A research topic is a broad area of interest — not yet a question, not yet a plan. Most of the effort that goes wrong in this stage isn’t a lack of ideas; it’s skipping the work of testing an idea before committing months (or years) to it. This guide is a method for doing both: finding real candidate topics from sources that actually surface unanswered problems, and testing each candidate against criteria that predict whether it will survive contact with a supervisor, an ethics board, a funder, or a peer reviewer.
Topic, question, aim, and hypothesis are not the same thing
These four terms get used interchangeably in casual conversation and that’s exactly where a lot of wasted early effort comes from. Each sits at a different level of specificity:
| Term | What it is | Example |
|---|---|---|
| Topic | A broad subject area with no specified population, variable, or boundary | Adolescent social media use |
| Research question | A single, answerable sentence with a population, variable/intervention, and outcome | Is daily short-form video use associated with self-reported sleep latency in 13-17 year olds? |
| Aim (or objective) | A statement of intent — what the study sets out to do, often the question restated as a goal | To determine whether daily short-form video use is associated with sleep latency in adolescents |
| Hypothesis | A testable, predicted answer to the question, stated as a claim | Adolescents who use short-form video daily will report longer sleep latency than those who do not |
Not every study needs a hypothesis — exploratory and much qualitative work is question-only. But every study needs a topic narrowed into a question before a methodology can be chosen. For the mechanics of that narrowing step and the FINER framework applied to a finished question, see How to Write a Research Question; for turning a question into a testable claim, see How to Write a Research Hypothesis. This guide is about the step before either of those: finding and stress-testing the topic itself.
Where topics actually come from
“Pick something you’re interested in” is true but not actionable. In practice, viable topics come from a small number of identifiable sources — and the strongest candidates usually show up in more than one of them at once.
Recent review articles and their “future directions” sections
A well-written review or systematic review doesn’t just summarize a field — its discussion and limitations sections routinely name what the authors themselves think is unresolved, understudied, or methodologically weak in the existing evidence. This is one of the highest-yield sources available, because the gap has already been identified and argued for by people who know the field. Read the last two to three paragraphs of recent reviews in your area specifically for phrases like “future research should,” “remains unclear,” “has not been examined in,” or “limited by.”
Systematic-review gap statements
Systematic reviews are more explicit than narrative reviews about what they didn’t find: a review’s own “implications for research” section (a required element under PRISMA-style reporting) states directly what evidence is missing, what populations weren’t represented in the included studies, and where the evidence base is too thin to draw a conclusion. Checking whether a systematic review already exists on your candidate area — and, if so, what it says is unresolved — also does double duty as the novelty check covered further down.
Funder priority calls and roadmaps
Funding bodies publish, in detail, what they want studied next: NIH funding opportunity announcements and Requests for Applications, NSF program solicitations, UKRI strategic delivery plans and cross-council themes, and Horizon Europe work programmes all name specific priority areas, often years in advance of the deadline. A topic that aligns with a published funder priority has a real advantage later, when the same topic needs to become a fundable proposal — see writing a research proposal for what comes after this stage.
Conference themes and track calls
Annual meeting themes, call-for-abstracts tracks, and keynote topics are a live, current-year signal of what a field’s active researchers consider pressing right now — often more current than published literature, which lags behind by the time it clears peer review.
Replication and extension opportunities
An underused source: taking a published study and asking whether its finding holds in a different population, setting, or time period, or whether a plausible moderating variable was left unexamined. Replication is not a lesser form of research — for fields with a documented reproducibility problem, replication and extension work is often more publishable and more useful than another novel exploratory study.
Supervisor and lab pipelines
For graduate students specifically, this is usually the most practical source, not the least rigorous one: an active lab or supervisor typically has an existing dataset, an ongoing programme of work, or a specific sub-question within a larger funded project that needs a person to take it on. A topic that plugs into existing infrastructure — data already collected, ethics approval already in place, a supervisor already expert in the method — clears the feasibility test discussed below almost automatically.
Testing whether a candidate topic is viable
Once you have one or more candidate topics, the next step is evaluation, not enthusiasm. The most widely taught framework for this is FINER — Feasible, Interesting, Novel, Ethical, Relevant — introduced by Stephen Hulley and colleagues in Designing Clinical Research and now taught across disciplines well beyond clinical research as a general-purpose viability checklist. Applied at the topic-selection stage, before a specific question has even been drafted, each criterion becomes a concrete question to answer honestly:
- Feasible. Is there realistic access to the population, data, materials, or archives this topic requires? Is the necessary technical or statistical expertise available on the team? Can it be scoped to fit the time and budget actually available — a semester, a funding cycle, a thesis timeline? A topic that would be scientifically ideal but requires resources nobody on the team has is not yet a usable topic.
- Interesting. Would answering it actually motivate you through the slow parts, and would it interest the supervisor, funder, or field whose engagement the work depends on? A topic chosen purely because it seemed easy rarely survives the length of a real project.
- Novel. Does it confirm, refute, or extend existing findings in a way that adds something — a new population, a new context, a new method, a genuinely unanswered question — rather than simply repeating a settled finding? See the novelty-checking section below for how to verify this concretely rather than assume it.
- Ethical. Can it be approved by an IRB or research ethics committee as designed? Topics involving vulnerable populations, deception, or sensitive data need this checked early, not after a design is already built around an approach that won’t clear review.
