Every dissertation stands or falls on a single sentence: its research question. Get it right and the rest of the project has a spine — your literature review has something to organise around, your method has something to answer, and your conclusion has something to return to. Get it wrong and everything drifts: chapters swell to cover too much, methods do not quite fit the aim, and the conclusion answers a question nobody asked. Learning how to write a research question is therefore not a preliminary chore. It is the most consequential writing decision in the whole dissertation.
This guide walks through the process step by step, from a broad interest to a question that is focused, researchable and worth answering. It uses worked examples across subjects, shows the common faults that markers penalise, and explains how your question should connect to your aims, objectives and method. If you have already chosen a topic but not yet phrased your question, start at the narrowing section; if you are still choosing, our guide on how to choose a dissertation topic pairs naturally with this one.
What a research question actually does
A research question does three jobs at once. First, it focuses the project: it draws a boundary around what you will investigate and, just as importantly, what you will not. Second, it directs the method: a question about how something happens points towards different evidence from a question about how often it happens or whether one thing causes another. Third, it sets the standard for the conclusion: at the end, you should be able to say, in plain language, how your findings answer the question — or why they only partly answer it.
Start broad, then narrow in stages
Strong questions are rarely found fully formed. They are narrowed, in stages, from a broad interest. Suppose your interest is student wellbeing. That alone is far too broad for a dissertation: it could cover anything from sleep to finance to assessment stress, across any population. Narrowing means making choices, and each choice should be deliberate and defensible.
First, choose an aspect: assessment-related stress, for example. Then choose a population: first-year undergraduates, or nursing students on placement, or international students in their first term. Then choose a setting or angle: the effect of a specific intervention, the experience of a particular group, or the relationship between two factors. A question begins to emerge: “How do first-year nursing students experience assessment-related stress during their first clinical placement?” That is still broad, but it is now a recognisable study rather than a theme.
Each narrowing decision should be justified in your introduction and proposal by reference to the literature or to a practical problem. Why nursing students, and why placement? Because the literature suggests placement is a distinct stressor, or because your placement experience raised the issue — either justification works, provided it is stated. A question that has been narrowed without reasons looks arbitrary to a marker; one narrowed with reasons looks like a research decision.
The tests a good question must pass
Before you commit to a question, run it through five tests. First, is it clear? A reader who knows nothing about your project should understand what you are asking without needing a paragraph of explanation. Remove jargon where plain language works, and define any term that could be read two ways.
Second, is it focused? A question that tries to cover three populations, two interventions and four outcomes is really three questions wearing one sentence. Split it into a main question and sub-questions, or cut it down. One main question with two or three sub-questions is a normal, healthy structure; one question with six clauses is a warning sign.
Third, is it researchable in the time and with the resources you have? A question requiring a national survey, access to a restricted population, or a year of observation cannot be answered in a final-year dissertation, however interesting it is. Check the practicalities early: can you recruit participants, access data, obtain ethical approval, and analyse the material within your deadline? If any answer is no, narrow the question until the answer becomes yes. Our guide on how to write a research proposal shows how to test feasibility before you commit.
Fourth, is it analytical rather than merely descriptive? Markers reward questions that require analysis, evaluation or explanation — those beginning with how or why, or asking to what extent, or comparing approaches. Purely descriptive questions — “what do students think about X” with no analytical angle — can still work, but they need a clear purpose: describing something in order to evaluate it, explain it, or inform practice.
Fifth, is it original enough? Originality at dissertation level does not mean discovering something nobody has ever studied. It can mean studying a familiar issue in a new population, combining two approaches, applying a theory to a new setting, or updating earlier work. Your literature review should show the gap your question fills, even if that gap is modest. A question identical to one already answered thoroughly, in the same population, will struggle to justify itself.
Question types and when to use them
Different kinds of inquiry need different kinds of question. Exploratory questions suit qualitative work where little is known: “How do mature students experience returning to study after a career break?” These questions are open, and they suit interviews, focus groups or case studies that explore experience in depth.
Explanatory questions suit work that tests a relationship or a cause: “To what extent does regular formative feedback improve essay grades among second-year undergraduates?” These questions suit designs that compare groups, track change over time, or test an association, and they often pair with hypotheses.
Evaluative questions suit work that judges a policy, intervention or practice: “How effective is peer mentoring in supporting first-year retention, and for whom does it work best?” These questions suit mixed methods, combining outcome data with participants’ experience, and they must define effectiveness before they can judge it.
Comparative questions suit work that contrasts groups, settings or approaches: “How do assessment experiences differ between campus-based and distance-learning students?” Comparison is a powerful analytical move, but it needs genuinely comparable groups and a clear basis for comparison, or the findings will be muddied by unrelated differences.
