How to write a research question for your dissertation (good + bad examples)
Info: 5204 words (21 pages) Study Guides
Published: 14 Oct 2026
A step-by-step method for turning a broad topic into a focused, answerable dissertation research question, with 12 before-and-after rewrites across disciplines, templates for qualitative and quantitative studies, and a checklist to run before you see your supervisor.
The short answer
A good dissertation research question is a single, open question that names who or what you are studying, in what context, and what you want to know about it, and that you can realistically answer with the data, time and skills you have. To write one: start with a topic, read to find a gap, narrow to a specific problem, draft the question with a stem that fits your method (How…, To what extent…, What is the relationship between…), test it against seven criteria, then split it into two to four sub-questions that map onto your objectives.
Weak: Does social media affect students?
Strong: How does daily use of short-form video apps relate to self-reported concentration during independent study among first-year undergraduates at UK universities?
UK Essays writing support
Struggling to find the right dissertation question?
Tell us your subject and any ideas you have. An experienced specialist in your subject will develop focused research question or title options tailored to your interests and course requirements.
All our writers hold at least a 2:1 degree from a reputable university. Most hold a master’s degree or PhD, and/or have teaching experience.
1. What a research question is (and what it isn’t)
A research question is the specific question your dissertation exists to answer. Everything else in the document hangs off it: the literature review explains why the question matters and what is already known, the methodology explains how you will answer it, the findings chapter answers it, and the conclusion states the answer and its limits. Examiners literally read the question first, then check whether the dissertation delivers on it.
Most students’ early drafts fail not because they are “bad questions” but because they are a different thing wearing a question mark. Here is how to tell them apart.
| Element | What it does | Example (same project) |
|---|---|---|
| Topic | The broad area. Too big to answer. | Remote and hybrid working |
| Research problem | The specific gap, tension or unknown within the topic. | Firms have adopted hybrid working at scale, but we know little about how it affects informal knowledge-sharing in technical teams. |
| Research aim | A one-sentence statement of purpose, usually starting “To explore / examine / evaluate…”. | To examine how hybrid working shapes collaboration in software development teams. |
| Research question | The precise question the study will answer. | How do hybrid working arrangements influence informal knowledge-sharing among software developers in UK mid-sized technology firms? |
| Research objectives | The concrete steps you will take to answer it. | 1. Map the hybrid patterns in use. 2. Interview 15 developers. 3. Identify practices that sustain knowledge-sharing. |
| Hypothesis | A testable prediction of the answer (mainly quantitative studies). | Teams with ≥3 co-located days per week report higher knowledge-sharing scores than teams with ≤2. |
| Interview / survey question | What you actually ask participants. Never the same as the research question. | “Can you describe the last time you learned something from a colleague without scheduling it?” |
The quickest diagnostic
Read your draft aloud and ask: “Could I describe the data that would answer this?” If the honest answer is “all of it” or “I’m not sure”, you still have a topic or an aim, not a question.
2. The 7 tests of a good research question
Different universities phrase these differently (you may see “clear, focused, answerable, relevant, significant” or the medical acronym FINER), but every list boils down to the same seven tests. Run each draft through all of them.
| Test | What it means | Fails | Passes |
|---|---|---|---|
| 1. Focused | One question, one phenomenon. Not two questions joined with “and”. | How do teachers and parents view homework, and does it raise attainment? | How do primary school teachers in Wales perceive the purpose of homework for Years 3–6? |
| 2. Specific | Every key term is defined or definable; population, context and time frame are named. | What effect does technology have on learning? | How does weekly use of adaptive maths software relate to end-of-year attainment among Year 7 pupils in two English academies? |
| 3. Answerable | It can be answered with evidence (data or sources), not opinion or belief. | Is capitalism fair? | How do gig-economy workers in Manchester describe fairness in platform pay-setting? |
| 4. Feasible | You can get the data, with your skills, in your time, within ethics approval. | How has childhood nutrition changed across sub-Saharan Africa since 1960? | How do mothers attending two Nairobi clinics make decisions about complementary feeding? |
| 5. Open | It cannot be answered with yes, no, a number, or a single fact found in one search. | Has youth unemployment risen since 2020? | How have changes in the apprenticeship levy shaped employers’ hiring of 16–24-year-olds since 2020? |
| 6. Neutral | It contains no built-in assumption, value judgement or desired answer. | Why does fast fashion damage workers’ lives? | How do garment workers in Dhaka describe the impact of fast-fashion production targets on their working conditions? |
| 7. Worth asking | Answering it adds something: a gap, a new context, a contested finding. It passes the “so what?” test. | What percentage of UK adults use Instagram? | How do UK adults over 65 who joined Instagram during 2020 use it to maintain family relationships? |
The medical sciences’ FINER criteria (Feasible, Interesting, Novel, Ethical, Relevant) cover the same ground and add “ethical”, which matters if you plan to collect data from people. See section 9.
