
The IB’s current examiner instructions state that assessment is criterion-related, not norm-referenced: student work is judged against identified levels of attainment, not against other candidates. Yet grade boundaries move from one exam session to the next. That leaves students wondering whether the rest of the cohort is quietly shaping their grade.
The answer lies in two separate decisions, and the marking decision comes first. Examiners use differentiated level descriptors as their primary tool and apply them on a best-fit basis, picking the descriptor that fits the response best overall. A strong aspect can make up for a weaker one. A response, therefore, doesn’t have to meet every indicator to earn a given level.
The grade-boundary decision comes after marking. In each exam session, the IB carries out grade-award processes to set the final boundaries, as described in its own research on statistical grade-boundary setting. Examiner judgment and statistical guidance, such as recommended boundaries, decide which total marks count as a 4, 5, 6, or 7 in that session. That call is made after the scripts exist, so no one can know the boundaries in advance. Boundaries can move without changing the question that decides your marks: did this response demonstrate the descriptor features?
The preparation habit that works against the model
Many students revise by covering the most ground, doing the most questions, or hoping for a forgiving paper. Those habits suit a contest where candidates are ranked against one another. The descriptor check rewards something narrower: analysis, structure, or justification that the response visibly shows. Wide coverage doesn’t guarantee any of the three.
The consequence is that a student can know a subject thoroughly and still lose marks because the response does not visibly demonstrate the features the descriptor requires. The gap is often less about knowledge and more about whether the answer speaks the language the examiner is checking against. Closing it calls for a different starting point: the criteria themselves.

What IB subject guides actually contain
The IB’s expectations are already written down, in detail, in its subject guides. A 2002 National Research Council review of the International Baccalaureate found that the program guides contain substantial detail about what students are expected to be able to do. Those outcome statements influence how teachers plan instruction and how learning and assessment tasks are designed. That detail is a long-standing structural feature of the program’s design.
The review’s account of the mathematics guides shows what that detail looks like. They list objectives such as organizing and presenting data in graphic, tabular, or diagrammatic form, formulating and communicating mathematical arguments clearly, recognizing patterns and structures, and using appropriate technology. Each is something a response either shows or doesn’t. The guides also carry teaching notes and suggested strategies, which are advisory rather than mandatory.
Students can hold their own work up against outcomes like these using their subject guides. In my view, that is the step most of them skip.
Turning descriptor language into recognizable patterns
The first move is to read the relevant descriptor or mark-point list and ask what a response would visibly contain or demonstrate. Take a descriptor that rewards well-developed analysis. The phrase itself cannot be observed, but its effects can: a specific, contestable claim rather than a summary, relevant evidence rather than generalities, an explanation of how that evidence supports the claim, and a structure that makes the reasoning easy to follow.
Best-fit marking changes how you use that list. A response can sit at a level even when one feature is weaker because examiners judge which level the whole response most resembles. The features help you gauge that overall resemblance. They aren’t boxes to tick, and they aren’t stock phrases to collect.
Then test your reading against marked work. International Baccalaureate past papers supply authentic exam questions, and their mark schemes show which descriptor features earn credit. Practicing past questions without the mark scheme is repetition: it builds familiarity with formats but leaves descriptor matching untouched. With the mark scheme beside it, a practice response becomes evidence of which features earned marks and which are still missing.
Carrying the habit across subjects
A new subject means a new paper, and it can feel like starting from scratch. The habit that carries over doesn’t depend on content. Read the subject’s criteria, decide what a strong answer in that subject must show, then check practice responses against that picture.
The mathematics guides are one illustration of how a subject’s criteria can be turned into features you can check, from presenting data clearly to communicating a mathematical argument. I’d expect the same step to work in any subject that publishes its criteria.
Students who build this habit stop experiencing each new paper as unpredictable and start reading it as a system with legible expectations. That shift compounds over the diploma cycle because every new paper arrives with the same question already in hand: what must a strong answer show here?
Putting the scoring model to work in your revision
The IB’s criterion-referenced model is consistent and transparent. The criteria are published, the descriptors define what examiners look for, and the distance between current work and a higher band can usually be traced to specific missing features rather than a vague shortfall in knowledge. Revising from the criteria works with that logic instead of around it. A student who learns early to spot missing features gets to use that skill in every later unit.
Boundaries will keep moving, and no one can see them in advance. What you control is narrower and more useful: making the descriptor features unmistakable in the response you write. On the day, the examiner will hold that response up against the descriptor, not against the candidate at the next desk.