TL;DR: How to Write a Research Question
- A good research question is specific, researchable, and neither too broad nor too narrow.
- Start with a broad topic, then narrow it using the FINER or PICO framework.
- A research question asks what you want to find out; a thesis statement states what you found.
- You can change your research question mid-process – but document why.
- Multiple research questions are fine, as long as they’re connected and manageable.
Most people think the hardest part of research is collecting data or writing it up. It isn’t. The hardest part – and the part that derails more studies than any other – is starting with the wrong question.
A vague, unmeasurable, or hopelessly broad research question doesn’t just slow you down. It shapes every decision that follows: which methods you pick, what data you collect, how you analyze it, and what conclusions you can actually draw. Get the question wrong, and everything downstream suffers.
This guide walks you through how to write a research question that’s focused, answerable, and genuinely useful – with real examples, frameworks, and the specific mistakes that most articles skip over.
What Is a Research Question?
A research question is the specific question your study sets out to answer. It defines the scope of your research, keeps your inquiry focused, and ultimately determines whether your findings are meaningful.
It’s not a topic (“social media and teenagers”) and it’s not a thesis statement (“social media harms teenage mental health”). It’s an open-ended question that you don’t yet have the answer to – one that requires systematic investigation to resolve.
According to Creswell’s foundational research design framework, a good research question sits at the intersection of what’s interesting, what’s feasible to study, and what actually matters to the field.
What Makes a Good Research Question for a Research Paper
Not all questions are research questions. “Is climate change real?” has a scientific consensus answer – it’s not a research question. “How do climate narratives in regional newspapers affect voter behavior in non-coastal districts?” is.
Strong research questions share five qualities, often remembered with the FINER framework – Feasible, Interesting, Novel, Ethical, and Relevant:
Feasible means you can actually answer it with the resources, time, and access you have. A question requiring 10 years of longitudinal data isn’t feasible for a semester-long dissertation.
Specific means there’s a clear, defined focus – particular population, phenomenon, context, or timeframe. Vagueness kills research.
Complex means the answer isn’t yes or no. Good research questions require analysis, not just a lookup.
Researchable means there’s data you can actually collect or access to answer it – primary or secondary.
Relevant means it contributes something, either filling a gap in existing literature or addressing a real-world problem.
The PICO framework (Population, Intervention/Issue, Comparison, Outcome) is particularly useful in health and social sciences for structuring these criteria into the question itself.
How to Write a Research Question Step by Step
Step 1: Choose a broad topic that genuinely interests you
Start with something you’re curious about – but make it a subject area, not a question yet. “Remote work,” “adolescent anxiety,” “sustainable packaging” – these are starting points, not research questions.
The interest factor matters more than people admit. Research takes time. If you picked the topic because it seemed safe or practical but you don’t actually care about it, your motivation will stall by month three.
Step 2: Do preliminary reading
Before you can narrow your topic, you need to understand its landscape. Read recent review articles, meta-analyses, and the introduction sections of key papers in the area. You’re looking for: what’s already been studied, what’s contested, and where authors say “future research should explore…”
Those “future research” sections are gold. They’re essentially invitations.
Step 3: Identify the gap
This is where your question starts to emerge. What hasn’t been studied? What’s been studied in one context but not another? What assumption has everyone made that nobody’s actually tested?
A useful exercise: take the existing literature and ask “but what about…?” repeatedly. “Studies show remote work increases productivity – but what about remote workers with caregiving responsibilities?” That pivot is the beginning of a research question.
Step 4: Narrow using the 5 W’s + H
Once you have a general direction, get specific by working through:
- Who is the population? (e.g., full-time remote workers aged 25–45 in the US)
- What is the phenomenon or variable? (e.g., perceived work-life balance)
- Where is the context? (e.g., tech industry)
- When is the timeframe? (e.g., post-pandemic, 2021–2024)
- Why does it matter? (e.g., for organizational policy)
- How will it be studied? (qualitatively, quantitatively, mixed methods?)
Not every question needs all six – but working through them forces specificity.
Step 5: Write a draft question
Put it in writing. At this stage, it’ll probably be messy. Something like: “How does remote work affect work-life balance for parents in tech companies?” That’s a real draft – not polished, but directional.