3. How to write a research question in 6 steps
The steps below run one worked example from a topic to a finished question. Expect to loop back: most students draft five to ten versions before the one that goes in the proposal.
Step 1: Start with a topic you can live with for months
Choose something within your degree programme that you are genuinely curious about, and that has enough published work for a literature review but not so much that the question has already been answered to death. If you have a choice of supervisors, pick a topic one of them actively researches.
Topic Remote and hybrid working
Step 2: Read to find the gap
Skim ten to fifteen recent sources (review articles and the most-cited papers from the last five years are the fastest route). You are not reading to learn everything; you are reading to answer three questions:
- What is already well established?
- Where do findings disagree?
- What populations, contexts, methods or time periods have been neglected?
Write the gap down in one sentence beginning “We know X, but we don’t know Y.” That sentence is your research problem, and it will become the opening of your proposal.
Niche Hybrid working in software teams
Gap Studies measure productivity, but few examine informal, unplanned knowledge-sharing, and almost none look at mid-sized firms rather than large corporations.
Step 3: Narrow with the four scoping levers
Broad questions become specific when you pin down four things. You will rarely need all four, but you need at least two.
| Lever | Ask yourself | Worked example |
|---|---|---|
| Population | Who, exactly? Age, role, group, sample. | Software developers |
| Context | Where? Which sector, organisation, country, setting? | UK mid-sized technology firms (50–250 staff) |
| Variables or phenomenon | What specifically are you looking at? Which outcome, experience, relationship or text? | Hybrid working patterns → informal knowledge-sharing |
| Time frame | When? A period, a cohort, before/after an event. | Current practice (2025–26) |
Step 4: Draft the question with the right stem
The first two or three words of your question signal what kind of study it is. “Does” and “Is” invite yes/no answers. “Why” often smuggles in an assumption. The stems below keep the question open and point to a method (more on this in section 5).
- How do / How does… (process, experience, mechanism)
- What are the… (characteristics, factors, perceptions, strategies)
- To what extent does… (degree of effect or association)
- What is the relationship between… (correlation)
- How does X compare with Y in terms of Z… (comparison)
- In what ways does… (interpretive or textual analysis)
First draft Does hybrid working hurt collaboration in tech companies?
Second draft How does hybrid working affect knowledge-sharing among software developers?
Third draft How do hybrid working arrangements influence informal knowledge-sharing among software developers in UK mid-sized technology firms?
Step 5: Stress-test it
Run the draft through the seven tests, then ask three harder questions that supervisors ask:
- Could I describe the data that answers this? (Here: interviews with developers across three to four firms, plus each firm’s hybrid policy documents.)
- Could I collect that data in my timeline, with ethics approval? (Fifteen one-hour interviews in six weeks: yes, if recruitment starts early.)
- If the data came back the opposite of what I expect, would that still be a legitimate finding? If only one answer is acceptable to you, the question is closed and you are about to write a confirmation exercise.
Step 6: Break it into sub-questions and align with objectives
A main question is usually still too big to answer directly. Two to four sub-questions divide it into parts you can answer from specific data, and each sub-question should map to one objective and one method. Section 8 shows the full structure for the worked example.
