Weighting: Pupil and parent data weighting

The weighting for pupils in academic year 2024-2025 followed the same approach as academic year 2023-24 (Year 7). State and independent schools were weighted together, instead of separately.

This therefore means that there were four sets of weights in total that were generated separately:

• Parent returns school years 1 and 2
• Pupil returns school years 1 and 2
• Pupil returns school years 3 to 6
• Pupil returns school years 7 to 11 

Each of these four groups were weighted up (i.e. grossed) to the corresponding population estimates from the 2022-23 School Census.

Some of the data reported in the wellbeing tables involves analysis of matched school year 1-2 pupil data with parent data for those year groups. This is only done when an individual match can be made based on gender, date of birth, year group and school.

For this analysis, the parent weight was used. This is the weight provided on the matched dataset of parents and pupils.

It was not possible to adjust for local authority, as the sample sizes for some local authorities were too small; indeed, there were 10 local authorities in which no schools participated at all.

Instead, counts of pupils in NUTS2 (Nomenclature of Territorial Units for Statistics) geographical areas were used, as there was at least one responding school in each area. In academic year 2024-25, independent schools were weighted with state schools, rather than being weighted separately (as they had been up to academic year 2022-23).

The pupil and parent data weighting for academic year 2024-25 (Year 8) followed the same approach as for academic year 2023-24 (Year 7).

Reserve sample schools which asked to opt in to the survey were treated as sampled for the selected year groups in the school and are included in the national data for academic year 2024-25 (Year 8).

Pupils in reserve schools which opted in who were not in the selected year groups were not included in the national data or weighted.

Furthermore, some schools accidentally had their pupils completing on non-sampled opt-in URLs rather than the correct sampled URLs. In these cases, pupils in the selected year groups were weighted, with trimming if necessary, and included in the national data. See final paragraph of this section for an explanation of trimming weights.

The weighting scheme does not take account of term because there are combinations of NUTS2, school type, gender, phase, rurality and free school meals where there are no cases in a term.

The only way to weight by term would be to stop weighting by another factor. Since the other factors are important for the weighting and consistency should be maintained where possible, it was not possible to include term in the weighting. This should be considered when comparing results between terms and between years.

The weighted sample within each term is not representative, whereas the weighted sample across the year is. Where possible, we advise considering school phases as part of any analysis by term given weighting by term is not possible, as detailed above.

Because the calibration weighting was carried out on proportions rather than actual counts, they were grossed up to regional counts when combined so that the total weighted sample count by region was the same as the population counts from the 2022-2023 school Census for the years covered.

The population estimates for weighting are produced from DfE pupil population estimates from the 2022-23 Get Information About Schools (GIAS) (previously referred to as Edubase).

The sample for academic year 2024-25 (Year 8) was drawn from 2022-23 school data because the sample was selected during the previous academic year so that APs could contact schools in advance of the fieldwork term.

For consistency, the weighting was carried out using the same data which had been used for sampling. The population counts include pupil counts from schools which have permanently closed.

This is because schools that have just closed will have pupils for the previous academic year that will not have been assigned to the new school yet and so removing them would underestimate the number of pupils in the education system.

As for previous years, we noted that a number of schools had substantially more responding pupils than the selected 30 per up to three selected classes.

In some cases, this appeared to be because schools which had opted in additional pupils (which is an option offered to schools) had asked these opted-in pupils to complete using the URLs for sampled pupils only.

Therefore, the weighting for academic year 2024-25 (Year 8) was designed to trim the number of weighted pupils to 40 per class in the school, to avoid these additional pupils affecting the national results. The same procedure has been applied in all previous academic years since 2018-19 (Year 2).

In academic year 2024-25 (Year 8) a boost was carried out in a local authority in London, by which all primary schools (44) were sampled instead of the usual 10. The weights in this area were adjusted by 10/44 to ensure that they were not over-represented in the national sample.