Checking submissions against the assignment brief and rubric

A marker with eighty scripts is doing two jobs at once: checking whether each essay has the length, style and sources the brief asked for, and judging whether it does what the task asks. Done in one read, the checking is uneven and the judging starts from a blank page. To mark essays faster without marking them worse, separate the two.

This guide, published by EdCitation, shows how to check submissions against the assignment brief as university marking policy and the research on rubrics and feedback suggest, with a real MIT brief worked through. EdCitation's free Check your paper reads a brief into a checklist, each rule beside the sentence it came from. Theoretics QA, part of Max, reads a submission against that brief and quotes the paper for every finding. No tool here gives a grade: the academic judgement stays with you. The sources were read on 23 September 2026.

What does marking against the brief mean?

Marking against the brief means judging each submission by what its own brief and published criteria asked for, not against the rest of the pile or a standard the students never saw. The University of Bristol's code of practice ties brief and marking to the same criteria: they are meant to make expectations clear within assessment briefs (University of Bristol, n.d.-a), and a programme's criteria must be precise enough for consistency across candidates and markers while leaving room for "a proper exercise of academic judgement" (University of Bristol, n.d.-b).

Rules and criteria are two different things

A rule is met or not: 2,500 to 3,000 words, MLA, at least ten academic sources. A criterion describes quality: the perspectives are presented fairly and comparatively. Brookhart (2018), reviewing 46 studies of rubrics in higher education, draws the same line: a checklist asks a yes-or-no question of each item, a rubric describes performance at several levels. She warns that a criterion about following directions, such as "has three sources" in place of a varied set of relevant, credible sources, misdirects students' effort and mis-measures learning. So tick the rules quickly from a checklist and spend the reading time on the criteria.

Why do markers disagree, even with a rubric?

Markers disagree because a rubric fixes the words of the criteria, not how each marker reads them against a particular script.

Jönsson and Svingby (2007) reviewed 75 empirical studies: rubrics can make scoring more reliable, especially analytic, topic-specific rubrics used with exemplars or marker training, but a rubric does not by itself make a judgement valid. In Bloxham et al. (2016), 24 experienced UK assessors from four disciplines described what separated strong work from weak. They chose, ranked and scored criteria differently, and the authors concluded that at this level variability is inevitable, something universities should be honest with students about.

Sadler (2009) describes how it happens inside one marker. Markers often form a view of the whole work first and fill in the level on each criterion afterwards, and two markers can agree on the grade while disagreeing on the criteria beneath it. The cost falls on feedback organised by criteria, which is only as precise as the markers' shared reading of them.

What the research means at the desk

The rules of a brief can be identical for every marker, because they are facts about the paper. The criteria cannot, but they can be tied to evidence. A criterion marked with the passage that earned it is something a second marker can read and disagree with, which is what moderation needs; a level on a grid with nothing under it is not.

What do marking and moderation policies ask of a marker?

They ask for a mark that can be explained to three readers: the student, a moderator and an external examiner. Section 21 of Bristol's code is specific enough to plan a marking week by (University of Bristol, n.d.-b):

• Show the reasoning. Markers are encouraged to use a pro forma showing how they reached the mark, with comments that let the student, moderators and external examiners see why (21.7).
• Benchmark first. With several markers, each marks the same small random selection, five scripts for example, and the team agrees how marks follow the criteria before marking the rest (21.12).
• Moderate a real sample. As a guide, at least eight scripts or 10% of them, whichever is greater, across the range of marks, and every script when seven or fewer were submitted (21.19).
• Keep a trail. The conclusions of the moderator's dialogue with the marker are recorded, for the exam board and as evidence in any appeal (21.22).

Its feedback section adds a deadline: feedback on summative work within 15 working days of the submission deadline, with exceptions such as final dissertations (University of Bristol, n.d.-c). Policies differ; where yours says something else, yours wins.

How do I check submissions against the assignment brief?

Build one checklist from the brief before the first script, then mark every script against it, rules first and criteria second.

1. Read the brief into rules. Paste it into EdCitation's Check your paper, free with no account, or list length, sources, style, format and sections by hand.
2. Check each rule against its sentence. Untick anything misread and add anything missed; the worked example shows why.
3. Add the rubric's criteria as judgement lines, each worded as something a paper shows: "sets two perspectives against each other", not "analysis".
4. Benchmark on five scripts with the other markers, against the list, before marking the rest.
5. Mark each script rules first, met or not met, then each criterion with the passage that decides it.
6. Write one action for each criterion missed, pointing at the passage.
7. Keep the checklist, the evidence and the mark together for moderation and any appeal.

