The viva, the defence and the next study

A viva is an examination of a document by readers who did not write it, conducted aloud, a format that also tests how easily a candidate speaks the examiner's language. The examiner must rebuild a thesis's reasoning from what the pages say.

The tenth of ten units argues that the viva examines reasoning rather than recall, closing a sequence that began with what a paper claims. An examiner needs to know why the study took the shape it did, which decisions were forced and which were convenient, and where the claim stops. A project with a dated decision record and an honest limitations section answers from the record. One that learned only to recite its results improvises. The criteria are only partly written down (Adunyarittigun et al., 2026). Experience changes how work is judged (Kiley and Mullins, 2004), and the examiner is a person chosen partly for temperament (Kiley, 2009).

Examiners weigh a few general criteria, not a rubric

Mullins and Kiley interviewed thirty experienced examiners about the judgements they make before writing their reports (Mullins and Kiley, 2002). The criteria used, the points of critical judgement, and the influence of the candidate's published work and the other examiner's views all shaped the verdict.

Holbrook and colleagues analysed the reports on 301 Australian PhD theses across all discipline areas and compared high-quality with low-quality theses (Holbrook et al., 2004). Examiners weighed many features against their expectations, and three general criteria separated the two groups. Golding, Sharmini and Lazarovitch synthesised thirty articles and identified eleven examiner practices (Golding et al., 2014). Examiners expect a thesis to pass, they want a coherent whole at the standard of publishable research, and they give instruction alongside the verdict.

The examiner is reconstructing an unfamiliar study

Tan analysed interviews with twelve doctoral examiners in Malaysia. He identified four purposes they described for the viva: gatekeeping, empowerment, dialogue, and enculturation into a disciplinary community (Tan, 2023). The examiners emphasised the thesis while seeing their own role as facilitative. The examiner must rebuild a study they did not run, and the candidate must make that rebuilding possible aloud. Three questions recur, and the record answers them: why this sample, why this analysis, and what result would have made the claim untenable.

The decision record makes the reasoning reconstructable

Consider the module body's case. Two participants into a think-aloud study, the team finds that the prompt invites commentary answering a different question. The team rewrites it, and the later sessions no longer use the same instrument as the earlier ones. That discrepancy is visible in the transcripts, so the examiner will ask about it. A dated decision log, the instrument file at each version and the date of the change answer before the question is put. Kiley and Mullins found that, compared with experienced examiners, novices leaned on institutional criteria and the summative dimension (Kiley and Mullins, 2004). They were unsure of the boundary conditions separating a very good thesis from a poor one, and they referred back to their own experience. The candidate who names the boundary supplies what the less experienced examiner is least certain about.

Naming the boundary is the answer that cannot be improvised

Bourke and Holbrook collected responses from 353 PhD and 74 research masters examiners on twelve indicators of importance (Bourke and Holbrook, 2013). The indicators spanned contribution, the literature review, approach and methodology, analyses and results, and presentation. Examiners rated their relative importance very similarly at both degree levels, and in almost the same order.

Adunyarittigun and colleagues surveyed 71 examiners and interviewed 19 of them about dissertations and oral examinations in Thailand (Adunyarittigun et al., 2026). Institutional documents set out procedures but carried no detailed rubrics. The examiners named seven criteria, among them argumentative rigour and research integrity, applied holistically. Criteria that are real but unwritten are criteria to state for yourself.

Who examines is part of what is measured

Kiley analysed interviews about the selection of examiners and found that experienced supervisors treated selection as a quality assurance step (Kiley, 2009). They protected their candidate from examiners they regarded as unsuitable, and they looked for empathy and understanding alongside high standards. Starfield and colleagues analysed one New Zealand university's examination criteria and 142 examiner reports from it through the appraisal framework (Starfield et al., 2015). The thesis is primarily what reports appreciate, but the candidate is often judged as well. The examiner is both a chosen person and a reader whose language evaluates a person as much as a text, and adjustments for a deaf, dyslexic or second-language candidate belong to what the examination measures.

The contested question: a common standard or a personal one?

The first position holds that examiners apply a shared and stable standard. Bourke and Holbrook (2013) found 427 examiners rating twelve indicators very similarly across two degree levels. Golding and colleagues (2014) distilled eleven practices consistent across thirty studies, and Adunyarittigun and colleagues (2026) found seven criteria applied holistically. The second holds that the examination is substantially a property of the examiner. Kiley and Mullins (2004) found novices weighing criteria and the summative dimension differently, and Kiley (2009) found examiners selected partly for temperament. Starfield and colleagues (2015) worked within one institution's criteria, and Tan (2023) found examiners describing four purposes for the same event.