From question to aims, objectives and method
Your question should generate your aim and objectives almost mechanically. The aim restates the question as a purpose: “This study aims to explore how first-year nursing students experience assessment-related stress during their first clinical placement.” Objectives then break that aim into steps: review the literature on placement stress, explore students’ experiences through interviews, identify the factors students associate with stress, and consider implications for placement support.
The method follows from the question in the same way. An exploratory question about experience points towards interviews or focus groups; a question about prevalence points towards a survey; a question about change over time points towards a longitudinal or before-and-after design. If your method cannot answer your question, one of the two is wrong — and it is usually easier, and more honest, to adjust the question than to force an unsuitable method onto it.
Write the question, aim and objectives together, on one page, and check them against each other before you write anything else. If an objective does not serve the question, cut it. If part of the question has no objective and no method, the question is still too broad. That one page is the skeleton of your whole dissertation. Our research proposal help page explains how tutors review that skeleton before the main writing begins.
Worked examples: weak questions made strong
Consider a weak question: “What is the impact of social media?” It fails every test: unclear about whose experience, which platform, what kind of impact, and measured how. A stronger version specifies all four: “How does daily Instagram use relate to body image concerns among female undergraduates aged 18 to 21?” That version can be investigated, because it names a behaviour, a population, and an outcome.
Another weak question: “Should universities do more for mental health?” That is a policy opinion, not a research question, and no study could answer it directly. A researchable version asks something evidence can address: “What barriers do undergraduate students report when accessing university counselling services, and how do those barriers vary by year of study?” The findings could then inform policy, but the question itself is answerable.
A third: “How can teachers improve learning?” That is too broad and too vague to guide a study. Narrowed, it becomes: “To what extent does structured peer feedback improve the quality of first drafts in Year 10 English lessons in two case-study schools?” The narrowed version is smaller, but it is a real study, and its findings will mean something. Breadth is not ambition; precision is.
Common faults to avoid
The most frequent fault is the double question: two inquiries joined by “and” that need different methods and different literatures. Split them, choose one as the main question, or demote the second to a sub-question. Another frequent fault is the leading question, which embeds the expected answer: “How does poor teaching damage student motivation?” assumes poor teaching and damage before any evidence is gathered. A neutral phrasing — “How do students perceive the relationship between teaching approaches and their motivation?” — lets the evidence speak.
Watch also for questions that are really titles in disguise, such as “Student engagement in higher education”, which asks nothing. Add a verb and a focus: “What factors do lecturers associate with student engagement in large lectures?” Finally, avoid questions that depend on access you do not have, such as interviewing senior executives or patients in a restricted setting, unless that access is already confirmed. A question you cannot investigate is not ambitious; it is unusable.
Refine your question with your supervisor early, and expect it to change. Most good questions go through several drafts as reading reveals what is already known and what is genuinely open. A question that survives contact with the literature — narrowed, clarified, made researchable — is stronger for having changed. Treat the first draft as a starting point, not a commitment.
Research question FAQs
How many research questions should a dissertation have?
Usually one main question, supported by two or three sub-questions that break it into manageable parts. More than that tends to fragment the study. If you find yourself with five or six questions, you probably have more than one project; choose the strongest and save the rest for future work.
Can my research question change during the dissertation?
Yes, within reason, and it often should. Reading, pilot work and early findings commonly refine a question. Small refinements are normal; a wholesale change of topic late in the project is not, because it invalidates the literature review and method already written. Agree any change with your supervisor and record why it was made.
Should a research question be a yes-or-no question?
Usually not. Yes-or-no questions close down analysis: the answer is a single word, and the interesting work is in the explanation behind it. Rephrase closed questions as open ones: instead of “Does feedback improve grades?”, ask “To what extent, and in what ways, does formative feedback affect essay grades?” That phrasing invites analysis rather than a verdict.
What is the difference between a research question and a hypothesis?
A question asks what you want to find out; a hypothesis predicts what you expect to find, and is then tested. Qualitative and exploratory studies often have questions without hypotheses. Quantitative studies often have both: a question, plus a hypothesis stating the expected relationship between variables, which the analysis then supports or refutes.
Where does the research question appear in the dissertation?
It appears in the introduction, usually near the end of the opening section after the background and rationale, and again in the methodology chapter where you show how your design answers it, and in the conclusion where you state how far it has been answered. Keeping it visible in all three places is what gives a dissertation its coherence.
What if my supervisor disagrees with my question?
Take that seriously: supervisors have seen many questions fail, and their objections usually point to scope, feasibility or originality problems. Ask what specifically concerns them, revise against that concern, and return with a revised version rather than a defence of the original. The question is a joint decision, and a supervisor who supports it will support the project that follows.