You cannot defend a question you only borrowed

You have found an idea you like in Research Idea, the free site where people leave research ideas for others to take up. Someone else brought it, the site shaped it, and it even suggests how it could work at Honours, master's or doctoral level. It is tempting to copy the question into your proposal and start.

Hold off for a week. Over the next year you will be asked "Why this question?" by your supervisor, a proposal committee, an ethics board and, at the end, an examiner. You will ask it yourself on the day your data does not do what you expected. A borrowed question gives you nothing to answer with.

Borrowing itself is normal. Many PhD topics start in someone else's words: in two French doctoral programmes, the subject "was often initially written by the supervisors" (Girard et al., 2025, p. 1011). The work is turning those words into yours.

You own a question when it can survive a no

Here is a test. Suppose your study finds the answer is no. Can you say what you would have learnt, what you would do next, and why you would still care? A borrowed question has one answer in mind. An owned one survives being wrong, because you chose it for reasons that still hold.

Before you test that, check the idea is worth a year. Ask the question from an earlier piece in this series: who would decide differently if you were right? If nobody would, keep looking.

You get there by choosing how to ask the question. Take a real idea from the registry: "Does AI screening mark down research summaries written in a second language?"

As a gap: nobody has tested this with African Englishes. This is the safest choice and the quickest to set up, which suits a single Honours year. It also passes the test most easily: a no is still a finding.

As a challenge to an assumption: "Studies like this assume there is one neutral English that second-language writing falls short of. What if the screener simply rewards one style of writing, whoever wrote it?" This is the most interesting choice and the hardest to argue, so it needs a supervisor willing to back it.

As a practical problem, worked out with the people involved: "How should a university's research office use AI screening so that second-language writers are not marked down?" This gives the clearest answer to who would decide differently, and it takes longest, because you have to work with the people who decide.

These three come from a list of approaches to research questions set out by Girard and colleagues (p. 1003), who ran courses on the subject for about 500 doctoral students over more than ten years. In their view, the student "is the one who makes the subject evolve, who reflects and chooses the components of the research question" (p. 1009), and the supervisor guides and then steps back. A survey of doctoral students points the same way. Those whose supervisors encouraged them to act on their own ideas, while still guiding the research, were the most confident in their ability to do research (Overall et al., 2011). That confidence is what you draw on when the data surprises you.

One of the Girard courses asks second-year PhD students to write their topic as it was worded in the original PhD offer, next to the wording they would use today, and to explain the difference (p. 1011). Start that page now: the registry wording on one side, yours on the other, and a sentence on each change. Add a line for your answer to "What would I do if the answer were no?"

Keep the page, because your question will move again. A study of five first-year PhD students in the United Kingdom found their approach to their project shifted as they gained experience and hit unexpected events (Inouye, 2023). Then talk your version through with two or three people outside your field. The Girard courses set one rule for listeners: question the student, but never rewrite the topic for them (p. 1013).

Give back what you grew

Someone left this idea for you. What do you owe them? You do not need their permission; they shared it so that someone would take it up. But a seed library asks borrowers to bring back seed from what they grew, and a registry can ask the same. Three things cost you little:

1. Say you are on it. Mark the idea as taken up (the "I'm researching this" button on its page), so the next student knows.
2. Cite it. The idea has a permanent link and a registry number. Cite it like any other source where your question started.
3. Bring back what you grew. When your thesis or article is out, connect it to the idea. The person who left it will see where it went, and the next reader will find your work.

It will not always be this tidy. The person who left the idea may want to work on it with you, or may publish on it themselves. The page shows a contributor code, not a name, but some contributors link their ideas from their own profiles. If you can find out who it is, a short, early email saves an awkward one later.

The registry gives you someone else's question. The page with two wordings on it shows it has become yours, and the link back shows where it went.

The Girard framework comes from two training courses at one French research institute, written up by the people who designed them. The survey shows an association, which may not be a cause. In South Africa a proposal is usually approved formally, so later changes to your question go back through your supervisor or committee.

References

Girard, N., Cardona, A., & Fiorelli, C. (2025). Learning how to develop a research question throughout the PhD process: Training challenges, objectives, and scaffolds drawn from doctoral programs for students and their supervisors. Higher Education, 89(4), 1001–1020. https://doi.org/10.1007/s10734-024-01258-2

Inouye, K. (2023). Developing the PhD thesis project in relation to individual contexts: A multiple case study of five doctoral researchers. Higher Education, 85(5), 1143–1160. https://doi.org/10.1007/s10734-022-00882-0

Overall, N. C., Deane, K. L., & Peterson, E. R. (2011). Promoting doctoral students' research self-efficacy: Combining academic guidance with autonomy support. Higher Education Research & Development, 30(6), 791–805. https://doi.org/10.1080/07294360.2010.535508