Step 6: Test it against quality criteria
Run your draft through these checks:
- Can it be answered empirically? (Not “Should companies allow remote work?” – should questions are normative, not empirical)
- Is it too narrow to generate meaningful findings? (Not “How does remote work affect Sarah’s Tuesday morning productivity?”)
- Is it too broad to study in one project?
- Does it lead to a yes/no answer? (If so, reframe it as “How,” “What,” or “To what extent”)
- Has it already been answered definitively? (If so, find a new angle)
Step 7: Refine and finalize
Take feedback. Show the question to a colleague, advisor, or peer. Ask: “What would you expect the answer to be?” If they say “I have no idea,” your question might be too vague. If they say “Obviously X,” it might be too settled.
The sweet spot is a question where thoughtful people genuinely disagree, or simply don’t know.
How to Tell If Your Research Question Is Specific Enough
This is one of the most common things researchers get stuck on – and most guides brush past it.
Here’s a practical test:
Can you draw a boundary around your study just from reading the question?
If someone reads “How does social media affect mental health?” they can’t tell you: whose mental health, which platforms, what kind of effect, measured how, in what context. That’s four missing dimensions – which means the question is doing zero work.
Now try: “What is the relationship between daily TikTok usage and self-reported anxiety levels in female university students in the UK?” Every dimension is specified. Someone could design a study from that sentence alone.
Another useful test: The title test. If your research question were the title of a study, would a peer reviewer know immediately what the study is about? If yes, it’s specific enough. If they’d need to ask clarifying questions, it’s not.
A third test: The impossible negative. Try to imagine a result that would definitively answer your question as “no.” If you can’t, the question is probably too vague or too normative.
How to Turn a Broad Topic Into a Focused Research Question
Most research questions start life as a topic. The narrowing process is iterative, and most people try to do it in one jump – which is why they end up either too broad or too narrow.
Here’s a staged narrowing approach:
- Level 1 (Topic): Climate change
- Level 2 (Angle): Climate change communication
- Level 3 (Population/Context): Climate change communication in rural American communities
- Level 4 (Phenomenon): How rural Americans’ trust in local versus national media sources shapes their interpretation of climate change messaging
- Level 5 (Research Question): “To what extent does source credibility (local vs. national media) predict climate change skepticism among rural American adults, and does this relationship differ by political affiliation?”
Each level adds a constraint. The discipline is in not skipping levels.
A tool worth knowing: concept mapping. Draw your broad topic in the center, then branch out with related sub-topics, populations, contexts, and phenomena. Where two or three branches intersect in a way that hasn’t been studied – that’s likely your research question.
Research Question vs. Thesis Statement: What’s the Difference?
This is one of the most searched questions on the topic, and it’s genuinely underexplained in most resources.
Here’s the clearest way to think about it:
A research question is the beginning. A thesis statement is the end.
A research question is what you set out to investigate. It’s open-ended, uncertain, and drives your methodology. You write it before you do the research.
A thesis statement is your answer to that question. It’s a claim you’re making based on your findings. You write it (or finalize it) after the research.
Example:
- Research question: “How do first-generation college students navigate financial aid processes compared to continuing-generation students?”
- Thesis statement: “First-generation college students experience significantly higher levels of financial aid confusion due to limited family knowledge networks, resulting in lower completion of available aid applications.”
The research question is neutral. The thesis statement takes a position.
One important nuance that most articles miss: in qualitative research, the thesis statement often emerges from the data rather than preceding it. You genuinely can’t write your thesis before you’ve done the study. In quantitative research, you might write a hypothesis (a predicted answer to your research question), but even then, the hypothesis and the thesis are different things.
Can I Change My Research Question After Starting My Research?
Yes – and you probably should if the evidence demands it.
The myth that research questions are fixed is a leftover from a model of research that treats the question like a contract. In practice, especially in qualitative and exploratory research, the question evolves as you learn more.
What matters is: document the change and explain why.
If you discover midway through that your original question rested on a false assumption – for instance, you assumed a relationship existed that your early data shows doesn’t – it’s methodologically honest to revise the question. A rigid attachment to the original question leads to motivated reasoning and cherry-picking.
