How to write a research question

What separates a strong dissertation research question from a weak one, with tests you can run on your own question before you commit to it.

9 min read · Reviewed 2026

Your research question is the single most load-bearing sentence in your dissertation. Every chapter is downstream of it: the literature you review, the method you choose, the data you collect and the argument you defend in your viva.

A good question is narrow, answerable with the evidence you can realistically gather, and open enough that the answer is not already obvious.

The four properties of a strong question

Test any candidate question against these four properties before you commit to it.

Specific: a reader can tell exactly what will be studied and where.
Answerable: the evidence needed exists and you can access it.
Non-trivial: the answer is not already settled in the literature.
Bounded: it can be answered within your word count and timescale.

Choose the question type deliberately

'What' and 'how' questions usually lead to descriptive or exploratory qualitative work. 'To what extent' and 'what is the relationship between' questions imply measurement and quantitative analysis. 'Why' questions are the most demanding because they require you to rule out alternative explanations.

Pick the stem that matches the method you are able to run, not the one that sounds most academic.

Sub-questions do the structural work

Two to four sub-questions break the main question into chapters and analytical steps. Each sub-question should map to an objective, and each objective should map to something you actually do — an interview set, a dataset, a document analysis.

Stress test it

Ask: if my data came back the opposite way, would that still be a legitimate answer? If the only possible result is the one you already believe, the question is closed and you are heading for a descriptive dissertation.

Common mistakes

Two questions welded together with 'and'.
A question that can be answered yes or no in a sentence.
Loaded wording that assumes the conclusion.
A question requiring data you have no route to.

Try this: Rewrite it three ways

Write your question as a 'what' question.
Rewrite it as a 'to what extent' question.
Rewrite it as a 'how do participants describe' question.
Choose the version whose method you can actually deliver, and say why in one sentence.

Do it in SuperDiss

Build the question, then stress test it against the four properties.

Open the Research Question builder