Aim and research question: how to narrow a dissertation topic
The aim states what your dissertation sets out to achieve; the research question states the precise question it answers. A topic is not a question — "social media and mental health" is a subject area, while "how do Swedish upper-secondary students describe the effect of Instagram use on their sleep?" is a question you can actually answer with data. A question is ready when it passes four tests: it is answerable with the method and time you have, it is narrow enough to be finished, it is genuinely open, and it has not already been definitively answered.
Key points
- A topic names a subject; a research question names what will be answered.
- Narrow along population, period, place and phenomenon — not by cutting words.
- Yes/no questions and questions with the answer built in are both unusable.
- The question determines the method, never the other way round.
- The question is allowed to change while you read; it must be fixed before you collect data.
What is the difference between aim and research question?
The aim is the purpose of the study and the research question is the specific question the study answers; a dissertation needs both, and they are not rephrasings of each other. The aim can be stated as an intention — "the aim is to examine how…" — while the question ends in a question mark and is answered explicitly in your conclusion.
A simple test: read your conclusion and see whether it answers the question you posed. If it answers something adjacent, the mismatch is between the aim and the question, and the marker will find it. This is the most common structural criticism in dissertation feedback, above any issue of method or referencing.
Why is a topic not a research question?
A topic identifies a field; a question specifies what you want to know within it, and only a question can be answered. "Remote work and productivity" tells the reader nothing about what you will conclude, because a hundred different dissertations would fit under it.
The consequence of starting with only a topic is a dissertation that describes rather than argues. You end up summarising what has been written, because with no question there is nothing to answer — and description without an argument is where a middling grade comes from, whatever the quality of the writing.
How do you narrow a topic that is too broad?
Narrow along four dimensions rather than by deleting words: population, period, place and phenomenon. Each one turns an unanswerable subject into something a single student can actually study in a term.
| Dimension | Broad | Narrowed |
|---|---|---|
| Population | Employees | Nurses in their first year of practice |
| Period | Recent years | 2020–2023 |
| Place | Sweden | Two hospitals in Region Skåne |
| Phenomenon | Wellbeing | Self-reported emotional exhaustion |
| Result | How do first-year nurses at two hospitals in Region Skåne describe emotional exhaustion in the period 2020–2023? | |
What makes a question answerable?
An answerable question names something you can observe, measure or interpret with the data you can realistically obtain. "Is inequality unjust?" is a philosophical question, not an empirical one; "how do recipients of income support describe their experience of the assessment process?" can be answered with twelve interviews.
Run three cheap checks before you commit. Can you name the data that would answer it? Can you obtain that data in the time you have? Would two different answers both be publishable, or is one of them the answer you already assume? A question that fails the third check is an argument in disguise.
Which questions should you avoid?
Four types fail at the proposal stage, consistently:
- Yes/no questions. "Does stress affect exam results?" is answered in one word. Ask how, or under what conditions, or for whom.
- Questions with the answer inside. "Why is remote work harmful to team cohesion?" assumes the finding, and the dissertation is then a search for confirmation.
- Double questions. Two questions joined by "and" turn into two dissertations, and the second one is always the one that gets rushed.
- Questions nobody disputes. If the answer is in the first textbook chapter, there is no contribution to make.
Alvesson and Sandberg (2011) argue that the strongest questions come from challenging an assumption in the existing literature rather than from spotting a gap in it — gap-spotting produces safe questions, while problematisation produces interesting ones.
How does the question decide the method?
The question determines the method, and choosing the method first is how dissertations end up answering the wrong thing. A question starting with "how do people experience" needs qualitative data; one starting with "to what extent is X associated with Y" needs quantitative data.
Watch the verb in your question. Describe, explore and understand point to interviews or observation; compare, measure and predict point to a survey or an existing dataset. If your verb and your planned method disagree, one of them is wrong, and it is usually the method that was chosen because it felt easier. Our article on choosing a research method goes through the trade-off in detail.
How many research questions should a dissertation have?
One main question, optionally with two or three sub-questions that together answer it. More than that and the dissertation loses its spine, because each question needs its own data, its own analysis and its own place in the conclusion.
Sub-questions should be components, not siblings. If you can answer sub-question two without having answered sub-question one, they are probably two separate studies. A good test is whether the conclusion can answer the main question by assembling the sub-answers; if it cannot, the structure is wrong regardless of how good each part is.
When is the question allowed to change?
The question can and should change while you read the literature, and it must be fixed before you collect data. Changing it during the reading phase is a sign the reading is working; changing it after the interviews means the data you have was gathered to answer something else.
Keep a dated note of each version and one line on why it changed. That record is useful twice: in supervision, where "why this question" is the first thing you will be asked, and in your method chapter, where the narrowing process is itself worth describing. If you are working through a stack of articles to find the argument you want to challenge, summarising the papers you have uploaded makes the comparison faster than rereading each one. Booth, Colomb and Williams (2016) treat this iteration between reading and question as the normal working method rather than a sign of indecision.
How do you test the question before the proposal?
Say it out loud to someone outside your subject and ask them what they think you will find. If they cannot guess what an answer would even look like, the question is too vague; if they can tell you the answer, it is too obvious.
Then write the conclusion you expect, in three sentences, before you have any data. This is not prejudging the result — it is checking that the question has an answerable shape. If you cannot write a plausible answer, you cannot design a study that produces one. The rest of the process, from proposal to submission, is covered in our step-by-step guide to writing a dissertation.
Frequently asked questions
How long should a research question be?
One sentence, usually between fifteen and thirty words. If it needs a semicolon it is probably two questions. Length is not the real test, though — whether a reader can restate it after hearing it once is.
Should the research question be in the introduction?
Yes, at the end of the introduction, after the background that makes it necessary and before the method that answers it. Placing it earlier leaves the reader without the context to see why it matters; placing it later makes the introduction read as though it is circling something.
Can a research question be a hypothesis?
They are different tools. A hypothesis states a testable expectation and belongs in quantitative work; a question asks something open. Some quantitative dissertations carry both — the question frames the study and the hypotheses operationalise it.
What if my question has already been answered?
That is usually good news, not a dead end. A study answered in one context, population or period can be genuinely open in another, and replication is undervalued. What does not work is repeating an existing study in the same setting without saying what your version adds.
How do I find a question in the first place?
Read recent articles in your area and go straight to the "limitations" and "future research" sections, where authors write down what they could not do. Then look for a claim that several papers repeat without evidence — the assumption everyone shares is often the most productive thing to question.