The practical rule: you can narrow, sharpen, or pivot your research question, but you shouldn’t widen it after data collection starts. Widening your question retrospectively to capture findings you didn’t plan for is called HARKing (Hypothesizing After Results are Known) – a recognized form of research misconduct.
For quantitative studies with pre-registered hypotheses, any change to the research question must be noted as exploratory rather than confirmatory.
How Do I Make Sure My Research Question Is Researchable?
A research question is researchable if there’s a realistic pathway from asking it to answering it. Three things determine this:
Data availability. Can you access the information needed? If your question requires interviewing active military personnel, classified documents, or patients in clinical settings, access barriers may make it unresearchable for most researchers.
Methodological fit. Does a method exist that can answer this kind of question? “Why do people feel love?” is a real question, but it’s not easily researchable with quantitative methods – you’d need qualitative phenomenological work. Make sure the method and question are compatible.
Ethical clearance. Some questions are researchable in principle but require ethical approval that’s difficult to obtain – for instance, research involving minors, deception protocols, or sensitive populations. Check IRB/ethics requirements early.
A question that passes these three checks is researchable. One that fails even one may need to be reformulated.
What Should I Do If My Research Question Is Too Broad?
Broad questions don’t just make research harder – they make it impossible to conclude anything meaningful. Research on cognitive load in academic writing shows that researchers with poorly scoped questions report higher rates of methodological confusion and abandoned projects.
If your question is too broad, use these strategies:
Add population specificity. “How does exercise affect mental health?” → “How does aerobic exercise affect depression symptoms in adults over 60 with a prior diagnosis?”
Add a timeframe. “How have social media platforms changed?” → “How did Instagram’s algorithmic changes between 2020 and 2023 affect small creator follower growth?”
Specify the relationship type. “What is the relationship between X and Y?” → “Does X predict Y, and is this relationship moderated by Z?”
Choose one variable. If your question has five independent variables, you’re writing five studies. Pick one.
Limit the context. “In UK secondary schools” is more specific than “in schools.” “In B2B SaaS companies” is more specific than “in companies.”
The goal isn’t to make your question small – it’s to make it answerable.
Can I Have More Than One Research Question in a Paper?
Yes, and it’s often appropriate. The key constraint is coherence: multiple research questions should address different facets of the same central problem, not pull the study in unrelated directions.
A useful mental model: think of your main research question as the trunk of a tree, and your sub-questions as branches. They all serve the same tree.
Example of coherent multiple questions:
- “What are the primary barriers to physical activity among low-income urban adolescents?”
- “How do these barriers differ by gender?”
- “What community-based interventions have been most effective in similar populations?”
These three questions form a connected investigation. They’re all answerable within one study using related data.
Example of incoherent multiple questions:
- “What are the barriers to physical activity among urban adolescents?”
- “How has sports broadcasting changed since the 1990s?”
Those belong in different papers.
The practical rule for most research papers: one primary research question, with two to three sub-questions maximum. More than that suggests a project that should be split into multiple studies or reformulated around a tighter focus.
Common Mistakes to Avoid When Writing a Research Question
Most of these get mentioned briefly in other articles but deserve more thorough treatment.
Asking a yes/no question. “Does exercise reduce anxiety?” can be answered “yes” by a single study. It doesn’t drive nuanced investigation. Reframe: “To what extent does aerobic exercise frequency predict self-reported anxiety reduction in adults with generalized anxiety disorder?”
Moralizing instead of investigating. “Should companies provide parental leave?” is a policy opinion, not a research question. “What is the relationship between paid parental leave duration and employee retention rates in mid-size US companies?” is.
Answering a question that’s already settled. The literature review exists precisely to check this. If three meta-analyses already converge on your question, your contribution is to add a replication or extend it – not to re-ask it from scratch.
Conflating the topic with the question. “My research question is about burnout in healthcare workers” is not a research question. It’s a topic. Add what you want to know about it.
The “kitchen sink” question. Trying to pack too many variables into one question. Every variable you add to a question multiplies the complexity of your analysis and the likelihood that you won’t be able to answer it.
