Sampling: Sample design

All schools are eligible for the survey regardless of whether they took part in all or some previous years. For secondary schools, this is necessary as most local authorities have so few schools which were not sampled in previous years.

There is more scope for selecting a ‘fresh’ sample of schools for primary schools, which are more prevalent, but removing the primary schools which were selected in a previous year would result in a sample that was not a true random sample.

This is because we select the sample using probability proportional to size (PPS) and so the larger primary schools are more likely to be sampled each year and would be disproportionately removed from the sampling frame for the following year.

At the end of the academic year 2021-22 (Year 5) survey and during the academic year 2023-24 (Year 7), analysis was carried out to explore whether repeated requests to participate led to lower response as a result of survey fatigue in schools.

No evidence was found that schools which had taken part before were less likely to participate. There was some evidence that as the survey becomes familiar to schools they are more likely to respond.

The Department for Education provides an official register of educational establishments across England and Wales (Get Information about Schools [GIAS]), which is kept updated continuously, and this was used to match on contact information and other school-level information not included in the School Census data.

The schools contained within the Schools Census and GIAS provide unparalleled coverage of children aged 5-15 in England.

When the survey sample was selected in academic year 2024-25 (Year 8), the mid-year population estimate from 2022-23 for children aged 5-15 in England was 7,538,713 and the 20,393 eligible open schools in the Schools Census included 7,138,827 children aged 5-15 (95% of children aged 5-15) across all year groups.

The starting sample frame was all 24,442 schools in England (independent and maintained).

The following school types were removed from the sample frame prior to selection: state funded nurseries, pupil referral units, state funded special schools, non-maintained special schools, independent special schools, establishments attended only by pupils aged 16-19, studio schools which are mainly school year 9-11, technology colleges, university technical colleges and schools containing fewer than two school years.

In academic years 2017-18 (year 1), 2018-19 (Year 2) and 2019-20 (Year 3), there was a requirement for schools to have at least 30 pupils to be included in the sample but this has no longer applied since academic year 2020-21 (Year 4).

These schools were removed from the sample frame for a variety of reasons, including the survey not being adapted for pupils in special schools and pupil referral units having a changing pupil population over the course of a term.

Schools which contain only or mainly nursery, reception or school year 12-13 pupils were excluded because most or all of the pupils would not have been eligible. Schools were removed from the sample frame if they were not recorded as open or had no pupils aged 5-15 in years 1-11. 

Special schools were excluded because, while children with special needs in mainstream school can participate with their usual support, many pupils in special schools would not be able to take part and a different approach is needed to understand their participation in sport.

During academic year 2020-21 (Year 4), development work was carried out to explore how the survey could be adapted for pupils in special schools. This included interviews with key informants, cognitive testing with pupils and a small pilot.

The conclusion from this was that further adaptation was needed for meaningful participation from special schools which did not place too great a burden on staff in those schools. Sport England are continuing to look at options to ensure that the voices of children in special schools are heard.

Removing these schools (3,503) reduced the sample frame of schools to 20,939. It should be noted that some five-year-olds are educated in reception classes, which were not included in our sample frame.

All pupils turn five before they start in year 1. We have taken the pragmatic decision to start with children who are five at the start of the academic year.

In addition, pupils turn 16 years old while they are in school year 11. The use of school year groups 1-11 for sampling means that the sample includes pupils who had turned 16 years old by the time they look part in the survey and excludes five-year-olds within reception classes.

The sample of schools selected from GIAS for the whole year consisted of:

• up to 10 state primary schools in each local authority
• up to 10 state secondary schools in each local authority
• 370 independent schools across the whole country.

In addition, up to five state primary and five state secondary schools were selected per local authority as a reserve sample.

In the sample frame, there was no local authority with fewer than 10 primary schools and there were 166 local authorities with 10 or fewer state secondary schools. In these local authorities, every secondary school was selected.

Selection of state primary schools: For state primary schools, up to 15 schools were selected in each local authority to form the main (10) and reserve (5) samples. The probability of selection was calculated for each school. Where schools had a probability of greater than 1, they were automatically included in the sample.

After this, 287 state primary schools had been forced into the sample. Then the remaining schools were selected with a probability proportional to the size of the school. The sample was stratified by local authority, urban/rural indicator and pupil count.

Once up to 15 schools had been selected per local authority, five of these were randomly selected to be in the reserve sample.

Selection of state secondary schools: For secondary schools, the main and reserve samples were selected separately. As there were so many local authorities with 10 or fewer state secondary schools (where all secondary schools therefore had to be selected), it was not possible to make the reserve sample probability of selection proportional to size of school.

The main sample was selected in the same way as the primary sample, with schools with a probability of selection greater than 1 forced into the sample, with 1,328 schools forced into the sample.

After this, the remaining schools were selected with a probability proportional to size with up to 10 main sample schools being selected per local authority (fewer where there were fewer than 10 schools in the local authority).

