Guides / academic careers
PhD viva and interview questions
Two different events get confused with each other: the interview that decides whether you are admitted to a PhD, and the viva voce that decides whether you are awarded one. This guide separates them, gives the questions each one actually asks, and shows weak answers rewritten into answers that hold under follow-up.
A viva opens with a broad question, usually a summary of your contribution, then moves systematically through the thesis: why this question, why this design rather than the alternatives, what the key decisions were and how you justified them, what the findings mean, what the limitations are, and what you would do differently. Examiners are not testing recall. They are establishing that the work is yours, that you understand its boundaries, and that you can defend the choices you made without either collapsing or refusing to concede anything.
Start here
Two different events
| Admission interview | Viva voce | |
|---|---|---|
| When | Before you start | At the end, after submission |
| Decides | Whether you are offered a place and funding | Whether the degree is awarded |
| Length | 20 to 45 minutes | 90 minutes to 3 hours, sometimes longer |
| Who is there | Prospective supervisors, sometimes a panel | Internal and external examiners, an independent chair |
| Based on | Your proposal, CV and transcript | Your submitted thesis, read closely |
| Testing | Potential, fit and feasibility | That the work is yours, original and understood |
| Preparation | Know the supervisor's work and your proposal | Know your own thesis better than anyone |
The confusion matters because preparation that works for one is wasted on the other. Reading widely around your field helps an admission interview and does almost nothing for a viva, where the questions come from the pages you wrote.
Unlike the public defence common in parts of continental Europe, a UK viva normally happens in a room with you, two or three examiners and sometimes an independent chair, with no audience. Your supervisor may attend as an observer if you and the examiners agree, and if they do they cannot speak on your behalf. Check your institution's regulations, because the detail varies and knowing the format removes one source of anxiety for free.
Before you start
The admission interview
The panel is answering three questions about you, and almost everything they ask maps to one of them: can this person do a PhD, is this project feasible in three or four years, and do we want to work with them for that long.
The questions that come up most
| Question | What they are checking |
|---|---|
| Tell us about your proposed research | Whether you can explain it in three minutes without jargon |
| Why this topic? | Genuine motivation, not a gap found in a literature search |
| Why this department, and why us? | Whether you have read their work |
| What methods would you use? | Feasibility, and whether you know what the methods require |
| What is the contribution? | Whether you can distinguish new from merely unstudied |
| How would you spend the first six months? | Realism about timescales |
| What might go wrong? | Whether you have thought past the plan |
| How do you handle setbacks? | That you will still be here in year three |
| What are your plans afterwards? | Fit with what the studentship is for |
The two that separate candidates
"Why us?" is answered badly more often than any other. "Your department has an excellent reputation" is a non-answer. Name the person, name the specific work, and say how it connects to yours: "Your 2024 paper on measurement invariance across occupational groups is the closest existing work to what I want to do, and I would want to extend it to a longitudinal design."
"What might go wrong?" rewards honesty and punishes both extremes. Saying nothing will go wrong reads as inexperience. Listing five catastrophic risks reads as a project that should not be funded. Name one real risk and the response: "Recruitment through NHS trusts is the main risk, since R&D approval timescales vary. I would start approvals in month one and have a second recruitment route through professional bodies as a fallback."
Have a question ready, and make it a real one
You will be asked if you have questions. Ask about the thing you would genuinely want to know before committing four years: how the supervisory team divides the work, what training the programme provides in the methods you will need, what the cohort looks like, how progression is reviewed. Avoid asking anything already answered on the webpage, which signals you did not read it.
