Choosing Your Research Topic and Question

From Curiosity to Question

Every research project begins with some form of curiosity — a puzzling clinical observation, an unexpected result in a preliminary experiment, a gap noticed while reading a review article, or a problem raised by a mentor or collaborator. But curiosity, however genuine, is not yet a research question. The transformation of a broad interest into a specific, answerable question is one of the most intellectually demanding stages of the entire research process, and it is one that new investigators often underestimate. Moving too quickly from a vague topic to a study design is a common mistake that leads to unfocused projects, wasted resources, and, ultimately, papers that reviewers struggle to evaluate because the central question was never clearly formulated.

A research topic and a research question are related but fundamentally different things. A topic is a domain of interest: the biology of synaptic plasticity, the epidemiology of type 2 diabetes in adolescents, the relationship between graduate student mental health and academic culture. A question is specific and answerable: Does chronic sleep restriction impair long-term potentiation in the hippocampus of adult mice? What is the prevalence of type 2 diabetes in adolescents aged 12 to 17 in urban areas of the United States between 2010 and 2020? Does the quality of mentorship relationships predict depressive symptoms in doctoral students in biomedical sciences? The difference matters enormously in practice. A topic tells you roughly which part of the library to visit; a question tells you exactly what data you need to collect, what comparisons you need to make, and what a successful answer would look like.

The process of moving from topic to question is iterative and requires engagement with the existing literature. Researchers who attempt to formulate a question in isolation — without reading widely in the relevant field first — often produce questions that have already been answered, questions that are unanswerable with available methods, or questions so broad that they encompass entire research programs rather than single studies. The literature search described in Chapter 3 is not a step that happens after the question is formulated; it is part of the process by which the question gets formulated. Preliminary reading reveals what is known, what is contested, and where genuine gaps exist, and this knowledge is indispensable for crafting a question that is both novel and tractable.

Characteristics of a Good Research Question: The FINER Criteria

One of the most useful frameworks for evaluating a proposed research question is the FINER criteria, an acronym that stands for Feasible, Interesting, Novel, Ethical, and Relevant NASEM 2018. Originally developed in the context of clinical research but applicable across the life sciences, FINER provides a systematic checklist that researchers can use to assess whether a question is ready for investigation or needs further refinement.

Feasibility is perhaps the most practical criterion. A question is feasible if it can be answered with the resources available: the time, funding, equipment, expertise, and access to research participants or samples that a given investigator can realistically mobilize. Many excellent scientific questions are not feasible for a particular researcher at a particular career stage. A medical student rotating through a clinical department cannot conduct a ten-year prospective cohort study. A graduate student with two years of funding cannot sequence the genomes of ten thousand patients. Recognizing the constraints within which one is working is not defeatism; it is the foundation of realistic project planning. The question needs to be scaled to fit the available resources, and if the full question cannot be addressed in a single project, it may be possible to address a tractable sub-question that advances the larger inquiry incrementally.

Interesting refers to the genuine intellectual engagement of the investigator and, equally importantly, of the broader scientific community. Research is hard, and projects that extend over months or years require sustained motivation. A question that does not genuinely interest the researcher is unlikely to sustain the effort that rigorous investigation demands. Beyond individual motivation, the question should be interesting to the field — it should address a problem that other researchers recognize as significant, even if they might approach it differently. A question that only one person in the world finds interesting may produce a publication, but it is unlikely to generate the citations, replications, and subsequent investigations that mark scientific progress.

Novelty means that the question has not already been definitively answered. This requires a thorough knowledge of the existing literature. Novelty does not require that nothing has been done on the topic before; most productive research questions build on prior work and are novel in a specific, delimited sense — they extend prior findings to a new population, apply a new method to a known problem, or test a hypothesis in a context where it has not previously been examined. What a novel question cannot be is a straightforward replication of a study that has already been conducted and reported, unless the replication itself is the explicit aim of the project (which, in the current climate of concern about reproducibility, is a legitimate and valuable goal in its own right).

Ethical feasibility means that the question can be investigated without violating the rights and welfare of research participants, whether human or animal. Ethics review processes — described in Chapter 5 — exist precisely to ensure that the scientific value of a question does not override the moral obligations researchers have to those whose cooperation makes the research possible. But ethics considerations should not be treated as bureaucratic hurdles to be cleared after the question is formulated; they should inform question formulation from the beginning. A question that requires exposing participants to unreasonable risk, or that involves deception that cannot be justified by the scientific value of the findings, or that disproportionately burdens already-vulnerable populations should be reformulated, not rationalized.

Relevance asks whether the question matters — to science, to clinical practice, to public health, to society. Relevant questions are those whose answers will make a difference: to the understanding of a biological mechanism, to the treatment of a disease, to the design of public health interventions, or to the development of scientific methodology. Relevance is closely linked to significance as assessed by funding agencies and journal editors; a question that cannot be framed in terms of its broader importance is unlikely to attract funding or find a publishing home. But relevance is not only a strategic consideration. Science consumes enormous resources — financial, intellectual, and human — and questions that are relevant to real problems justify that investment more readily than questions chosen primarily for their technical elegance or their alignment with available methods.

The PICO Framework for Clinical and Translational Research

For researchers working in clinical, epidemiological, or translational contexts, a complementary tool for question formulation is the PICO framework, which stands for Population, Intervention, Comparison, and Outcome. PICO provides a structured template for operationalizing a clinical research question in terms that map directly onto study design choices.

