How to Write a Research Question: A Complete Guide With Frameworks and Examples

A research question is the focused, specific question that drives a research study. It identifies what the researcher wants to find out, defines the scope of the investigation, and shapes every subsequent decision about methodology, data, analysis, and writing. A strong research question is the single most important element of a research project. Get it right, and the rest of the work has a clear path. Get it wrong, and even brilliant methodology can't rescue a study that's asking the wrong thing. This guide covers what a research question is, the criteria a strong one meets, the established frameworks for developing one (FINER, PICO, PICOT, SPIDER, PEO), an eight-step process for writing one from a broad topic, the major types of research questions across qualitative and quantitative studies, discipline-specific guidance, and the most common mistakes researchers make.

Whether you're a doctoral candidate developing a dissertation question, an early-career researcher preparing a grant application, a master's student framing a thesis, or an undergraduate writing a research paper, the principles are the same. The question you write determines what kind of study is possible, what evidence will count, and what the contribution of the work will be. This article gives you the practical tools to develop a research question you can defend, study, and answer.

What Is a Research Question?

A research question is a clearly stated, focused, answerable question that a research study sets out to investigate. It's narrower than a topic and more specific than a research problem. The topic might be "social media and adolescent mental health." The research problem might be "the relationship between adolescent social media use and depression symptoms is poorly understood and inconsistently measured across studies." The research question, drawn from that problem, might be "Does daily duration of social media use predict depression symptoms among US adolescents aged 13 to 17, controlling for prior depression history and offline social engagement?"

A research question does several things at once. It tells you what data you need to collect. It tells you what methods you should use. It tells you what kind of answer would count as a finding. And it tells the reader what the study is for. A good research question is the spine of the entire research process. A vague or unfocused question produces a vague and unfocused study; a sharp, specific question produces work that contributes meaningfully to its field.

What Makes a Good Research Question? Six Criteria

A strong research question meets six criteria. These criteria are widely recognized across academic disciplines and form the basis of the FINER framework discussed in the next section.

• Focused. The question addresses a single problem or relationship rather than a tangle of related issues. "What factors affect college student outcomes" is too broad. "Does participation in a learning community in the first year predict four-year graduation rates among first-generation college students at large public universities" is focused.
• Specific. The question uses precise terms with clear definitions. Vague terms ("affect," "good," "successful") signal weak questions. Specific terms ("cause a 10 percent increase in," "meet criteria for major depressive disorder," "achieve a four-year graduation rate of 60 percent or higher") signal strong ones.
• Researchable. The question can be answered using primary or secondary data, established methods, and resources you can realistically access. A question that requires data nobody collects, or methods nobody has developed, isn't researchable as written.
• Feasible. The question can be answered within the constraints of your timeline, budget, sample access, ethical approvals, and skills. A perfectly researchable question that would take fifteen years and a major grant to answer isn't feasible for a doctoral dissertation. Feasibility narrows what counts as a good question for you specifically.
• Complex enough to merit investigation. The question can't be answered by a quick search or a simple yes or no. "Is exercise good for health" can be answered by reading a few review articles. A research question pushes beyond what's already known.
• Relevant. The question matters to your field, contributes to ongoing scholarly conversations, or addresses a real-world problem. A question that's specific, focused, and feasible but trivial doesn't justify the investment of a research project.

Frameworks for Developing a Research Question

Several established frameworks help researchers transform a broad interest into a specific, well-formed research question. Different frameworks suit different fields and different types of studies. The most commonly used frameworks are FINER, PICO, PICOT, SPIDER, and PEO.

The FINER framework

The FINER framework, developed for clinical research but widely used across disciplines, evaluates a research question against five criteria: Feasible, Interesting, Novel, Ethical, and Relevant. A strong research question scores well on all five.

• Feasible. Adequate sample size, expertise, time, and resources are available to answer the question.
• Interesting. The answer matters to the researcher, to the field, and to potential funders. A research question you don't care about will not sustain you through the years a serious project takes.
• Novel. The question has not been answered, or has only been answered partially or in ways that current evidence challenges. Confirmatory and replication studies have legitimate scientific value, but most research questions need to add something new.
• Ethical. The study can be conducted in ways that protect participants, follow disciplinary standards, and meet institutional review board requirements.
• Relevant. The answer would advance knowledge, change practice, or inform policy in ways that justify the investment.

FINER is the most widely applicable framework across disciplines. Use it as a final checklist regardless of which other framework you use to structure the question itself.

