How to Write a Research Question (With 30+ Examples)
To write a research question, start with a topic, read enough to see what is still unknown or debated, and narrow it to one group, place or time. Then ask a how, why or to what extent question about it, and test that it is clear, focused, open-ended and answerable with the time and sources you have.
If every question you write comes out either huge ("How does social media affect people?") or finished in one line ("Do teens use Instagram?"), you are not doing anything wrong. The gap between a topic and a good question is where most projects stall, and a simple process closes it. Below, you will follow one example from a loose topic to a finished question, get a fill-in formula for each kind of question, and see more than 30 examples you can adapt.
What is a research question?
A research question is the one question your paper, thesis or study sets out to answer. It is narrower than your topic, and it comes before your answer. It also does a lot of quiet work. It tells you what to read, what evidence to collect, which method fits and when you are done. Without one, a paper tends to drift into an "all about" report that summarizes a topic instead of answering anything.
It helps to see a research question next to the things students often mix it up with. Here is one project at each stage.
Term | What it is | Example |
|---|---|---|
Topic | The broad area you are interested in | Phones and sleep |
Research question | The one open question you will answer | How is phone use after lights-out related to how long first-year students take to fall asleep? |
Hypothesis | A testable prediction of the answer, used in studies | First-year students who use their phones after lights-out take longer to fall asleep than students who do not. |
Thesis statement | Your answer and main claim, used in essays and papers | Late-night phone use slows sleep for first-year students, so residence programs should build in phone-free wind-down time. |
The question stays open, the hypothesis predicts, and the thesis states what you found or argue. Our guide on how to write a thesis statement picks up from there.
What makes a good research question?
Strong research questions pass five tests.
Clear. A reader in your field understands it without asking what you mean, and every key term has one meaning.
Focused. It covers one problem and is narrow enough to answer well in the space you have.
Open-ended. A yes, a no or a quick search cannot answer it. It needs evidence and analysis.
Researchable. Sources or data exist, you can reach them, and you can finish in your time frame.
Worth answering. Someone in your field would care about the answer, not only you.
The last test is the one most people skip. A quick way to check it comes from The Craft of Research by Wayne Booth, Gregory Colomb and Joseph Williams, a classic guide for student researchers. Finish one sentence: I am studying [topic] because I want to find out [your question], so that my reader understands [why it matters]. If you cannot fill in the third part, your question may be tidy but not yet worth a paper.
Most strong questions also share the same parts. Once you can see them, you can build or repair any question.
How to write a research question in 5 steps
Here is the process with one example carried all the way through, so you can watch the question take shape.
1. Choose a topic and read the background
Pick something you are curious about, because you will live with it for weeks. Then read two or three overview sources, such as a recent review article or a textbook chapter, to learn the main terms and debates. No topic yet? Start with how to choose a research paper topic.
Example: Phones and sleep.
2. Narrow it
Ask who, where, when and what. Limit the topic to one group and one setting, and add a time period if it helps. Each limit makes the question easier to answer and easier to search.
Example: Bedtime phone use among first-year students living in university housing.
3. Find what is still unknown or debated
Start by looking for a gap: a group nobody has studied, a result nobody has tested in a new setting, or findings that disagree. It is by far the most common route. When Mats Alvesson and Jörgen Sandberg studied 52 articles from four management journals, they found that every one built its question by spotting a gap. They argue for a second route that they say is more likely to produce interesting work: question an assumption the existing research takes for granted.
Example gap: Suppose most studies you find measure total screen time, but few look at phone use after the lights go out. Example assumption to question: Studies may assume that phones cause poor sleep. What if students who already struggle to sleep are the ones reaching for their phones?
4. Draft three or four versions
Write several questions with different openings, then keep the one that is clearest and most answerable. Drafting more than one shows you both your options and the traps.
Draft | Verdict |
|---|---|
Does phone use hurt students' sleep? | Yes or no, and "hurt" assumes the answer |
How does technology affect sleep? | Far too broad for one paper |
How is phone use after lights-out related to sleep onset time among first-year students in university housing? | Keep. Clear, focused and measurable (sleep onset time is how long it takes to fall asleep) |
How do first-year students describe the role of their phones at bedtime? | Also strong, if you want interviews instead of numbers |
5. Test it, then commit
Run a quick search on your two or three key terms in Google Scholar or a library database. A Duke library guide puts the signals simply: very few results suggest your topic is too narrow, while something like 500,000 results suggests it is too broad. Next, check that you can get the data. Here, a short survey or a one-week sleep diary would work. Finally, make sure you can finish the "so that my reader understands" sentence.
One more fast check: write a one-sentence guess at the answer and paste it into Writing Studio's free Citation Finder. If it comes back with real sources and matching passages, your question has evidence to work with.
