered by academic mentors was approximately 4% higher for the sample used in the new methodology compared to the sample used in the old methodology.
 

Academic Mentors (2021/22) - School Participation in All Publications

We counted a school as participating in the academic mentor route of the NTP if Randstad reported that a mentor had been placed with that school. Not all schools were eligible for a mentor, but all schools were included in this metric for comparison against other school participation statistics. 

Academic Mentors (2021/22) - Hours of Tutoring Delivered and Hours per Course in the ‘2021/22 Academic Year Publication’

Estimates of hours per course per academic mentor were also based on the sample of academic mentors, as outlined above. As mentioned, there was a lot of underreporting in the data generally. We have mitigated the impact of this through testing an extensive variety of different samples. There still remains a high degree of uncertainty in our estimate; as our approach was cautious this is likely to mean our estimates understates the true number of hours per course. 

Academic Mentors (2021/22) - Number of Pupils Receiving Tutoring in the ‘2021/22 Academic Year Publication’  

We estimated the number of unique pupils tutored using a similar approach to calculating courses delivered and hours delivered. This figure may contain some duplication of pupils who may have received tutoring from more than one AM.  Using our method, such a pupil would appear as two ‘unique’ pupils. This is necessitated by the limited data uploaded by AMs. We have assessed the effect of this overcounting on a subsample of schools who have sufficient data coverage. Based on this analysis, we expect this method to increase the estimate of unique pupils by 2%.

Academic Mentors (2021/22) - Pupil Characteristics in the ‘2021/22 Academic Year Publication’

We estimated the characteristics of pupils by calculating percentages for those pupils tutored by academic mentors in the data sample. This information was sourced from schools.

Other Notes

Metrics

Where we have provided local authority, regional or phase breakdowns:

• estimates of courses started by pupils or courses delivered to pupils may not sum perfectly due to rounding, and
• estimates of school participation are weighted by the number of schools. 

Schools

For the number of schools, we have used Get Information About Schools Data (GIAS). For example, for the publication on 31st March 2022, GIAS data was downloaded on the 23rd February 2022. We include the following list of Open schools in England: 

• Academy alternative provision converter
• Academy alternative provision sponsor led
• Academy converter
• Academy special converter
• Academy special sponsor led
• Academy sponsor led
• City technology college
• Community school
• Community special school
• Foundation school
• Foundation special school
• Free schools
• Free schools alternative provision
• Free schools special
• Non-maintained special school
• Pupil referral unit
• Studio schools
• University technical college
• Voluntary aided school
• Voluntary controlled school

From the ‘Academic Year 2021/22 publication’ onwards, we also filter on ‘School Phase’ to include the phases defined below. 

For the ‘Academic Year 2021/22’ publication, we used the GIAS on 1st September 2022 to maximise consistency with the ‘Year-End Statement’ data that was submitted at that time. We included activity on the Tuition Partner route related to other schools types (e.g. virtual schools) but those schools were not included in school participation statistics. 

Activity related to Virtual Schools or Independent Special Schools on the School-Led Tutoring Route has not been included in this release or previous releases and will be subject to revision for statistics in April 2023. 

Phase of School

The table below outlines the types of GIAS school phase included in the NTP Statistics school phase:

GIAS School Phase	NTP Statistics School Phase
Primary	Primary
Middle-deemed primary
Secondary	Secondary
All-through
Middle-deemed secondary
Not applicable	Other