Rubric-based feedback with the marker's name on it
Rubric-based feedback ties every comment to a criterion the marker scored, so a learner can act on it or contest it. How it is drafted, edited and released.
Rubric-based feedback is feedback in which every comment is tied to a criterion of the rubric the work was marked against, so a learner can act on it and a panel can check it. On NUADU it is drafted after marks are signed off, anchored to the criteria the marker scored, edited by the marker, and released with the marker's name attached and a route to appeal. For programme directors and heads of assessment, the thing to remember is short: a comment with no criterion behind it can be neither contested nor defended.
What is rubric-based feedback?
It is feedback written criterion by criterion: for each criterion the work was scored on, what met the descriptor, what fell short, and what would move the response up a level. The rubric supplies the vocabulary, so the comment and the score describe the same thing.
Most feedback on summative assessment arrives as a paragraph of goodwill instead. "Good effort, but the argument needs work." The learner cannot tell which criterion that sentence refers to, the marker cannot show how it connects to the score, and an appeals panel cannot test it against anything. It is kind. It is not evidence.
On NUADU the rule is structural: one rubric serves authoring, marking and feedback. An item written against criteria and marked against them is explained against them too. Writing those criteria well is a discipline of its own, covered in how to write a rubric that markers can apply; this post starts once the marks exist.
Generic praise cannot be appealed. A criterion-anchored comment can.
How do you write feedback for a student?
Write it against the criteria, in your own voice, after the mark is settled. In practice:
- One comment per criterion the response was scored on, not one paragraph for the whole piece.
- Each comment says what the response did and what the next level asks for.
- The comment matches the score the marker signed, including any criterion where a draft score was overridden.
- The marker's name goes on it, so the learner knows who stands behind it.
The order matters. Feedback written before the marks are final has to be rewritten when a score moves; feedback written after sign-off describes the decision that was actually made. NUADU's signed loop fixes that order: marks are signed off first, and feedback is released afterwards — rubric-anchored, marker-approved.
Can AI write feedback for students?
It can draft feedback for the marker; it should not write to the student. NUADU's Feedback assembler is a workflow assistant triggered when marks are signed off. It drafts rubric-anchored feedback for the marker to edit before release — designed to keep the marker's voice — and its output goes to the marker and nobody else. The one thing it never does is invent criteria outside the rubric.
That boundary answers the usual worry about AI-generated feedback on writing. Any comment in the draft maps to a criterion the institution wrote. The marker can rewrite it, add a note the descriptor does not capture, or cut a line that does not sound like them. Release is the marker's act: nothing reaches the learner until the marker approves it, and the approval is logged. The marking side of the same workflow — draft scores, notes per criterion, overrides — is set out in AI-assisted marking against your rubric.
How do you give feedback on assessment results?
Release it as part of the result: connected to the criteria, under a name, with a way to contest it. What the learner receives on NUADU:
- Feedback connected to specific rubric criteria, not a free-standing paragraph.
- The marker's name attached to the result.
- The criterion-by-criterion result alongside the marker's notes.
- A documented route to contest the grade through the institution's appeals process.
When a candidate disputes a grade, the feedback is often the first document everyone reads. The institution's appeals process applies, and the audit trail is there for the review: each AI output, each override, each sign-off. The marker can override any AI score. The candidate can contest any released grade. The audit trail supports both. A name on the result says who decided; a name on the feedback says who explained the decision and stands behind it.
A result can also open a practice route, if the institution has switched that assistant on and the learner opts in. It is drafted from approved content, addresses the criteria where the learner scored lowest, and stays a suggestion: the learner chooses whether to follow it, and it is never a label. The learner-controlled side of that is the subject of an AI tutor for higher education.
Three lines for the procurement review
Three lines belong in the file:
- No feedback is released to a learner without the marker's approval.
- The assistant never invents a criterion that is not in the rubric.
- In institutional deployments, no decision about a named learner is taken solely by automated means: the marker reviews, may override and signs every result, and the platform records who decided what and when. NUADU's AI Transparency Notice draws that line under Article 22 GDPR.
All of it also works with every AI capability switched off: the marker writes the feedback against the same criteria and releases it under the same name. The assistant shortens the drafting. The judgement, the voice and the name stay with the marker. The Grade pillar on marking, rubrics and moderation follows the workflow from draft score to signed result, and NUADU's AI capabilities set out what each workflow assistant does and never does.