- Relevant. Does it matter to scientific knowledge, clinical or policy practice, or future research directions — the kind of relevance a funder, journal, or examiner will actually recognize, not just relevance to the researcher personally?
A topic that fails Feasible or Ethical outright should usually be dropped or substantially reshaped before more time is invested. A topic that’s Feasible and Ethical but weak on Novel or Relevant can often be saved by narrowing (see below) rather than discarded — the topic itself may be fine; the scope was just too broad or too well-trodden as originally framed.
Checking a topic hasn’t already been answered
“Novel” in FINER is a claim that has to be checked, not assumed. Before committing to a topic, run an actual search strategy against it:
- Search systematically, not casually. Run the same core search terms across at least two to three relevant databases (subject-specific plus a broad multidisciplinary index), not just a single Google Scholar query, and track what you searched so the check itself is defensible later.
- Check PROSPERO for registered systematic reviews. PROSPERO is the international registry of prospectively registered systematic reviews (health and social care, and increasingly other fields); searching it tells you whether someone has already committed to systematically answering a closely related question, even if their review hasn’t published yet. If a live protocol already covers your exact question, that’s a strong signal to narrow further or shift angle rather than duplicate the effort. A related but distinct question can still be worth pursuing — check the registered protocol’s scope carefully rather than assuming an overlapping title means full overlap.
- Check preprint servers for very recent, not-yet-indexed work. Published, peer-reviewed literature lags behind the actual state of a field by months to years. Preprint servers such as bioRxiv, medRxiv, SSRN, and arXiv surface work that exists but hasn’t cleared peer review yet — worth checking specifically for very current or fast-moving topics where the published literature will understate what’s already being studied.
- Use citation-forward searching, not just citation-backward. Once you find one relevant paper, use “cited by” tools (Google Scholar, Web of Science, Scopus) to see who has cited it since — this surfaces the most recent work building on that finding, which a standard keyword search on an older paper will miss.
If a systematic review, meta-analysis, or several recent studies already answer your exact question with consistent findings, that’s not necessarily a dead end — it may mean the more valuable question now is a boundary condition, a mechanism, or a population the existing evidence doesn’t cover, which is itself a narrower, more defensible topic than the one you started with.
Narrowing broad to specific: worked examples
The mechanics of narrowing are the same regardless of field: start broad, then constrain along one axis at a time — a specific population, a specific variable or intervention, a specific outcome, or a specific context — rather than trying to narrow everything simultaneously.
| Discipline | Broad topic | Narrowed, testable topic |
|---|---|---|
| Public health | Vaccine hesitancy | Vaccine hesitancy among parents of school-age children in rural primary-care settings |
| Education | Remote learning outcomes | Formative-assessment engagement in asynchronous undergraduate STEM courses |
| Computer science | Large language model reliability | Hallucination rates of retrieval-augmented models on domain-specific factual queries |
| Environmental science | Soil carbon sequestration | Cover-crop effects on topsoil organic carbon in temperate no-till systems |
Each narrowed version in the right-hand column is still a topic, not yet a question — it has a bounded population and context, but no question mark and no specified relationship between variables. The next step, turning a narrowed topic like these into an actual answerable question and checking it against FINER as a finished question, is covered in full in How to Write a Research Question; for discipline-specific worked examples of the finished question itself, see Quantitative Research Question Examples by Study Design.
A quick pre-commitment checklist
- Can you state the topic in one sentence, bounded to a specific population, context, or scope?
- Have you checked for existing systematic reviews and registered protocols (PROSPERO) on this exact area?
- Have you checked preprint servers for recent unpublished work?
- Does it pass Feasible and Ethical without a major redesign?
- Does it connect to a funder priority, active lab, or supervisor’s programme of work?
- Can you defend, in one sentence, what it adds that the existing literature doesn’t already say?
A topic that clears all six is ready to be narrowed into a formal research question — the next real step in the process.
Frequently asked questions
How is a research topic different from a research question?
A topic is a broad subject area described in a phrase, with no specified population, variable, or relationship — “adolescent social media use.” A research question is a single, answerable sentence that specifies a population, a variable or intervention, and an outcome, and requires a question mark: “Is daily short-form video use associated with self-reported sleep latency in 13-17 year olds?” A topic can’t be tested directly; a well-formed question can.
What is the FINER framework and when should I use it?
FINER (Feasible, Interesting, Novel, Ethical, Relevant) is a checklist for evaluating whether a candidate topic or question is worth pursuing, introduced by Stephen Hulley and colleagues in Designing Clinical Research. Use it as early as possible — at the topic stage, before investing time drafting a full question or proposal, and again once the topic has been narrowed into a specific question.
How do I check that a research topic hasn’t already been answered?
Run a systematic search across two to three relevant databases with tracked search terms, check PROSPERO for registered or in-progress systematic reviews on the same area, check preprint servers (bioRxiv, medRxiv, SSRN, arXiv) for very recent unpublished work, and use citation-forward (“cited by”) searching from key papers to catch work published after them.
Do I need a hypothesis to have a viable research topic?
No. A hypothesis is a predicted, testable answer to a research question, and not every study needs one — exploratory and much qualitative work proceeds on a question alone. Every study does need a topic narrowed into a question before a methodology can be chosen.
What if my topic is too broad?
Narrow along one axis at a time — a specific population, a specific variable or intervention, a specific outcome, or a specific context — rather than trying to constrain all of them simultaneously. See the worked examples above, and How to Write a Research Question for the full narrowing-to-question process.