A note on timing
Get the question to third-draft quality before writing your literature review in full. Students who write the review first tend to end up with a question shaped by whatever they happened to read, rather than a review shaped by the question. A rough review (Step 2) is enough to find the gap; the full review comes after the question is agreed.
4. Research question templates by type
Most dissertation research questions fall into one of seven types. Identify which type your project is, use the template as scaffolding, then edit it until it reads naturally. Square brackets mark the parts you fill in.
| Type | Template | Example | Typical design |
|---|---|---|---|
| Descriptive | What are the [characteristics / patterns / prevalence] of [phenomenon] among [population] in [context]? | What are the most common coping strategies reported by final-year nursing students in Scottish universities during clinical placements? | Survey, content analysis, secondary data |
| Comparative | How does [group / case A] compare with [group / case B] in terms of [outcome]? | How do Finnish and English primary curricula compare in the time allocated to unstructured play for 5–7-year-olds? | Comparative case study, cross-sectional survey |
| Relational (correlational) | What is the relationship between [variable A] and [variable B] among [population]? | What is the relationship between sleep duration and exam performance among second-year undergraduates at one UK university? | Cross-sectional survey, regression |
| Causal / effect | To what extent does [intervention / exposure] affect [outcome] in [population], compared with [comparison]? | To what extent does a six-week mindfulness programme reduce self-reported exam anxiety among Year 11 pupils compared with a waiting-list control? | Experiment, quasi-experiment, longitudinal |
| Exploratory / experiential | How do [population] experience / understand / make sense of [phenomenon] in [context]? | How do international students at UK universities experience academic feedback during their first semester? | Interviews, focus groups, phenomenology |
| Evaluative | How effective is [policy / programme / tool] at achieving [stated goal] for [population], and what factors shape its effectiveness? | How effective is the Kickstart Scheme at moving 18–24-year-olds in the North East into sustained employment, and what factors shape its outcomes? | Mixed methods, programme evaluation |
| Interpretive / textual | In what ways does [text / artefact / discourse] [represent / construct / negotiate] [concept], and with what effect? | In what ways do UK broadsheet editorials between 2016 and 2019 construct “the will of the people”, and how does this framing shift over time? | Discourse analysis, close reading, archival research |
Design-based disciplines (engineering, computer science, architecture) often add a design question: “How can [artefact] be designed to achieve [performance goal] under [constraints]?” This is legitimate, provided the dissertation evaluates the artefact against the goal rather than simply building it.
5. Matching your question to your method
This is the alignment problem that most guides skip and most examiners check first. The wording of a research question commits you to a kind of evidence. If the question says “to what extent” but your design is six interviews, or the question says “how do participants experience” but your design is a 200-person survey, the proposal is misaligned before you collect anything.
| Quantitative | Qualitative | Mixed methods | |
|---|---|---|---|
| What the question asks about | Relationships, differences, prevalence, prediction, effect size | Experience, meaning, process, interpretation, how something is done or understood | Both: typically a measurable pattern plus an explanation of it |
| Typical stems | To what extent… · What is the relationship between… · What is the effect of… · How does X differ between… | How do… · What are the experiences of… · In what ways do… · How is X understood by… | A quantitative main question with a qualitative sub-question (“…and how do participants explain this pattern?”), or vice versa |
| Needs | Named variables, a defined population, an implied measurement | A named group, a phenomenon, a context | A clear statement of which strand leads and how they integrate |
| Usually paired with | Hypotheses (H1, H0) | No hypotheses; sometimes a conceptual framework | Hypotheses for the quantitative strand only |
| Example | To what extent does the number of co-located days per week predict knowledge-sharing scores among developers in hybrid teams? | How do developers in hybrid teams describe the ways informal knowledge-sharing happens, or fails to happen? | What is the relationship between co-location days and knowledge-sharing scores in hybrid development teams, and how do developers account for that relationship? |
If your question and method don’t match
Rewrite the question, not the method. Your underlying interest is usually clearly qualitative or quantitative; the question just hasn’t been worded to reflect it yet. A student drawn to interviews who has written “To what extent does X affect Y?” almost always means “How do people experience X in relation to Y?”
