To find out how a pupil is reasoning, ask them to represent one taught idea, explain how the representation works, then test it against a changed case. Keep the first explanation before correcting it. Compare the prediction with what follows and invite a revision. A right answer alone can conceal a fragile rule; an unclear drawing can conceal sound reasoning.
Draw, explain, test is an editorial planning routine for teachers, assessment leads and SENCos in England. It isn't a validated assessment instrument or a programme with a measured effect. Use it for a specific classroom decision about mathematical or scientific understanding.
One fractions claim, three pieces of evidence
This worked record is entirely synthetic. Ms Patel, the class teacher and reviewer, is a fictional owner; no pupil, school or customer is described. Her question is whether a pupil understands equal parts of one whole well enough to interpret fraction diagrams.
The mathematics programme of study connects representations, reasoning and discussion. The EEF's mathematics guidancerecommends purposeful representations and assessment that investigates pupils' understanding. Neither source validates this particular three-part routine.
Draw: Show one quarter of a rectangle. Make a second representation that looks different but still shows one quarter of an equal-sized whole.
Explain: Show how you know the shaded part is one quarter. What must be true about the parts and the whole?
Test: A new equal-sized rectangle has four unequal regions, with one shaded. Must the shaded part be one quarter? Explain what you would check or change.
The pupil makes two correct equal-part drawings. Their explanation refers only to counting four pieces. That leaves uncertainty: did they deliberately preserve equal area, copy a familiar diagram, or overlook the condition that matters? Ms Patel records the explanation before offering a hint. Dylan Wiliam's rich-questioning articleexplains why a correct response can fit several different conceptions.
In the changed case, four regions alone are insufficient: the shaded area must be checked against the whole. Equal-area partitioning makes the original quarter representation clear; unequal regions require further reasoning. Ms Patel invites the pupil to repartition the rectangle or compare areas, then keeps the revised model beside the first explanation. These examples check the claim in specific cases. They don't constitute a general mathematical proof.
Keep the reasoning demand accessible
Offer prepared shapes, movable pieces, pointing, speech, an established communication aid or a teacher-scribed explanation. Preserve the mathematical question across response modes. Don't rate drawing fluency, handwriting or speed when the intended evidence is reasoning about equal area. Record who positioned pieces or supplied language, including any cue that disclosed the answer.
A missing explanation stays missing. Silence can have several explanations, and an adult's polished paraphrase can hide them all. Check the meaning with the pupil and distinguish their contribution from the adult's interpretation. These are classroom arrangements; this routine supplies no statutory test judgement. Formal access arrangements for regulated assessments follow their own assessment-specific rules.
In science, test a prediction against an observation
For a separate synthetic science task, the question is whether pupils can use a complete-loop model to predict a bulb lighting. After teaching simple circuits, ask them to draw or build a cell, bulb and connecting wires, then explain the path. Before trying it, ask what they predict will happen if a switch opens a gap in that loop, and why.
Use suitable low-voltage classroom equipment under the school's safety procedures. Compare the prediction with the observed bulb, then ask the pupil to revise the representation or explanation. If it doesn't light after the gap is closed, check the equipment and connections. One observation cannot distinguish a faulty component from every possible misunderstanding.
The science programme of study places complete circuits in Year 4, with pictorial representations appropriate before conventional symbols are introduced in Year 6. The EEF's primary science guidance recommends explaining thinking, working scientifically and assessment that informs teaching. The observation can challenge a prediction; it doesn't confirm an entire causal model or establish understanding of current and voltage.
In this science example, the same fictional Ms Patel keeps the pupil's initial prediction, observation and revision. She reviews them on 30 October 2026. An unexplained result prompts an equipment check and another observation; a revised explanation or pupil correction can change her provisional interpretation. The pupil or parent can question the account with her.
Add the teaching history before choosing the follow-up
Return to Ms Patel's fractions record. The synthetic attendance entry shows a missed lesson on equal-area partitions; the curriculum log shows the pupil has met strip diagrams but hasn't practised alternative partitions. The first response now prompts a narrower question: what happens after that teaching opportunity is restored?
Ms Patel teaches the missed comparison and plans another unfamiliar partition task in the next mathematics lesson on 28 October 2026. She retains the usual response support and records any new prompts. On Friday 30 October 2026, she reviews the original representation, explanation and changed case alongside the follow-up with the mathematics lead. Attendance alone cannot establish which content was missed or explain the response; the teaching record and pupil account still matter.
Leave one linked record that can change
Keep a small annotated sample for the teaching decision. Record the purpose; source and date; task version and curriculum exposure; response conditions, support and adjustments; first representation and attributed explanation; prediction, test and revision; missing evidence; provisional interpretation; reviewer and review point. Preserve the original when an interpretation changes, adding the correction, author, date and reason as a new version.
Ms Patel marks this interpretation provisional. Independent recognition of the equal-area condition in a different task would strengthen it. A pupil correction, an overlooked prompt or evidence of unfamiliar task language could change it. The pupil can explain a mismatch; a parent can contact the class teacher with relevant context; another professional can ask the teacher or subject lead to revisit the evidence. Record disagreement and the response.
This record cannot diagnose, infer effort, assign a fixed label, automatically group pupils, trigger a referral or justify a high-stakes recommendation. Later improvement cannot establish that the routine caused it: teaching, attendance, practice effects, maturation, support, task differences and measurement error may contribute. Small-cohort patterns may be unstable and identifiable. Suppress or aggregate where needed and avoid precise percentages that suggest the pattern generalises.
Where the relevant school access is enabled, Student Radar's Question-Level Analysis shows topic and strand patterns from recorded paper-test diagnostics. Check the underlying paper and coverage: its analysis is school-wide, and subject, date-range and question-set filters aren't available. Digital Markbook can keep follow-up outcomes and evidence near class summaries. Keep the linked record above in the school's approved evidence system; these features don't supply the complete record automatically or decide what a pupil's explanation means.
For whole-class pacing, use the hinge routine. For evidence after feedback, use compare, respond, revisit. To inspect the product boundaries, request an assessment-focused walkthrough.
Sources and further reading
- National curriculum in England: mathematics programmes of study, Department for Education, published 11 September 2013; updated 28 September 2021. England.
- National curriculum in England: science programmes of study, Department for Education, published 11 September 2013; updated 6 May 2015. England.
- Improving Mathematics in Key Stages 2 and 3, Education Endowment Foundation, first published 3 November 2017; report design updated November 2022.
- Improving Primary Science, Education Endowment Foundation, first published 28 November 2023.
- Formative Assessment in Mathematics Part 1: Rich questioning, Dylan Wiliam, Equals: Mathematics and Special Educational Needs, spring 1999, volume 5(2), pages 15 to 18; original author article held by UCL Discovery. Historical questioning analysis, not an evaluation of this routine.