At the end
The viva: format and outcomes
How it usually runs
The question categories
| Category | Typical form |
|---|---|
| Contribution | What is original here? What do we know now that we did not before? |
| Motivation | Why this question? Who needs the answer? |
| Design | Why this design and not a cohort study? What would a trial have added? |
| Method decisions | Why that sample size? Why that model? Why did you exclude those cases? |
| Interpretation | Does the data support that claim? What else could explain it? |
| Literature | How does this sit against Author X, who found the opposite? |
| Limitations | What is the weakest part? What would you do differently? |
| Forward | What next? What would you publish first, and where? |
The outcomes
| Outcome | What it means |
|---|---|
| Pass, no corrections | Rare. Not the realistic target. |
| Minor corrections | The most common result. Typically weeks to three months, approved by the internal examiner. |
| Major corrections / revise and resubmit | Substantial rework, often 6 to 12 months, sometimes a second viva. |
| MPhil awarded | The work does not meet doctoral standard but merits a master's. |
| Fail | Very rare, and almost never a surprise to the supervisory team. |
Minor corrections is the normal outcome of a successful viva, and a list of corrections is not a criticism of you. Candidates who expect a clean pass sometimes read a correction list as failure, which it is not.
Worked example one
Answers rewritten
Four exchanges as they commonly go, then rewritten. The facts are the same in each pair; what changes is whether the answer survives the follow-up question.
1. "What is the contribution of this thesis?"
Weak: "No one has studied this population before, so it fills a gap in the literature."
Why it fails: an unstudied population is not in itself a contribution. The obvious follow-up is "and why does that matter?", which the answer has not anticipated.
Stronger: "Three things. The measurement work shows that the standard wellbeing scale is not invariant across occupational groups, so published comparisons between those groups have been comparing different constructs. Second, once invariance is modelled, the gap between the groups is about a third of what was previously reported. Third, the method I used to test this is now available as an R package, so other people can check their own instruments. The first is the one I would defend as the main contribution."
Three specific claims, ranked, with the consequence of each stated. Note the final sentence: nominating your own main contribution stops the examiner having to guess which one you think matters.
2. "Why did you choose this design?"
Weak: "It was the most appropriate design for my research questions."
Why it fails: it restates the conclusion as the reason, and it shows no awareness of the alternatives.
Stronger: "I considered three. A cohort study would have given me temporal order, which I wanted, but following people for the two years needed was not feasible within the studentship. A cross-sectional design loses temporal order but allowed the sample size the measurement work required, which was the priority. I chose cross-sectional and I am explicit in chapter five that the associations cannot be read causally. If I had the resources again I would run a two-wave panel, which would get me both."
This names the alternatives, states the trade-off, acknowledges the cost and says what you would do differently. It also pre-empts the limitation question by raising the limitation yourself.
3. "Author X found the opposite. How do you explain that?"
Weak: "Their sample was smaller than mine."
Why it fails: it dismisses rather than engages, and sample size rarely explains a reversal of direction.
Stronger: "I think there are two candidate explanations and I cannot fully separate them. They measured the outcome with a single item where I used the full nine-item scale, and a single item attenuates the association substantially. They also sampled a workforce with a much lower baseline rate, and the relationship may genuinely be weaker there. My data cannot distinguish those two, which is why I present it in chapter six as a discrepancy to be resolved rather than as evidence that their finding is wrong."
Engaging with a contradictory finding and saying honestly that you cannot resolve it is a strong answer. Claiming to have resolved it when you have not is where candidates get into trouble, because the next question tests the claim.
4. "What is the weakest part of this thesis?"
Weak: "I do not think there is a weak part." Or, equally weak: "The whole sampling strategy is a problem, and the sample size, and the measure, and I ran out of time on chapter four."
Why both fail: the first is not credible and invites the examiners to find one for you. The second hands them a list.
Stronger: "Chapter four. The response rate in the smallest occupational group was 41% against 73% elsewhere, so that group's estimates rest on 58 people and the invariance test for it is underpowered. I report it and I flag it, but if I were doing it again I would over-sample that group deliberately from the start rather than relying on proportional allocation."
One weakness, named precisely, quantified, with the consequence and the fix. This is the answer examiners are looking for, because the ability to locate the limits of your own work is most of what a doctorate certifies.
Worked example two
Handling the question you feared
Every candidate has one. A decision that was made under pressure, defended thinly in the thesis, and that a good examiner will find. The structure below handles it, and it also handles a question you genuinely cannot answer.
