Why This Matters

"I gave them a practice test two weeks before the real one — that's my formative assessment."

If the practice test looks exactly like a smaller, unrecorded version of the summative, and you didn't actually change anything about your teaching after marking it, you've done a rehearsal, not formative assessment. It's not a wasted exercise — students got practice — but it missed the entire point. Formative assessment exists to tell you something you didn't know, in time for you to do something about it. If nothing in your next lesson changed because of what you found, the checkpoint didn't do its job.

Understand the Idea

Formative assessment is any check, during the unit, whose purpose is to reveal where students currently are — so you can adjust your teaching before it's too late to matter. Summative assessment, by contrast, is the final, recorded judgement of achievement against the criteria at the end of the unit — the moment described across the rest of this chapter. The two aren't different in format so much as different in purpose and timing. The same task type (a short written response, a lab, a discussion) can be formative or summative depending on what you do with the result.

This purpose distinction has a practical consequence: a formative checkpoint doesn't need to cover everything. In fact, it shouldn't. The most useful checkpoints target one specific strand (see Article 21) or one specific ATL skill — an Approaches to Learning skill, the IB's umbrella term for transferable skills like organizing evidence, structuring an argument, or collaborating effectively, which sit alongside subject content rather than replacing it. A checkpoint that tries to formatively assess everything at once is really just a smaller summative, and it will overwhelm you with data you can't act on before the real deadline arrives.

Formative work is also, ideally, low-stakes: it shouldn't carry grading weight that follows a student into their reported achievement level, because the entire point is to let students take risks and be wrong safely, in a way that a summative record would discourage. This is where self-assessment (students judging their own work against the same descriptors you use) and peer-assessment (students judging a classmate's work, with guidance) earn their place — not as a break from "real" teaching, but as one of the fastest ways to surface a misunderstanding before it hardens.

The job on your end is what's sometimes called responsive teaching: you don't just collect the checkpoint data, you use it to make a specific, nameable decision about what happens in the next lesson. "I noticed X, so I'm doing Y differently tomorrow" is the whole discipline in one sentence. If you can't finish that sentence after a checkpoint, the checkpoint didn't tell you anything useful — or you collected the data and didn't look at it closely enough.

🧠 Teacher insight: A good test for whether something is genuinely formative: could you cancel the summative entirely and the checkpoint would still have been worth doing? If the answer is no — if its only value was as rehearsal for the real thing — it's summative-adjacent, not formative.

See It in Practice

Here's a three-week formative sequence built for the Grade 8 Individuals & Societies "change and industrialisation" unit, targeting the same investigating-and-evaluating-sources skill area from Article 23's rubric example.

Week 1, Tuesday — Checkpoint: exit ticket. After introducing two primary source excerpts (a factory owner's account and a labour organizer's account), students write two sentences: "One reason I'd trust this source" and "One reason I'd question it." What it reveals: Most students can name surface reasons ("he's the boss so he's biased") but very few connect bias to a specific limitation on what the source can be used to claim. What the teacher does next: Wednesday's lesson opens with a five-minute mini-lesson explicitly modelling how to move from "this source is biased" to "this source can tell us X but not Y" — something that wasn't originally planned as a separate lesson segment.

Week 2, Thursday — Checkpoint: peer-assessment on a short source-evaluation paragraph. Students trade paragraphs and, using a simplified version of the strand language from Article 23, mark whether their partner's paragraph evaluates reliability or merely describes the source. What it reveals: About a third of the class is still writing description ("this source was written in 1842 by a mill owner") and calling it evaluation. What the teacher does next: The teacher pulls that third of the class into a small group during the next independent work block for direct modelling, while the rest of the class moves on to drafting — a differentiated response the whole-class pacing wouldn't have triggered without this data.

Week 3, Monday — Checkpoint: self-assessment against the task-specific clarification. Students highlight their own draft paragraph against the top-band clarification language from Article 23 and write one sentence identifying their own gap. What it reveals: Individual, specific gaps — useful for one-on-one conferencing in the final days before the summative, rather than whole-class reteaching. What the teacher does next: The teacher runs short individual conferences instead of another whole-class lesson, because by this point the gaps are too individualized for group instruction to be efficient.

Weak version of this same sequence, for contrast: three unrecorded mini-essays, same format each time, no change to instruction between them, feedback limited to a level number with no comment. Students got three attempts at what amounts to the same test. Nothing the teacher learned in week 1 changed what happened in week 2.

Apply It

🛑 Do this now. Open your own unit plan for the unit currently in progress (or the next one you're planning).
  1. Identify three points in the unit — roughly spaced across its length — where you'll check in on student progress before the summative.
  2. For each checkpoint, name the one strand or ATL skill it targets. Resist the urge to make it comprehensive.
  3. For each checkpoint, write the specific teaching decision it will inform — not "I'll see how they're doing," but "if fewer than half the class can do X, I will do Y in the next lesson; if most can, I will move on to Z." Write both branches. A checkpoint without a stated decision is just data collection.
  4. Decide, for each checkpoint, whether it will involve self-assessment, peer-assessment, or teacher-only checking — and whether it will be low-stakes (no grade attached) or informally scored for your own planning only.

Artefact to produce: A three-checkpoint formative map for your current unit — checkpoint, target skill/strand, what it will reveal, and the specific decision each outcome triggers.

Avoid This

Running a checkpoint and not changing anything afterward. Symptom: your lesson plans look identical whether the checkpoint went well or badly. Fix: write the "if this, then that" decision before you run the checkpoint, so you're forced to act on the result.

Making every checkpoint comprehensive. Symptom: your formative tasks look like mini versions of the whole summative. Fix: target one strand or skill per checkpoint.

Grading formative work as if it were summative. Symptom: students panic about a low mark on a practice task, or you find yourself averaging formative scores into a final level. Fix: keep formative feedback focused on descriptive comments and next steps, not levels that follow students to their report.

Treating formative assessment as extra work bolted onto the unit. Symptom: checkpoints feel like an add-on you skip when time is tight. Fix: build checkpoints into your unit planner from the start (Chapter 03), not as an afterthought once teaching begins.

Check Yourself

Pick your most recent classroom check-in — quiz, discussion, exit ticket, whatever it was. Can you state, in one sentence, what specifically changed in your next lesson because of what you found? If you can't, that's the gap this article is asking you to close.

Continue Learning

Next step → Article 25 closes the loop: what you do with everything a formative sequence and a summative task have told you, in the feedback you give and the level you defend. Downloadable resource: Formative Sequence Map. 📌 Official source: for the current definition and role of Approaches to Learning (ATL) skills in your subject, consult your subject guide and current MYP: From Principles into Practice guidance.