Scoring Guide¶
This page contains criteria and scales, not worked answers. Read it before an assessment so the evidence standard is transparent.
Three separate scales¶
| Instrument | Scale | Maximum | What the score describes |
|---|---|---|---|
| Entry and exit diagnostics | Eight competencies, each 0–3 | 24 | Judgment across eight different situations |
| Chapter checkpoint | Four task dimensions, each 0–3 | 12 | Quality of one bounded performance artifact |
| Capstone | Seven evidence areas, each 0–2 | 14 | Coverage of one integrated project |
Do not convert one scale into another or average the three totals. The entry and exit diagnostics are a paired measure: they use different situations but the same eight competency anchors. Chapter and capstone scores supply additional evidence, not replacement diagnostic scores.
Diagnostic competency scale¶
Use the detailed competency anchors.
| Level | Meaning | Observable pattern |
|---|---|---|
| 0 — Not yet evidenced | Missing, unsafe, or unrelated | No usable decision or artifact; major risk ignored |
| 1 — Emerging | Names part of the idea but cannot yet apply it reliably | Generic advice, incomplete steps, or verification by appearance alone |
| 2 — Capable | Applies the competency to the situation with relevant evidence | Bounded decision, workable artifact, appropriate check, and material risks addressed |
| 3 — Adaptive | Explains tradeoffs and adjusts for stakes or failure | Multiple evidence types, explicit limitations, recovery or escalation, and a justified alternative |
Chapter checkpoint rubric¶
Score each dimension from 0–3. Maximum: 12 points.
| Dimension | 0 — Missing or unsafe | 1 — Emerging | 2 — Capable | 3 — Adaptive |
|---|---|---|---|---|
| Judgment and framing | No relevant outcome, or material risk ignored | Goal stated but scope, user, evidence, or tradeoff is vague | Bounded outcome, observable criteria, relevant constraints, and justified choice | Ambiguity and stakes are surfaced; alternatives, stop conditions, and portability are reasoned |
| Execution and artifact | No usable artifact | Partial artifact or result depends on author coaching | Reproducible artifact satisfies the chapter evidence standard | Artifact handles an edge case and can be used or reviewed by another person without coaching |
| Verification and safety | No check, or unsafe data/action | Appearance or agent self-report is the main check | Independent check plus relevant data, permission, provenance, and human-review controls | Seeded failure, limitation, recovery/escalation, and proportionate responsible-use evidence are shown |
| Reflection and transfer | No reflection or unsupported claims | Describes activity rather than learning | Records feedback/failure, revision, limits, and one transferable principle | Compares alternatives, calibrates claims to evidence, and defines a specific next experiment |
| Total | Interpretation |
|---|---|
| 0–3 | Rework before continuing if the gap involves unsafe action or data. |
| 4–7 | Partial evidence. Revise the lowest dimension and ask for review. |
| 8–10 | Capable chapter performance. Continue and carry the noted limitation forward. |
| 11–12 | Adaptive evidence for this bounded task. Test in another domain before claiming transfer. |
A task with 0 in Verification and safety cannot pass, regardless of total. For Chapters 3, 8, 9, and 11, a score below 2 in that dimension means the next attempt remains disposable, synthetic, private, and supervised.
Capstone scale¶
The capstone rubric scores seven evidence areas as 0 = missing, 1 = partial, 2 = evidenced, for a maximum of 14. A capstone with any zero needs revision; its total is not converted to a diagnostic level or chapter score.
Updating a competency profile¶
Each chapter has primary competencies and reinforced competencies in the course map.
- Evaluate the checkpoint with the four-dimension rubric above.
- Record directly observed checkpoint evidence under every primary competency.
- Record reinforced evidence only when the artifact actually demonstrates the corresponding competency anchor; name the evidence and reviewer rationale.
- Keep chapter evidence beside, rather than averaged into, the entry/exit scores.
- Use the paired exit diagnostic for the formal before/after comparison.
When evidence conflicts, keep both observations and identify the next task that could resolve the difference.