A worked example: a real MIT brief, read into a checklist

The brief is Essay #2, the research or investigative essay set by Walsh (2015) for MIT's 21W.011, Writing and Rhetoric, which MIT OpenCourseWare publishes under a Creative Commons licence. Students present different perspectives on a social or ethical issue, in a first version and then a revision.

What Check your paper read

We pasted the brief's requirements, in its own words, into the instructions reader that Check your paper runs. It returned seven rules, each with the words of the brief it came from, shortened here:

It also listed ten sentences no rule can test, among them the heart of the task: working with and comparing perspectives and sources, choosing between educating the reader and arguing one side while presenting the other fairly, and citing every fact in the history and background section.

What the marker changes before marking

Each correction shows at a glance, because the sentence sits beside the rule.

• The word count came from the wrong draft. "At least 2,000" is the first version's minimum. Marking the revision, you untick it and hold the essay to its suggested 2,500 to 3,000 words by hand.
• The count needs a judgement beside it. The rule says ten; its sentence says secondary academic sources, such as books and journal articles. Counted alone, ten newspaper pieces would pass, which is Brookhart's warning in one line.
• Three things are added by hand. No rule or listed item asks for the revision's cover letter or for its changes in bold, and in the reader's output the ban on Wikipedia appears only inside the history-section sentence, not as a rule for the whole essay.

The judgement lines, written for marking

The brief's own list of writing challenges gives the criteria. Three carry most of the mark, and each needs a different kind of evidence:

What does Theoretics QA give a marker?

Theoretics QA gives a marker a start made of evidence: each finding arrives with the passage it rests on, so you open a script with the criteria located, not a blank page. For that job it is the best tool we know of, and Sadler's observation is the reason: markers who agree on a grade often part company on the criteria, and a criterion with its quotation attached is one two markers can settle by reading the same words.

What goes in, and what comes back:

• Two uploads: the instructions (the brief, the rubric or both) and the submission.
• Every rule on the confirmed checklist marked met or not met, with the sentence of the instructions it came from.
• A reader's judgement: for each thing the instructions ask, whether the submission does it, the passage that shows it, and what is missing.
• Formatting, language, citations and references, in the same report.
• Nothing written: not a word of the submission is written, rewritten or paraphrased, and no grade is given.

It comes with Max: $24 a month, or $216 paid yearly, with the first 14 days free (see the plans). No finding of its appears here, because it could not be run for this guide.

What stays with the marker

The mark stays with you, and every judgement behind it. Bristol's code builds academic judgement into its criteria (21.6), and the guidance on AI-assisted marking from King's College London (2025) says such feedback must not independently decide marks or academic judgements, which remain with human educators. Treat the quoted passages as places to look, then decide.

The quotations help the feedback too. Carless and Boud (2018) argue that students often cannot decode comments well enough to act on them, and need the opportunity and the means to act. "The comparison is thin" is a verdict. "Your fourth paragraph summarises both positions and never sets them against each other; do that for the two strongest arguments" is something a student can do next.

How do I take formatting and references off my desk?

Hand them to checks that look things up and measure, and keep your reading for the criteria.

• Verify references, free with no account. Paste a reference list or upload a paper: each entry comes back verified, doubtful ("check this") or not found, retracted papers are flagged, and "could not check" is never shown as "not found". For a class, run each list in turn; marking reference lists faster builds a five-minute routine around it.
• References from a file, Pro at $8 a month: every reference in the uploaded paper looked up, and every in-text citation matched to the list, with the place a mismatch happens.
• Mechanics QA, Pro: formatting against the APA 7 student paper, comments and tracked changes left in the file, citations, numbers and language, each finding with the rule it broke and the fix. No model reads the paper.

Both Pro tools allow three checks a month with a free account. For an MLA brief like Walsh's, Mechanics QA's formatting rules are APA 7's, so there its use is the leftovers, the citations and the numbers.

What a run looks like

We uploaded the site's own sample paper, a short APA 7 nursing essay, to the reference check that runs on an uploaded paper. Four of its five references came back verified, each because its DOI resolved to a record with a matching title. One, an entry for a 2021 journal article by Thompson and Alvarez, was in no publisher's, registry's or library's record, and Mechanics QA made it its one must-fix finding. Five in-text citations matched five entries. That tells a marker where the hand check starts, not what happened; how to spot fabricated references in student work covers the fair next steps.

