Stop editing the document
This is the hardest instruction to follow and the most valuable. Seven days out, further revision produces changes nobody will notice and a version you have not rehearsed with. Committees do not arrive having compared your Tuesday draft to your Thursday one.
What they arrive with is a set of questions formed while reading, and your job this week is to be ready for those rather than to improve the thing that generated them. Print the version you are defending, put the file away, and treat it as fixed. Students who keep editing right up to the morning walk in less prepared and more anxious, having spent their last week on the only part of this that was already finished.
Cut the slides to what fits
Almost every first draft of a defense deck is built for a presentation twice as long as the one you have been given. Then it gets delivered at speed, the closing section is rushed, and the strongest material arrives while everybody is checking the time.
Find out your actual speaking slot and build to two-thirds of it. The remainder is not wasted; it is the buffer that lets you answer a mid-presentation question without losing the thread, which is a thing committees do. A useful test: if a slide would take more than ninety seconds to explain, it is two slides or it is an appendix.
- One slide on the problem, in language a non-specialist would follow
- One on why it matters, with the gap or the practice need stated plainly
- Two or three on method, because this is where questions come from
- The findings, one per slide, with the table readable from the back of a room
- One on limitations, written by you before anybody asks
- Appendix slides for everything you cut, held in reserve for questions
The two questions that always come
Why this method rather than the obvious alternative, and what would you do differently. They arrive in some form at nearly every defense I have sat in, and both are answerable in advance.
The first is not an accusation. A committee is checking that the choice was made rather than defaulted into, so the answer is a short comparison: what you considered, what it would have given you, and why the constraint of your setting or your timeline made your choice better. Thirty seconds, unhurried.
The second catches people because it feels like an invitation to apologise. It is not. Name one genuine thing, explain what you would change and why, and stop. A candidate who says nothing would change reads as somebody who has not thought about it, and a candidate who lists six is dismantling their own work in public.
Rehearse out loud, at least twice
Silently reading slides is not rehearsal. Say the whole thing aloud, standing, to an empty room or a patient friend, with a timer running. The first attempt is always worse than you expect and that is the entire point of doing it early enough to fix.
Then run a hostile version. Ask somebody to interrupt with the questions you least want, in whatever order they like. If nobody is available, write your six worst questions on cards, shuffle them, and answer at random. The value is not in polishing answers but in discovering which ones you cannot yet give without hesitating. That hesitation is the thing to spend the week on. Our mock defenses are built to be harder than the real room for exactly that reason.
The day itself
Sleep matters more than one more pass through chapter three. Tired recall is worse recall, and a defense is a two-hour exercise in retrieving things you already know under mild pressure.
Arrive early enough to test whatever technology is involved, because a projector that will not connect costs you the calm you spent a week building. Bring the printed document with your own tabs in it, since being able to turn to a page while answering is worth more than remembering the page number.
And when a question comes that you cannot answer, say so. Committees respond far better to a candidate who says that is outside what my data can support than to one who improvises. You are being examined on judgement as much as on findings, and knowing the edge of your own work is part of the judgement.
Questions people actually ask.
Should I keep revising the document this week?
No. Changes made now are ones nobody will notice, and they leave you defending a version you have not rehearsed. Committees read the copy they were sent and arrive with questions formed from it. Print what you are defending, close the file, and spend the week on the room instead of on the text.
How much of my slot should the talk fill?
Build to about two-thirds of your allotted slot. The rest is buffer for questions arriving mid-presentation, which committees do routinely. Anything requiring more than ninety seconds of explanation is either two slides or belongs in an appendix you can turn to if somebody asks about it.
What if I cannot answer a question?
Say so plainly and name why: that it falls outside what your data supports, or outside the scope you set. Improvising is worse, because a committee that catches one invented answer starts testing the others. Knowing the boundary of your own work is part of what is being assessed, not evidence against you.
Is it a bad sign if they interrupt a lot?
Usually the opposite. Interruptions mean people are engaged with the work rather than waiting politely for it to end. What tends to precede a difficult outcome is silence followed by a small number of very fundamental questions. Build buffer into your timing so interruptions cost you nothing.
