Collecting more sources can make a broad topic feel productive without making it answerable. Before you build a library, turn the topic into a question, add only the boundaries that change the research, and run a small search to test whether the evidence and method can support it.
You will waste less time on irrelevant reading now. More importantly, the question will guide later searching, note selection, argument structure, and the decisions you make when the evidence changes your first idea.
Start with the assignment, not a question formula
There is no single research-question template that works across every university subject. A history essay, laboratory proposal, policy review, and literary analysis ask different things of evidence.
Read the brief and identify:
- the task word, such as analyse, compare, evaluate, or explain;
- the required kind of evidence;
- whether primary research is expected, allowed, or unnecessary;
- the word count and deadline;
- any required period, case, text, theory, or method; and
- the marking criteria or learning outcome.
Monash University Library's research-question guidance stresses that useful questions depend on the discipline and assignment purpose. Treat the brief as your first boundary.
Turn the topic into something evidence must answer
A topic names an area. A research question gives the evidence a job.
- Topic: social-media use and sleep
- Broad question: How does social media affect sleep?
- More focused question: How have studies published since 2020 explained the relationship between late-night social-media use and sleep duration among first-year university students?
This is a constructed example, not a universal model. The focused version defines a population, behaviour, outcome, period, and evidence base. It can still fail if those boundaries do not fit the assignment or if suitable studies are unavailable.
Ask what your evidence must help you do:
- explain a cause or mechanism;
- compare two interpretations;
- evaluate a policy or argument;
- trace change over time;
- analyse how a text creates an effect; or
- examine a relationship under named conditions.
Do not write the final answer into the question. A research question directs inquiry. Your thesis is the answer you can defend after reading the evidence, and it may change.
Add boundaries only when they change the research
A question becomes focused when its limits make the project more answerable—not when its title becomes longer.
Test these possible boundaries:
- Population or object: Which group, text, organization, material, or case?
- Issue or relationship: What exact effect, disagreement, mechanism, or comparison?
- Place: Does location change the evidence or context?
- Period: Which years, event, stage, or historical moment matters?
- Evidence: Which source types or data can answer the question?
- Method: How will you analyse or collect that evidence?
- Purpose: What will the comparison, explanation, or evaluation establish?
Keep a boundary only if removing it would change the sources, method, interpretation, or scale. Adding “in the United Kingdom between 2020 and 2025” is not useful if neither location nor period matters to your argument.
Run a small preliminary search
Do not wait for perfect wording before searching. Cornell University Library's introduction to research begins with a question and keywords, then treats early searching as a reason to revise terms or the topic when useful information is difficult to find.
Run a small search using:
- the main idea in your question;
- one or two synonyms for each important term;
- the population, case, text, or period; and
- a method or evidence term when relevant.
Open a small number of recent and influential sources. At this stage, your goal is not to finish the literature review. Look for:
- the terms researchers actually use;
- the main disagreement or conversation;
- which evidence is available;
- whether the same narrow question has already been answered;
- access problems; and
- a better boundary suggested by the field.
Monash recommends bounded preliminary reading to identify the key debates before fixing the question. This keeps the search exploratory without letting it grow into an uncontrolled pile of PDFs.
Test scope, evidence, method, and time together
A narrow-sounding question may still be impossible to complete. Check the whole project:
Can evidence answer it?
You need enough relevant, credible, accessible material to support analysis. A high result count proves very little; many results may be duplicates, tangential, inaccessible, or unsuitable.
Does the method fit?
If your question asks about an effect, do you have evidence that can examine effects? If it asks how a novel creates an impression, does your method include close analysis of the text? If it requires interviews, are they permitted and realistic within the assignment?
Does it fit the limits?
Can you answer it with the available time, word count, tools, data, language skills, and ethics approval? University of Toronto guidance on academic proposals connects a worthwhile question to its relationship with existing work, proposed method, resources, timeline, and feasibility.
Does it leave room for analysis?
A question can also be too narrow. Warning signs include:
- one source answers it completely;
- the answer is a simple fact or yes/no;
- there is no meaningful disagreement or interpretation;
- the evidence is too thin for the required length; or
- the wording only asks for description when the brief expects analysis.
The right scope is not the smallest possible question. It is the smallest question that can still support the assignment's required thinking.
Rewrite the question after the pilot search
Research develops through revision. The ACRL information-literacy framework describes research as iterative inquiry and searching as strategic exploration, not a straight line from first keywords to final answer.
After the pilot search, rewrite the question using what you learned:
- replace everyday wording with the field's real terms;
- remove a boundary that produced no useful distinction;
- add a case, period, or comparison the evidence makes important;
- change an unavailable method;
- narrow a large population or evidence base; or
- reopen a question that became too easy to answer.
Then test it again. Two or three short revisions are usually more useful than collecting dozens of sources under wording you have not challenged.
Know when to ask for help
Bring your candidate question and pilot-search notes to a librarian, tutor, or instructor when:
- you cannot find the field's main terms;
- relevant sources exist but you cannot access them;
- the method may require approval or specialist tools;
- the question crosses disciplines;
- you cannot tell whether the scope fits the word count; or
- the assignment rules conflict with general advice.
Ask a specific question such as: “My search finds good evidence for the outcome but not this population. Should I broaden the population or change the relationship?” That gives the other person a decision they can help you make.
Your question is ready for deeper research when it fits the brief, names a real analytical task, has an evidence and method path, fits the limits, and remains open to what the sources may show.