Guide

Exam Marking Moderation: A Practical Guide for Schools

Plan a clear internal marking moderation process with calibration, representative samples, independent review, decisions and an audit trail.

exam marking moderation, internal standardisation of marking, moderate teacher marking, assessment moderation process

Marking moderation is a structured comparison of assessment judgements to check that criteria have been applied accurately and consistently. Internal moderation helps a department find differences, agree the appropriate standard and apply any resulting change fairly across the full cohort.

For regulated qualifications, this general guide does not replace the current instructions from the awarding body or regulator. JCQ’s current coursework instructions, for example, require internal standardisation where more than one teacher marks a component and require centres to retain evidence that it took place. (JCQ)

Standardisation happens before or during marking. Teachers use shared examples and criteria to establish a common understanding of the standard.

Moderation checks marked work. A moderator reviews a sample to determine whether the criteria were applied accurately and consistently, then records the outcome and any required action.

Both matter. A department that only moderates at the end may discover a problem after hundreds of decisions have been made. A department that only discusses exemplars at the beginning may never test whether the agreed standard survived the full marking run.

Define the process before marking begins

Write a short moderation plan that answers:

  • Which assessment and cohort are in scope?
  • Which version of the mark scheme or rubric is authoritative?
  • Who is the lead marker or internal standardisation lead?
  • How will teachers calibrate?
  • What sample will be reviewed and when?
  • Will the moderator see the original mark before making an independent judgement?
  • What difference counts as within tolerance for this local assessment?
  • Who resolves disagreement?
  • How will changes be applied beyond the sample?
  • What evidence of the process must be retained?

For an externally regulated component, do not invent a local tolerance or sampling rule where the awarding body already specifies one.

Calibrate with real responses

Choose several anonymised scripts or answers spanning the likely attainment range. Include a response that exposes a known ambiguity in the mark scheme.

Each marker should apply the criteria independently before the group discussion. Compare both the final marks and the reasons behind them. A shared total can hide very different interpretations of individual criteria.

Record agreed clarifications in one controlled note. Do not create separate personal versions of the mark scheme, as they will diverge during the marking period.

Select a representative moderation sample

For an ordinary internal assessment, the sample should help detect both systematic and isolated differences. Include:

  • work from every marker;
  • high, middle and low attainment;
  • marks near important boundaries;
  • manually overridden rubric totals;
  • flagged or unusual responses;
  • a spread of questions or criteria;
  • any group whose access arrangements change how evidence is presented.

Random selection alone may miss boundary and exception cases. Judgemental selection alone may overrepresent problems. A combined sample is usually more informative.

For public qualifications, follow the sample specified by the awarding body. Ofqual’s 2026 guide explains that awarding organisations use a sufficient range of work to check whether marking is accurate and consistent. (Ofqual)

Review independently before comparing

Where the process permits, the moderator should record an independent mark or criterion decision before viewing the original. This reduces anchoring on the first marker’s number.

For each sampled response, capture:

  • original decision;
  • moderator decision;
  • difference;
  • criterion or question affected;
  • reason for the difference;
  • agreed final outcome;
  • whether the issue might affect work outside the sample.

Comments such as “too generous” are not enough. State which evidence or criterion was interpreted differently.

Distinguish systematic from isolated differences

An isolated difference may come from one ambiguous response. A systematic difference appears repeatedly: one marker may be consistently generous, one rubric criterion may be interpreted too narrowly or one alternative answer may be treated inconsistently.

The response should match the pattern.

  • Isolated case: correct and document the sampled decision if policy allows.
  • Clear rule clarification: identify every affected response and apply the clarification consistently.
  • Marker-level shift: review or adjust all work marked to that standard, not just the sample.
  • Unstable mark scheme: pause marking, resolve the scheme and recalibrate before continuing.

Never change sampled marks while leaving known comparable errors elsewhere in the cohort.

Keep an audit trail that another teacher can understand

A useful moderation record contains:

  • assessment and version;
  • date and participants;
  • sample selection method;
  • original and moderator decisions;
  • agreed outcomes and reasons;
  • action applied to the wider cohort;
  • completion check and sign-off;
  • links or identifiers for the retained evidence.

Keep student data in approved systems and retain it only for the required period. For regulated work, follow the specific retention and review rules that apply to the qualification.

Moderate the workflow as well as the marks

If differences recur, examine the conditions around marking:

  • Was the relevant mark scheme visible?
  • Did every marker use the same version?
  • Were rubric descriptors distinguishable?
  • Were markers switching constantly between questions?
  • Were manual overrides visible?
  • Did scanned pages hide evidence?
  • Was the session long enough for fatigue to affect judgement?

The consistent marking guide and rubric marking guide address these upstream causes.

Close the loop

After moderation, tell markers what changed and why. Confirm that all required updates were applied, totals recalculated and exports regenerated. Add the agreed example or clarification to the next standardisation session.

Moderation should improve future judgement, not exist only as a compliance record.

Structured moderation with Nib

Nib keeps the original script, question scores, feedback, annotations and rubric evidence connected. Structured moderation rounds can sample work, compare decisions and record an agreed outcome without passing loose PDFs and spreadsheets between markers. The purpose is a visible, defensible process in which teacher judgement remains central.

TestFlight invitation coming soon

Nib's TestFlight beta is nearly ready. Check back soon for the invitation link.