The PICO and PICOT frameworks

The PICO framework is the standard for clinical research questions and evidence-based practice in medicine, nursing, public health, and the health sciences. PICO stands for Population, Intervention, Comparison, and Outcome. PICOT adds Time as a fifth element. A PICO-structured question makes each element explicit.

• Population. Who is being studied? Specify age, sex, condition, setting, and any other relevant characteristics. "Adults aged 65 and older with type 2 diabetes" is more useful than "older adults."
• Intervention. What treatment, exposure, or condition is being studied? Specify the dose, duration, and form. "Twelve weeks of supervised resistance training, three sessions per week" is more useful than "exercise."
• Comparison. What is the alternative? A standard treatment? Placebo? No intervention? Specify what the intervention is being compared against.
• Outcome. What measurable result is being assessed? Specify the construct, the measure, and the timing. "HbA1c at 12 weeks measured by standard laboratory assay" is more useful than "blood sugar."
• Time (PICOT). Over what period is the outcome measured? "At 12 weeks and 6 months post-intervention" is more useful than "after the study."

An example PICOT question: "Among adults aged 65 and older with type 2 diabetes (P), does twelve weeks of supervised resistance training (I) compared to standard care alone (C) reduce HbA1c (O) at 12 weeks and 6 months post-intervention (T)?" That question has a clear study design implied (a randomized controlled trial with an active intervention and a comparator), a clear data collection plan (HbA1c measurements at specified time points), and a clear answer format (a difference in HbA1c between groups).

The SPIDER framework

The SPIDER framework was developed for qualitative and mixed-methods research where PICO doesn't fit well. SPIDER stands for Sample, Phenomenon of Interest, Design, Evaluation, and Research type. It's particularly useful for systematic reviews of qualitative studies and for framing qualitative research questions.

• Sample. Who is being studied? In qualitative work, "sample" rather than "population" reflects the interpretive, in-depth nature of the inquiry rather than statistical generalization.
• Phenomenon of Interest. What experience, behavior, or process is being explored?
• Design. What study design is being used? Interviews, focus groups, ethnography, case study, narrative inquiry?
• Evaluation. What outcomes or experiences are being evaluated?
• Research type. Qualitative, quantitative, or mixed methods?

SPIDER is the standard for qualitative research question development in nursing, public health, education, and the social sciences when in-depth interpretive work is the primary mode of inquiry.

The PEO framework

The PEO framework is used for qualitative research questions, particularly in nursing, social work, and education. PEO stands for Population, Exposure, and Outcome (or sometimes Population, Phenomenon of Interest, and Context). It's a simplified alternative to PICO and SPIDER for studies where the comparison element of PICO doesn't apply.

Eight Steps to Write a Research Question

The steps below take you from a broad topic to a specific, defensible research question. The process is iterative; expect to revise the question multiple times as your reading deepens and your thinking sharpens.

Step 1: Choose a broad topic that genuinely interests you

Start with a research area you find compelling. The topic should be relevant to your field and broad enough that you can read into it and find subtopics. Don't worry about narrowness yet. "Household financial decision-making" or "adolescent mental health and social media" or "membrane proteins in cancer biology" are appropriately broad starting points. The topic doesn't need to be specific yet; it needs to be a domain you're willing to spend years thinking about.

Step 2: Conduct preliminary reading to map the field

Before narrowing, read widely in the topic area. Read recent review articles, leading textbooks, and the most-cited papers in the field. Use Google Scholar, Web of Science, Scopus, PubMed (for health sciences), or PsycINFO (for psychology) to identify the conversations your field is having right now. Take notes on what's known, what's contested, what methods are commonly used, and what populations or contexts have been studied. This preliminary reading typically takes weeks or months. Skipping it is the single most common mistake new researchers make. Without it, you can't know whether your eventual question is novel or already settled.

Step 3: Identify a specific gap or problem

As you read, look for what's missing or unresolved. Gaps come in several forms: an understudied population (effects shown in adults but not in adolescents), an understudied context (research conducted in high-income countries but not in low- and middle-income settings), an understudied mechanism (the relationship is documented but not explained), a methodological gap (existing studies use cross-sectional data and longitudinal evidence is needed), or a contradictory finding gap (some studies find an effect, others don't, and the conditions under which it appears are unclear). Note these gaps as you encounter them. The most promising research questions live in these gaps.

Step 4: Narrow your topic to a specific niche

Use the gaps you've identified to narrow your broad topic. "Household financial decision-making" might narrow to "gender differences in financial risk tolerance among married couples." "Adolescent mental health and social media" might narrow to "the relationship between TikTok use and depression symptoms in US high school students." "Membrane proteins in cancer biology" might narrow to "the role of EGFR mutations in non-small cell lung cancer treatment resistance." The niche is narrow enough to focus a study but still has multiple possible questions within it.

