The Five Questions Defense Professors Ask Most Often—I Wrote Out My Answers in Advance
Before my defense, I prepared my slides, not answers to questions.
As a result, ten of the fifteen minutes were spent answering questions. The first question stumped me: “Compared with the ×× method, what is the advantage of your method?” I answered, “It’s simpler,” and the professor followed up with “In what way is it simpler?” I couldn’t go on.
Later, when I helped junior students with mock defenses, I organized the commonly asked questions into five categories and wrote a 30-second answer for each.
Question 1: Why did you choose this topic?
This question isn’t testing your topic choice; it’s testing whether you know why you’re doing it.
Answer in three parts: what the phenomenon is, where existing research falls short, and therefore why I took this step.
Don’t say “My professor told me to” or “This direction is popular.”
Question 2: What is your innovation?
This is the most common question and the easiest one to mess up. Answering “It fills a gap” is as good as not answering.
My approach is: “Compared with ××, under ×× conditions I took one extra step of ××, and the result was ××.”Your sentence must include something to compare against.
If there really is no methodological innovation, clearly state the differences in “experimental conditions/samples/data,” and don’t force claims of theoretical innovation.
Question 3: Is your data reliable?
This question is really asking, “How do you prove you didn’t miscalculate?”
Answer with three things: sample size, number of repetitions, and sources of error. If you had a control group, mention it; if you didn’t do something, say so directly.
“I didn’t do this part” is more credible than “It should be fine.”
Question 4: What are the limitations?
Many people dread this question, but it’s actually a freebie—as long as you bring it up yourself.
I usually mention two: limitations in conditions (I only covered a certain temperature range) and limitations in methods (I didn’t do in situ observation).
After saying that, add, “This is also what needs to be done next,” connecting the limitation to your plan.
Question 5: What will you do next?
Give two or three items, each with a timeline. Don’t answer “continue in-depth research.”
I usually divide next steps into “what can be done immediately” and “what requires certain conditions.” For the former, I give a timeline; for the latter, I say what support is needed.
What to Do When You Don’t Know How to Answer
A three-sentence template:
1. “I haven’t worked on this question, but my understanding is…”
2. “I haven’t filled in this part of the data yet, so at present I can only say…”
3. “I’ll look into the angle you mentioned when I get back.”
**Admitting your limits doesn’t cost you points; making things up does.** Professors will generally follow up along your stated limits rather than keep pressing you.
How I Practiced
I pasted my paper’s abstract, methods, and conclusion into MuchSci’s chat + agent, had it play the role of a defense professor, ask five questions at a time, and I answered aloud; after each answer, I had it point out “which sentence has no evidence.”
The questions MuchSci asks aren’t necessarily the same as those in a real defense, but **the rhythm of “following up with one more why” is right**, and after two rounds of practice, you won’t panic.
After practicing, I wrote the five categories of answers on cards and saved them in MuchSci’s personal knowledge base. The day before my defense, I only reviewed those five cards.
One Detail
When answering, state the conclusion first, then the reasons.
When the professor says “time is limited,” it means “don’t beat around the bush.” My habit is: first sentence gives the answer, second sentence gives the evidence, third sentence gives the boundary.
A Few Reflections
A defense doesn’t test “how much you know”; it tests “whether you can clearly explain what you did in thirty seconds.”
Writing out the five questions in advance is like rehearsing the hardest part of the fifteen minutes. What MuchSci helped me do was “simulate questioning”; I still had to verify every number in my answers myself—after all, I’m the one being questioned, not it.