Global Academic Publishers Logo
GAP Insights · Publishing Guides · 2026

How to Write a Research Question: A Step-by-Step Guide

By Muaz Dave · 8 November 2026, 6:50 AM PKT

How to Write a Research Question: A Step-by-Step Guide

Introduction

Ayesha, an MPhil student in Lahore, spent four months reading about social media and student life. Then her supervisor asked one question: "What exactly are you measuring, and in whom?" Ayesha could not answer in one sentence. Her topic was wide, her outcome was unclear, and her four months of reading had no centre. She went back, rewrote her question, and only then did her reading start to make sense. A research question is not a formality at the top of a proposal. It is the decision that organises everything else: what you read, whom you study, what data you collect, and what your paper can honestly claim.

This guide shows you how to write that sentence. You will work through a real example from a broad topic to a focused question, test it with the FINER criteria, build clinical questions with PICO, choose the right question type, and check your draft three ways before you commit to it.

Ayesha's Four Lost Months: Why the Question Comes First

Ayesha began with the topic "social media and students". It sounded like a research question, but it was only a subject area. She read about screen time, sleep, grades, anxiety, and friendships without a filter, because her topic did not tell her what to keep and what to ignore. When her supervisor asked her to name the population, the behaviour, and the outcome in one sentence, she realised she had three different studies mixed together: one about sleep, one about grades, and one about mood.

She fixed it in one afternoon by answering four questions on paper: who exactly (first-year university students in Lahore), what behaviour exactly (daily use of short-form video apps), what outcome exactly (sleep duration), and over what time (one semester). Her new question was: "How does daily use of short-form video apps relate to sleep duration among first-year university students in Lahore over one semester?" Suddenly her reading had a filter. Papers about grades or anxiety went aside. Papers about sleep and screen time stayed. The lesson is simple: write the question before you read widely, not after. Reading without a question is collecting; reading with one is researching.

The FINER Test, Applied to a Real Question

FINER is a five-point test widely used in health and social research: Feasible, Interesting, Novel, Ethical, and Relevant. Apply it to one real question rather than memorising the letters. Bilal, a veterinary postgraduate, drafted this question: "What is the prevalence and what are the risk factors of mastitis in dairy herds in Punjab, and how do farmers manage it?" Run it through FINER and it fails twice. Feasible? No — prevalence, risk factors, and farmer practices are three studies, not one, and a single student cannot do all three well in one year. Novel? Partly — prevalence data may exist already, so he must check what is already published before claiming a gap.

Bilal split the question and tested the first half alone: "What is the prevalence of subclinical mastitis in dairy herds in three districts of Punjab, and which herd factors are associated with it?" Now it is feasible (three districts, one outcome, one season of sampling), interesting (mastitis costs farmers milk and money every day), novel (no recent prevalence figures for those districts), ethical (milk samples only, with farmer consent), and relevant (the answer guides local control advice). A question that fails even one FINER point needs changing before you invest months of work.

PICO: Building a Clinical Question in Four Slots

PICO builds questions about treatments, tests, or care. P is the Population you will study. I is the Intervention being tested. C is the Comparison it is measured against. O is the Outcome you will measure. Fill the four slots and the question writes itself. A nurse researcher, Sara, wanted to study phone support for diabetes patients. Her slots were: Population — adults with type 2 diabetes attending a district hospital; Intervention — a weekly nurse phone call; Comparison — usual care with no calls; Outcome — hospital admissions over six months.

Her question became: "In adults with type 2 diabetes, does a weekly nurse phone call, compared with usual care, reduce hospital admissions over six months?" Notice what PICO forced her to decide: the exact group, the exact action, the exact alternative, and the exact result, including the time period. If your question is about a treatment or a test, write out the four slots on paper first. If any slot is blank, your question is not ready. Our guide to how to write a research introduction shows how the same four slots become the opening paragraphs of your paper.

Three Question Types — and the Mistake of Mixing Them

There are three basic types, and each one promises a different kind of answer. Descriptive questions ask what is happening: how common is a problem, who is affected, what does it look like? Comparative questions ask how groups differ. Causal questions ask whether one thing causes another. The mistake is promising one type and delivering another — usually promising a causal answer from a study that can only describe.

Take one topic, mastitis in dairy herds, and see all three. Descriptive: "What is the prevalence of subclinical mastitis in dairy herds in three districts of Punjab?" Comparative: "Do herds milked by machine have lower mastitis rates than herds milked by hand in the same districts?" Causal: "Does a new teat disinfectant reduce new mastitis cases in dairy herds over six months?" Choose the type that matches what you truly want to know, and check that your method can deliver it. A survey can answer the first two; only a controlled comparison can answer the third.

From a Broad Topic to One Sentence: The Narrowing Ladder

Narrowing works in stages, and each stage removes part of the topic. Usman started with "animal health" — a subject for a library, not a study. Step one: choose one part, poultry disease. Step two: choose one problem, Newcastle disease in backyard chickens. Step three: add a place and a group, smallholder farmers in one district. Step four: add a time, one year. Step five: name the outcome, the proportion of flocks affected and the factors linked to outbreaks. His question became: "What is the prevalence of Newcastle disease in backyard chicken flocks kept by smallholder farmers in one district over one year, and which flock factors are associated with outbreaks?"

At each step ask four narrowing questions: who exactly, where exactly, when exactly, and what outcome exactly? If you cannot answer one of them, the question is still too broad. A focused question can still matter a great deal; it is simply one that a single study can honestly answer.