What would settle it is a study that holds the thesis fixed and varies the examiner. It would take one corpus of theses and examiners of differing experience and discipline, each examining the same work against a rubric registered before the exercise. It would measure both their writing and their recommendation. No existing study does this, since each holds one of the two variables still.

Read by kind, the common-standard side rests on several hundred examiners rating the same twelve indicators and on a synthesis of thirty studies, while the personal-standard side rests on interviews with novice examiners and on interviews about how examiners are chosen. Both are reports about practice, and the study that would settle the question, one corpus of theses examined by several examiners under a registered rubric, has not been run. While the dispute is open, a project should keep a dated decision log, write its limitations as answers rather than admissions, and rehearse the three questions the record answers.

A related disagreement concerns whether the examination measures the work or the author. Tomkins, Zhang and Heavlin (2017) had every submission to a selective 2017 conference scored by two single-blind and two double-blind reviewers. They found that single-blind review conferred a significant advantage on papers with famous authors and high-prestige institutions. The conference was Web Search and Data Mining, which accepted 15.6 per cent of submissions. Each paper went to four committee members drawn from two disjoint pools, two with author information and two without. The reviewers with author information bid for 22 per cent fewer papers and preferred those from top institutions. They were more likely to recommend acceptance for papers by famous authors, with an odds multiplier of 1.63. They were also more likely to recommend papers from top universities and top companies, with multipliers of 1.58 and 2.10. Squazzoni and colleagues (2021) analysed 145 journals covering about 1.7 million authors and 740,000 referees. They found manuscripts by women treated even more favourably, so peer review did not penalise them. Lee and colleagues (2013) found that the limitations of much work on bias raise questions about the existence and extent of many hypothesised biases. Neither measurement settles the other's question, because the two differ in venue, field and design. A multi-field randomised comparison with registered outcomes would.

The strongest objection to this unit's thesis is this. If the criteria are stable, as the first position claims, the written thesis decides the outcome, and the viva merely confirms it. The reply is that coherence and argumentative rigour are tested in speech, not by the presence of a section (Holbrook et al., 2004; Golding et al., 2014).

What the next study would be

A defence ends with the examiner's next question, usually about what the study could not do. Tan (2023) found enculturation among the viva's purposes. The closing section should name the study that would test what this one could not.

Where this is heading

The direction of travel is towards examinations whose criteria are published and whose purposes are stated. Tan (2023) argues for making the purposes of the oral examination explicit. Adunyarittigun and colleagues (2026) conclude that stating the criteria to candidates may demystify assessment. Researchers now study examiner reports as a genre rather than a secret (Starfield et al., 2015). Whether a published rubric improves the examination or narrows it is open.

The sources this unit rests on

Each note states what the cited work did and found and what it adds to the argument. Identifiers were checked against Crossref, OpenAlex and the publisher record on 31 October 2026.

