Two sentences, same topic. “I am writing about social media and sleep.” And: “How does evening smartphone use relate to the sleep duration of first year students at one UK university?” The first is a topic, the second is a research question.
You write a research question by narrowing your topic to a single group, period and setting, choosing one type of question, and then cutting the wording until you can answer it with the time, the data and the skills you actually have. By the end you will have one question that fits your introduction and your supervisor’s expectations.
📌 Key Takeaways
- A research question narrows the topic, the aim states the purpose.
- Originality at dissertation level sits well below PhD level.
- Yes or no questions do not work as research questions.
- Five types: description, explanation, prediction, design, evaluation.
- Feasibility decides: time, literature, access and skills.
Create a survey for free
With empirio.ai you can create a modern online survey in minutes — with hosting in the EU.
- AI-built survey
- Adjust by drag & drop
- Real-time analysis
What is a research question?
A research question is the one question your project answers: it narrows your topic to a section you can actually work through, fixes what will be examined, and decides which methods and which literature come into play at all.
Skills@Library at the University of Leeds makes the practical case in “The Final Chapter”, its guide to final year projects: “It is helpful to turn your topic into a research question, or set of research questions, to answer even if you do not explicitly state these in your project write up.” Many dissertations never print the question at all, and it still decides what goes into the literature review.
Research question, thesis and hypothesis end up in one pot more often than not. The differences between research question, thesis and hypothesis are covered separately, and the wider view sits in our overview of empirical research.
Research question, aim and objectives
Your research question says what you want to find out, your aim says what you want to achieve with the answer, and your objectives are the steps that get you there. British module handbooks may ask for all three, so check yours early.
Skills@Library at the University of Leeds defines the aim as a statement that expresses the purpose of your research, and the objectives as the steps needed to reach it, each one specific, measurable, realistic, achievable and time constrained. The difference shows up the moment you draft your introduction, because all three sit next to each other there.
| Element | Answers | Example |
|---|---|---|
| Topic | What are you writing about? | Smartphone use and sleep among students |
| Research question | What do you want to find out? | How does evening smartphone use relate to sleep duration? |
| Aim | What is the answer for? | Starting points for first year sleep advice |
| Objectives | Which steps get you there? | Review the literature, run a survey, analyse results |
An aim is usually written with a verb that names a result: describe, compare, explain, test, derive. “The aim is to describe the relationship” works. “The aim is to engage with the topic” does not, because no result follows. The wider sequence is laid out in our walkthrough of the empirical research process.
How to write a research question in five steps
A research question rarely arrives in one go. Most take several rounds, and only when the question stands do you decide on method and material.
Order matters here. Anyone who starts with the method builds the question around the tool instead of finding the tool that fits the question. Leeds puts one step before all of them: “Discuss any topic ideas with your supervisor as early as possible to ensure that you are researching an appropriate area.”
- Narrow the topic. Set three boundaries: a group, a period and a setting. “Social media and sleep” becomes “first year students, this academic year, one university”.
- Run through the question words. Put who, what, how, why, where and when to the narrowed topic in turn.
- Choose a type. Decide whether you want to describe, explain, predict, design or evaluate. The type fixes the shape of the sentence.
- Cut the wording. Write the question as one sentence and delete every word that settles nothing. Over two lines, two questions are hiding inside it.
- Check that it is manageable. Leeds asks whether enough background literature exists, whether you have the skills, and whether there is time to gather and analyse the data.
💡 Tip
Take two versions to your supervisor, one narrow and one wide. Feedback gets more concrete when there is something to compare, and you find out what scope your school expects.
How original does your research question have to be?
Your research question does not have to close a gap in the literature, at least not at undergraduate or taught postgraduate level. Skills@Library at the University of Leeds heads its originality criteria with a bracket that says the quiet part out loud: “Original (to some extent - this is not a PhD!)”.
Leeds spells out what counts instead, and the list is far more forgiving than the research gap advice circulating online. An original angle can mean challenging or corroborating an existing interpretation, using existing sources in new ways, employing new methodologies, or applying an approach from another field. Leeds then adds a sentence most guides leave out: your project does not have to do all of these things.
