Use discussion to open the work, not interrogate the student
An interim viva is most informative when tied to real releases. The student explains why a question narrowed, a source was selected or a method fits the data. Staff can observe understanding and transfer. A surprise knowledge test detached from the announced competence serves a different purpose.
The discussion is not standalone proof of authorship. Anxiety, disability, additional-language use, communication conventions and circumstances affect immediate answers. A rehearsed presentation can also conceal limited understanding. Read the conversation alongside documentary evidence and scholarly decisions.
If summative, the viva needs an approved basis, transparent weighting and comparable conditions. If formative, later uses of its record should still be stated. Institutions differ; the relevant regulations and authorised process govern local implementation.
Ask about decisions rather than matching wording
Invite the student to explain one choice and its rejected alternative. Useful follow-ups change a parameter, introduce contrary evidence or ask for a source location. This moves beyond repeating a stored talk while remaining inside the announced subject area.
Avoid traps and comparisons between spoken and written style. People express the same thought differently. The useful question is whether concepts, evidence and conclusion cohere. A student may accurately say that a date is in the analysis log rather than guess from memory.
For team projects, separate individual and joint work. Each student explains their decisions and interfaces. Shared knowledge does not prove identical contribution; different specialisms do not prove absence.
Questions should not seek confidential participant identities or require students to expose protected data. A data inventory, method description or controlled demonstration may supply the relevant evidence. Agree alternatives before the session.
Use a short observable rubric
Possible criteria include disciplinary accuracy, quality of reasoning, source connection, handling of uncertainty and transfer to the follow-up. Write performance descriptions for each. Rhetorical polish belongs in marking only where communication is itself an outcome.
Use more than one area and at least one transfer prompt so a single moment does not dominate. Keep duration and permitted materials comparable. Calibrate assessors with sample answers and record evidence rather than impressions such as “seemed unsure”. A second assessor can support high-stakes use where the approved process provides for one.
Arrange reasonable adjustments and accessible formats in advance. Extra time, visible written prompts or assistive technology must not be interpreted as weaker independent work. Language and translation rules remain explicit.
Keep a factual record rather than total surveillance
The short record identifies date, participants, purpose, prompts, core responses, evidence references, feedback and next actions. Where marked, include criteria and reasons. Audio or video is not automatically necessary and introduces additional legal, technical and educational questions.
Give the student the record or an authorised review route and allow factual corrections. Date later additions. Institutional rules determine access, retention and deletion. Do not upload the discussion record to an external AI service without an approved arrangement.
A later integrity process must not retroactively redefine the viva. It establishes what the student explained and showed in this setting. It cannot demonstrate every step in the dissertation's creation.
Start narrowly within formal authority
- Define whether the aim is formative or summative.
- Confirm approval, weighting and reassessment rules.
- Publish areas, duration, permitted material and rubric.
- Prepare staff on examples and fairness risks.
- Arrange privacy, accessibility and access to the record.
- Review clarity and workload after delivery without inventing effectiveness claims.
Before wider use, review whether the timing genuinely supports learning. A viva too early has little evidence; one placed after completion cannot guide the work. The programme documents observed implementation problems rather than attributing all grade changes to the new format.
The process portfolio supplies working releases. The redesign matrix shapes follow-ups. More models sit in the higher education practice hub. The sole foundation route is the guide to verifiable academic evidence.