Case A — new rubric
5 criteria × 4 levels, 40 points total (scaled to 35% of the course grade across stages). Two criteria are process criteria — they grade the capabilities the stress test showed the original rubric couldn’t see.
The rubric
| Criterion | 10 — Accomplished | 7.5 — Proficient | 5 — Developing | 2.5 — Beginning |
|---|---|---|---|---|
| 1. Argument construction (10) | Thesis is precise and contestable; every section advances it; implications are specific and earned | Clear thesis; argument consistent; implications stated | Thesis present but general; argument drifts or restates | No sustained thesis; summary replaces argument |
| 2. Source verification & evidence integrity (10) | All five sources verified with access route and page-anchored quotation; annotations show the specific content of each source; references flawless | All five verified; annotations accurate; minor reference slips | Some sources verified; one annotation is generic or unverifiable | Sources unverified or missing; annotation pattern suggests the texts weren’t read |
| 3. Engagement with counter-arguments (10) | Strongest version of the counter-case presented, then answered with reasons, not dismissal | Counter-argument engaged and answered | Counter-argument acknowledged; answer is thin | Counter-argument absent or strawmanned |
| 4. AI-use judgment & disclosure (10) | Log is specific and honest; at least one substantive rejection or verification of AI output, explained; disclosure statement complete | Log honest and complete; judgments mostly explained | Log present but vague; little evidence of critical AI use | No log, or log contradicts the work; undisclosed AI use |
| 5. Clarity & presentation (10) | Prose carries the argument; signposting, APA 7 consistent throughout | Clear writing; minor APA issues | Readable but uneven; frequent slips | Disorganized; obscures the argument |
Stage weighting feeds these criteria: stages 1–2 are graded with checklists against criteria 2 and 4 (their stage marks convert into the course grade separately); the essay and defence are graded on all five criteria.
Alignment to the course outcome
| Course outcome element | Rubric criterion | Where the evidence comes from |
|---|---|---|
| Select and verify scholarly evidence | 2 (source verification) | Annotated source list + verification notes + defence question (a) |
| Reason from evidence | 1 (argument) + 3 (counter-arguments) | Final essay |
| Defend argumentative choices | 1 + 3 + defence | Defence questions (b) and (c) |
| Evaluate AI use critically (course AI-literacy outcome) | 4 (AI-use judgment) | Draft + AI-use log + disclosure statement |
| Communicate to academic standards | 5 (clarity) | Final essay |
Why each criterion replaced its predecessor
- Criterion 1 keeps the original “thesis & argument” intent but raises the specificity bar: “sustained, logically ordered” becomes “every section advances it”.
- Criterion 2 replaces “use of evidence (≥5 sources, APA)” — the criterion the stress test flagged as highest-risk because it rewards citation formatting. It now rewards verification, which generic AI output cannot fake without reading the sources.
- Criterion 3 keeps counter-argument engagement but requires the strongest version of the counter-case — raising the floor above generic “some argue…, however…”.
- Criterion 4 is new: it grades the honesty and judgment of AI use. An honest log full of rejected suggestions scores 10; a silent AI-polished essay scores 2.5 — not for using AI, but for hiding it.
- Criterion 5 keeps fluency as a communication criterion but is now the smallest part of the assurance story, since fluency alone is no longer diagnostic.
Gradebook mapping
| Component | Weight | Graded via |
|---|---|---|
| Stage 1 proposal | 5% | Checklist (complete/revise) |
| Stage 2 annotated sources | 10% | Checklist on criterion 2 elements |
| Stage 3 draft + log | 0% (completion) | Completion check; log informs criterion 4 |
| Stage 4 final essay | 15% | All 5 criteria (scaled to 40 pts → 15%) |
| Stage 5 defence | 5% | Criteria 1–4 in oral/written form (pass/revise/redo per question) |
| Total | 35% |