# Teacher-only key and next teaching

Keep this separate from public check sheets. Save the child's first complete answer and participation route before prompting, correcting or using after-check swaps. These are low-stakes formative tasks, not secure or standardised exams. Passing is allowed; offer a later chance. If an adult moves cards, distinguish child-directed decisions from adult suggestions.

## Check A · Paper Display Depot

From the five-card source, SHAPE groups are **SQUARE 1, 3, 5** and **OVAL 2, 4**. Under the *different* BORDER question, **DOTTED 1, 2**, **PLAIN 3, 5**, **CANNOT TELL 4**. ID 4 is an oval with border **NOT SHOWN**. “Not shown” is unavailable data, not evidence for plain. One correct three-form match may be any ID: for example physical broad square-dotted token 1 → source drawing square/dotted → `QD`; ID 4 requires physical oval with unknown tag → picture with hidden border → `O?`. Alternatives for 2=`OD`, 3=`QP`, 5=`QP`. Duplicate symbols are expected because IDs 3 and 5 share the same stated features; ID remains a separate identifier.

Look for both rule labels, all five IDs in each grouping, unknown kept distinct, and a real **physical** stand-in plus picture plus symbol. A child can use an unambiguous alternative tactile/word key if they define it. If they move ID 4 to PLAIN because no dots are visible, revisit a different covered-border practice token after saving. If they can group shapes but not borders, change only one question at a time with source alongside. No Science observation follows from this invented source.

## Check B · Paper Tool Trial

The stated fictional outcomes group as **HELD M, N**; **SLIPPED K**; **NOT TESTED L**. L is unknown for held/slipped, not a failure. Accept a source-backed three-form link for any one ID; for example physical M token → source HELD picture/word → `H`, or physical L token → NOT TESTED picture/word → `?`. The empty category **MY VOICE RECORDING · EMPTY** could be a child's own personal data *if it contained their recording*; the neutral fictional holder symbol is not their own voice. Neither value is filled, requested, captured or shared.

If a learner places L under SLIPPED, show on another source that NOT TESTED lacks an outcome. If they call any named sample a material, remind them K/L/M/N are invented IDs with no stated materials. P01 category recognition is a narrow opportunity; do not infer consent/privacy mastery. This paper task says nothing about child device exploration. For K01, separately log actual child-controlled hardware input, display/software response and purpose under [the device gate](../DEVICE-GATE.md), or write **NOT OBSERVED**.

## Daily next-step ledger

| Day | First response to keep | If a new invitation is useful |
| --- | --- | --- |
| 21 | Source-backed badge field and honest hidden-border statement. | Mask all but one large badge, read ID/shape/border separately. |
| 22 | Shape groups and reason from source, with route. | Use two broad physical tokens before extending to five. |
| 23 | New border groups with badge 5 CANNOT TELL. | Contrast an actual dashed mark with a covered or missing field. |
| 24 | One object/picture/symbol alignment and empty own-data category. | Place a genuine physical stand-in beside its drawing, then agree code. |
| 25 | Fresh A first complete two-rule grouping and three-form match. | Practise a different source after the check, never the A source as a worked example. |
| 26 | P/Q/R/S outcome representation with R untested. | Cover all but one story row and ask what it states. |
| 27 | HELD P/S, SLIPPED Q, NOT TESTED R with source citation. | Make three large bins, including a real cannot-tell space. |
| 28 | Titled child display, legend and any actual reader response. | Ask a willing reader what mark was unclear and revise once. |
| 29 | Rule/result on paper or, conditionally, observed device action/purpose. | Choose paper if needed; plan a locally suitable actual device chance. |
| 30 | Fresh B first groups, three-form link and own-data category. | Use a new fictional outcome set on a later day. |

**Evidence boundary:** A source-backed grouping can sample K02; an empty category may sample P01. It cannot establish K01 without observed child engagement with a running system, a real living thing's feature, a holder's material property or a child's personal-data handling in everyday life.