Final question: How is phone use after lights-out related to sleep onset time among first-year students living in university housing?
Research question format: a formula for each type
The first words of your question decide what kind of answer you can give, and often which method you will use. William Trochim's Research Methods Knowledge Base sorts questions into three basic types that build on each other: descriptive, relational and causal. You have to describe things before you can relate them, and relate them before you can claim that one causes another, which is why causal questions are usually the most demanding. Add questions about lived experience and the analytical questions used in the humanities, and you get the six formulas below.
Type | Fill-in formula | Example | Usual method |
|---|---|---|---|
Descriptive | What is the [level, pattern or share] of [X] among [group] in [place or time]? | What share of final-year nursing students in Ontario hold paid jobs during clinical placements? | Survey or existing records |
Comparative | How does [X] differ between [group A] and [group B]? | How do reading scores differ between third graders taught with phonics-first and balanced literacy programs in one school district? | Comparison of groups |
Relationship | What is the relationship between [X] and [Y] among [group]? | What is the relationship between weekly hours of paid work and GPA among community college students? | Correlational study |
Causal | What effect does [change] have on [outcome] for [group], compared with [alternative]? | What effect does weekly low-stakes quizzing have on final exam scores in first-year chemistry, compared with weekly homework? | Experiment or trial |
Experience (qualitative) | How do [group] experience or describe [phenomenon] in [setting]? | How do first-generation PhD students describe asking their supervisors for help? | Interviews or focus groups |
Analytical (humanities) | How does [text, artist or source] use [feature] to [effect or argument]? | How does the nested narration in Frankenstein shape readers' sympathy for the creature? | Close reading or archival work |
The link between question and method is more than a writing tip. In a step-by-step guide to forming research questions, Ratan and colleagues show how the question sets the study design: asking how common a condition is leads to a survey, asking about risk factors leads to a case-control or cohort study, and a question about a treatment's effect may lead to a clinical trial. So choose your opening words with your method in mind, and read our guide to research methodology before you plan your data.
The first four types are usually quantitative: you answer them with numbers. Experience and analytical questions are qualitative: you answer them with interviews, texts or observation.
Is your research question too broad or too narrow?
Most weak questions fail on scope, so it is worth checking on its own. A question is too broad if it could fill a book, if its key words could mean ten different things, or if your searches bury you. It is too narrow if one fact or one source answers it, or if you cannot find anything to compare it with. Here is one topic at three scopes:
Too broad: How does climate change affect animals?
About right: How have spring arrival dates of one migratory songbird in the Great Lakes region changed since 2000?
Too narrow: On what date did the first warbler of 2024 arrive at one city park?
Research question examples: weak vs strong
Each pair below fixes one common problem. Every stronger version names a group, a setting or a measure, and none can be answered with a yes or a no.
Weak question | Problem | Stronger question |
|---|---|---|
Is social media bad for teenagers? | Yes or no, and "bad" is vague | What is the relationship between daily time on image-sharing apps and body satisfaction among girls aged 13 to 17? |
Why are dating apps harmful? | Assumes the answer | How do dating app users aged 18 to 25 describe the effect of the apps on their self-esteem? |
What is the effect of global warming on the oceans? | Far too broad | How have marine heatwaves since 2015 changed coral cover on reefs in the northern Great Barrier Reef? |
How are doctors treating diabetes in the US? | A quick search answers it | Which community-level factors predict whether adults with prediabetes complete a diabetes prevention program? |
Is remote work good? | Opinion, no group or measure | How do promotion rates for junior software engineers in fully remote roles compare with those in hybrid roles? |
What do students think about AI? | Too broad, and "AI" is unclear | How do undergraduates decide when using generative AI on an assignment crosses their course's integrity rules? |
Should the minimum wage go up? | Asks for an opinion, not evidence | How did California's 2024 increase in the fast food minimum wage affect staffing levels at franchise restaurants? |
How did World War II change women's lives? | Too broad for one paper | How did wartime factory work change married women's paid employment in Detroit between 1940 and 1950? |
Research question examples by subject and level
Use these as models, not as finished questions. Swap in your own group, setting and measure.