Writing a question you already know the answer to. This is subtler than it sounds. Confirmation bias often shows up at the question-writing stage – framing the question to point toward a conclusion you’ve already reached. A useful self-check: could your question, in principle, produce a result that surprises you?
Ignoring feasibility. Writing a great question you can’t answer with available time, data, or access. Ambition is good; wishful thinking isn’t.
Research Question Examples by Field
Seeing examples across disciplines is one of the best ways to calibrate what specificity actually looks like in practice.
Psychology: “What is the relationship between sleep duration and emotional regulation ability in adolescents aged 13–17, and does this relationship differ between those with and without ADHD diagnoses?”
Business/Management: “How do remote-first companies in the tech sector manage informal knowledge transfer, and what communication tools are most associated with high knowledge-sharing behavior?”
Public Health: “What is the association between neighborhood walkability scores and physical activity levels in adults aged 40–65 in US metropolitan areas with populations above 500,000?”
Education: “How do flipped classroom models affect academic performance and self-reported engagement among first-year community college students in STEM courses?”
UX Research / Product: “What decision-making patterns do first-time B2B software buyers exhibit during the evaluation phase, and how do those patterns differ from repeat buyers?” (This type of question – when it sits within a UX or user research process– shapes everything from participant recruitment to interview design.)
Environmental Science: “To what extent has urban tree canopy cover in Chicago changed between 2010 and 2023, and what socioeconomic factors predict canopy loss at the census tract level?”
Each of these specifies a population, a phenomenon, a context, and an implicit method. That’s the template.
How Articos Can Help You Build Better Research Questions Faste
If you work in product, UX, or business research, your research questions don’t live in a vacuum – they live inside a research process with timelines, stakeholders, and budget constraints.
One of the less-discussed problems in applied research is that poorly framed questions often don’t surface until after you’ve already recruited participants or started data collection. By then, pivoting is costly.
Articos helps teams work through the research design phase faster. Its AI-powered research framework lets you test whether your research question is specific enough before you commit to a method – by running it against synthetic user personas that surface assumptions, ambiguities, and gaps in your framing early.
Rather than spending weeks on recruitment and scheduling only to realize your question was too broad or asked the wrong population, you can pressure-test your framing in a fraction of the time. Teams using AI for user research report being able to validate or refocus their research questions in hours rather than weeks – which changes what’s actually feasible in a given sprint cycle.
If you’re curious what that looks like in practice, start a free trial here.
Internal Resources Worth Exploring
If this guide is part of a broader research project – especially one involving users, customers, or product decisions – these resources map to the next steps:
- User Research Methods – how to choose the right approach once your question is set
- How to Conduct User Interviews – translating a research question into an interview guide
- Qualitative vs. Quantitative Research – matching your question type to your method
- User Research Process – the full workflow that your research question sits inside
- User Interview Questions – practical question design once your research question is finalized
Final Thoughts
A research question isn’t administrative paperwork you write before the “real” work starts. It is the work – at least in the beginning. Every methodological choice, every data point you collect, every claim you make in your conclusion traces back to it.
The researchers who produce the clearest, most actionable findings aren’t necessarily smarter or better resourced. They’re usually the ones who spent longer than anyone else on the question before they started answering it.
Start there.
FAQs: How to Write a Research Question
Apply the “boundary test”: can someone design a study from your question alone, without asking you clarifying questions? If not, it needs more specificity. Use the 5W+H framework to add population, context, and phenomenon detail until the question draws its own perimeter.
Yes – narrowing and sharpening are legitimate and often necessary. Document the change and the reason for it. Avoid widening the question post-data-collection (HARKing), which undermines research integrity.
Check three things: (1) Is the data accessible? (2) Does a methodological approach exist to answer it? (3) Can it pass ethical review? A question that fails any of these needs reformulation, not just polishing.
Add at least one constraint: a specific population, a defined timeframe, a limited context, or a more precise relationship type. Broad questions produce unmanageable studies. Narrow questions produce answerable ones.
Yes, but they must be coherent – different facets of the same problem, not separate studies stitched together. One primary question plus two to three sub-questions is a practical maximum for most papers.