The four-move structure
| Move | What you do |
|---|---|
| 1. Concede the point | Say plainly what is true about the criticism |
| 2. Explain the constraint | Why the decision was made, at the time, with what was available |
| 3. Bound the damage | What it does and does not affect in the thesis |
| 4. State the alternative | What you would do with the knowledge you now have |
Applied
Examiner: "You dichotomised your main predictor at the median. Why would you discard that much information?"
Answer: "You are right that it costs power, and the arithmetic is not trivial: a median split multiplies the correlation by about 0.80, which is equivalent to losing around a third of the sample. I did it because the policy audience the chapter was written for works with a high and low banding and a continuous coefficient was not usable to them. What I should have done, and what I would do now, is report the continuous analysis as the primary result and the banded version alongside it for that audience. The direction and significance are the same in both, which I checked, but the headline estimate in chapter five understates the relationship and I would present it the other way round."
That answer concedes, explains, bounds and corrects. It would very likely produce a minor correction asking for the continuous analysis to be added, which is a good outcome.
When you do not know
Say so, and then do something with it.
"I do not know. My instinct is that it would attenuate rather than reverse the effect, because the error is unlikely to be correlated with the outcome, but I have not tested that and I would not want to claim it. It is checkable with the data I have, and it is the first thing I would look at."
"I do not know, and here is how I would find out" is an acceptable viva answer. Guessing confidently is not, because the examiner usually knows the answer and is now watching you construct something unsupported. Few candidates fail for not knowing something; some get into difficulty for defending a position they cannot hold.
Examiners are not trying to trap you, but they will press a weak defence until it breaks. A point conceded in the first sentence ends that line of questioning and moves the viva forward. A point defended for ten minutes and then conceded has cost you ten minutes and some credibility. Knowing in advance which three or four points you intend to concede is one of the most useful things a mock viva produces.
Mid-degree
The transfer or upgrade viva
Most UK programmes have a formal progression point in the first or second year, called a transfer, upgrade, confirmation or progression review depending on the institution. You submit a report and defend it to a panel that usually excludes your supervisor.
The question being answered is narrower than at the final viva: is this project on track to become a doctorate? Not whether it is a doctorate yet.
| Question | What they want |
|---|---|
| What is the research question now? | That it has narrowed since the proposal, which is normal and good |
| What have you done so far? | Concrete completed work, not reading |
| What is the plan and the timetable? | A realistic chapter-by-chapter schedule to submission |
| Where is the original contribution? | That you can articulate it before the work is finished |
| What are the risks? | That you have identified them and have responses |
| Is the scope right? | Often that it is too large and needs cutting |
Scope is the issue the panel raises most. Being told your project is too ambitious at transfer stage is a favour, and the right response is to engage with which part to cut rather than defend all of it. A panel that cuts your third study in year two has saved you a year.
Treat the transfer as the rehearsal it is. The questions are a subset of the final viva's, the stakes are lower, and the feedback is the most useful you will get before submission. Write it down.
Preparation
Running a mock viva that is useful
A mock with your supervisor, who already knows and likes the work, is of limited value. The useful version has three properties.
Prepare these in advance
| Item | Form |
|---|---|
| The three-minute summary | Spoken aloud until fluent, not memorised word for word |
| Your contribution, in three claims | Ranked, with the consequence of each |
| A one-page map of the thesis | Chapter, question, method, finding, on one sheet |
| Four limitations | Each with its consequence and what you would do instead |
| The points you will concede | Decided in advance, not in the room |
| Your examiners' relevant work | Read it. They will ask how yours relates to theirs. |
| Three errors you have already found | Typos and small mistakes you can acknowledge immediately |
That last one is worth the effort. Having your own list of known errors means that when an examiner points one out, you can say you had found it too, which changes the dynamic from being corrected to agreeing.
Read your examiners' relevant papers properly. It is the highest-return preparation available, because examiners tend to ask about the things they work on, and because being able to say how your finding sits against theirs is the kind of answer that is remembered.