Write It in One Sentence: A 20-Minute Drafting Exercise

A research question should fit in one sentence, usually between fifteen and twenty-five words. One sentence forces you to choose your key terms and drop extra ideas that belong in a second study. Set a timer for twenty minutes: write your draft, count the words, then cross out every word that does not name the population, the key factor, or the outcome. Read what is left aloud. If it still feels crowded, move the extra idea into a sub-question.

Compare these pairs and notice what changed. Weak: "What is the impact of social media on students?" Strong: "How does daily use of short-form video apps relate to sleep duration among first-year university students in Lahore?" The strong version names the group, the behaviour, and the outcome — and drops the vague word "impact". Weak: "Is exercise good for health?" Strong: "Does a twelve-week walking programme reduce blood pressure in adults with hypertension attending a rural clinic?" The weak version can be answered yes or no without any research; the strong version must be studied. Replace vague words such as impact, effect, or role with the exact thing you will measure, for example test scores, milk yield, or admission rates.

Three Ways to Test Your Question Before You Commit

First, search the literature. If ten studies have already answered your question exactly, in the same group and setting, you need a new angle, a different group, or a different setting. If the answer is already settled, change the question now — it costs a week, not a year. Second, check the data. Can you reach the people, records, or samples you need? Do you have permission? A question you cannot study is a dead end, however elegant it looks on paper. Phone one clinic, one school, or one farm office before you commit, and ask whether access is realistic.

Third, explain your question to a colleague or supervisor in one minute and ask them to repeat it back to you. If they repeat something different, your wording is unclear. Fixing a question now costs days; fixing it after data collection costs months. Keep a short note of each test and its result, so your supervisor can see the question was checked, not just chosen.

Matching the Question to the Method

Your question and your methods must match, because the question decides what counts as evidence. A question about how patients experience a long illness needs interviews, not a survey with tick boxes. A question about whether a vaccine works needs a comparison group, not a single case report. A question about how common a problem is needs a defined sample and a clear case definition, not a handful of convenient examples. When the design does not fit the question, even perfect data cannot produce a trustworthy answer.

Write the question first, then choose the design, the sample, and the measures that answer it — not the other way round. Our guide to how to write the methodology section shows how to describe that match so reviewers can see it. If your method cannot answer your question, change the method or the question.

Seven Mistakes That Sink Good Projects

First, asking a question that is too broad, so the study tries to cover everything and proves nothing. Fix it with the narrowing ladder above. Second, asking several questions at once and calling them one question — split them and choose the main one. Third, choosing a question because the data are easy to get, even though the question is dull or already answered. Fourth, using vague words that cannot be measured, such as impact or quality, without saying what will be counted. Fifth, writing a question that is really a statement of opinion, such as "why is this treatment the best?" — a question must be open to an answer you do not already know.

Seventh, asking a yes-or-no question that needs no research at all. Before you start, read our guides on how to write a literature review and how to publish a research paper so your question fits the paper it will become. A good question is a small sentence that carries a whole study — write it carefully, test it honestly, and the rest of the project becomes far easier.

FINER Criteria at a Glance

CriterionWhat it asksWeak exampleStrong example
FeasibleCan I answer it with my time and access?Does this drug cure all cancers?Does this drug lower blood sugar in adults with type 2 diabetes?
InterestingDoes it matter to me, and to others?A question chosen only because data were easy to findA problem nurses raise every week
NovelDoes it add something new?Repeating a study with nothing newTesting a known treatment in a new group
EthicalCan it be studied without harm?A study risking harm without approval or consentA study with approval and informed consent
RelevantWill the answer be useful?A question nobody in practice needs answeredA question that could change local guidelines

Further Reading and Useful Resources

These authoritative external resources explain the standards and tools mentioned in this guide:

Frequently Asked Questions

What is a research question?
A research question is the single question your study sets out to answer. It states the problem, the group or setting, and the outcome you care about, and it guides your methods, your data collection, and the structure of your paper.

How long should a research question be?
One sentence, usually between fifteen and twenty-five words. A short question forces you to choose your key terms and to leave extra ideas for sub-questions or future studies. If your draft runs longer, cut words that do not name the population, the key factor, or the outcome, and move extra ideas into sub-questions instead of crowding the main sentence.

What is the difference between a research question and a hypothesis?
A research question asks what you want to find out. A hypothesis is a specific, testable prediction about the answer, such as that a treatment will lower blood pressure. Not every study needs a hypothesis, but every study needs a research question.

Can I change my research question after I start?
Yes, small changes in the early weeks are normal, for example when the literature or pilot data show the question needs narrowing. Record each change and its reason, but do not change the question after seeing the results just to make the findings look stronger.

What do FINER and PICO mean?
FINER is a test for any research question: Feasible, Interesting, Novel, Ethical, and Relevant. PICO is a framework for clinical questions: Population, Intervention, Comparison, and Outcome. Use FINER to judge a question and PICO to build one about a treatment or test.

How do I know if my research question is good enough?
Test it against FINER, check that the literature has not already answered it, and confirm you can reach the data you need. If a colleague can repeat your question back to you correctly after hearing it once, and your methods can answer it, it is ready.

Submit your manuscript  ← Back to GAP Insights

More Resources for Researchers

Research Resources Directory

Useful publishers, databases, funders, tools and institutions for researchers:

Publishers

Databases and Indexes

Preprint Servers and Repositories

Ethics and Reporting Standards

Funders

Research Tools

Health and Veterinary Organisations

Veterinary Schools

Leading Universities

Major Journals