The 2026 results show attainment. They do not isolate what an independent school caused

01

The important question comes after the celebration

Examination results deserve recognition as evidence of what pupils achieved. The analytical difficulty begins when the same results are used to establish how much a school improved learning or whether one educational model is superior. Those are different claims. A strong independent-school account should make the distinction clear without diminishing a young person’s achievement.

The May curriculum editorial asked what evidence could support a school’s curriculum choices. Its qualification companion separated level, size and recognition. August supplied the next stage of evidence. The question now is how to connect results with the education that produced them while resisting a conclusion the data cannot identify.

02

The new evidence has a defined population and unit

Ofqual first published its 2026 results release on 13 August for level 3 qualifications and added level 1/2 and level 2 on 20 August. It covers results issued to students in schools and colleges in England. The publication history matters: the first results-day version did not yet contain the later qualifications.1

In its all-subject A-level table, the self-reported independent/private-school category, including city technology colleges, contains 107,940 results, with 48.6% at grade A or above. This is a proportion of results in that stated centre category, not the proportion of independent-school pupils obtaining all high grades or an estimate of the effect of attending those schools.2

Ofqual explains that centre types are self-classified and that some selective schools are reassigned to the grammar category. It also distinguishes results from students and provisional results from certificates. These qualifications are essential to the meaning of a comparison, rather than footnotes that can be removed from its public presentation.2

03

A broad difference still leaves the mechanism unresolved

A descriptive gap can be real without showing which part arose from teaching. Intake, prior attainment, subject choice, age, retention and the number of qualifications taken are examples of factors an investigation would need to consider. This is a methodological argument, not a finding that any one factor explains the reported difference.

The missing step is a suitable comparison of pupils and educational experience. A school with high attained grades can still ask whether pupils progressed as expected, which groups encountered barriers and where a course did not work as intended. A headline cannot answer those questions merely by becoming more prominent in the prospectus.

04

More detailed geography improves description, not causation

On 7 August, Ofqual announced new local-authority-level results and subject-uptake breakdowns, with 2026 data to follow on results days. It also explained why its figures can differ slightly from JCQ: Ofqual receives awarding data earlier, when some results are still processing. A small discrepancy is therefore not automatically evidence that one account is wrong.3

A more detailed map can help identify a local question. It cannot turn the area surrounding a school into a matched comparison group. Pupils may travel across boundaries, and different schools may offer different courses or recruit different cohorts. Leaders should identify what the geographic comparison is intended to explain before treating proximity as equivalence.

05

The UK and international references need their own frame

JCQ’s A-level dashboard, published on 13 August, describes data from five awarding organisations across its stated UK nations. Its notes distinguish those nations’ regulators and explain that this dashboard initially focuses on GCSE and A-level statistics. It should not be relabelled as an England-only account of every qualification.4

The IB’s bulletin page, updated on 26 August, separately presents summary statistics for the May 2026 Diploma and Career-related Programme assessment session. The assessment session is not the date on which the bulletin was published. A global session summary also supplies a different comparison population from England’s A-level centre table.5

For a school offering more than one route, the educational responsibility is to explain each route’s outcomes accurately. Converting unlike grades or programme totals into one invented common ranking would require a defensible method. A family needs to understand the actual qualification and cohort, rather than a synthetic number whose apparent precision conceals the differences.

06

A useful school-level account connects the cohort to the course

Imagine a department with a strong overall grade profile and a smaller subject group whose outcomes changed sharply. The headline may remain excellent while the smaller group needs serious investigation. This hypothetical example shows why leaders need an account of course experience and progression alongside the aggregate. It does not suggest that any published 2026 school results conceal a problem.

Such investigation should ask a precise educational question. Was the course accessible to the pupils who started it? Did assessment evidence identify difficulty in time? Did advice about continuation match the demands pupils subsequently faced? Where pupils changed route or left, the public account should avoid implying that the final examination cohort describes everyone’s earlier experience.

07

The comparison should survive an ordinary explanation

One useful editorial test is whether a reader can explain the comparison back without changing its meaning. If the statement begins with examination entries but the reader hears pupils, or begins with a selected centre category but the reader hears every independent school, the wording has failed.

A clear caption and a short methodological explanation can prevent that error. The objective is not to burden a celebration with every statistical detail. It is to keep the detail that determines what the central claim actually says.

08

What the August evidence can support

Our assessment is that results should be presented as attained outcomes within a defined population, then connected with appropriate evidence of progress and participation. A strong result is meaningful. A causal claim about the institution requires additional work and should remain clearly identified as an interpretation unless the research design supports it.

This analysis does not rank schools or calculate a sector value-added effect. The published results were provisional, and different sources have different scopes. Corrected figures, a better matched longitudinal analysis or evidence of previously obscured cohort changes could alter the assessment. The next reporting step is to investigate that educational explanation, rather than repeat the same grade headline.

Leadership questions

Questions that turn the development into a governing conversation.

What we are monitoring next

Publication is the beginning of the watch.

• Corrections, certification and later official analytical releases.
• School-level evidence connecting starting points, participation, route changes and attained outcomes.
• Corrected results or a rigorous comparable pupil-level analysis could change the interpretation; documented changes in the final cohort would require the school narrative to be revised.