What should I check before uploading students' work?

Check your institution's rules on student data before any student's paper goes into an outside service, this one included. King's College London (2025) asks staff to use institution-approved tools that meet data protection law, and says student details and work should not go into generative AI tools without explicit permission. Other universities set their own rules: ask your data protection team first.

Some of the work involves no student data at all. The brief goes into Check your paper with nobody's name on it, and a pasted reference list carries none of the student's own writing. Whole papers are the case to clear first.

For a department, the Institution licence is the route: every student of a programme covered, integration with the LMS and sign-in, white label, training and one invoice, scoped with you on a call. Students can then run the same EdCitation checks on their drafts before they submit, when a missed rule costs least. Their side is in does my paper answer the assignment question, and reading the brief in how to read an assignment brief and rubric.

Quick questions

Can a tool mark essays for me?

No. EdCitation's tools check rules, look references up and point to the passages that matter; none gives a grade. The judgement and the mark stay with the marker.

How do I mark essays faster without losing consistency?

Build one checklist from the brief before the first script, tick the rules first, and give each criterion the passage that decides it. EdCitation's free Check your paper builds the rule half from the brief.

What is the difference between a checklist and a rubric?

A checklist asks a yes-or-no question of each item, such as a word range or a citation style. A rubric describes quality at several levels, and Brookhart's 2018 review warns against rubric criteria that only count things.

Why do two markers give the same essay different marks?

Because people read criteria differently: Bloxham and colleagues found 24 experienced assessors choosing, ranking and scoring them differently. Benchmarking on shared scripts can narrow the gap, and evidence for each criterion makes the rest discussable.

Can I upload student work to a checking tool?

Only as your institution allows. King's College London, for one, says student work should not go into generative AI tools without explicit permission.

References

• Bloxham, S., den Outer, B., Hudson, J., & Price, M. (2016). Let's stop the pretence of consistent marking: Exploring the multiple limitations of assessment criteria. Assessment & Evaluation in Higher Education, 41(3), 466–481. https://doi.org/10.1080/02602938.2015.1024607
• Brookhart, S. M. (2018). Appropriate criteria: Key to effective rubrics. Frontiers in Education, 3, Article 22. https://doi.org/10.3389/feduc.2018.00022
• Carless, D., & Boud, D. (2018). The development of student feedback literacy: Enabling uptake of feedback. Assessment & Evaluation in Higher Education, 43(8), 1315–1325. https://doi.org/10.1080/02602938.2018.1463354
• Jönsson, A., & Svingby, G. (2007). The use of scoring rubrics: Reliability, validity and educational consequences. Educational Research Review, 2(2), 130–144. https://doi.org/10.1016/j.edurev.2007.05.002
• King's College London. (2025). Guidance for AI-assisted marking and feedback. https://www.kcl.ac.uk/about/strategy/learning-and-teaching/ai-guidance/ai-assisted-marking-and-feedback
• Sadler, D. R. (2009). Indeterminacy in the use of preset criteria for assessment and grading. Assessment & Evaluation in Higher Education, 34(2), 159–179. https://doi.org/10.1080/02602930801956059
• University of Bristol. (n.d.-a). 20: Marking criteria and scales. Regulations and Code of Practice for Taught Programmes, Academic Quality and Policy Office. https://www.bristol.ac.uk/academic-quality/assessment/regulations-and-code-of-practice-for-taught-programmes/marking-criteria/
• University of Bristol. (n.d.-b). 21: Marking, moderation and anonymity. Regulations and Code of Practice for Taught Programmes, Academic Quality and Policy Office. https://www.bristol.ac.uk/academic-quality/assessment/regulations-and-code-of-practice-for-taught-programmes/marking-and-moderation/
• University of Bristol. (n.d.-c). 24: Feedback to students on their work. Regulations and Code of Practice for Taught Programmes, Academic Quality and Policy Office. https://www.bristol.ac.uk/academic-quality/assessment/regulations-and-code-of-practice-for-taught-programmes/feedback/
• Walsh, A. (2015). Essay #2 sequence: Research/investigative essay [Assignment]. 21W.011 Writing and Rhetoric: Rhetoric and Contemporary Issues, MIT OpenCourseWare. https://ocw.mit.edu/courses/21w-011-writing-and-rhetoric-rhetoric-and-contemporary-issues-fall-2015/resources/mit21w_011f15_essay2/