Lock down the material and applicable standard
Consistency starts with case identity, not the percentage displayed on a dashboard. Record the submitted file, version, submission time, report date and known settings. Exclusions for quotations, bibliographies or short matches affect what the reviewer sees. If a setting is unavailable, label it unknown rather than reconstructing it from the result. Preserve the original report and annotate a working copy.
Next identify the standard that applied: the assignment brief, communicated referencing style, module material and any relevant examination rule. A clarification published later should not silently become the earlier standard. The review first describes evidence and academic relevance. Only the authorised body can decide whether a formal breach occurred or which consequence is available under the institution’s process.
Make the second view genuinely independent
Reviewer B should not merely approve Reviewer A’s conclusion. For the selected difficult matches, they need access to the same primary material without being steered towards a result. Both reviewers use shared field definitions, while retaining short written reasons. Labels such as “minor” or “relevant” are not self-explanatory and should never stand in for reasoning.
The appropriate second reviewer depends on subject, organisational structure and local authority. It might be another assessor, a trained academic-integrity adviser or a properly constituted panel member. Conflicts of interest and access permissions require attention. Adding a reviewer does not justify distributing student work more widely. Each person receives only the material needed for the defined role.
Double review of every harmless standard phrase may consume resources without improving the decision. An institution might instead select disputed source attributions, passages central to a method or conclusion, or cases with potentially serious consequences. The selection approach should be settled before an individual outcome is known. That protects the second review from becoming a tool used only when somebody wants a particular answer.
Preserve disagreement instead of averaging it away
Different conclusions reveal uncertainty in a rule, source or relevance judgement. They are useful evidence, not a failed review. Keep both explanations, then ask whether additional material can resolve the point: the whole chapter, an original edition, an earlier draft, assignment guidance or the student’s factual account of how the passage was produced.
Running the software again is not an independent academic review. A different database, setting or product release can change the report while leaving the substantive question unanswered. If a rerun is necessary, give it a separate identifier and preserve the first report and reason for the change. Equally, agreement between two people does not prove intention. It records convergence on an academic interpretation of specified evidence.
If reviewers cannot resolve the difference, refer both positions through the established route. Do not invent a casting vote or improvised majority rule. The responsible institutional body applies the real procedure. “Unresolved on the available evidence” is more accurate than forcing a confident category which the record cannot support.
Close a proportionate and reviewable case record
The final note lists the case identifier, matches examined, sources consulted, applicable standard, each reviewer’s reasoning, resolved differences, open questions and referral decision. Exculpatory evidence is recorded alongside concerns. Avoid retaining large copies of third-party works when a bibliographic record, precise location and necessary extract will support review.
The affected person receives information and an opportunity to respond in accordance with the applicable process. The checklist cannot replace that opportunity. If an error is corrected, consequential records and communications should be corrected consistently. Institutions may use carefully de-identified categories to improve training, but a selected set of cases cannot support an invented general error rate.
The Higher education practice hub connects related institutional workflows. The guide to evaluating a similarity report develops the match-level analysis. Exactly one foundation page is linked: academic integrity foundations.
Calibrate the template before live use
A team can trial the fields on a small set of synthetic or explicitly authorised examples. The purpose is not to manufacture an agreement benchmark. Reviewers instead compare how they understood each field, whether they located the same source context and whether their written reasons can be followed by somebody else. Divergent interpretations prompt a clearer definition or worked example.
The checklist itself carries a version, date, owner and change note. A later template does not overwrite earlier case records. Periodic case sampling may reveal missing fields, excessive copying or conclusory language. Those observations support training and procedural refinement; they do not automatically reopen completed matters. Used with these limits, a second view can strengthen traceability without pretending to deliver mathematical certainty or a compulsory process for every university.