Release learner checks on Days 15 and 20 only. Keep this key separate from the learner pages. A neutral source reading, font change, tactile description, AAC or scribe changes access, not the content criterion. Note prompt level. First responses are formative and are not a validated scale. Do not infer a learner's ability from one missing response or from drawing/voice.
Daily core and choice evidence
| Day | Exact core/check result | A/B/C route evidence and next move |
|---|---|---|
| 11 | T1→S2, T2→J2; T3 is unmatched because S3's stem is short, T3's long; T4 mixed edge matches no baseline edge. | A table, B script and C comparison all name edge/tip/stem. If “part jagged” is treated as “entirely jagged”, reread the card text with two edge strips. |
| 12 | V1 T3: smooth→pointed→S3, then stem mismatch; V1 T4: includes jagged→rounded→J1, then mixed edge and short stem mismatch. V2 returns unmatched for both. | A flowchart and pseudocode, B ordered acted trace, C branch table all show condition→candidate→verification. If a route gives only final label, request intermediate states. |
| 13 | V1 8/10=.8=80%, V2 10/10=1=100%; gap 2/10=.2=20 percentage points in the same invented set. “All real plants” unsupported. |
A chart, B spoken report, C critique each need 8/10, 10/10, criterion and limit. If “20% increase” appears, distinguish relative change (25%) from 20 percentage points. |
| 14 | An effective order: overview→one rule→full three-trait match→unmatched/limit. T4 should not be forced into J1. | A panels, B script, C instruction card have equal accuracy and reading order. If peer test did not happen, record self-test. If “unmatched=species X”, revisit model limits. |
| 15 | See fresh check below. After-check route V1→S1; V2→J3; C says one case cannot establish universal performance. | Give the route after securing first check. If a label is given without all three fields, have learner compare a deliberate one-trait mismatch. |
| 16 | 2³=2×2×2=8; 3²=3×3=9; 8×9=72; 3²×2³=9×8=72. Draft B's two power values are wrong. |
A factors, B explanation and C edit all contain expansion and verification. If 2³=6, use three groups of 2 and distinguish repeated multiplication from addition. |
| 17 | Four-frame order from goal to split 8×9, expansion, check; complete linear description. | A print, B audio script, C tactile/text sequence prove the same 72 product. If accurate math but no overview, ask an absent reader what the first frame promises. |
| 18 | A leaves $25−$18=$7; B leaves $25−$24=$1. Under B>A choosing B forgoes A's possible immediate feedback; under A>B choosing A forgoes B's possible repeat practice. Neither is measured. |
A board, B note, C reversed ranking show price, funds left and next-best benefit separately. If “opportunity cost=$24”, use a ranking card to ask what could have been done instead. |
| 19 | Draft A is correct but underexplained. Draft B asserts 2³=6 and 3²=6, both false. Font/visual appeal does not validate arithmetic. |
A annotation, B critique, C comparison quote a particular line and show old/new version. If critique is personal, rephrase as observable text evidence. |
| 20 | See fresh check below. After-check route uses 90=2×3²×5=2×9×5=90; 2,3,5 are prime. |
Keep 90 practice after fresh check. If presentation masks an arithmetic error, require expanded factors before layout revision. |
Day 15 fresh answer and formative decisions
Item 1 (U1). Version 1: edge includes jagged→tip rounded→J1, unsafe: U1's short stem conflicts with J1's long stem. Version 2: entirely jagged→rounded→candidate J1→three-trait verification fails on stem→unmatched, safe under the given six-card criterion. A pupil may put the mismatch beside either trace if the reasoning is clear.
Item 2 (U2). Entirely smooth→tip rounded→stem long→candidate S1→smooth, rounded, long all match; output S1. Do not accept only “S1 because it looks right” as a verified trace.
Item 3. Old 7/8=.875=87.5%; revised 8/8=1=100%; gap 1/8=.125=12.5 percentage points. Supported: “On these eight invented profiles, the revision returned eight safe outputs under the stated criterion.” Unsupported: “It is 100% accurate for plants or future inputs.”
Next moves. If Item1 branch wrong, practise a two-step decision with known edge and tip. If branch right but mismatch missed, compare U1/J1 side by side with all fields. If Item2 label right without proof, insist on a three-trait verification. If Item3 arithmetic wrong, use an 8-cell strip and show one cell is 12.5%; if scope wrong, supply a new untested card and ask what remains unknown. Record independent / neutral access / content prompt / not observed separately.
Day 20 fresh answer and formative decisions
Item 1. Draft C's 2⁵=10 is false: 2⁵=2×2×2×2×2=32. Its implied 3×10=30 is true arithmetic but does not equal 96 and does not support the factorisation. Correct 96=2⁵×3=32×3=96. 2 and 3 are prime. A learner who explicitly says 3×10=30 is true but irrelevant shows strong source critique.
Item 2. P leaves $30−$22=$8; Q leaves $30−$27=$3. Both fit cap. Ranking can reasonably be P>Q or Q>P if criterion is named and speculative benefits remain possible, not proven. If P>Q and P chosen, opportunity cost is Q's possible repeat-practice benefit; if Q>P and Q chosen, it is P's possible immediate-feedback benefit. Price ($22/$27) and money remaining ($8/$3) are different quantities from the next-best forgone benefit. An internally consistent alternative ranking and choice earns credit; do not score adult preference.
Item 3 exemplar. “Here is a way to explain 96 using prime powers. 2⁵ means five 2s multiplied, which makes 32; then 32×3=96, so 96=2⁵×3. Multiply back to check before sharing it. We have not tested this wording with a younger reader.” An equivalent honest limit can refer to the unmeasured optional outreach benefits. Do not demand the exemplar's phrasing. Check overview, power meaning, product and bounded statement.
Next moves. If exponent meaning is wrong, build five equal factor cards and contrast with 5×2; if power right but product absent, multiply back. If budget arithmetic wrong, use a $30 number line. If opportunity cost is treated as money, ask what useful option was next best under that pupil's ranking. If the prose is mathematically sound but hard to follow, ask for a first sentence and a final check. Do not report uptake, outcome or classroom trial from this desk-created material.
Observation record
For each code/goal, write one evidence quote or trace, then mark I independent, A neutral access support, P content prompt, or N not observed. Note the specific next instruction, not a fixed learner type. A teacher can adapt timing, but the first check must remain genuinely fresh to support transfer inference.