# Separate staff key · after first capture

Never app-loaded. Observe chosen meaning/design/control/input/state and actual operator/help; paper and directing evidence remains valuable but distinct. No full-band grade/attainment or speed/neatness/speech/motor judgement. If an exact fresh arrangement/meaning/answer was pre-exposed, record that actual exposure; do not call it unfamiliar.

## E independently calculated key

Candidate states per record: SET3 thenADD2→5; SET3 thenADD4→7; SET3 thenADD1→4; SHOW4. This drops earlier contributions. Valid repair: SET3 once; FOR EACH ADD input; SHOW. States3→5→9→10; SHOW10. Another correct input-responsive algorithm is accepted. Hardcoding10 is not a general repair; a changed batch must still follow the rule. A prediction alone supports reasoning, not actual implementation. SET outside the loop with manual ADD records can work for this exact batch but evaluate whether it handles the changed batch/general criterion; do not reject a useful partial design merely for not yet generalising. E has no branch requirement and earns no multiple-alternative evidence.

## F independently calculated key

One valid design SET2; FOR EACH IF input<3 THEN ADD1 ELSE IF input<6 THEN ADD2 ELSE ADD4; SHOW. Batch[1,4,7]: totals2→3→5→9, selected outcomes first/middle/last. Changed[3,6]:2→4→8, middle/last (equality matters). Empty[] leaves2; final SHOW2. Reverse conditions or other valid structures accepted when only the intended update applies once per record and start persists. Two independent IFs can wrongly apply both increments; inspect trace rather than infer control from final number. One correct final number is not evidence of all boundaries, iteration, variable input and change/run.

## Descriptive evidence dimensions and next teaching

| Dimension | Available evidence | Useful next move |
| --- | --- | --- |
| Story/criteria | Own purpose/check connects to a design; absent evidence remains open | Revisit what the total should mean, one criterion at a time. |
| Algorithm/control | Ordered steps, loop body and alternatives reasoned; incomplete or supported work attributed | Trace one input and its enclosing body; compare another outcome. |
| Variable/input | Start/reset/add meaning and before/after state explained | Use unequal records or empty batch to separate reset from accumulation. |
| Actual implementation | Who created/entered/changed blocks, actual run and trace/output | Offer usable access/support and one meaningful actual edit; do not infer device use from diagram. |
| Evaluation | Evidence against story/criterion, useful edit and conditional possible impact | Test a revealing boundary/changed batch; keep real reader impact open until observed. |

For each dimension describe absent, emerging, supported or independently evidenced task evidence with actual help and first/later response. These are descriptive teaching notes, not a mechanical score or QCAA five-point scale. A learner may show independently reasoned design with adult entry; record both facts. Stop at sufficient formative evidence; no checklist quota. Fresh sources are unfamiliar relative to supplied ordinary material, not a claim about a real learner's prior experience.