Avoid these
Six mistakes that cost candidates
| Mistake | Why it matters |
|---|---|
| Defending an indefensible decision | Examiners press until the defence breaks. Conceding in the first sentence ends it; conceding after ten minutes costs credibility. |
| Claiming there are no limitations | Not credible, and it invites the examiners to find them for you and name them in the corrections. |
| Listing every limitation you can think of | Hands them a charge sheet. Name the real ones, with their consequence, and stop. |
| Not having read the examiners' work | They will ask how your findings relate to theirs, and not knowing reads as incuriosity. |
| Guessing confidently | "I do not know, and here is how I would check" is acceptable. A confident invention usually is not, and the examiner often knows. |
| Treating it as an interrogation | It is the first conversation about your work with people who have read it all. Defensiveness closes down the discussion you want. |
A seventh, which is procedural: not knowing your own institution's regulations. Whether your supervisor may attend, whether an independent chair is appointed, what the outcome categories are called and how long corrections are allowed. These vary and candidates are sometimes surprised by their own regulations on the day.
The questions that are hardest to answer in the room are usually the ones about a decision made under pressure in year two. A statistician reads the analysis and tells you which choices will be pressed, before anyone presses them.
See results chapter review
Elaine HalliburtonProfessor of Applied StatisticsPracticalities
The week before, and the day
The week before
On the day
If the discussion becomes genuinely difficult, that is not a signal of failure. Vivas in which the examiners argue with you at length are often the ones that pass with minor corrections, because sustained argument is what examiners do with work they take seriously.
If you want the methods chapter pressure-tested before you submit, that is what results chapter review and independent statistical review are for. Interview and viva preparation is available through the quote form, and the academic CV guide covers the stage before the interview.
Common questions
Frequently asked questions
How long does a PhD viva last?
Usually 90 minutes to three hours in the UK, occasionally longer. Length is not a signal of how it is going. A short viva can mean the examiners were satisfied early or that they had decided before it started, and a long one often means they are engaging seriously with the work.
What is the most common viva outcome?
Minor corrections, by a wide margin. Typically a list of specific changes completed in weeks to three months and signed off by the internal examiner. A pass with no corrections at all is rare, so a correction list is the normal result of a successful viva rather than a criticism.
Can I fail a PhD viva?
It is possible but very rare, and it is almost never a surprise to the supervisory team. The more common adverse outcomes are major corrections with a resubmission, or the award of an MPhil. If your supervisors have agreed you are ready to submit, outright failure is an unlikely outcome.
What should I say when I do not know the answer?
Say you do not know, then add what you would do to find out. "I have not tested that. My expectation is that it would attenuate rather than reverse the effect, but I would want to check it in the data before claiming so" is a perfectly good answer. Guessing confidently is the riskier option, because the examiner usually knows.
Should I memorise my answers?
No. Memorised answers break on the second follow-up and sound rehearsed, which invites harder probing. Prepare the three-minute summary until it is fluent, prepare your contribution as three ranked claims, and prepare your limitations. Everything else should be a conversation.
What is a transfer or upgrade viva?
A formal progression review in the first or second year, where you defend a report to a panel that usually excludes your supervisor. It asks whether the project is on track to become a doctorate, not whether it is one yet. Scope is the issue raised most often, and being told the project is too large is useful feedback rather than a setback.
How do I prepare for a PhD admission interview?
Differently from a viva. Be able to explain your proposal in three minutes without jargon, read the specific work of the people interviewing you so "why us" has a real answer, have a realistic plan for the first six months, and be ready to name one genuine risk with your response to it. Have a question of your own that is not answered on the webpage.
Can my supervisor attend my viva?
At most UK institutions, as a silent observer, if you and the examiners agree. They cannot answer on your behalf or intervene. Whether to invite them is a personal choice; some candidates find it reassuring and others find it adds pressure. Check your institution's own regulations, since the detail varies.
Related services
If you want it read by someone who sits on panels
Academic CV review
Your CV read against the specification, marked up, £95 fixed.
£95 fixedResults chapter review
A statistician checks the chapter you wrote before it is submitted.
Fixed quoteIndependent statistical review
A fixed-price check of analysis already done, before it reaches a journal or a board.
Fixed quoteEvery piece of work is done and signed by a named statistician. Send the data or describe the project and you get a price and a deadline, usually within one working day.