# Draft research instruments · Version 3.0

These original instruments operationalize the proposed study. They require independent pilot testing, accessibility review, rater calibration and a frozen scoring manual before allocation. No validity, reliability or treatment effect is claimed for them.

## A. Student transfer-and-revision rubric

Each dimension is scored independently from 0 to 4. The sum is 0–20. The primary score is the mean of two independently rated totals. Assess disciplinary reasoning and evidence; do not award points for agreement with the adult, length, visual polish, number of citations, or technology use.

| Dimension | 0 | 1 | 2 | 3 | 4 |
| --- | --- | --- | --- | --- | --- |
| Problem representation | No relevant problem is represented. | Names a topic without a decision or constraint. | Identifies a decision and one relevant constraint. | Represents the decision, constraints and affected people coherently. | Justifies the framing and acknowledges an important boundary or unresolved trade-off. |
| Warranted evidence | No relevant evidence is used. | Repeats an assertion or number without explaining its relevance. | Links one piece of relevant evidence to a claim. | Uses relevant evidence accurately and explains the inference. | Weighs conflicting evidence, preserves its source and calibrates the claim to its limits. |
| Alternatives | No alternative is considered. | Names an alternative without analysis. | Gives a relevant alternative or objection. | Compares credible alternatives against explicit criteria. | Explains when the preferred account could fail and what evidence would change the decision. |
| Substantive revision | No revision or reason is evident. | Changes wording or presentation only. | Makes a relevant change with a partial reason. | Changes the substantive recommendation in response to evidence and explains why. | Traces how the new information changes the model or decision, including what remains uncertain. |
| Transfer under a changed constraint | Ignores the changed condition. | Mentions it without adapting the plan. | Adapts one relevant feature. | Revises the plan coherently for the new condition. | Generalizes a defensible principle while identifying a limit to its transfer. |

Every administration includes an initial response and a revision opportunity. Baseline and endline forms must be comparably demanding. Do not compare a first draft at baseline with a fully supported final product at endline.

## B. Illustrative task packet: a heat-resilient school courtyard

**All numbers below are invented assessment inputs. They are not observations, prices or recommendations about an actual school.** Pilot testing must examine reading demand, numerical demand and cultural accessibility.

The student advises a fictional school committee deciding how to improve a courtyard used at lunchtime. Its budget is 100 planning units. The committee values usable shade, accessible routes and responsible water use. Choose a defensible plan, explain its limits and identify what else you would want to know.

| Fictional option | Cost units | Added shaded seats | Water-use units per week | Access note |
| --- | --- | --- | --- | --- |
| Shade canopy | 65 | 40 | 0 | Requires checking clearance on an accessible route. |
| Planted shade area | 45 | 20 after establishment | 12 | Does not provide immediate mature shade. |
| Movable shade and seating | 35 | 18 | 0 | Requires a storage and safe-movement plan. |
| Path and drainage improvements | 30 | 0 | 0 | Improves a currently difficult access route. |

Initial phase: interpret the problem, compare at least two plans, use the evidence and explain a recommendation. Revision phase: the fictional school learns that its water allowance is reduced to six units per week and that an accessible route must remain usable during installation. Revise the recommendation and explain which original premises changed. Reflection phase: identify an observation or stakeholder response that would lead you to revise again.

This task cannot establish future employment readiness. It is an instrument-development example for reasoning, revision and transfer. A parallel form should use a different substantive setting with comparable reasoning demands. Students receive equivalent approved accommodations and language access across conditions; the pilot determines which changes preserve the intended construct.

## C. Teacher professional-interpretation task

Provide a de-identified, independently selected student artifact and its instructional context. Ask the teacher to identify the learning difficulty, distinguish observation from inference, give a credible alternative explanation, propose an instructional change and state what subsequent evidence would test that decision. Rate the five elements separately on a pilot-calibrated 0–4 rubric. Do not infer quality from the teacher's confidence or the number of technical terms used.

Separate raters score this task from the student primary outcome, masked to condition and school. Keep student and teacher research identifiers in different access-controlled maps. Record which context was removed during de-identification so that scoring limitations remain inspectable.

## D. Candidate organizational measures

Use the published SILS only after inspecting its documented form, referent and permissions. It measures implementation leadership; it is not a psychological-safety or Movement Thinking effectiveness scale. Do not rename or modify items and carry over the published reliability as though it were obtained in this sample.

The following newly authored prompts are a **draft team-voice measure**, with 1–5 agreement responses and a separate not-applicable option. They are not Edmondson scale items and must not be labeled a validated psychological-safety scale:

1. Our instructional team makes room for a member to explain why a current teaching approach may be failing.
2. A teacher can ask for help with an instructional difficulty without that request being used as evidence of incompetence.
3. When team members disagree about student work, they can examine the disagreement without a penalty for raising it.
4. The team considers evidence that challenges a preferred explanation, including evidence offered by a less senior colleague.
5. A member who notices an error has a usable route to change the team's next action.

Examine content coverage, ordinal distributions, factor structure, aggregation and longitudinal comparability in pilot and confirmatory work. Until those checks support a scale, report the items separately. Collective-efficacy measurement requires a separately selected, justified instrument; individual confidence items are not a substitute.

## E. Fidelity, time and community evidence

For every cycle, record scheduled and delivered paid minutes, the instructional question, participants' roles, the decision charter version, the sampled artifact IDs, the instructional decision and one observation planned to test it. Cap the artifact sample and avoid collecting material simply because storage is inexpensive.

Time diaries distinguish contracted work, paid additional work, unpaid work and research-only activities. Record actual minutes in designated windows. Community feedback records who was invited, which access supports were offered, which concerns changed a decision and which concerns remain unresolved. Attendance counts alone do not establish influence.

## F. Human and privacy safeguards

Teaching, research and public publication involve separate permissions. No student needs to publish an artifact or obtain an ORCID to receive ordinary instruction or participate in the proposed assessment. Never place identifiable artifacts or sensitive teacher responses in the public paper-agent interface. Record research consent, age-appropriate assent and publication permissions under the approved institutional process; this document does not grant them.