How to write a research question
To write a good research question, start with a broad topic, narrow it to one population, one setting and one issue, then phrase it as a single question that your study can actually answer with the data, time and access you have. Check the draft against the FINER criteria (feasible, interesting, novel, ethical, relevant) and, for health and intervention questions, structure it with PICO. A strong question is specific enough to choose a design from, and open enough that the answer is not already known.
Get the free Research Question & Gap Builder
What makes a good research question
A research question is the one sentence your whole study exists to answer. It decides your design, your sample, your data and your analysis, so a vague question produces a vague thesis. A good question is focused (one main issue), answerable (data could settle it), specific (the people, setting and concept are named), and worth asking (the answer would add something).
Questions also differ by the kind of answer they need. Descriptive questions ask what is happening or how common something is. Comparative or relational questions ask whether groups differ or whether two things are linked. Causal questions ask whether one thing changes another. Exploratory questions ask how people experience or explain something. Knowing which kind you are writing is the quickest route to the right research design.
Two tools for testing a question: FINER and PICO
FINER comes from Hulley and colleagues’ textbook Designing Clinical Research (Hulley et al., 2013). It is a checklist of five properties a good question should have. Although it was written for clinical research, it works in most fields.
| Letter | Criterion | Ask yourself |
|---|---|---|
| F | Feasible | Can I get enough participants or data, with the skills, time and money I have? |
| I | Interesting | Do I, my supervisor and my field care about the answer? |
| N | Novel | Does it confirm, extend or challenge earlier work, rather than repeat it with nothing new? |
| E | Ethical | Could an ethics committee approve it? |
| R | Relevant | Would the answer matter to knowledge, practice or policy? |
PICO was set out by Richardson and colleagues as a way to build well-formed clinical questions (Richardson et al., 1995). It breaks a question into four parts:
- Population or problem: who you are studying.
- Intervention or exposure: what is done to them or what they experience.
- Comparison: what it is compared with (usual care, no exposure, another option).
- Outcome: what you will measure.
Use FINER to judge whether a question is worth pursuing, and PICO to make the parts of an intervention or comparison question explicit. PICO fits less well for exploratory qualitative questions, which often have no comparison or measured outcome; in that case simply make sure the population, setting and phenomenon are named.
How to create a research question, step by step
- Start with a topic you can sustain. Pick an area you will still care about in a year, and that your supervisor can support.
- Read to find what is unsettled. Recent reviews and the limitations or future-research sections of studies show where questions remain. See how to find a research gap.
- Name the problem. Write one or two sentences on what is known, what is not, and why it matters. This becomes your problem statement.
- Narrow. Choose one population, one setting, one concept or intervention, and (if relevant) one time frame.
- Decide the kind of answer you need. Description, comparison, relationship, cause or experience.
- Draft the question. Use PICO parts if it is an intervention or comparison question. Start with how, what, whether, to what extent, or is there a difference.
- Test it with FINER. Be honest about feasibility in particular: access to participants is the most common reason a question has to change.
- Revise with your supervisor. Expect several drafts. Keep each version so you can show how the question developed.
Worked example: from topic to question
Suppose you are a master’s student in public health with a broad interest in physical activity.
| Stage | Draft | What changed |
|---|---|---|
| Broad topic | Physical activity in young people | Too wide to study |
| Narrower topic | Active travel to school | One behaviour chosen |
| Focused topic | Walking and cycling to school among secondary pupils in one city | Population and setting named |
| First question | Why don’t teenagers walk to school? | A question, but loaded and vague |
| Revised question | What factors do secondary school pupils and their parents in one city describe as influencing whether pupils walk or cycle to school? | Neutral, specific, answerable by interviews or focus groups |
If you wanted a quantitative study instead, PICO gives a different version: Among secondary school pupils in one city (P), is living within a set distance of school (I), compared with living further away (C), associated with walking or cycling to school (O)? The topic is the same; the kind of answer, and therefore the design, is different.
Now test the qualitative version with FINER. It is feasible if you can reach schools and families through a contact; interesting to local transport and health planners; novel if local views have not been gathered; ethical with parental consent for under-18s; and relevant to how active travel is encouraged. Each answer should be something you can explain in your proposal.
