The key to defending predicted grades is to make each prediction the documented outcome of a consistent assessment process, not an impression formed near a deadline. A defensible prediction uses representative student work, current IB standards, the complete assessment model, and departmental standardisation.
The IB describes a predicted grade as the grade a teacher expects a student to achieve, based on all evidence of the student's work and the teacher's knowledge of IB standards. Professional judgement is therefore required, but it must be anchored in attainment. A predicted grade is not a simple average, a reward for effort, or the grade needed for university admission.
What makes an IB predicted grade defensible?
A prediction is defensible when another informed teacher can review the evidence and understand the decision. They may not select exactly the same grade in a borderline case, but they should recognise that the reasoning is fair, consistent, and aligned with IB expectations.
The IB explanation of predicted grades and assessment emphasises all available evidence and knowledge of IB standards. The official Diploma Programme grade descriptors clarify the qualities associated with each grade.
A robust prediction should be:
- Representative: It covers the components, objectives, knowledge, and skills assessed in the subject.
- Recent: It reflects the student's current level while recognising persistent performance patterns.
- Authentic: The work was completed by the student under known conditions.
- Comparable: Equivalent evidence rules and standards were applied across the cohort.
- Moderated: Borderline or unusual decisions were reviewed by another specialist where possible.
- Documented: The reasoning can be reconstructed without relying on memory.
Build a representative body of evidence
Not every mark has equal predictive value. A timed mock usually provides stronger evidence of examination performance than open-book homework, while an assessed IA measures criteria that a mock may not cover.
| Evidence source | Predictive value | Main limitation |
|---|---|---|
| Full mock examinations | Show breadth, timing, recall, stamina, and exam technique | One sitting may be affected by illness, anxiety, or paper difficulty |
| Timed paper sections | Test particular examination components | May not represent the complete syllabus |
| Criterion-assessed tasks | Show achievement against objectives or markbands | Conditions and task difficulty may vary |
| Internal assessment | Provides evidence against subject-specific criteria | Represents only one part of most subjects |
| Topic tests | Show knowledge and application in taught units | Narrow coverage can distort the overall picture |
| Oral, practical, or performance work | Measures skills written examinations cannot capture | Reliability depends on common criteria and standardisation |
| Homework and formative work | Reveal trends and misconceptions | Authentication and working conditions may be uncertain |
Create an evidence map listing each final component, its weighting, the skills assessed, and the school tasks that provide relevant evidence. This exposes gaps, such as predicting a language grade without enough oral evidence or a science grade without data-analysis tasks.
The IA should inform the prediction, but it cannot substitute for the whole subject grade. Predicted grades forecast overall subject performance, while internally assessed work normally represents one component. The IB assessment principles emphasise reliability, fairness, comparability, and construct relevance, which are useful tests for a school's process.
Use a transparent decision model
Prediction is not perfectly mathematical, but a transparent model reduces arbitrary differences between teachers. A department can begin with an indicative weighted mark based on estimated component performance and the current assessment model. This is a local starting point, not an official IB formula.
A practical process is:
- Confirm the assessment construct. Use the current subject guide and assessment model.
- Select representative evidence. Cover different objectives, components, and task conditions.
- Apply IB standards. Use current criteria, markschemes, and grade descriptors.
- Estimate component performance. Determine what the evidence suggests the student would currently achieve in each component.
- Test plausible boundaries. Historical boundaries can support sensitivity analysis, but future boundaries are unknown.
- Record professional judgement. Explain how trends, missing evidence, task conditions, or anomalous results affected the estimate.
- Standardise borderline cases. Review relevant work with a colleague and record the agreed rationale.
The purpose is not to eliminate judgement. It is to make that judgement visible and reviewable.
Standardise predictions across the department
When several teachers teach the same course, standardisation helps prevent one class receiving optimistic predictions while another is judged conservatively. Teachers can independently mark a small selection of high, middle, and borderline responses, compare their decisions against the criteria, and agree how standards will be applied.
Historical school outcomes are useful as a reasonableness check, not a quota. A sharp departure from previous patterns may deserve investigation, but an unusually strong cohort can legitimately receive stronger predictions. IB assessment is criterion-related, so past distributions must not override current evidence.
The RevisionDojo guide to how schools calculate predicted grades can help departments explain the roles of mocks, coursework, trends, and moderation. Publishing the process reduces misunderstandings before individual grades are discussed.
Keep a concise prediction record
For each student, retain a prediction record in accordance with school retention and data-protection policies. It should identify:
- the prediction, date, course, level, and examination session
- the assessments used, including dates and conditions
- relevant component or criterion marks
- the guide, criteria, and descriptors applied
- any weighting or calculation used
- excluded evidence and the reason for exclusion
- performance trends and standardisation outcomes
- a short rationale for borderline or exceptional decisions
- the reviewer and review date, where applicable
A useful note might state: "Predicted 5. The IA shows grade 6 characteristics, but three recent timed tasks covering both papers remain securely within the grade 5 range. A grade 6 would depend on improvement not yet demonstrated under examination conditions. Reviewed by the subject team on 12 October."
This is stronger than "average was 68%" or "student works hard" because it identifies the evidence, explains conflicting indicators, and connects the decision to likely overall attainment.
Handle challenges as evidence reviews
A question from a student or parent should trigger an explanation, not a negotiation. Restate the purpose of the prediction, show which evidence categories were used, and explain what distinguishes the assigned grade from the next grade.
Correct omitted evidence, data-entry errors, or inconsistent marking. Where the disagreement concerns professional judgement, use the school's published review procedure. The RevisionDojo predicted-grade appeals guide helps separate legitimate evidence review from pressure to increase a grade.
Do not change a prediction because a student needs another point for an application. Entry requirements explain why the student wants the grade, not why they are likely to achieve it. Equally, systematic under-prediction is no more accurate than grade inflation.
Common practices that weaken predictions
Several approaches are difficult to defend:
- using one mock as the entire prediction
- taking a simple mean of assessments with different scope and conditions
- treating the IA grade as the overall subject grade
- rewarding effort or behaviour instead of demonstrated attainment
- applying historical boundaries as fixed future cut-offs
- forcing a cohort into previous grade distributions
- accepting automated marks without teacher review
RevisionDojo guidance for IB teachers and departments explains how question sets and assessment data can support consistent evidence collection. Its IB study and assessment features may help teachers identify patterns, but Jojo AI feedback and automated grading remain supporting information rather than official IB marking.
Conclusion
Defending predicted grades requires representative evidence, current IB standards, appropriate weighting, departmental standardisation, and a concise record of each decision. The strongest prediction is neither the highest nor the most cautious grade. It is the grade that best reflects likely overall performance at that point.
RevisionDojo's Questionbank, Mock Exams, class analytics, and Jojo AI can support evidence collection and feedback. Final responsibility must remain with the teacher and the school's agreed process.