6. 12 bad vs good research question examples, by discipline
Each example shows a typical first draft, the specific fault, and a rewrite that would pass at proposal stage. Notice that the rewrites are not simply longer: they each add a population, a context, and a defined phenomenon, and they swap closed or loaded stems for open ones.
Business & Management
WeakDoes remote working affect productivity?
Fault: yes/no stem; “productivity” undefined; no population or sector; answered dozens of times already.
StrongHow do hybrid working arrangements influence informal knowledge-sharing among software developers in UK mid-sized technology firms?
Why it works: open “how” stem; names the phenomenon (informal knowledge-sharing), population, firm size and country; targets a documented gap.
Psychology
WeakWhy is social media bad for teenagers’ mental health?
Fault: leading: assumes the answer (“bad”) before collecting data; “social media” and “mental health” are both too broad to measure.
StrongWhat is the relationship between daily time spent on image-based social media platforms and body-image satisfaction among 16–18-year-olds in England?
Why it works: neutral relational stem; two measurable variables; defined platform type, age range and country.
Education
WeakWhat is the best way to teach maths?
Fault: value judgement (“best”); no level, topic, or learner group; unanswerable within one study.
StrongTo what extent do weekly retrieval-practice quizzes improve GCSE mathematics attainment among Year 10 pupils in two English secondary schools, compared with traditional homework?
Why it works: names an intervention, outcome, population, setting and comparison; “to what extent” leaves room for a nuanced answer.
Nursing & Health
WeakHow can we stop nurse burnout?
Fault: demands a solution rather than understanding; “we” is undefined; scope is global.
StrongWhat factors do newly qualified nurses working on NHS acute medical wards identify as contributing to burnout during their first year of practice?
Why it works: exploratory stem suited to interviews; specific career stage, setting and time period; findings can inform solutions without promising one.
Computer Science
WeakIs AI good for cybersecurity?
Fault: yes/no; “AI” and “cybersecurity” are entire fields; “good” is a value judgement with no metric.
StrongHow does the detection accuracy and false-positive rate of a transformer-based intrusion detection model compare with a signature-based system when evaluated on the CIC-IDS2017 dataset?
Why it works: comparative stem; two named metrics; defined model class, baseline and dataset, so the study is reproducible and feasible.
History
WeakWhat caused the First World War?
Fault: a century of historiography already answers it; no sources a student could realistically add; far too broad.
StrongHow did three provincial English newspapers frame the July Crisis of 1914 in the weeks before Britain’s declaration of war, and what does this suggest about public opinion beyond London?
Why it works: defined sources, period and place; original angle (provincial vs metropolitan opinion); answerable from accessible digitised archives.
English Literature
WeakWhat are the themes in Jane Eyre?
Fault: purely descriptive; answered by any study guide; no argument to make.
StrongIn what ways does Brontë’s use of retrospective first-person narration in Jane Eyre complicate the novel’s treatment of female autonomy?
Why it works: names a technique and a concept and asks how they interact; requires close reading and argument rather than summary.
Sociology
WeakHow has society changed because of the internet?
Fault: “society” and “the internet” are unbounded; no population, mechanism or time frame; presumes causation.
StrongHow do first-generation university students in the UK use online peer communities to build social capital during their first year of study?
Why it works: specific group, specific practice, specific concept (social capital) with an established literature, bounded time frame.
Environmental Science
WeakIs climate change affecting birds?
Fault: yes/no; already answered in general; no species, region, or variable.
StrongHow have the first-egg dates of blue tits (Cyanistes caeruleus) in southern England shifted in relation to mean March–April temperatures between 1990 and 2020?
Why it works: one species, one measurable outcome, one predictor, defined region and period; answerable from existing nest-record datasets.
Law
WeakShould the UK reform its defamation laws?
Fault: asks for a policy opinion; no specific provision, claimant type or evidence base.
StrongTo what extent has the “serious harm” threshold in section 1 of the Defamation Act 2013 altered the outcomes of defamation claims brought by corporate claimants in England and Wales?
Why it works: names a specific provision, a class of claimant and a jurisdiction; answerable through doctrinal analysis of reported cases.