The reserve sample was selected from among the schools remaining in the sample frame after the main sample selection, with up to five schools in each local authority randomly selected with equal probability of selection.

This is different from the primary approach, which would not work for secondary schools. In secondary schools the priority is getting the main sample as close to equal probability as possible.

Selection of independent schools: The sample consists of 370 independent schools, in addition to the state schools. The sample was stratified by local authority and pupil count.

Sixteen schools were forced into the sample because they had a probability of selection of greater than 1. All other schools were selected with a probability proportional to size.

It is good practice to stratify by local authority to ensure we have a good spread of schools geographically.

In some years, the allocation of schools across terms is adjusted in such a way as to avoid overlap with the Smoking, Drinking and Drugs (SDD) Survey which Ipsos carries out for NHS Digital. This was not necessary in academic year 2024-25 (Year 8) as SDD did not run in that academic year. 

The main issued sample of schools was 1,884 in the Autumn term, 1,884 in the Spring term and 1,884 in the Summer term. The sample was designed to ensure that, where possible, year groups affected by summer exams were oversampled in earlier terms.

In the spring term, 418 reserve schools in 186 local authorities were issued. In addition, six reserve schools had already asked to take part during the Autumn term, of which one extended into the spring term, and 11 reserve schools chose to opt in during the spring term, of which four extended into the summer term.

In the summer term, 263 reserve schools in 124 local authorities were issued. In addition, 30 reserve schools chose to opt in during the summer term.

If schools requested to take part in a different term to the one they were sampled for, or if reserve schools requested to take part earlier in the academic year, this was permitted, providing it did not imbalance the sample in the local authority or Active Partnership area.

It should be noted, however, that the Schools Census was from the academic year 2022-23 but the survey took place in 2024-25.

In most schools the school years available in a school were the same from year to year. However, schools which are in the process of growing from the bottom up will have year groups in 2024-25 which did not exist in 2022-23.

For the purpose of sampling, these new higher years which emerged in 2023-24 were not included in the sample frame. Only year groups which existed in 2022-23 were selected.

Within each selected school, three year groups between school year 1 and school year 11 were selected. The only exceptions to this were in infant schools, school year 10-11 only schools, or other schools where there are only two year groups where both year groups were selected. Details on how year groups were selected for each school type are given below.

Selection of state primary years: All state primary schools were randomly assigned to one of two patterns of selection; school years 1, 3 and 5 or school years 2, 4 and 6. These patterns worked well for schools with six primary years 1-6.

Where the primary schools were infant (1-2), junior (3-6) or another combination, the years were selected at random.

At the end of this process, approximately equal numbers of each year group had been selected and schools were only set to select year groups which actually exist in their school based on the latest GIAS data available at the time the sample was drawn.

Selection of state secondary years: All state secondary schools were randomly assigned to one of 10 patterns of selection. Where the pattern assigned did not fit with the years available in the school, the school years were sampled at random.

In a very small number of schools, which had both primary and secondary year groups, only secondary year groups were selected. For all through schools which already cover school year 1-11, the random selection of three year groups was done across all the years in the school.

At the end, approximately equal numbers of each year group had been selected.

Selection of independent school years: In independent schools, three year groups in each school were selected at random across all the years covered by the school. 

To select a class in a year group, a school would select the class with the tutor/teacher or class letter that is latest in the alphabet.

In Year 7 it was earliest in the alphabet and it was changed in Year 8 to reduce the chance of the same teacher’s class being selected as in the previous year, so in Year 9 it will revert to earliest.

The classes from which the selection should be made should be mixed ability (e.g. tutor groups, PSHE groups) and unrelated to ability or engagement with PE. They should not be setted by gender, ability or age.

If a school has a single class for the selected year group which also includes children from another year group, this class should be selected and the whole class can complete the survey, even if some children are not in the year group originally selected, except in the case of mixed Year 1/reception classes, when only the Year 1 pupils would be eligible. 

If a school has a combination of classes with pupils from a single year group and mixed year groups, the selection method depends on how the mixed classes are formed:

• If children are assigned randomly to the single and mixed classes, then one of the single year group classes should be selected using the selection method described above. In most cases this will just be a single class so the selection method won’t be required.
• If children are assigned to the mixed classes based on age or ability, the selection method should be used for all classes that include children of the year group of interest. If a mixed class is then randomly chosen, the whole class (all year groups) can complete the survey. 

If any of the selected year groups participated in the survey in academic year 2023-24 (Year 7) and the children are still in the same class with each other, if the school wished they could remove that class from the list when carrying out the selection.

This is not necessary, but the school can opt to do this if they did not wish to burden the same children with participation in academic year 2024-25 (Year 8).

This could also apply if a year 1 class has moved to year 2 and is selected again – the school may wish to select a different class (if one exists) to avoid burdening parents.