Flagged exam sessions: a fair review process after an alert

A flag is a starting point

Proctoring systems flag gaze movement, background sound, a second voice, a network drop, and dozens of other events, most of which have innocent explanations. A process that sanctions on flags alone will punish candidates for a doorbell, a child, or a neighbour's radio, and will not survive the first serious appeal.

Write down, in advance, which flag combinations justify which level of review. Reviewers working from the same criteria produce consistent outcomes, and consistency is the property that makes decisions defensible when two similar cases are compared side by side.

Review the whole session, not the clip

A thirty second clip can make any candidate look suspicious. Reviewers should watch the surrounding minutes, check the session log timings, and consider the candidate's conduct across the whole exam. Behaviour that is part of a pattern tells a story. A single odd moment, in context, usually tells a smaller one.

Use two reviewers for any finding that could void a result, with the second reviewing blind to the first conclusion where possible. Agreement between independent reviewers is the strongest practical evidence that the finding is in the recording and not in one viewer's expectations.

Let the candidate answer the allegation

Before a final decision, tell the candidate plainly what behaviour is in question and invite their account, with a deadline and a named contact. Candidates explain flags credibly more often than teams expect: a carer entering, assistive technology, a medical need that should have had an adjustment in place.

Where an adjustment or a technical failure explains the flag, the remedy is usually a free resit under correct conditions, not a sanction. Recording that outcome as a process fix, rather than a candidate failure, improves the next sitting for everyone with the same need.

Record reasoning, close the loop

Every reviewed flag should end with a short written decision: what was reviewed, what was found, and why the outcome followed. That record serves the appeal, the audit, and the programme's own learning about which flags waste reviewer time.

Feed the outcomes back into flag tuning. If a category of flag never survives review, lower its weight or fix the guidance candidates get before the exam. A review process that learns keeps sanctions rare, accurate, and trusted by the honest majority.

Publish the process to candidates before they sit, in plain language: what gets flagged, who reviews it, what outcomes are possible, and how to appeal. Candidates who understand the process behave better during the exam, panic less at normal events like a doorbell, and accept adverse outcomes more readily when the steps they were promised were visibly followed. Transparency is not softness. It is what makes the strict end of the process, voiding a result for real misconduct, hold up when it is challenged.

Keep reading: Exam item banks: rotating questions without breaking fairness, Remote exam room scans: what the camera check is really looking for, and Exam accommodations in online proctoring: getting approved support to work on the day cover the nearby cases in this series.

Frequently asked questions