Step 5: Formulate a draft research question

Within your niche, draft an initial question. Use the appropriate framework for your field: PICO/PICOT for clinical studies, SPIDER for qualitative work, FINER as a general checklist. The first draft will be imperfect. That's expected. Drafting a question, then reading more, then revising, then drafting again is the normal process. Most research questions go through five to ten drafts before they're submission-ready. For example, a draft on the financial decision-making topic might read, "Are there gender differences in how married couples make investment decisions?" That's a starting point, but it's still vague.

Step 6: Test your question against the FINER criteria

Run your draft question through FINER. Is it Feasible given your data, time, and skills? Is it Interesting enough to sustain the work? Is it Novel given what you've read? Can it be conducted Ethically? Is it Relevant to your field? If any of the criteria fail, revise the question. FINER often pushes you to specify the population, narrow the construct, or sharpen the comparison.

Step 7: Specify the question

Make every term in the question precise. Replace vague verbs with specific ones. Replace abstract constructs with measurable variables. Replace general populations with defined groups. The earlier draft becomes: "Among married couples in the United States, do husbands and wives differ in their willingness to take financial risks, as measured by the standard financial risk tolerance scale, after controlling for income, education, and age?" Now the question implies a study design, specifies the variables, defines the population, and identifies the controls.

Step 8: Review the question with mentors and peers

Bring your refined question to your advisor, your committee members, or experienced peers. Ask them to identify weaknesses. Are there obvious confounders the question doesn't address? Is the population definition too narrow or too broad? Has the question already been answered in literature you missed? Mentor and peer feedback is essential and often produces revisions that wouldn't occur to you. Plan to revise once more after this review. The final question is the one you can defend in your literature review, methods section, and eventual discussion.

Types of Research Questions

Research questions come in several types, and the type determines what kind of study you'll design and what kind of evidence will answer the question. Most research projects have one main question of one of the types below; many also have sub-questions that may be of different types.

Descriptive research questions

Descriptive questions ask what something is, how it's distributed, or how it's changing. They don't propose causal relationships. "What proportion of US adolescents aged 13 to 17 use TikTok daily, and how does usage vary by age, sex, race/ethnicity, and household income?" is descriptive. Descriptive questions are answered with surveys, administrative data, or cross-sectional observational studies. They're foundational; you can't ask sophisticated causal questions about a phenomenon you haven't first described.

Comparative research questions

Comparative questions ask whether two or more groups, conditions, or contexts differ on some outcome. "Do husbands and wives in US married couples differ in financial risk tolerance?" is comparative. Comparative questions are answered with between-groups analyses (t-tests, ANOVAs, regression with categorical predictors). The Fisher and Yao (2017) study examining gender differences in financial risk tolerance is an example of well-developed comparative research using nationally representative US data.

Correlational research questions

Correlational questions ask whether two variables are associated, without claiming causation. "Is daily duration of social media use associated with depression symptom severity in US adolescents?" is correlational. Correlational questions are answered with regression analyses, structural equation modeling, or other association analyses. They're appropriate when randomization isn't possible or when the goal is to map relationships before testing causal hypotheses.

Causal and explanatory research questions

Causal questions ask whether one variable causes another, and explanatory questions ask why or how the relationship works. "Does limiting social media use to two hours per day cause a reduction in depression symptoms in US adolescents who currently use social media four or more hours per day?" is causal. Causal questions are answered with randomized controlled trials, quasi-experimental designs (difference-in-differences, instrumental variables, regression discontinuity), or structural causal analyses with strong identifying assumptions. They're the most demanding to answer well, requiring careful design to rule out alternative explanations.

Exploratory research questions

Exploratory questions ask about phenomena that haven't been well studied and don't yet have established theoretical frameworks. "How do first-generation college students at large public universities describe their experience of belonging on campus during their first year?" is exploratory. Exploratory questions are typically answered with qualitative methods (in-depth interviews, focus groups, ethnography, narrative inquiry, grounded theory) and produce rich descriptions and theoretical insights rather than statistical estimates.

Evaluative research questions

Evaluative questions ask whether a program, policy, or intervention is working, and how well. "Does the federal SNAP program reduce food insecurity among eligible low-income households, and by how much?" is evaluative. Evaluative questions sit at the intersection of academic research and policy, and they're answered with program evaluation methods, often using quasi-experimental designs to estimate causal effects of policy interventions.