Marketing
WeakHow much do influencers affect sales?
Fault: requires proprietary sales data you cannot get; “influencers” undefined; no product category or market.
StrongHow does the perceived authenticity of micro-influencers affect purchase intention for sustainable fashion brands among Gen Z consumers in Ireland?
Why it works: swaps unobtainable sales data for a measurable proxy (purchase intention); defines influencer tier, category, audience and market.
Engineering
WeakHow can bridges be made safer?
Fault: an entire profession’s remit; no structure type, failure mode or material.
StrongHow does the fatigue life of additively manufactured titanium lattice joints compare with conventionally machined joints under cyclic loading representative of pedestrian footbridges?
Why it works: comparative stem; one material, one property, one loading case; testable in a university lab within a project timeline.
Three more faults worth recognising
The double-barrelled question
WeakHow do students perceive online learning, and does it improve their grades?
Fault: two different questions needing two different datasets (interviews and grade records). Either becomes the main question and the other a sub-question, or one is dropped.
StrongHow do second-year undergraduates at one UK university perceive the impact of online seminars on their learning?
The trivially narrow question
WeakHow many students use the campus library on Tuesdays?
Fault: answerable with a headcount; passes “feasible” but fails “worth asking”. Narrow is good; trivial is not.
StrongHow do commuter students’ patterns of library use differ from those of campus residents, and what do these patterns reveal about barriers to access?
The interview question in disguise
WeakWhat do you find most stressful about your job?
Fault: addressed to a participant, not about a phenomenon. This is something you ask in an interview to help answer the real research question.
StrongWhat sources of occupational stress do secondary school teachers in their first three years of practice identify, and how do they manage them?
7. Common mistakes and how to fix them fast
Use this as a troubleshooting table: find the symptom that matches your draft, then apply the fix.
| Symptom | What it usually means | Fix |
|---|---|---|
| Starts with “Does”, “Is”, “Has”, “Can” | Closed question; invites yes/no | Rewrite with “To what extent”, “How”, or “What is the relationship between” |
| Starts with “Why” | Often embeds an assumption that the thing is true | Check the premise. Replace with “What factors…” or “How do participants explain…” |
| Contains “best”, “good”, “bad”, “should”, “better” | Value judgement or policy demand | Replace with a measurable criterion: “effective at achieving X”, “associated with Y” |
| Contains “and” joining two clauses | Double-barrelled | Split. Promote one to the main question and demote the other to a sub-question |
| Could be answered in one search | Fact-finding, not research | Ask about relationships, mechanisms, experiences or change over time |
| Covers a country, a decade and an industry | Too broad for your word count | Apply the four scoping levers: population, context, variables, time |
| Needs data a company or government won’t give you | Infeasible | Substitute a measurable proxy (intention, perception, publicly available records) |
| Uses “impact”, “effect” or “cause” with a cross-sectional survey | Causal language, correlational design | Change to “relationship between” or “is associated with”, or change the design |
| Terms like “success”, “engagement”, “wellbeing” are undefined | Vague constructs | Define each in your scope statement, or replace with the specific measure you will use |
| You already know the answer you want | Confirmation exercise | Ask: would the opposite result still be a legitimate finding? If not, reword until it would |
| Longer than about 35 words | Carrying detail that belongs elsewhere | Move method, sample size and definitions into objectives and scope; keep the question lean |
| Five or more research questions | Unfocused design | Merge any two questions answered by the same data. Aim for one main question plus 2–4 sub-questions |
8. Sub-questions, aims and objectives: building the “golden thread”
How many research questions should a dissertation have?
Fewer than you think. The dominant pattern at every level is one main research question supported by a small number of sub-questions. The guide numbers below are typical, not rules; follow your handbook if it specifies otherwise.
| Level | Main questions | Sub-questions | Notes |
|---|---|---|---|
| Undergraduate (8,000–12,000 words) | 1 | 2–3 | One dataset or one body of sources is usually all you can handle |
| Master’s (12,000–20,000 words) | 1 | 2–4 | More than four sub-questions almost always means one is doing no work |
| PhD / doctoral | 1–2 | 3–6 | Often one question per empirical chapter or study |
The test for whether a sub-question earns its place: it must require its own data or its own analysis. If two sub-questions would be answered by the same table of results, they are one question.
