Writing an Answerable Dissertation Research Question: A Practical Workshop

Every dissertation supervisor has met the student with a fascinating topic and no question. "I want to study social media and mental health" is a topic. "Among first-year undergraduates in the UK, is time spent on image-based social media associated with body dissatisfaction, and does this association differ by gender?" is a question. The difference decides whether the next year is productive or painful. A good research question tells you what to read, what data to collect, how to analyse it, and when you are finished. This workshop-style guide walks through how to move from a broad interest to an answerable question, with exercises you can do in an afternoon and notes on where a source-linked research assistant such as Yila can speed up the background reading.

What makes a question answerable

An answerable question has five properties.

1. It is specific. It names the population, the phenomenon and, where relevant, the comparison and the outcome.
2. It is feasible. You can realistically get the data, participants or sources within your time, budget and ethical constraints.
3. It is grounded in the literature. It responds to something already known, whether a gap, a contradiction or an untested assumption.
4. It matches a method. The wording points to a design: "how much" suggests measurement, "how do people experience" suggests qualitative inquiry, "what explains" suggests theory-testing or comparison.
5. It matters to someone. An examiner should be able to see why the answer would be useful to a field, a profession or a community.

Most weak questions fail on specificity or feasibility. Students fear that narrowing the question will make the work trivial. In practice the opposite happens: a narrow, well-justified question produces deeper analysis and a stronger thesis.

Step 1: Map your interest

Start by writing a paragraph about what interests you and why. Do not edit. Then underline the nouns. These are your candidate concepts. In the social media example, you might underline "social media", "mental health", "young people", "comparison", "self-esteem" and "Instagram".

Next, for each concept, write down what you already know and what you are unsure about. This exposes vague terms early. "Mental health" covers depression, anxiety, wellbeing, body image and much more. Choosing one is the first act of narrowing.

Step 2: Do a fast scoping read

Before committing to a question, spend a few days scanning the literature. You are not doing a full review yet; you want to know:

• What questions have already been asked and answered convincingly?
• Where do findings disagree?
• Which populations, settings or methods are under-represented?
• What do recent reviews list as directions for future research?

Recent systematic reviews and their "future research" sections are gold for this. They tell you where experts think the gaps are. Read abstracts widely and full texts selectively.

This stage benefits from tools that let you query several papers at once and see exactly where each answer comes from. With Yila, you can ask which populations a set of studies included or what limitations the authors reported, and each answer links back to the passage in the paper. That makes it easier to build a quick, trustworthy picture of the field without re-reading every PDF, and it gives you citations you can check later.

Step 3: Choose a question frame

Frameworks help you name the parts of a question. Pick the one that suits your type of study.

• PICO (Population, Intervention, Comparison, Outcome) for effectiveness questions.
• PEO (Population, Exposure, Outcome) for observational questions about associations.
• SPIDER (Sample, Phenomenon of Interest, Design, Evaluation, Research type) for qualitative and mixed-methods questions.
• SPICE (Setting, Perspective, Intervention, Comparison, Evaluation) for service evaluation and policy questions.

Fill in each element for your topic. If you cannot fill one in, you have found a decision you still need to make.

Exercise

Take your interest paragraph and complete a PEO or SPIDER table. Then write three different questions using the same table, each emphasising a different element. Compare them for feasibility and interest.

Step 4: Test feasibility honestly

Ask yourself:

• Access. Can you reach the population? Do you need gatekeepers? How long will ethical approval take?
• Data. If you plan to use secondary data, does a dataset exist that measures your variables? Can you get access in time?
• Sample size. For quantitative work, do a rough power calculation. For qualitative work, think about how many interviews you can conduct, transcribe and analyse well.
• Skills. Do you have, or can you learn in time, the methods required?
• Time. Map the question onto your timetable. Data collection nearly always takes longer than planned.

If a question fails on feasibility, adjust the population, the setting or the method rather than abandoning the topic. "UK undergraduates" can become "undergraduates at my university". A national survey can become a secondary analysis of an existing dataset.

Step 5: Align question, aims and objectives

A dissertation usually has one main question supported by a few objectives. The objectives break the question into tasks. For example:

Question: Among first-year undergraduates at a UK university, is time spent on image-based social media associated with body dissatisfaction, and does this association differ by gender?

Objectives:

1. To measure self-reported time on image-based platforms.
2. To measure body dissatisfaction using a validated scale.
3. To estimate the association between the two, adjusting for relevant confounders.
4. To test whether gender moderates the association.

Each objective should map to a section of your methods and results. If an objective does not, it is either unnecessary or the methods are incomplete.

Step 6: Get feedback early

Share a one-page summary with your supervisor and, if possible, with peers. Include the question, a short rationale with three to five key references, the intended design, and the main feasibility risks. Ask specifically: Is the question clear? Is it too broad? What would make it more interesting?

Expect to revise. Most strong questions go through several drafts.

Common problems and fixes

The question is really several questions. Split it, and choose one as primary. The others may become secondary objectives or future work.

The question presumes its answer. "How does social media damage students' mental health?" assumes damage. Rephrase neutrally: "Is there an association between...?"

The question is unanswerable with data. Questions about what people "should" do are normative. You can study what people believe they should do, or what outcomes follow from different choices, but you cannot measure "should" directly.

The question is too descriptive for the level. At master's or doctoral level, examiners usually expect explanation, comparison or theory, not just description.

The question drifts. As you read, it is tempting to chase new ideas. Keep a "parking lot" document for interesting tangents, and revisit your question formally at agreed points.

Keeping the literature organised as you go

Your question will rest on dozens of sources. Record, for each key source, what it found, how, in whom, and what it says about gaps. Note the page or section of every claim you might cite. When your supervisor asks where a statement came from, you should be able to find it in seconds. Source-linked notes, whether in a reference manager, a spreadsheet or a research assistant like Yila, save hours in the writing-up stage and protect you from accidental misquotation.

Closing thoughts

An answerable research question is not a constraint on curiosity; it is the tool that turns curiosity into findings. Spend the time to get it right. Map your interest, read strategically, use a frame, test feasibility, and align your objectives. The rest of the dissertation will be easier for it, and your examiners will see a project that knew what it was trying to find out and found it.