Theoretical research questions

Theoretical questions ask about the structure, properties, or implications of a theoretical framework, often without immediate empirical reference. "Under what assumptions does the Modigliani-Miller capital structure irrelevance theorem hold, and which empirical violations are best explained by which assumption failures?" is theoretical. These questions are common in mathematics, theoretical economics, theoretical physics, and parts of philosophy and political theory. They're answered with formal proofs, mathematical analysis, or rigorous conceptual argument.

Research Question Examples Across Disciplines

Strong research questions look different in different fields, but they share the underlying criteria. The examples below show how the same principles apply across diverse disciplines. For a much larger collection of strong research questions across topics and methods, see our companion article on research question examples.

Examples in the social sciences and economics

"Among married couples in the United States, do husbands and wives differ in financial risk tolerance, after controlling for income, education, and age?" This question has clear population, comparison, and outcome elements, and it identifies the controls that matter for interpreting the comparison. The Fisher and Yao (2017) study addressed this question using nationally representative US data, finding gender differences that persist after controlling for plausible confounders.

"Does increasing the federal Earned Income Tax Credit for childless adults reduce poverty rates in this group, and by how much?" This is an evaluative question with policy relevance. It can be answered with quasi-experimental analysis of the 2021 EITC expansion or international comparison data.

Examples in the health and biomedical sciences

"Among adults aged 65 and older with type 2 diabetes (P), does twelve weeks of supervised resistance training (I) compared to standard care alone (C) reduce HbA1c (O) at 12 weeks and 6 months post-intervention (T)?" This is a textbook PICOT question with a randomized controlled trial design implied.

"What proportion of US adults aged 18 to 64 received a diagnosis of major depressive disorder in 2024, and how does prevalence vary by age, sex, race/ethnicity, and household income?" This is a descriptive epidemiological question answerable with national survey data such as the National Survey on Drug Use and Health.

Examples in education research

"Does participation in a first-year learning community increase four-year graduation rates among first-generation college students at large public universities, compared to first-generation students who don't participate?" This is a causal question with a comparison group, a defined population, a measurable outcome, and a specified setting. It's answerable with a quasi-experimental design or a randomized controlled trial.

Examples in engineering and computer science

"Does a transformer-based model with 7 billion parameters fine-tuned on legal contract data outperform a general-purpose 70-billion-parameter model on a benchmark of 500 contract clause classification tasks, measured by F1 score?" This is a comparative engineering question with specific models, a specific task, a specific benchmark, and a specific outcome metric.

Examples in the humanities

"How does the conception of moral responsibility in Aristotle's Nicomachean Ethics differ from the Kantian account, and which conception better accommodates the moral significance of luck?" This is a theoretical question in philosophy. It's answered through close textual reading, conceptual analysis, and engagement with the secondary literature.

"How did Appalachian women writers from 1970 to 2010 represent the relationship between place and identity, and how did these representations change across the period?" This is an exploratory humanities question answerable through close reading of primary sources, contextualization in literary and historical scholarship, and thematic analysis.

Qualitative vs Quantitative Research Questions

Qualitative and quantitative research questions differ in form, language, and what they ask the study to do. Most research projects use one approach or the other; mixed-methods projects use both, often with separate questions for each component.

Quantitative research questions ask about measurable variables, relationships, and effects. They use language like "how many," "to what extent," "is there a relationship between," "does X cause Y," and "what is the difference between." Quantitative questions imply statistical analysis and produce numerical estimates with associated uncertainty. They're well suited to PICO, PICOT, and FINER frameworks. The financial risk tolerance and resistance training examples earlier in this article are quantitative.

Qualitative research questions ask about experiences, meanings, processes, and contexts. They use language like "how do," "what is the experience of," "how do participants describe," and "in what ways." Qualitative questions imply interview, observation, or document analysis methods and produce rich descriptions, theoretical insights, and conceptual frameworks rather than numerical estimates. They're well suited to SPIDER and PEO frameworks. The Appalachian women writers and first-generation college students belonging examples earlier in this article are qualitative.

A common mistake is writing a quantitative question for a study you're conducting qualitatively, or vice versa. If your question asks "how many" but your method is in-depth interviews with 15 participants, the question and method are mismatched. Either change the question or change the method.

Common Mistakes When Writing a Research Question

The mistakes below appear repeatedly in draft research questions across disciplines. Awareness of them helps you self-edit before bringing the question to mentors or committees.

The question is too broad

"What factors affect mental health" is unanswerable as a research question. So is "how does technology change society." Both are topic descriptions, not research questions. The fix is to specify the population, the construct, the relationship, and the context until the question implies a study design.

The question is too narrow

"Did Mary Smith experience improved mood after using a meditation app for two weeks in March 2024" is too narrow to be a research question. It's a question about a single individual at a single time. The fix is to broaden the population and time frame until the question is a meaningful contribution rather than a single observation.

The question is yes/no answerable

"Is exercise good for health" can be answered yes by reading any of dozens of review articles. The fix is to ask which type, dose, or context produces which specific outcome in which specific population, transforming a yes/no question into a question that requires sustained investigation.

The question contains undefined or vague terms

"Does technology affect well-being" uses two terms ("technology," "well-being") that mean radically different things in different studies. The fix is to specify the technology (which platform, which device, which use pattern) and the well-being construct (which measure, which time frame, which dimension).

The question assumes the answer

"How does social media damage adolescent mental health" assumes the conclusion. The fix is to ask the relationship as a question rather than as a presumption: "Is social media use associated with adolescent mental health, and if so, in what direction and under what conditions?"

The question is not researchable with available methods

"What did Aristotle really think about justice" is not researchable; we can only access what Aristotle wrote, not what he thought. The fix is to reframe in terms of textual evidence: "How do Aristotle's accounts of justice in the Nicomachean Ethics, the Politics, and the Eudemian Ethics differ, and what do those differences suggest about the development of his thought?"

The question doesn't connect to existing literature

A research question developed without reading the field often re-asks a settled question or asks something incoherent. The fix is preliminary reading. There's no shortcut.

Research Questions for Different Project Types

The form of a research question depends partly on the project it's framing. Below are notes on what's appropriate for the most common project types.

Undergraduate research papers

Undergraduate research papers typically have a single research question that's narrow enough to be answered with secondary sources or modest primary data within a semester. The question should be focused, specific, and feasible for the page length and timeline. A reasonable undergraduate question is: "How did US monetary policy respond to the inflation surge of 2021 to 2023, and how did this response differ from the responses to inflation in the 1970s?"

Master's theses

Master's theses typically have one main research question and may have two or three sub-questions. The main question should be answerable with primary or secondary data collection that's feasible within the program timeline (typically 12 to 24 months). The sub-questions break the main question into testable components.

Doctoral dissertations

Doctoral dissertations may have one main research question with multiple sub-questions, or they may consist of three or four related studies (the "three-paper dissertation" format), each with its own research question. The questions should be sophisticated enough to merit four to six years of doctoral training and should make a clear contribution to the field.

Grant applications

Grant applications, including NIH R01s, NSF proposals, ERC grants, and Horizon Europe applications, typically include specific aims that function as research questions. Strong specific aims are framed as testable questions with clear hypotheses, defined methods, and measurable outcomes. The aims must collectively justify the requested funding and demonstrate that the research team can answer them.

Journal articles

Journal articles typically address one research question (or one closely related set of questions). The introduction motivates the question, the methods describe how it was studied, the results report what was found, and the discussion interprets the answer. A journal article that doesn't have a clearly stated research question is significantly harder to publish; reviewers and editors look for the question explicitly.

From Research Question to Research Manuscript

Once you have a strong research question, you've completed one of the hardest parts of the research process. The remaining work, designing the study, collecting data, analyzing results, and writing the manuscript, all flows from the question. But the writing matters too, and a strong study with weak writing often gets desk-rejected before reviewers consider the substance.

Editor World's academic editing service works with researchers from initial draft through journal submission. Our editors hold advanced degrees across the social sciences, the natural and physical sciences, medicine, engineering, computer science, and the humanities, and you select your editor by subject matter expertise before submitting. Editors average 15 years of professional experience and have edited manuscripts that have appeared in top-tier journals across disciplines. Our journal article editing service focuses specifically on preparing manuscripts to the standard international peer reviewers expect, and our dissertation editing service reviews full-length doctoral and master's theses as a unit. Every document is reviewed entirely by a qualified native English editor; no AI tools are used at any stage. A certificate of editing is available as an optional add-on for any manuscript.

For specific guidance on related stages of the research process, see our articles on how to outline an essay, how to improve essay writing, ideal paragraph length and structure, and what to do after journal rejection. For ESL researchers, our article on common English writing mistakes non-native speakers make covers the patterns that most often affect manuscript clarity. For a much larger collection of model research questions across disciplines and methods, see our companion article on research question examples.

Frequently Asked Questions

Content reviewed by Editor World editorial staff. Editor World provides professional academic editing, dissertation editing, journal article editing, and book editing services for researchers, doctoral candidates, faculty, and authors worldwide. Native English editors from the United States, the United Kingdom, and Canada with subject-matter expertise across the social sciences, the natural and physical sciences, medicine, engineering, computer science, and the humanities. No AI tools are used at any stage.