GradewingSubject Insights
Back to 16 to 19 performance in England

ENGLAND · A LEVELS · OFQUAL RESEARCH

Subject difficulty:
General Ability

For students with similar ability estimated from their A level results, which subjects are harder to achieve a given grade in?

Read this first

We reproduce the results of Astera Brylka and Qingping He’s Ofqual study, Inter-subject comparability in GCSEs and A levels in summer 2024, which compares 2019, 2023 and 2024. The authors use Rasch modelling to estimate subject difficulty, linking a student’s chances of higher grades to the difference between their estimated general ability and the subject’s grade-specific difficulty.

The model estimates the statistical difficulty for each grade in each subject. The Overall difficulty averages the estimates at D, C, B and A for each subject; A* is shown separately.

Subjects ranked by difficulty

Most difficult subject first (rank 1).

2024 2019 baseline
  1. 1. Further mathematics
  2. 2. Physics
  3. 3. Chemistry
  4. 4. Computer science
  5. 5. Biology
  6. 6. Mathematics
  7. 7. Philosophy
  8. 8. Music
  9. 9. French
  10. 10. German
  11. 11. Physical education
  12. 12. Accounting
  13. 13. Geology
  14. 14. Spanish
  15. 15. Economics
  16. 16. Psychology
  17. 17. Politics
  18. 18. History
  19. 19. Religious studies
  20. 20. Geography
  21. 21. Law
  22. 22. Classical civilisation
  23. 23. English literature
  24. 24. Design & technology: Product design
  25. 25. English language
  26. 26. Business studies
  27. 27. English language & literature
  28. 28. Drama & theatre studies
  29. 29. Sociology
  30. 30. Dance
  31. 31. Media studies
  32. 32. Art & design: Art, craft & design
  33. 33. Art & design: Three-dimensional design
  34. 34. Art & design: Fine art
  35. 35. Chinese
  36. 36. Film studies
  37. 37. Art & design: Textiles
  38. 38. Art & design: Graphics
  39. 39. Art & design: Photography

Download CSV

Method and sources

Ranks stay relative to all 39 subjects when searching. The tracks show ordinal positions from rank 1 (most difficult) to rank 39 (least difficult), not differences in grade widths. Filled markers show the selected year; hollow markers show the 2019 baseline. Change is movement relative to 2019.

General Ability method overview

These are published-model results for 2019, 2023 and 2024, not estimates from our DfE transition matrices or results for 2026. The overall measure averages Rasch difficulties at D, C, B and A; it does not include A*. Each grade has its own ranking.

How the ranks were reconstructed. Figure 11 (p. 23) supplies the ascending 2019 subject order. Table A4 (pp. 36–37) supplies changes from 2019 to 2023 and 2024. We add those changes to the baseline positions, then reverse the convention so rank 1 here means most difficult. All twelve reconstructed rankings contain each position from 1 to 39 exactly once.

Rank calculation. Let a be the subject’s ascending 2019 position in the 39-subject comparison set, and Δ the appendix change from 2019 to the selected year (zero for 2019). Displayed rank = 40 − (a + Δ). The movement label is 2019 rank − selected-year rank: a positive value means movement towards the more difficult end of the comparison set, rather than an absolute change in difficulty.

Coverage. Music technology appears in Figure 11 but is absent from Table A4. It is excluded from every year here; the 39 remaining subjects form the comparison set. Ranks describe only this set.

Figure 12 is about change, not rank. Its vertical scale is the change in Rasch difficulty in grade widths (GW). A subject can become less statistically difficult while retaining the same rank. We have not digitised those plotted values or treated rank movement as a change in GW.

Source inconsistency. For Geology at A, Table A4 prints −3 for 2023−2019, −2 for 2024−2019 and −1 for 2024−2023. The first two imply +1 for the last comparison. This page consistently uses the two comparisons with 2019 and retains the original entries in the downloadable source data.

The model assumes a common underlying ability across subjects. Subject-specific strengths, selection, teaching and other factors can affect outcomes. Small differences in rank are not evidence of statistically meaningful differences; the page does not infer confidence intervals.

Source: Astera Brylka and Qingping He, Inter-subject comparability in GCSEs and A levels in summer 2024, Ofqual, published 24 July 2025. Read the source PDF · Publication page.