How to Write a Research Question

Almost every problem that surfaces in month four was present in the research question in week one.

The research question determines the design, the data, the analysis and the shape of the write-up. Getting it right is the highest-leverage hour in a research project, and getting it wrong is expensive in a way that becomes apparent only once substantial work has been done.

Move from topic to question

A topic is an area of interest; a question is something that can be answered. The transition requires specifying a relationship, a population and a context. “Remote working” becomes “is the frequency of scheduled team contact associated with reported isolation among fully remote knowledge workers in their first year?” — which tells you what data you need and what analysis will answer it.

If you cannot write your project as a single question, you do not yet have a project. The effort of the sentence is diagnostic.

Use a framework to force precision

Structured frameworks exist because vagueness in a question propagates through everything downstream. PICO — population, intervention, comparison, outcome — is standard in health research. PEO suits observational work. SPIDER suits qualitative studies. SPICE suits service evaluation.

• Population — who exactly, with what inclusion criteria?
• Intervention or exposure — what is being done or experienced?
• Comparison — against what alternative or baseline?
• Outcome — measured how, and over what period?

Even where your discipline does not use these formally, working through them exposes which element you have left undefined.

Match the question type to the method

Questions imply designs, and mismatches are the most common structural fault in student proposals. Descriptive questions ask what is happening and suit surveys and observational work. Relational questions ask whether two things vary together and suit correlational designs. Causal questions ask whether one produces the other and require a comparison and, ideally, control.

Exploratory questions asking how people experience or make sense of something require qualitative approaches. A question asking “how do nurses experience escalation decisions” cannot be answered by a questionnaire with fixed responses, however large the sample.

Make it answerable within your constraints

A good question you cannot answer is worse than a modest one you can. Test each candidate against access, skills, time, ethics and data availability before committing. Access is where enthusiasm most often overrides judgement — a study of hospital staff needs permissions that may take longer than your entire timeline.

Halve whatever recruitment estimate you make. It is routinely optimistic.

Keep the scope narrow

Most first drafts are too broad, and narrowing happens along several axes at once: population, setting, timeframe, outcome and mechanism. A question containing “and” twice is usually two questions that need separating, one of which should be dropped.

Narrow questions produce better dissertations. Breadth is available later; a study that cannot be completed is not.

Check it is genuinely open

If you already know the answer, there is nothing to research. Questions that can only be answered one way — because the alternative is implausible or because the phrasing presupposes the finding — produce work that reads as confirmation rather than inquiry.

A useful test: can you describe a plausible result that would surprise you? If not, reconsider the question.

Distinguish aims, questions and hypotheses

These are frequently conflated. An aim states what the study intends to achieve. A research question states what it will answer. A hypothesis states a specific testable prediction, and is appropriate only where the design tests one — much qualitative work has questions but no hypotheses, and inventing them to appear rigorous signals the opposite.

Write sub-questions that decompose the main one

Where a main question is too large for a single analysis, two or three sub-questions can break it into answerable parts. Each should be a genuine component of the main question rather than a related topic, and together they should exhaust it. Sub-questions that add material the main question does not require signal that the scope has drifted.

Test it against the literature

Before committing, search for your question as though someone else had already answered it. Two outcomes matter: if the exact study exists, you need a different angle; if nothing at all appears, ask why, because unexplored questions are sometimes unexplored for practical reasons others discovered first.

Expect it to change, and record why

Research questions evolve as reading and early data complicate them, and that is a sign of thinking rather than failure. What matters is that revisions are deliberate and documented — a methodology chapter that explains why the question narrowed after the pilot is stronger than one that pretends the final question was the original.

Write it down and leave it visible

A surprising number of projects drift because the question is never written in one place. Put the final wording at the top of your working document and reread it whenever you make a design decision. Anything that does not serve it is scope creep, and scope creep is the main reason dissertations run long and finish thin.

The same sentence becomes the spine of the write-up: the introduction poses it, the methods explain how it was answered, the discussion answers it, and the conclusion states how confidently.