# A09 · Spatial transfer task and human scoring guide

Version 1.0 · Original draft instruments · Not yet piloted or validated

## Purpose and use

This instrument measures a particular form of orbital-model reasoning and evidential revision. It is not a diagnostic spatial-ability test. A curriculum specialist, astronomy specialist, accessibility specialist and pilot raters must review the frozen tasks before recruitment. Do not use a study score for course grades, admissions or student ranking. No completed learner responses accompany the instrument.

## Case-manifest requirements

For every case retain: case ID; training/immediate/delayed designation; planet and host identifiers; source table DOI 10.26133/NEA12; frozen response SHA-256; parameter-reference link and complete source citation; period in days; semi-major axis in astronomical units; eccentricity without units; limit flags; signed upper/lower errors; reasons for any assumptions; viewing transform and units; model/kernel version. Copying a literature figure requires its own permissions. This package includes no raw archive row.

Training and assessment cases must have disjoint host stars. Select cases from the strict uncertainty subset for evidence-update tasks, inspect original parameter papers, and freeze the final list before registration. Do not select cases using participant outcomes. Reject a case if an astronomer cannot explain a fixed circular fit, unresolved parameter contradiction or relevant model failure. Viewing orientation, phase/time origin and marker size are stipulated teaching conventions, not reconstructed observations.

## Timing and equivalent delivery

Baseline: 25 minutes before allocation; prerequisite items plus short reasoning form. Three 45-minute lessons in one week, with the same cases, narration, variable changes, numerical precision and available pause time in every condition. Immediate task: 25 minutes after lesson three. Delayed primary task: 35 minutes, 21 calendar days after the final lesson, within plus or minus three school days.

No headset is used in either assessment. Provide accessible text and diagrams, captions where spoken information is used, keyboard/switch-compatible input, non-color-only cues, magnification and an equivalent supported response route. A scribe may record dictated reasoning faithfully without prompting disciplinary content; retain the response route for audit and masking checks. The scoring target is reasoning, not handwriting or controller fluency.

## Training sequence: explicit revision condition R=1

For each of three lesson cases:

1. State a specific prediction about where observations at equal time intervals would lie. Identify the information you used and what is unknown.
2. Select a comparison that could challenge that prediction. Write what result would make you reconsider.
3. Run the model or inspect its matched tabular/diagrammatic output.
4. Describe any discrepancy between prediction and evidence.
5. Retain or revise the claim and explain why. State at least one model assumption that remains consequential.

The instructor uses neutral prompts such as “Which observation would distinguish these explanations?” and does not reveal another condition's worksheet.

## Time-matched comparison condition R=0

Use the same cases, model, manipulations and scientific explanations. Students classify diagrams, inspect the worked examples, identify the quantities in an output table and summarize the supplied explanation. Provide ordinary corrective feedback. Match writing and reflection time, without requiring a prior personal prediction followed by a documented discrepancy-and-revision commitment. Do not call this a no-teaching control.

## Delayed primary task: participant-facing scaffold

**Initial brief, 12 minutes.** You advise a team choosing observations of an unfamiliar system. Read the provided parameter card, model legend and viewing diagram. Decide whether the proposed equal-time observation schedule provides evidence for the claim “the planet moves through equal angles in equal times.” Explain your reasoning, distinguish what the diagram establishes from what it assumes, and identify an alternative explanation for an apparent speed pattern.

**Evidence update, 13 minutes.** The administrator supplies a second viewing representation, a bounded parameter scenario and an updated observation constraint. Which parts of your explanation remain defensible? Which should change? Show how at least two pieces of evidence support that decision. A thoughtful decision to retain a claim can earn full credit when justified; changing an answer merely because an update was supplied does not.

**New decision, 10 minutes.** Recommend a revised observation schedule under the changed constraint. Explain how a different viewing orientation could make the displayed path misleading, and describe what additional information you would need before treating the model as an actual sky prediction. State a limitation of the supplied uncertainty information.

The final classroom form must include the specialist-approved case card and evidence panel; this scaffold is not a complete classroom-ready validated instrument. Administrators must not improvise missing scientific information during the main study.

## Original 0–20 analytic rubric

Score five dimensions independently, each from 0 through 4. Apply the rubric to the full initial-plus-revised response, not only its polished final sentence. Technical vocabulary alone does not earn credit. Alternative correct explanations receive credit when consistent with the declared model and evidence.

| Dimension | 0 | 1 | 2 | 3 | 4 |
|---|---|---|---|---|---|
| Reference-frame coordination | No interpretable frame claim | Names a viewpoint without using it | Distinguishes object motion from one view, with a substantive error | Correctly relates two views with a minor omission | Reconciles both views and explains why projection alone cannot establish the claimed orbital motion |
| Geometry and timing | No relevant relation | Treats visual spacing as sufficient evidence | Uses period or shape correctly but conflates angular and temporal intervals | Correctly distinguishes shape, phase and timing in the supplied model | Integrates geometry and timing to justify the schedule while identifying a consequential stipulated convention |
| Evidential revision | No interpretable judgment | Changes or retains claim without evidence | Refers to one relevant observation with an incomplete implication | Uses relevant evidence to justify a coherent revision or justified retention | Traces the initial claim, discriminating evidence and final judgment, explaining which premises changed and which did not |
| Rival explanation | No alternative | Names an irrelevant alternative | Plausible rival without a discriminating observation | Plausible rival plus a useful comparison | Evaluates at least one rival against evidence and identifies what still cannot distinguish the accounts |
| Model/uncertainty limits | Treats rendering as direct observation | Generic disclaimer with no consequence | Identifies one real assumption or missing quantity | Explains how a named assumption/error bound limits a conclusion | Distinguishes reported parameters, viewing conventions and marginal uncertainty, and specifies needed information before extending the inference |

Raters must not deduct for a response route, dialect or writing style unless the disciplinary meaning is genuinely absent. If an image is inaccessible, the administration record must identify the access failure; a low reasoning score cannot be treated as evidence that the learner lacked the construct.

## Human rater procedure

Two raters independently score every de-identified response; condition, school, instructor and exposure history are masked. Training uses pilot responses outside the analysis cohort. Raters freeze anchors after pilot calibration and before viewing main-study responses. The primary score is the mean of the two total scores. Preserve both raw scores.

If any dimension differs by more than one point or totals differ by more than four, a third reader examines the disagreement for calibration and descriptive reporting. This does not selectively overwrite the primary mean. Report total-score absolute agreement and dimension-wise agreement with intervals. A separate audit checks whether response artifacts unintentionally reveal condition. Report the number and nature of masking failures.

## Synthetic worked geometry check, not a learner exemplar

A deliberately fictional case has a=1 AU, e=0.4 and a stipulated phase origin at periapsis. At eccentric anomaly zero the model position is x=0.6 AU, y=0; at eccentric anomaly π it is x=−1.4 AU, y=0. Equal increments of mean anomaly correspond to equal fractions of a period. Equal increments of plotted true angle generally do not. A rotated view may compress a displayed axis without changing the stipulated underlying orbit. These are mathematical checks of the teaching kernel, not evidence about any real learner or an unnamed actual exoplanet.

## Local symptom and access items

Before and after each lesson: “Do you currently feel nausea, dizziness, eyestrain or headache?” Record each separately as none/mild/moderate/severe, plus whether the learner wants a break or different route. These original monitoring questions are not a validated clinical scale. Any request to stop is honored immediately. Record device failure, inaccessible information, adult assistance, actual mode and reasons for switching; do not infer a diagnosis from these items.