Weak vs strong research questions
| Weak | Problem | Stronger |
|---|---|---|
| What is the impact of social media? | No population, setting or outcome; far too broad. | How do first-year undergraduates at one university describe the role of social media in their sense of belonging? |
| Is exercise good for depression? | Yes/no; the general answer is already widely studied. | Among adults on a waiting list for talking therapy, is a supervised group walking programme, compared with usual waiting-list care, associated with lower depressive symptom scores after eight weeks? |
| Should schools ban phones? | A value judgement, not an empirical question. | How do teachers in schools with a phone restriction policy describe its effect on classroom routines? |
| How can nurse burnout be solved? | Assumes a solution exists and is too wide for one study. | What workplace factors do nurses on night shifts in one hospital associate with emotional exhaustion? |
| What do managers think about leadership? | Concept and population too loose. | How do newly promoted first-line managers in small retail firms describe learning to give feedback to former peers? |
How the question connects to the rest of the thesis
- Problem statement. The question responds directly to the gap your problem statement describes.
- Aims and objectives. The aim restates the question as a purpose; each objective is a step towards answering it. See research aims and objectives.
- Hypothesis. In quantitative studies a relational or causal question is often paired with a testable prediction. See how to write a research hypothesis.
- Design and methods. The kind of answer the question needs decides the design.
- Conclusion. Your final chapter should answer the question in the same words you asked it.
Common mistakes
- Too broad. If the question could be the title of a textbook, narrow it.
- Too narrow or trivial. If one fact or a quick search would answer it, widen it.
- Several questions in one. “How and why do X and Y affect Z” is three studies. Split it into a main question and sub-questions.
- Built-in assumptions. “Why is the policy failing” assumes it is failing.
- Mismatch with design. Causal wording such as “effect” with a design that can only show association.
- Ignoring feasibility. A question you cannot get data for will have to change later, at greater cost.
Checklist
- One main question, with sub-questions if needed.
- Population, setting and concept are named.
- The kind of answer (describe, compare, relate, cause, explore) is clear.
- For intervention or comparison questions, P, I, C and O can be identified.
- It passes each FINER criterion, and you can explain how.
- It is neutral and does not assume the answer.
- Your aim, objectives and design all follow from it.
Use the tools as you plan
- Research Question Validator: check your draft question against the criteria on this page.
- Hypothesis Testability Checker: if your question leads to a hypothesis, check that it can be tested.
- Citation Gap Finder: look for unanswered areas in the literature behind your question.
Get the free Research Question & Gap Builder
A fill-in sequence from a broad topic to a narrowed angle, a written gap statement, a FINER-scored question, and a check that problem, purpose, and question line up. We’ll email you the download link.
Frequently asked questions
How do you develop a good research question?
Start with a broad topic, read to find what is still unsettled, name the problem, then narrow to one population, setting and concept. Draft the question, test it against the FINER criteria and revise it with your supervisor.
What are the FINER criteria?
FINER stands for feasible, interesting, novel, ethical and relevant. It is a checklist for judging research questions described by Hulley and colleagues in Designing Clinical Research (2013).
What is PICO in a research question?
PICO stands for population, intervention, comparison and outcome. Described by Richardson and colleagues in 1995, it helps you build clear clinical and intervention questions by naming each of those parts.
How specific should a research question be?
Specific enough that you could choose a design, sample and data source from it alone. It should name who, where and what, but not be so narrow that one fact would answer it.
What are examples of good research questions?
For example: How do first-year undergraduates at one university describe the role of social media in their sense of belonging? Or: What workplace factors do nurses on night shifts in one hospital associate with emotional exhaustion?
Can a research question be a yes or no question?
Usually it should not be. Yes or no questions give thin answers. Reword them with how, what, to what extent, or whether there is a difference, so the study can describe or measure the answer.
How many research questions should a thesis have?
Many theses have one main question with a small number of sub-questions, but the norm varies by field and institution. Check your programme guidance and keep every question answerable within your study.
The Research Question & Problem Statement Workbook
You have a topic you care about. You do not yet have a study. This workbook takes you from one to the other.
57 fill-in templates. Paperback $27.00 on Amazon
See inside the workbookJust want the templates? The Research Question & Problem Statement Kit: every template as a fillable PDF, $19. Or all 17 kits for $99.