The Population element asks: who are the participants in this study? A well-specified population includes the relevant demographic characteristics, clinical characteristics, and any inclusion or exclusion criteria that define the study sample. Specifying the population precisely prevents the common problem of over-generalization, in which conclusions drawn from a highly specific sample are applied too broadly. A study conducted in adult men with moderate-to-severe heart failure cannot straightforwardly be generalized to women, to patients with mild heart failure, or to patients with heart failure complicated by renal disease.

The Intervention element asks: what is being done to or for the participants, or what exposure is being examined? In clinical trials, this is typically a treatment: a drug, a surgical procedure, a behavioral intervention, a public health campaign. In observational research, it may be an exposure: smoking, a dietary pattern, a socioeconomic condition. Specifying the intervention precisely includes specifying the dose, duration, route of administration, or other parameters that define exactly what is being studied Higgins et al. 2019.

The Comparison element asks: what is the intervention or exposure being compared against? In a randomized controlled trial, this is typically a placebo or standard-of-care control. In an observational study, it may be a comparison group defined by the absence of the exposure of interest. Some research questions do not involve a comparison — a purely descriptive study documenting the natural history of a disease, for example — but most analytical questions do, and specifying the comparison explicitly clarifies the causal or associational claim the study is designed to evaluate.

The Outcome element asks: what is being measured? Outcomes should be specified in advance, should be measurable with available instruments, and should be clinically or scientifically meaningful. The choice of outcome is not a minor technical detail; it is a fundamental determinant of the study’s significance. A study powered to detect a statistically significant difference in a surrogate outcome (a biomarker, an imaging finding) may or may not tell us anything about outcomes that patients actually care about (mortality, quality of life, functional status). Primary and secondary outcomes should be distinguished, and the rationale for the choice of primary outcome should be explicitly stated.

Scoping the Literature and Working with Advisors

The preliminary literature search that precedes and informs question formulation differs in character from the comprehensive systematic search conducted before a formal study begins. At the question-formulation stage, the goal is orientation: getting a sense of what the field knows, who the major investigators are, which methods are currently in use, and where the debates and uncertainties lie. This preliminary reading typically involves review articles, systematic reviews and meta-analyses, and the most recent primary literature in the relevant area. It need not be exhaustive, but it should be broad enough to prevent the common error of assuming that a question is novel when it has in fact been well-studied.

Working closely with an advisor or mentor during question formulation is invaluable. Experienced researchers have internalized knowledge of the field that does not appear in any single publication — knowledge of unpublished results, of failed approaches, of funding priorities, of methodological controversies, and of the personalities and perspectives that shape how the community is likely to receive different kinds of questions. A mentor can tell a graduate student that the question they are proposing has already been answered by a paper published two months ago, or that the approach they are considering is technically feasible but practically very difficult, or that the question touches on a contentious debate in which they will need to take a defensible position. This kind of guidance is not a substitute for independent intellectual engagement with the literature; it is a complement to it.

Avoiding Common Traps

New researchers encounter several characteristic pitfalls in question formulation that are worth naming explicitly. The question that is too broad is perhaps the most common: a question that encompasses so much territory that it could not possibly be answered by a single study, or by any study at all. “What causes cancer?” or “How does stress affect health?” are not research questions in the operative sense; they are research programs. Breaking such a question down into tractable sub-questions, each of which could in principle be answered by a specific study with specific data, is the essential intellectual work of question formulation.

The question that is too narrow is the opposite trap: a question so specific that its answer would be of interest to almost no one. Hyper-specific questions sometimes arise from an investigator’s desire to guarantee novelty — if the question has never been asked before, by definition no one has answered it. But novelty without significance is not a virtue. A study documenting the expression of a specific gene in a specific cell line under a specific set of experimental conditions may be technically rigorous and genuinely novel, but if no one outside the investigator’s immediate circle has reason to care about the result, the study will have little impact.

The question that has already been answered presents a different kind of problem. It is not always easy to know when a question has been definitively settled; sometimes the literature contains numerous studies that appear to address a question without any of them providing a truly convincing answer. But when a high-quality systematic review with consistent findings from multiple well-designed studies has addressed a question, the appropriate response is usually to build on that foundation rather than to replicate it. Understanding the difference between a question that has been studied and a question that has been answered requires careful critical reading of the literature and the judgment that comes with experience.

Finally, the question that is unanswerable deserves attention. Some questions are genuinely important but currently unanswerable because the necessary methods do not exist, the relevant populations are inaccessible, or the ethical constraints on the necessary study design are too severe. Recognizing this limitation is not a failure of imagination; it is an accurate assessment of the current state of the field. Sometimes the productive response is to ask a related question that can be answered now, with the explicit aim of building toward an answer to the larger question in the future.

Committing to a Question and Moving Forward

At some point, the iterative process of reading, discussing, and refining must produce a commitment to a specific question that will guide the design and execution of the study. This commitment does not mean the question can never change — research projects routinely evolve as preliminary data come in and as the investigator’s understanding deepens — but it means that the researcher is ready to move from exploration to execution. A well-formulated research question is specific enough to be answered by a real study, important enough to justify the investment of time and resources, and grounded enough in the existing literature to build meaningfully on what is already known. With that foundation in place, the next step is to search the literature comprehensively and systematically — a process taken up in the following chapter.