Consistent marking means applying the same published criteria to comparable evidence throughout a class set. It does not mean forcing every response into an identical pattern or removing professional judgement. It means that differences in marks can be explained by differences in the work.
Consistency is easier to protect when the marking process makes criteria, comparisons and exceptions visible.
Begin with a markable assessment
Some inconsistency begins in the assessment rather than the marking. Before the papers are sat, check that:
- each question measures the intended knowledge or skill;
- maximum marks match the amount of evidence requested;
- the mark scheme covers credible alternative answers;
- rubric criteria are distinct and observable;
- page layout makes it clear where students should respond;
- totals and grade boundaries are correct.
After the assessment, record unexpected but valid approaches instead of quietly treating them differently from script to script.
Calibrate before settling into the class set
Calibration establishes what the criteria look like in real work. Select a small sample with varied quality and, where possible, remove student identity. Mark each response against the scheme before comparing outcomes.
For an individual teacher, calibration can be a short self-check:
- Read three to five varied responses.
- Apply the mark scheme or rubric.
- Note ambiguous points and likely alternative answers.
- Revisit the first response after the sample.
- Record the interpretation you will use for the full class.
For a department, teachers should mark the same sample independently before discussing it. Agreement reached after one person announces a score is less informative than genuinely independent judgements.
Group similar decisions together
Marking one question across every student keeps the relevant criteria and maximum in working memory. It also makes it easier to notice if the standard is drifting.
Use question-by-question marking for discrete items and rubric sections. Retain paper-by-paper review for extended responses where the whole script changes the interpretation.
Whichever order you choose, avoid switching constantly. Every change of question or assessment requires the marker to reload a different standard.
Make criteria easy to consult
A mark scheme hidden in another folder will be consulted less consistently than one visible beside the answer. Keep the relevant section available, not merely the entire document.
For complex work, a well-designed marking rubric can turn broad expectations into observable criteria and performance bands. The rubric should support judgement, not create artificial precision. If a final mark differs from the calculated rubric total, record the override explicitly with a reason.
Separate a difficult case from the flow
Unusual answers can consume attention and change the standard for the next few scripts. When a response is genuinely uncertain:
- record the provisional decision if your process allows it;
- flag the answer;
- add a short note explaining the uncertainty;
- continue the first pass;
- review flagged cases together against the same references.
Grouped review allows boundary responses to be compared with clear examples. It also distinguishes one difficult answer from a problem in the mark scheme that affects the whole class.
Use feedback consistently, not identically
Comparable misconceptions should receive compatible guidance, but the wording should still fit the student’s evidence. A teacher feedback comment bank can hold stable explanations and prompts. Edit the selected comment so it points to the actual response.
Consistency does not require equal comment length. One answer may need a short correction; another may need an explanation of the judgement. Focus on equivalent usefulness, not visual symmetry.
Watch for predictable sources of drift
Order effects
The first few scripts may be marked cautiously while the final few are handled against an internal picture of the class. Recheck an early sample after the standard has settled.
Handwriting and presentation
Neat work is easier to process, but presentation should not gain credit unless the criteria require it. Zoom, rotate or use the full page before deciding that evidence is missing.
Halo effects
Prior knowledge of a student can influence expectations. Use candidate numbers or hide identity when policy permits, especially for high-stakes internal assessments.
Fatigue
Long sessions narrow attention. Work in meaningful blocks, stop at a clear question boundary and use progress tracking so a break does not create uncertainty about where to resume.
Mark-scheme memory
Once a marker feels familiar with a question, it is tempting to stop consulting the scheme. Keep it visible and check alternative-answer guidance throughout the run.
Run a consistency check at the end
Do not re-mark everything. Use targeted checks:
- compare a small sample from the beginning and end of the session;
- inspect scores immediately above and below a grade boundary;
- review every manual override and flag;
- check questions with an unusually wide or narrow score distribution;
- confirm that totals are calculated from recorded question scores;
- verify that no script or question remains unmarked.
Question-level patterns can reveal a marking issue as well as a learning issue. If almost every student received zero on a plausible question, inspect the scripts and scheme before concluding that the class did not understand it.
Extend consistency across several markers
When more than one teacher marks an assessment, agree the process as well as the scheme:
- common calibration sample;
- version-controlled mark scheme or rubric;
- shared interpretation notes;
- clear allocation of scripts or questions;
- a sampling plan for moderation;
- a named decision-maker for unresolved cases;
- a record of changes applied after moderation.
The exam marking moderation guide provides a full workflow.
Consistent marking in Nib
Nib brings the answer, question maximum, mark scheme, rubric, feedback and review state into one iPad workspace. Question-first navigation keeps comparable decisions together, while flags and explicit overrides make exceptions easier to revisit. The purpose is not to standardise teachers—it is to give professional judgement a dependable working context.