Skip to main content
Questions holds what your project is trying to find out, and it is a small surface with an outsized effect, because a research question is what everything else gets judged against. Your library is relevant or not relative to a question, your methods are appropriate or not relative to a question, and your advisor reads your corpus against your questions instead of in the abstract. Questions nest into sub-questions, every revision is kept, and each one can be scored for design alignment, which gives you a structured second opinion on whether what you are asking is answerable the way you plan to answer it.

Writing a question

Hit New question and type it, adding context or detail underneath when the question alone does not carry it. Write the question you are actually asking rather than the one that sounds most respectable, because nobody can help you with a question you have hedged into safety, and that includes your advisor, human or otherwise.

Sub-questions

Any question can have sub-questions beneath it, so Create sub-question nests one under a parent and the child shows you its parent. Most real projects have one or two questions that matter and a handful of operational questions underneath. Making that hierarchy explicit helps for the obvious reason, since it is usually the skeleton of your chapters, and for a less obvious one: it is where you notice that a “sub-question” is really a separate project.

Design alignment

Analyze scores a question against four dimensions:
  • Research design. Is the overall approach right for what is being asked?
  • Sampling design. Will who or what you study support the claim?
  • Measurement design. Can the constructs actually be measured this way?
  • Analytic design. Will the planned analysis answer the question?
Analyze all runs the whole set. Understand two things about the scores. They are read against your designs, which you set in Project settings, so scoring a qualitative project against a rubric that assumes randomised sampling produces nonsense and the fix belongs in settings. And they are diagnostic rather than a grade, so watch the weakest dimension, because that is where the question fails under questioning. This is what drives “{n} needs sharpening” on Overview, flagging a question with no score yet or one whose weakest dimension sits below the midpoint, and giving you a count and a next move rather than a percentage.

Evidence

Open a question and it shows the sources linked to it, split three ways into Supports, Pushes back and Mixed. The split does not happen on its own, since linked sources start unweighed and the page says so with “These sources haven’t been weighed yet.”, so Assess the evidence reads them against the question and sorts them while Re-assess runs again once you have added more. This is the most useful thing on the page, because the instinct while gathering literature is to collect what agrees with you, which means a question showing eight supporting sources and nothing pushing back is either settled or the product of a biased search, and it is usually the second. Read the “pushes back” column first.

Version history

Every revision of a question is kept, and you can read back through them. This matters more than it sounds, because research questions drift, usually for good reasons and occasionally because you have quietly retreated from something hard, so putting what you asked in month four next to what you are asking now is one of the few honest checks available on your own thinking. It also supplies the “how did this project evolve” paragraph that every thesis needs and nobody can remember.

Archiving

Questions can be archived instead of deleted, and archived ones stay searchable. Prefer archiving, because a question you abandoned is part of the record of the project and “why didn’t you look at X?” is a normal thing to be asked in a viva, which an archived question with its history answers completely.

Refining with the advisor

Refine in the advisor opens a conversation scoped to this question, and it will pressure-test clarity, scope and testability, arguing the counter-case at higher Pushback. It will not write your research question for you, though it will tell you which of two phrasings commits you to a claim you cannot support, which does you more good anyway.

How questions change everything else

Worth being explicit, because this is why the tab deserves upkeep:
  • Your advisor reads your library against these questions, so vague questions get you vague observations.
  • Sources link to questions, so a question shows you the evidence you have gathered for it, and more usefully, that you have gathered none.
  • The check-in offers to pressure-test questions you added recently.
  • Design alignment hands you the shape of a methods defence before you have to give one.
A project with a good library and no questions gets you description, whereas a project with both gets you an argument. Next: Models.
Last modified on August 29, 2026