• Mullins, G. and Kiley, M. (2002). 'It's a PhD, not a Nobel Prize': How experienced examiners assess research theses. Studies in Higher Education, 27(4), 369-386. doi:10.1080/0307507022000011507 Interviewed 30 experienced examiners about the judgements they make before writing their reports, finding that the criteria used and the influence of the candidate's published work and the other examiner's views shaped the verdict. Establishes that examination has stages rather than being one reading of a text.
• Kiley, M. (2009). 'You don't want a smart Alec': Selecting examiners to assess doctoral dissertations. Studies in Higher Education, 34(8), 889-903. doi:10.1080/03075070802713112 Analyses interviews about how examiners are chosen and finds that experienced supervisors treat selection as quality assurance, protecting the candidate from unsuitable examiners and looking for empathy and understanding alongside high standards. Shows that the person who examines the work is chosen with the candidate's welfare in view.
• Kiley, M. and Mullins, G. (2004). Examining the examiners: How inexperienced examiners approach the assessment of research theses. International Journal of Educational Research, 41(2), 121-135. doi:10.1016/j.ijer.2005.04.009 Reports interviews with 26 novice examiners and supervisors. Found novices leaned on institutional criteria and the summative dimension, were unsure of the boundary conditions separating a very good thesis from a poor one, and referred back to their own experience. Shows the standard is partly a property of the examiner.
• Holbrook, A., Bourke, S., Lovat, T. and Dally, K. (2004). Investigating PhD thesis examination reports. International Journal of Educational Research, 41(2), 98-120. doi:10.1016/j.ijer.2005.04.008 Analysed the reports on 301 Australian PhD theses across all discipline areas and found that examiners weighed many features against their expectations, with three general criteria separating high-quality from low-quality theses. Shows that the verdict rests on a few general criteria, which makes a coherent account of the work decisive.
• Golding, C., Sharmini, S. and Lazarovitch, A. (2014). What examiners do: What thesis students should know. Assessment and Evaluation in Higher Education, 39(5), 563-576. doi:10.1080/02602938.2013.859230 Synthesised thirty articles and identified eleven examiner practices, including the expectation that a thesis will pass and the demand for a coherent whole at the standard of publishable research. Establishes that examination is partly formative, since examiners advise as well as judge.
• Tan, W. C. (2023). Purpose-driven oral examination: Insights from doctoral viva examiners. Discover Education, 2(1), Article 52. doi:10.1007/s44217-023-00083-6 Analysed interviews with twelve doctoral examiners in Malaysia and identified four purposes they described for the viva: gatekeeping, empowerment, dialogue, and enculturation into a disciplinary community. The examiners emphasised the thesis while seeing their own role as facilitative, which supplies the unit's account of what the examination is for.
• Adunyarittigun, D., Byram, M., Sriharuksa, K., Vathanalaoha, K. et al. (2026). Examiners' perspectives on assessment criteria for doctoral dissertations and oral examinations in English language education in Thailand. Journal of Studies in the English Language, 21(2), 56-88. doi:10.64731/jsel.v21i2.290735 Surveyed 71 examiners and interviewed 19 about dissertations and oral examinations in Thailand, finding institutional documents without detailed rubrics and seven criteria, among them argumentative rigour and research integrity, applied holistically. Shows that the criteria a defence is measured against are real but partly tacit.
• Starfield, S., Paltridge, B., McMurtrie, R., Holbrook, A., Bourke, S. et al. (2015). Understanding the language of evaluation in examiners' reports on doctoral theses. Linguistics and Education, 31, 130-144. doi:10.1016/j.linged.2015.06.004 Analysed one New Zealand university's criteria and 142 examiner reports from it through the appraisal framework. Found the thesis is primarily what reports appreciate, but the candidate is often judged too, so a report's language evaluates a person as well as a text.
• Bourke, S. and Holbrook, A. P. (2013). Examining PhD and research masters theses. Assessment and Evaluation in Higher Education, 38(4), 407-416. doi:10.1080/02602938.2011.638738 Collected responses from 353 PhD and 74 research masters examiners on twelve indicators of importance covering contribution, the literature review, method, analyses and presentation. Found them rated very similarly at both degree levels and in almost the same order, which is the evidence behind the strongest counter-argument to the unit's thesis.
• Tomkins, A., Zhang, M. and Heavlin, W. D. (2017). Reviewer bias in single- versus double-blind peer review. Proceedings of the National Academy of Sciences, 114(48), 12708-12713. doi:10.1073/pnas.1707323114 Scored every submission to the 2017 Web Search and Data Mining conference, which had a 15.6 per cent acceptance rate, with two single-blind and two double-blind reviewers. Found single-blind review conferred a significant advantage on famous authors and high-prestige institutions, so identity changes how a piece of work is judged.
• Squazzoni, F., Bravo, G., Farjam, M., Marusic, A., Mehmani, B. et al. (2021). Peer review and gender bias: A study on 145 scholarly journals. Science Advances, 7(2), Article eabd0299. doi:10.1126/sciadv.abd0299 Analysed peer review data for 145 journals covering about 1.7 million authors and 740,000 referees, reconstructing three possible sources of bias. Found manuscripts written by women as solo authors, or coauthored by women, were treated even more favourably, concluding that peer review did not penalise them, which sets a limit on the bias argument.
• Lee, C. J., Sugimoto, C. R., Zhang, G. and Cronin, B. (2013). Bias in peer review. Journal of the American Society for Information Science and Technology, 64(1), 2-17. doi:10.1002/asi.22784 Reviews research on bias in peer review and its empirical, methodological and normative claims. Found that the limitations of much of this work raise questions about the existence and extent of many hypothesised biases, which makes the unit's contested question genuine rather than settled.