Why the misconception is expensive: a student chasing a genuine gap reads for weeks before the question exists at all. Replication is an accepted route at this level, and Leeds says so directly: repeat a study but apply different criteria, limiting it to a particular country, gender or industry.
⚠️ Warning
Originality is judged against the marking criteria in your module handbook, not against a journal. Where the handbook stays silent, your supervisor decides what counts as an appropriate area. No article can settle that for your department.
Create a survey for free
With empirio.ai you can create a modern online survey in minutes — with hosting in the EU.
- AI-built survey
- Adjust by drag & drop
- Real-time analysis
The five types of research question
Research questions fall into five basic types, and each calls for a different shape of sentence. Description, explanation, prediction, design and evaluation cover almost everything a dissertation asks. Once you know your type, most of the wording work is done.
| Type | Typical wording | What it suits |
|---|---|---|
| Description | What is the case? What does it look like? | Capturing states and distributions |
| Explanation | Why or in what way is something the case? | Testing and reasoning about relationships |
| Prediction | How will something develop? | Estimating change over time |
| Design | Which measures achieve the goal? | Deriving recommendations |
| Evaluation | How does something rate against criterion X? | Placing what already exists |
Description and explanation are the most common types in a dissertation with its own data collection. Prediction questions need data from several points in time, and design questions need a setting where you may propose measures. How the types line up with research designs sits in our article on exploratory, descriptive, explanatory and causal research.
A research question is finished when you know which data will answer it and where that data comes from.
From research question to your own survey
Your research question decides whether you need data of your own, not your timetable. Questions about attitudes, habits, satisfaction or experiences rarely have matching statistics behind them, and a survey of your own is the usual route. Questions about official figures can be answered from existing material.
Structure stays the same either way. One main question sits at the centre, and sub questions specify it. Moving into data collection, the main question does not go into the questionnaire. Each sub question turns into several items your participants can answer without subject knowledge. How a full questionnaire grows out of it sits in our guide to questionnaire design.
⚠️ Warning
A question about causes cannot be answered with a survey, only a relationship can. A causal claim needs an experiment with random allocation, rarely feasible inside a dissertation. Write the question as a relationship question from the start.
Cost is rarely the obstacle for a student survey. empirio.ai, an online survey tool from Germany, allows 50 responses per month on the free plan (as of August 2026). EU hosting is included on the free plan (as of August 2026), which matters if your ethics form asks where responses are stored.
Common mistakes when writing a research question
Most weak research questions come from time pressure rather than ignorance. A student who gets the topic late writes the question in one sitting and only notices in the methodology chapter that it will not carry. Four patterns turn up again and again.
All four show up early if you read the question aloud to someone outside your subject. Everywhere that person has to ask a follow up question, something in your wording has not been settled.
Mistake 1: two research questions in one sentence
Two questions in one sentence is the classic, usually joined by an “and”. “How do students use social media, and what effect does that have on their sleep?” calls for two studies with two methods. Pick one and demote the other to a sub question.
Mistake 2: the research question contains its own answer
Leading questions give the result away in advance. “Why does evening smartphone use harm sleep?” assumes the harm, and the dissertation can only confirm it. Put neutrally: “How does evening smartphone use relate to sleep duration?” A finding of no relationship then stays usable.
Mistake 3: evaluative words are left undefined
Words such as successful, useful, effective or sustainable sound precise and are not. If “a successful degree” appears in your question, decide beforehand whether you mean marks, completion within the standard programme length, or satisfaction. Without that decision nothing can be measured, and a viva panel asks it first.
Mistake 4: question and method do not match
A why question calls for a different design than a how many question. Motives need open answers or interviews, frequencies need a sample that matches the target group. Where the two do not match, the problem surfaces at the analysis stage.
Conclusion
A research question is good when it saves you work instead of making it. Your question tells you which literature to read, which to leave out, and whether you need data of your own. Cut it twice before it goes into your proposal, and remember that originality at this level means an angle, not a gap nobody has ever touched!
What to read next
- Want testable assumptions instead? How to write a hypothesis
- Short of ideas for a question? Research questions generator
- Still stuck on your title? Formulating your dissertation title
Your research question is settled and you need data of your own?
With empirio.ai you turn your sub questions into a questionnaire and start for free.