Worked example: the complete structure
Here is the running example from section 3, fully built out. Notice how each sub-question maps to exactly one objective and one method, and how together the sub-questions fully answer the main question.
Aim
To examine how hybrid working arrangements shape informal knowledge-sharing in software development teams.
Main research question
How do hybrid working arrangements influence informal knowledge-sharing among software developers in UK mid-sized technology firms?
Sub-questions, objectives and methods
| Sub-question | Objective | Method |
|---|---|---|
| SQ1. What hybrid working patterns are currently in use in the participating firms? | To document the hybrid policies and actual attendance patterns in four firms | Policy document analysis; short staff questionnaire |
| SQ2. How do developers describe the ways informal knowledge-sharing happens, or fails to happen, under these arrangements? | To explore developers’ accounts of informal learning in hybrid settings | 15 semi-structured interviews; thematic analysis |
| SQ3. What practices do developers and team leads use to compensate for reduced co-location? | To identify compensating practices and assess how developers evaluate them | Same interviews; cross-case comparison |
This mapping is what supervisors mean by the golden thread: aim → main question → sub-questions → objectives → methods → findings chapters → conclusion. If you can lay your project out in a table like the one above, the hardest structural work of the dissertation is done. Many students use the sub-questions as the section headings of their findings chapter, which guarantees the question gets answered.
Need your dissertation proposal written?
An experienced specialist in your subject will research and write your complete model dissertation proposal from scratch, including your research question, aims, objectives, literature review and proposed methodology.
Send us your topic or question and, if you wish, any university guidance or marking criteria.
All our writers hold at least a 2:1 degree from a reputable university. Most hold a master’s degree or PhD, and/or have teaching experience.
9. Question frameworks: PICO, PEO, SPIDER and FINER
If you are in health, nursing, psychology or evidence-based social science, your department may expect you to structure your question with a named framework. These are scaffolds for making sure you have not forgotten a component; they are not mandatory in other fields, and forcing a humanities question into PICO produces awkward results.
| Framework | Components | Best for | Example |
|---|---|---|---|
| PICO(T) | Population · Intervention · Comparison · Outcome · (Time) | Quantitative, clinical and intervention studies; systematic reviews | In adults with type 2 diabetes (P), how does a structured walking programme (I) compared with dietary advice alone (C) affect HbA1c (O) over six months (T)? |
| PEO | Population · Exposure · Outcome | Qualitative questions about experience of a condition or situation | What are the daily living experiences (O) of mothers (P) with postnatal depression (E)? |
| SPIDER | Sample · Phenomenon of Interest · Design · Evaluation · Research type | Qualitative and mixed-methods studies | What are the experiences (E) of young parents (S) attending antenatal classes (PI), explored through interviews (D) in a qualitative study (R)? |
| SPICE | Setting · Perspective · Intervention · Comparison · Evaluation | Service evaluation, social care, library and information science | In NHS community pharmacies (S), from patients’ perspective (P), how does a medicines-review service (I) compare with standard dispensing (C) in perceived usefulness (E)? |
| FINER | Feasible · Interesting · Novel · Ethical · Relevant | Not a structure but a quality check; apply to any question | Use alongside the seven tests; “Ethical” is the addition worth noting |
When a framework helps, and when it gets in the way
Frameworks help most when you keep writing questions that are missing a component: no comparison group, no defined outcome, no clear population. Filling in the boxes exposes the gap. They get in the way when you let the acronym dictate a design that does not fit your interest. A student interested in how nurses experience moral distress does not need a Comparison; forcing one in turns a good PEO question into a bad PICO one.
10. Where the research question goes in your dissertation
Your research question appears, word for word, in at least three places:
- Introduction (end of the chapter, after the problem and aim): “This dissertation addresses the following research question: …” followed by the sub-questions as a numbered list.
