How to Write a Strong Research Question for Your Thesis: Step-by-Step Guide 2026

Your research question is the foundation of your entire thesis. Every other decision — your methodology, your literature review scope, your sample, your analysis approach — flows from it. A weak research question produces a weak thesis, no matter how much work you put in after. Yet many students finalise their research question too quickly, discover its problems mid-study, and spend months trying to recover. This guide shows you exactly how to formulate a strong, researchable question from the start — with worked examples across disciplines and the frameworks that PhD supervisors at leading universities use to evaluate research questions.

The FINER Criteria for a Strong Research Question

The FINER framework (Hulley et al., 2013), originally developed for clinical research, is now widely used across social sciences, education, and humanities to evaluate research questions:

• Feasible: Can this be answered with the resources, time, access, and skills you have? A master’s student cannot conduct a 10-year longitudinal study.
• Interesting: Does the question matter to the field? Would a journal publish an answer to it? Would practitioners care?
• Novel: Does answering this question add something new to knowledge? (This connects directly to your research gap — see our guide on finding a research gap.)
• Ethical: Can the study be conducted without harm to participants, and will it gain ethics approval?
• Relevant: Does it address a real problem in the field, or contribute to policy, practice, or theory?

Run your proposed research question through all five criteria before finalising it. A question that fails any one criterion will undermine your thesis.

Types of Research Questions

Research questions follow predictable types, and the type you choose determines your methodology:

Your question type and your methodology must align. An exploratory “what are the experiences of…?” question cannot be answered with a questionnaire and regression analysis — it requires qualitative methods. See our guide on choosing the right research methodology for the full alignment framework.

Step-by-Step: Going from Topic to Research Question

1. Start with a broad topic. “AI tools in academic writing.”
2. Identify what interests you most within that topic. “How students use them and whether universities can tell.”
3. Review the literature. What do you know is already studied? What is missing? (See the literature review methodology guide.)
4. Identify a gap. “Most studies examine what students do; few examine students’ own understanding of the ethical boundaries.”
5. Draft an initial question. “How do postgraduate students understand the ethical implications of AI tool use in thesis writing?”
6. Apply FINER. Feasible? Yes, via interviews. Interesting? Yes. Novel? Check literature — yes, underexplored. Ethical? Sensitive but approvable. Relevant? Highly.
7. Specify the population and context. “…among postgraduate students at Russell Group universities in the UK.”
8. Add a timeframe if relevant. “…in the context of the 2024–2026 policy environment.”
9. Final research question: “How do postgraduate students at UK Russell Group universities understand the academic integrity implications of using AI writing tools in thesis writing, and how do they navigate institutional policy ambiguity?”

15 Worked Research Question Examples by Discipline

Education: “What is the effect of peer feedback on thesis completion rates among first-generation doctoral students at research-intensive UK universities?”

Psychology: “To what extent does academic perfectionism predict thesis avoidance behaviour in master’s students, and what role does self-compassion play as a moderating variable?”

Public Health: “How have UK university mental health services adapted their provision to address the specific wellbeing needs of postgraduate research students since 2020?”

Business: “How do small and medium enterprises in the UK manufacturing sector perceive the barriers to implementing sustainability reporting frameworks?”

History: “How did women’s writing collectives in post-war Britain construct narratives of domestic labour as political action in the period 1945–1965?”

Computer Science: “What are the security vulnerabilities introduced by large language model API integrations in academic writing platforms, and what mitigation strategies are most effective?”

Environmental Science: “What is the relationship between urban heat island intensity and income deprivation across English local authorities, and how has it changed between 2010 and 2025?”

Nursing: “What are the experiences of UK community nurses in managing palliative care for patients with dementia, and what training gaps do they identify?”

Law: “How have UK tribunals interpreted the concept of ‘legitimate expectation’ in immigration cases since the Supreme Court ruling in Pham v Secretary of State for the Home Department [2015]?”

Sociology: “How do second-generation immigrant women in Glasgow construct hybrid cultural identities through social media practices?”

Refining a Weak Research Question

Most first drafts of research questions have identifiable problems. Common issues and fixes:

Too broad: “How does social media affect mental health?” → Fix: “What is the relationship between passive Instagram use (scrolling without posting) and self-reported loneliness among UK undergraduate women aged 18–22?”

Not researchable (a value question): “Should universities ban AI writing tools?” → Fix: “What are the views of academic integrity officers at UK universities on effective AI policy for postgraduate thesis writing?”

Already answered: Check with Google Scholar + Elicit + ResearchRabbit before finalising. If the question is already well answered, you need to find a new angle or population.

Too narrow: “What did Participant 3 experience during her second year of PhD study?” → This is a case description, not a research question. Broaden the population.

Frequently Asked Questions