Goal/codes: Use an own question to choose a small safe follow-up, preserving prediction/result. Partial/conditional opportunities: AC9SFH01 / AC9SFI01–AC9SFI04.
Prepare: Actual earlier row if available, otherwise new labelled try/plan; adult preflight. Local preflight/access required; all physical actions adult-owned/approved.
- Child chooses · 3 min. “Would you like to keep this setup, change one thing, or ask another question?” Keeping is a decision, no requirement for a bigger/better roof.
- Predict the next try · 4 min. Read own earlier result if real; child predicts a genuinely new action/recheck from that experience. If no earlier sample, say new evidence, not a repeat of invented history.
- Try gently · 7 min. Adult positions approved parts. Verify matched conditions when claimed; otherwise name differences. Child observes/directs or accesses another safe actual property/report. One action only if sufficient.
- Compare real rows · 5 min. “Did our later try look/feel the same, different, or unclear?” Describe the accessed property/source. Don’t substitute prior fictional notebook result.
- Choose a bounded sentence · 4 min. “With this paper and this setup we noticed ___.” “We need another look” is useful. Another question is not proof of its suggested cause.
- Care · 2 min. Save real rows/interruptions as needed. Stop with the useful result, no fixed repeat count or strength test.
If not observed: keep the child’s question and access/contribution truthful. Offer a smaller welcomed route or later opportunity; pass/unavailable is not “cannot do.” Model/report can support reasoning, not invented direct sensory observation.