The setting
After exams were cancelled, England used teacher-supplied centre assessment grades and student rankings in a statistical standardisation process. This is a historical automated-decision case, not generative AI.
The decision context
A retrospective comparison of approaches described in the official record, after the 17 August 2020 reversal. Do not pretend the later analysis was available to the original decision makers. The calculated-grade policy is the historical comparator, not today’s business as usual.
The evidence to examine
Reported evidence
Rationale and reversal
Ofqual described standardisation as addressing inconsistency and likely optimism in centre assessments while maintaining national outcomes. After calculated grades failed to command public confidence, students received the higher of their centre assessment or calculated grade.
Source: Ofqual, Grading gaps in summer 2020: who was affected? · Introduction: grading method and 13–17 August 2020 chronology ↓Reported evidence
Retrospective analysis
Ofqual analysed differences between centre assessment and calculated grades. Centre assessments are estimates; there are no observed outcomes from the cancelled exams against which to measure true individual performance.
Source: Ofqual, Grading gaps in summer 2020: who was affected? · Executive summary and Introduction: meaning and limits of grading gaps ↓AIPEX analysis
AIPEX reading question
Which distributions and errors matter? How should exceptional circumstances and timely challenges be handled? These are institutional choices as well as modelling questions.
What this does not establish
Differences between estimates do not reveal each student’s true exam result. This case establishes neither a universally fair grading policy nor equivalence with every automated decision.
Try the decision first
Your task.
Compare policies when national consistency and individual treatment pull in different directions. What essential conditions and evidence would each need?
- Compare the calculated-grade approach, centre assessment grades and the higher-of-the-two policy. Separate what each was intended to achieve from its observed effects.
- Examine who bears which uncertainty: individual candidates, small subject cohorts, centres and the admissions system. Do not infer an individual’s true result from a difference between estimates.
- State what evidence could change the comparison and what cannot be recovered because exams did not occur. Include timing, challenge and implementation constraints.
Read one defensible approach
AIPEX analysis for discussion. Other conclusions can be defensible if their evidence, assumptions and trade-offs are explicit.
- State competing objectives explicitly. Aggregate alignment cannot settle the acceptability of an individual outcome; examine distribution and exceptional cases.
- Compare the three documented approaches against the same objectives and historical time constraints. A higher grade is not proof of greater accuracy; the higher-of-the-two policy also changes the national distribution. Keep estimates, observed grade changes and unknowable exam results separate.
- Include a timely, usable route to challenge and correction. Name who can change the approach and what signals trigger reconsideration.
Read the evidence / Reviewed 18 October 2026
Sources & scope.
AIPEX’s prompts and worked approaches are our own synthesis. Use the linked material to examine the method and its limits.
- Ofqual
Standardisation of grades in summer 2020: outliers ↗
December 2020 report: analysis of students for whom standardisation might be unreliable
A retrospective official account of statistical grade standardisation. This was not a generative AI system.
- Ofqual
Grading gaps in summer 2020: who was affected? ↗
Executive summary and Introduction: grading method, 13–17 August 2020 chronology and limits of grade comparisons
Official retrospective analysis. Centre assessment grades are a comparator, not observed examination outcomes.
- HM Treasury
The Green Book (2026) ↗
Chapters 4–7 and 10: objectives, options, distribution and evaluation
UK appraisal guidance. Our comparison aid covers an early options discussion; it does not replace a full business case or economic appraisal.