- Methodology (opening section): restated to justify the design. “To answer RQ1, which asks how…, a qualitative design was adopted because…”
- Conclusion (opening section): restated and answered directly. “This study asked how… The findings indicate that…”
Many markers also expect to see it echoed in the abstract and in the opening of the discussion chapter. Use identical wording every time. If the question shifts between chapters, examiners notice, and it reads as though the project changed direction without the author realising.
Title vs research question
Your title is a label; your research question is a commitment. A title can be shorter and catchier (“Together, apart: knowledge-sharing in hybrid software teams”), but it should clearly signal the same population and phenomenon. Students often fix the title first and then write a question to match it; do it the other way round.
11. Final checklist before you show your supervisor
Tick every box. If you cannot, the troubleshooting table tells you what to change.
- It is one question (no “and” joining two separate enquiries).
- It names a specific population, group, text or case.
- It names a context (place, sector, setting) and, where relevant, a time frame.
- Every key term is either self-evident or defined in my scope statement.
- It cannot be answered with yes, no, a number, or a single search.
- It contains no “best”, “should”, “good”, “bad” or other value judgement.
- It does not assume its own answer; the opposite result would still be a legitimate finding.
- I can describe exactly what data would answer it.
- I can collect that data within my timeline, budget, skills and ethics approval.
- Its wording matches my method (relational/extent → quantitative; how/experience → qualitative).
- I can state in one sentence why the answer would matter to anyone besides me.
- It is supported by 2–4 sub-questions, each requiring its own data or analysis, each mapped to an objective.
FAQs:
Usually one main question and two to four sub-questions. Undergraduate projects sit at the lower end (one main, two or three subs); master’s dissertations at one main and two to four; doctoral theses may have one or two main questions and three to six subs, often one per empirical chapter. If two of your questions would be answered by the same data, merge them.
It is best avoided. “Does X affect Y?” is answered by yes or no and gives you nothing to analyse. “To what extent does X affect Y?” or “How does X relate to Y?” asks the same thing but demands a nuanced answer. The one exception is a tightly specified hypothesis-testing study, where a closed main question is paired with quantified sub-questions and formal hypotheses.
The research question asks what you want to find out; the hypothesis predicts what you expect to find, in testable form. Quantitative studies usually state both (“RQ1: What is the relationship between…? H1: Higher X will be associated with higher Y.”). Qualitative studies have research questions but not hypotheses, because the aim is to explore rather than to test.
The question is what you want to know; the objectives are the steps you will take to find out. Objectives start with verbs (“To identify…”, “To compare…”, “To evaluate…”) and should be specific enough that a reader could check whether you did each one. A common error is writing an aim, a question and an objective that all say the same thing in slightly different words.
Yes, and most students do. Refinement as your understanding deepens is normal; so is adjusting when access to participants or data changes. Agree any substantial change with your supervisor, then check that the literature review, methodology and conclusion still answer the revised question rather than the original one.
One sentence, typically 15 – 35 words. If it is longer, it is probably two questions or it is carrying detail (sample size, method, definitions) that belongs in your objectives or scope section.
A question, ending in a question mark. Some disciplines allow a “research statement” or thesis statement instead, particularly in the humanities and in argument-led dissertations. If your handbook asks for a research question, give it one.
Yes, though it is a much rarer problem than being too broad. A question is too narrow when answering it would be trivial, when the answer would interest no one beyond the immediate setting, or when there is not enough literature to build a review around it. The fix is usually to connect the narrow case to a broader concept or debate (“…and what does this reveal about…”).
Yes. A library-based or systematic-review dissertation still needs a focused question; your “data” are the published studies. Frameworks such as PICO and SPIDER were in fact designed for exactly this purpose.
Summary
A dissertation research question is a single, open, specific question that names who or what you are studying, in what context, and what you want to know about it, and that you can answer with evidence you can realistically obtain. Get there by starting broad, reading to find the gap, narrowing with the four scoping levers (population, context, variables, time), drafting with an open stem that matches your method, stress-testing against the seven criteria, and then breaking it into two to four sub-questions that each map to an objective. Do that, and the rest of the dissertation has a spine to hang from.
Cite This Work
To export a reference to this article please select a referencing stye below:
