A research question is one of the most important components of an academic research project.
It gives the investigation direction and helps determine what information needs to be collected, what literature needs to be reviewed, what methodology may be appropriate, and what conclusions the study can reasonably reach.
A strong research question is not simply a topic written in question form.
For example:
“What is social media?”
is a question, but it is unlikely to provide sufficient direction for a substantial research project.
A more focused question might be:
“How does social media use influence purchasing decisions among university students?”
The second question identifies a subject, a relationship of interest, a population, and a context.
This guide explains how to develop research questions that are clear, focused, researchable, and appropriately aligned with academic research.
For broader academic writing principles, see the Academic Writing Basics Guide.
A research question is a clearly formulated question that a research project is designed to investigate and answer using appropriate evidence.
It identifies what the researcher wants to understand, explain, compare, explore, measure, or evaluate.
Depending on the research project, a research question may concern:
The question should provide enough direction to make the research manageable without unnecessarily restricting the investigation.
Research questions influence many parts of a research project.
They can help determine:
A well-designed question therefore contributes to research coherence.
A useful alignment can be represented as:
Research Problem → Research Question → Research Objectives → Methodology → Data → Analysis → Findings → Conclusion
If the question is unclear, problems can appear throughout the project.
These concepts are related but different.
The topic is the broad area of interest.
Examples:
A topic is usually too broad to serve as the final research question.
The research problem identifies an issue, uncertainty, gap, contradiction, difficulty, or situation that warrants investigation.
For example:
Employee turnover remains a significant concern for some organisations, but the factors influencing retention among early-career employees in small technology companies are not sufficiently understood.
This provides a reason for research.
The question converts the problem into something that can be investigated.
For example:
What factors influence employee retention among early-career employees in small technology companies?
The three components can therefore work together:
Topic: Employee retention
Problem: Factors affecting retention in a particular context are not sufficiently understood.
Question: What factors influence employee retention among early-career employees in small technology companies?
Consider:
Artificial intelligence in healthcare
This is a topic.
Possible research questions include:
How are healthcare professionals using artificial intelligence to support diagnostic decision-making?
or:
What factors influence healthcare professionals’ adoption of artificial intelligence tools?
or:
What are the perceived benefits and risks of artificial intelligence-assisted diagnosis among healthcare professionals?
Each question investigates a different issue.
This demonstrates why defining the question is more useful than simply selecting a broad topic.
A strong research question is generally:
It should also be realistic within the available:
The reader should understand what the question means without excessive interpretation.
Weak:
What is the effect of technology on society?
This could involve almost anything.
Stronger:
How has the adoption of mobile payment technology affected small-business transaction practices in urban Kenya?
The second question provides greater direction.
A research question should be narrow enough to investigate properly.
Consider:
How does climate change affect the world?
This is too broad for most individual academic projects.
A more focused version might be:
How has changing rainfall variability affected smallholder maize production in selected agricultural regions of Kenya?
The second question provides boundaries around:
A question should be capable of being investigated using available evidence.
For example:
What will the global economy look like in 2100?
may be difficult to answer meaningfully because of the enormous number of variables and uncertainty involved.
A more researchable question might be:
How do selected economic forecasting models estimate the long-term effects of demographic change on economic growth?
The second question defines a researchable analytical problem.
A question can be academically interesting and still be too large for the available resources.
A doctoral research programme may investigate a question that would be impossible to answer adequately in a short undergraduate assignment.
Always consider:
A good question fits the project.
Not every research question should have the same form.
A closed question may produce a yes-or-no answer.
For example:
Does employee training improve productivity?
This can be useful in certain research designs, but it may not provide sufficient depth by itself.
An open question allows greater exploration.
For example:
How does employee training influence productivity among manufacturing workers?
This allows the research to investigate mechanisms, experiences, relationships, and contextual factors.
The appropriate form depends on the purpose of the research.
Different research questions serve different purposes.
These ask what is happening.
Examples:
What types of cybersecurity practices are used by small businesses?
What are the most common challenges experienced by remote workers?
Descriptive questions can establish the characteristics of a phenomenon.
Exploratory questions investigate an issue that may not yet be well understood.
Examples:
How do first-generation university students experience online learning?
How do small businesses perceive the use of generative AI for customer service?
Exploratory questions are often suitable for qualitative research.
Explanatory questions investigate why something happens or how factors may be related.
Examples:
Why do some employees leave organisations despite receiving competitive salaries?
How does organisational culture influence employee retention?
These questions often require more than simple description.
Comparative questions examine differences between groups, contexts, methods, periods, or cases.
Examples:
How does employee satisfaction differ between remote and office-based workers?
What differences exist between public and private university students’ use of digital learning platforms?
These investigate whether and how variables or phenomena are related.
Examples:
What is the relationship between financial literacy and personal savings behaviour among university students?
How is employee engagement associated with staff turnover?
A relationship question does not automatically establish causation.
Evaluative questions examine the effectiveness, value, suitability, or outcomes of an intervention, policy, programme, or strategy.
Examples:
How effective is a workplace mentoring programme in improving employee retention?
To what extent has a digital marketing campaign achieved its stated objectives?
Causal questions examine whether one factor produces or contributes to a change in another.
For example:
Does targeted employee training improve workplace productivity?
Causal claims generally require an appropriate research design and careful control of alternative explanations.
A research question should not imply stronger causal evidence than the methodology can provide.
Predictive research questions examine whether available information can be used to estimate future outcomes.
For example:
Can customer transaction data be used to predict repeat purchases among online retail customers?
These questions may involve statistical modelling or machine learning.
See the Machine Learning glossary for background on machine learning approaches that may be relevant to some predictive studies.
Qualitative research often explores:
Questions may begin with:
Examples:
How do first-time entrepreneurs experience access to business finance?
What factors shape nurses’ experiences of workplace stress?
How do university students perceive the use of AI tools in academic learning?
The wording should reflect the intended qualitative inquiry rather than force the study into a quantitative framework.
Quantitative questions often involve:
Examples:
What is the relationship between employee engagement and staff turnover?
Does participation in professional training predict employee productivity?
Is there a significant difference in customer satisfaction between online and physical service channels?
The exact wording should correspond to the research design and statistical analysis.
Mixed-methods research combines quantitative and qualitative approaches.
A project might investigate:
What factors are associated with employee turnover, and how do employees explain their decisions to leave?
The first component may be examined quantitatively.
The second may be explored qualitatively.
The two components can then be integrated.
A complex project may require one central research question supported by several sub-questions.
For example:
How does remote work influence employee performance in small technology companies?
The main question establishes the overall direction.
The sub-questions divide the investigation into manageable components.
There is no universal number.
A small research project may have one main question.
A larger project may have one main question and several sub-questions.
The important issue is not the number but whether the questions are:
Too many questions can make a project unnecessarily broad.
Consider:
How does technology affect business?
This includes:
A project attempting to address all of these areas would probably lack sufficient focus.
Narrowing the question may involve specifying:
For example:
How does the adoption of cloud-based accounting software affect financial reporting practices among small businesses in Nairobi?
This is much more manageable.
The opposite problem is making the question so narrow that meaningful research becomes difficult.
For example:
What did five employees in one department think about one software update implemented on Tuesday?
This may be answerable but could have limited academic significance depending on the research purpose.
The appropriate level of specificity depends on the project.
A double-barrelled question asks two substantially different things at once.
For example:
How does remote work affect employee productivity and job satisfaction?
Productivity and job satisfaction are separate outcomes.
A clearer approach may be:
How does remote work affect employee productivity?
and:
How does remote work affect employee job satisfaction?
Alternatively, if both outcomes are essential, the research design should explicitly accommodate both.
A leading question assumes a particular answer.
For example:
Why does social media negatively affect students?
The wording already assumes that social media has a negative effect.
A more neutral question might be:
How does social media use affect students’ academic experiences?
The revised version allows the evidence to determine whether the effect is positive, negative, mixed, or insignificant.
Research questions should not unnecessarily assume facts that have not been established.
Weak:
Why are online courses less effective than classroom teaching?
The question assumes online courses are less effective.
A more neutral alternative might be:
How do students perceive the effectiveness of online and classroom-based learning?
The second question permits comparison.
Some questions can be answered by finding information in a textbook or reference source.
For example:
When was the internet invented?
This is a factual question rather than a substantial research question.
A research-oriented question could instead examine:
How did the development of the internet transform information-sharing practices in higher education?
The second question requires investigation and analysis.
A research project should ideally address an issue where investigation can produce meaningful understanding.
For example:
Do people need food to survive?
is unlikely to provide a useful research project.
A stronger question would involve a genuine issue requiring evidence.
Some questions may be technically interesting but impossible or inappropriate to investigate because of ethical constraints.
Consider:
How can employees be manipulated into working longer hours without noticing?
The problem is not merely methodological.
The research purpose itself raises serious ethical concerns.
Academic research should comply with applicable institutional and professional ethical standards.
The literature review helps refine the research question.
You may begin with a broad interest such as:
AI in education.
After reviewing existing literature, you may discover that:
The research question can then become more focused.
For example:
How do secondary-school teachers perceive the use of generative AI for lesson planning?
The literature therefore does not simply support the final paper.
It can help identify what should actually be investigated.
See the Literature Review Guide.
The question should respond directly to the research problem.
Suppose the problem is:
Small businesses increasingly use digital payment systems, but there is limited understanding of how these systems affect cash-flow management among small retailers.
A suitable question might be:
How does the use of digital payment systems affect cash-flow management among small retailers?
The connection is clear.
A weak question would investigate an unrelated issue such as:
What are the most popular social media platforms among consumers?
That may be interesting, but it does not address the stated problem.
Research questions and objectives are closely connected but serve different purposes.
A research question asks what the research wants to find out.
An objective describes what the research intends to accomplish.
For example:
What factors influence customer loyalty among online retail customers?
To identify factors influencing customer loyalty among online retail customers.
Another:
How does customer service quality affect customer loyalty?
To examine the relationship between customer service quality and customer loyalty.
The question and objective should correspond closely.
See the Research Objectives Guide.
The research question should influence methodological decisions.
For example:
How do employees experience workplace burnout?
A qualitative approach may be appropriate if the goal is to understand experiences and meanings.
What is the relationship between workload and burnout scores?
A quantitative approach may be appropriate if the study requires measurement of variables and statistical analysis.
How common is workplace burnout among employees?
A descriptive quantitative design may be appropriate.
The question does not mechanically determine one method, but it provides an important starting point for methodological reasoning.
See the Research Methodology Guide.
Ask:
What information would I need to answer this question?
For example:
What factors influence employee retention?
Potential information could include:
The question should therefore guide data collection rather than being written independently of it.
The planned analysis should be capable of answering the research question.
For example, if the question asks about a statistical relationship, the researcher needs an analytical method capable of examining that relationship.
If the question asks about people’s experiences, numerical analysis alone may not capture the information required.
This is why research planning should connect:
Question → Data → Analysis → Answer
See the Data Analysis glossary.
Begin with an area that interests you or relates to your discipline.
Examples:
Ask:
What problem, uncertainty, gap, or debate interests me?
Look for:
Specify where, when, or among whom the research will take place.
Are you trying to:
Do not expect the first version to be perfect.
Write several alternatives.
Check them against your available time, data, resources, methodology, and academic requirements.
Choose the question that best addresses the research problem.
Convert the selected question into practical research objectives.
Make sure the question, objectives, literature, methodology, data, analysis, and conclusions can work together.
Suppose the broad topic is:
Remote work.
The topic is too broad.
How does remote work affect employees?
Still broad.
How does remote work affect employee productivity?
Better, but the context remains unclear.
How does remote work affect productivity among employees in small technology companies?
More focused.
How does the frequency of remote work influence self-reported productivity among employees in small technology companies?
Now the researcher has identified:
The final question is substantially more specific.
Some health and applied research projects use frameworks such as PICO.
PICO can represent:
For example:
Among adults with a particular condition, does intervention A compared with intervention B improve outcome C?
PICO is particularly useful in evidence-based health research and systematic review contexts.
It is not necessary for every academic discipline.
Some qualitative and mixed-methods research projects use frameworks such as SPIDER.
The components may include:
Such frameworks can help researchers develop more focused questions when conventional variable-based approaches are not appropriate.
The framework should be selected because it improves the research design, not because every project needs a formal acronym.
Business research questions often investigate:
Examples:
What factors influence customer retention in subscription-based businesses?
How does digital transformation affect operational efficiency among small enterprises?
What factors influence the adoption of cloud accounting systems among small businesses?
The question should identify a meaningful business issue rather than simply describe a commercial topic.
Education research may investigate:
Examples:
How do students perceive the use of digital simulations in science education?
What factors influence university students’ engagement with online learning?
Does structured peer feedback improve students’ academic writing performance?
Technology-related questions should avoid being so broad that they become impossible to investigate.
Weak:
How is AI changing the world?
Better:
How do small businesses use generative AI to support customer communication?
More specialised:
What factors influence the adoption of generative AI tools for customer-service communication among small businesses?
Research involving AI may also require careful consideration of:
See the Artificial Intelligence glossary.
Data-related projects often involve measurable variables and analytical relationships.
Examples:
What factors predict customer churn among subscription-service users?
Is there a relationship between advertising expenditure and online sales?
Can historical sales data be used to forecast monthly demand?
These questions should be matched with an appropriate data source and analytical method.
A literature review may itself have a research question.
For example:
What does existing research reveal about the factors influencing employee retention in small businesses?
A systematic review may have a more narrowly defined question based on the review methodology and eligibility criteria.
The review question should determine what literature is included and excluded.
When preparing a proposal, the research question should be sufficiently developed to show that the proposed study is feasible.
A strong proposal should demonstrate:
See the Research Proposal Guide.
A research question establishes boundaries.
Scope can be defined by:
For example:
How do small retailers in Nairobi perceive digital payment systems?
is substantially narrower than:
How do businesses perceive digital payment systems?
The appropriate scope depends on the research purpose.
Geographic limitations can make a question more manageable.
Examples:
How do small businesses in Kenya use social media for customer acquisition?
or:
How do small businesses in Nairobi use social media for customer acquisition?
The second question is narrower.
However, the geographic boundary should have a legitimate research rationale rather than being added merely to make the title appear specific.
Some research questions need a time boundary.
For example:
How did remote-working practices change among employees between 2020 and 2025?
A time boundary can be particularly useful for:
The population should be defined where relevant.
Compare:
How does financial literacy influence saving behaviour?
with:
How does financial literacy influence saving behaviour among university students aged 18 to 25?
The second is more specific.
However, unnecessarily detailed population definitions can make the study difficult to conduct.
Before finalising a question, ask:
Can another student understand exactly what I want to investigate?
Is the question narrow enough for the available resources?
Does it address a meaningful academic or practical issue?
Can evidence be collected to answer it?
Can I complete the research within the available time and resources?
Can the study be conducted responsibly?
Can an appropriate research method address the question?
Can I realistically obtain the required information?
Can I analyse the information in a way that answers the question?
Could answering the question provide useful knowledge, insight, evaluation, or understanding?
Before approving a research question, check:
“Artificial intelligence in education” is a topic, not necessarily a research question.
A question covering an entire industry, country, population, or technology may be impossible to investigate adequately.
A question can become so specific that it produces little meaningful information.
Separate substantially different research problems where necessary.
Avoid wording that presumes the outcome.
Knowing what happens is different from understanding why it happens.
A theoretically excellent question is not useful if the required evidence cannot realistically be obtained.
The question and research method should be compatible.
If the research question changes, review the:
Research documents need internal alignment.
Take your current question and perform five revisions.
Make the question direct.
Specify who or what is being studied where necessary.
Identify the industry, institution, location, period, or setting where appropriate.
Determine whether you are describing, exploring, comparing, explaining, evaluating, measuring, or predicting.
Ask whether the research can actually be completed.
Once the research question has been established, it should influence the structure of the academic document.
A typical research project may progress as follows:
Research Problem
↓
Research Question
↓
Research Objectives
↓
Literature Review
↓
Methodology
↓
Data Collection
↓
Data Analysis
↓
Findings
↓
Discussion
↓
Conclusion
This relationship is one reason research questions should be developed before writing the full research document.
For more information about organising research documents, see the Academic Writing Structure Guide.
Research questions sometimes evolve as researchers learn more about the subject.
This is particularly common during exploratory research.
However, a change should not simply be made in one location.
Review all related components.
For example:
Old Question
How does social media affect business performance?
New Question
How do small retailers in Nairobi use Instagram to acquire new customers?
The revised question may require changes to:
Research alignment should be checked after substantial changes.
A research question should represent the genuine purpose of the research.
Do not create a question merely because it contains popular search terms or sounds sophisticated.
The question should arise from a legitimate research interest or problem.
Similarly, research sources should be evaluated rather than selected simply because they support a preferred answer.
Good research allows evidence to challenge assumptions.
AI tools can potentially help generate preliminary ideas or identify alternative ways of wording a question.
However, an AI-generated question should be treated as a starting point rather than an automatically suitable research design.
The researcher should independently evaluate:
AI should not replace the researcher’s responsibility for defining the research problem.
A practical workflow is:
Choose an area relevant to your discipline.
Determine what is uncertain, disputed, under-researched, inefficient, unexplained, or important.
Read relevant academic literature.
Ask what remains unanswered.
Determine who, what, where, or when is being studied.
Decide whether the study is descriptive, exploratory, explanatory, comparative, evaluative, causal, predictive, or another appropriate type.
Do not settle immediately on the first wording.
Consider time, data, resources, ethics, and methodology.
Choose the question that best addresses the problem.
Create objectives that correspond to the question.
Determine how the question will be investigated.
Make sure the complete research project points toward answering the question.
A good research question is clear, focused, researchable, relevant, feasible, and aligned with the research problem and methodology.
Begin with a broad topic, identify a specific problem or uncertainty, review existing research, and then formulate a question that can realistically investigate the issue.
A topic identifies a broad area of interest. A research question specifies what the researcher wants to investigate within that area.
There is no universal number. A dissertation may have one central question supported by several sub-questions, depending on the research design and institutional requirements.
Usually, a concise question is easier to understand. However, complexity should not be sacrificed merely to achieve a particular length.
It can be useful when geography is relevant to the research problem or necessary to make the project manageable.
Where the population is important to the research, identifying it can make the question considerably clearer.
It can, but combining separate research problems can make a project unnecessarily broad. If two outcomes or issues are being investigated, ensure that the methodology can adequately address both.
Common openings include:
The best opening depends on the research purpose.
It can, but many academic projects benefit from questions that permit deeper investigation and analysis. A yes-or-no question can be appropriate for particular research designs.
Often, yes. Qualitative questions commonly focus on experiences, meanings, perceptions, or processes, while quantitative questions often focus on measurable variables, differences, relationships, or effects.
Yes. Research questions may evolve as the researcher develops a better understanding of the literature or research setting. Significant changes should be reflected throughout the research design and document.
The research question states what the study seeks to investigate. Research objectives describe the specific actions or outcomes through which the study will address that question.
A hypothesis is a testable proposition about an expected relationship or outcome. Not every research question requires a hypothesis.
AI can assist with brainstorming and alternative wording, but the researcher should evaluate the resulting question for relevance, feasibility, accuracy, originality, ethics, and methodological suitability.
Continue building your research and academic writing knowledge with:
Research & Consulting Services can support research planning, analytical projects, research documentation, data interpretation, reports, and evidence-based consulting work.
Education & Training Services provides resources and professional support relevant to learning, training, educational development, and professional development.
Documents & Content Services covers professional document development, editing, research documentation, reports, and related content requirements.
AI & Data Services covers data-driven technology, artificial intelligence, analytics, automation, and related professional applications.
A research question provides the central direction of an academic investigation.
The strongest questions do not attempt to cover everything about a subject. Instead, they identify a meaningful issue and establish boundaries that allow the researcher to investigate it rigorously.
A useful development path is:
Broad Topic → Research Problem → Research Gap → Research Question → Objectives → Methodology → Data → Analysis → Findings → Conclusion
Before finalising a question, make sure it is:
Clear. Focused. Researchable. Relevant. Feasible. Ethical. Aligned.
A well-designed question can make the rest of the research project substantially easier to plan because it provides a reference point for deciding what information belongs in the literature review, what data needs to be collected, what methodology is appropriate, and what conclusions the research can legitimately support.