Subject | Level | Research question | Type |
|---|---|---|---|
Psychology | Undergraduate | How does the number of hours slept the night before relate to mood ratings among first-year students during exam week? | Relationship |
Education | Undergraduate | How do primary school teachers in rural districts decide when to use group work? | Experience |
Public health | Master's | What is the relationship between distance to the nearest supermarket and fruit and vegetable intake among low-income households in Atlanta? | Relationship |
Business | Undergraduate | How do independent coffee shops in one city adjust their prices after a rise in wholesale coffee costs? | Descriptive |
Marketing | Master's | What effect does showing a price per use instead of a monthly price have on sign-up intent for a subscription app? | Causal |
Economics | PhD | To what extent did broadband expansion change self-employment rates in rural US counties between 2010 and 2020? | Causal |
Computer science | Master's | How accurately do open-source speech recognition models transcribe accented English, compared with commercial models? | Comparative |
Environmental science | Undergraduate | How does tree canopy cover relate to summer surface temperatures across neighborhoods in Phoenix? | Relationship |
History | Undergraduate | How did local newspapers in Alabama frame the Montgomery bus boycott in 1955 and 1956? | Analytical |
Literature | Undergraduate | How does the narrator's unreliability in The Remains of the Day reveal what Stevens cannot admit to himself? | Analytical |
Sociology | PhD | How do gig delivery workers in Seoul build informal networks to manage risk on the job? | Experience |
Political science | Master's | What is the relationship between local newspaper closures and voter turnout in US county elections? | Relationship |
Law | Master's | How have US courts treated works made with generative AI in copyright disputes since 2022? | Analytical |
Engineering | PhD | How does replacing part of the sand with recycled plastic change the compressive strength of concrete over 90 days of curing? | Causal |
PICO, SPIDER and FINER: frameworks that sharpen a question
If you study health, nursing or medicine, you will meet PICO. It splits a question into Population, Intervention, Comparison and Outcome, and PICOT adds Time.
Example: In adults over 65 recovering from hip surgery (P), how does walking within 24 hours (I), compared with later mobilization (C), affect length of hospital stay (O) during the first admission (T)?
Qualitative researchers often use SPIDER instead: Sample, Phenomenon of Interest, Design, Evaluation and Research type. The choice matters more than it looks, because the framework you pick shapes your literature search. When Methley and colleagues searched three databases for qualitative studies using each framework, PICO returned 23,758 results and SPIDER returned 239. PICO found 18 relevant studies and SPIDER found 13. SPIDER was far more efficient, but it missed five studies. The authors recommend PICO when you need to be thorough, and PICOS (PICO plus study design) when time is short.
FINER is a final check that works in any field. Ask whether your question is Feasible, Interesting, Novel, Ethical and Relevant. If one letter fails, fix the question now rather than halfway through the project.
Main question and sub-questions for a thesis
A paper usually needs one research question. A thesis or dissertation often has one main question and two to four sub-questions. The sub-questions break the main one into parts you can answer one at a time, and together they should answer it fully.
Main question: How is phone use after lights-out related to sleep onset time among first-year students living in university housing?
How much do first-year students use their phones after lights-out, and for what?
How does sleep onset time differ between students with high and low after-dark phone use?
How do students explain their own bedtime phone habits?
Each sub-question points to its own piece of analysis, and the last one adds a qualitative angle that numbers alone would miss.
Common mistakes and how to fix them
Mistake | Looks like | Fix |
|---|---|---|
Two questions in one | How do stress and diet affect grades and sleep? | Split it, or keep one cause and one outcome. |
Promises a result | Can this program prove that tutoring causes significant gains? | Drop "prove" and "significant" and let your results decide: How large are reading gains for students who get weekly tutoring, compared with those who do not? |
Wrong type for your data | What effect does part-time work have on grades? (with one survey at one point in time) | A single survey can show a relationship, not an effect: What is the relationship between weekly hours of part-time work and grades among second-year students? |
Out of reach | How do all US universities handle AI misuse? | Match the scope to your data: one university, or a sample of 20 published policies. |
Checklist: is your research question ready?
It is one sentence and ends with a question mark.
It opens with how, why, what, to what extent or what is the relationship, not is or does.
It names a group, and a place or time where it matters.
Every key term has one clear meaning you could measure or observe.
It does not assume its own answer.
A quick search finds enough sources: not none, and not a flood.
You can get the data or sources and finish in your time frame.
You can complete the sentence "so that my reader understands..."
Once your question passes, the hardest part is behind you, because you now know what to read and what to look for. If you want one place to search sources, check each citation against the exact passage and draft your paper, Writing Studio is free to start.
Frequently asked questions
How many research questions should I have?
One for most essays and papers. A thesis or dissertation often has one main question plus two to four sub-questions that together answer it.
Can a research question be a yes or no question?
Usually not, because a yes or no leaves little to analyze. Even clinical questions in PICO form work better framed as how much or compared with what.
Where does the research question go in a paper?
Usually at the end of the introduction, right after you show what is unknown. In a thesis it appears in the first chapter and returns in the methods and conclusion.
What is the difference between a research question and a research objective?
The question says what you want to know. The objective says what you will do to find out, for example "to measure sleep onset time in 200 first-year students over one week."
What are the three types of research questions?
Descriptive, relational and causal. Descriptive questions describe what exists, relational questions look at how two things vary together, and causal questions test whether one thing changes another.
