03 · What You Need to Know
Specificity Should Make the Study Determinate, Not Overloaded
A useful research question does two jobs at once. It establishes the intellectual problem the study will address, and it places enough boundaries around that problem to make systematic investigation possible.
Research-methods literature commonly describes question formulation as iterative rather than instantaneous. Researchers may begin with an initial question, review relevant literature, clarify important parameters, seek feedback, and progressively refine the question. This process helps connect the question to subsequent design, analysis, and reporting decisions.
The important issue is therefore not whether the first version of your question is perfectly specified. The more consequential issue is whether unresolved ambiguity remains when you reach decisions that depend on the question.
Start by distinguishing the research question from the research protocol
A research question identifies what the study is trying to find out. A protocol explains how the study will be conducted.
Those documents overlap conceptually, but they are not interchangeable. A protocol may specify recruitment procedures, sampling strategies, instruments, operational definitions, data-management procedures, statistical models, interview procedures, ethical safeguards, and many other details that would make a research question unreadable if placed directly inside it.
Research question
Defines the inquiry sufficiently to establish what needs to be answered and the evidence relevant to that answer.
Research protocol
Specifies how the researchers will generate, collect, manage, analyze, and interpret the evidence needed to address the question.
A question can therefore be methodologically informative without becoming a miniature methods section.
The central phenomenon or relationship should be clear
Before beginning the study, you should be able to say what you are actually investigating. This sounds obvious, but broad concepts often conceal several possible studies.
“How does AI affect students?” is not specific enough to determine whether the study concerns learning, assessment, motivation, academic integrity, writing, cognitive processes, employability, or something else.
By contrast, “How do undergraduate students describe using generative AI while revising academic essays?” identifies a recognizable phenomenon and form of inquiry. It still leaves many methodological decisions for the protocol.
The required specificity therefore begins with conceptual clarity. If two researchers could read the question and reasonably plan entirely unrelated studies, the question probably needs further refinement.
The population or cases should be sufficiently bounded when they matter
Many research questions need some indication of whom or what the study concerns. This may be a patient population, students, teachers, organizations, documents, communities, events, datasets, cases, or another unit of inquiry.
That does not mean every inclusion criterion belongs in the question. “Undergraduate students” may be sufficient for one question, whereas another may genuinely concern first-generation first-year engineering students because those characteristics are integral to the problem.
The key test is substantive rather than grammatical: would changing the population change the meaning of the question or the inference you want to make?
This distinction becomes especially useful when deciding whether the population, variables, context, and outcome need to appear explicitly in the question. Not every study requires the same elements in the same form.
Specify the outcome when the question depends on a particular outcome
Some questions become meaningful only when the outcome is clear. “Does the intervention work?” is usually inadequate because “work” could refer to several outcomes, measured over different periods and with different implications.
For intervention and other relational quantitative questions, frameworks such as PICO or PICOT can help clarify the population, intervention or exposure, comparison, outcome, and sometimes time. Research-question guidance notes that these components can be specified at different levels of detail and refined iteratively rather than merely inserted mechanically into a template.
The outcome named in the question also need not always be the exact instrument used to measure it. “Depressive symptoms,” for example, identifies a construct. The particular validated instrument and scoring procedure may belong in the methods unless the measurement itself is central to the research question.
Specify comparisons when the comparison defines the question
If the purpose is to compare groups, conditions, interventions, exposures, periods, or approaches, the relevant comparison should generally be clear before the study begins.
“Is method A more effective?” is incomplete if the intended comparator remains uncertain. More effective than method B? Existing practice? No intervention? Participants' own baseline performance?
Comparison choices affect sampling, data collection, analysis, interpretation, and sometimes sample-size requirements. Leaving them unresolved can mean that the study question itself remains unresolved.
Time should be specified when time changes the meaning of the answer
Not every question needs a date or duration in its wording. Time becomes important when the outcome depends on when it is measured or when the phenomenon is inherently temporal.
An intervention might produce a short-term change that disappears six months later. Student experiences during the first weeks of university may differ from experiences near graduation. A policy may operate differently before and after a major institutional change.
Methodological guidance on structured research questions recognizes that timing can affect outcome development, attrition, recall, and other design considerations.
If changing the observation period could materially change the answer, the time frame should at least be settled in the study design and may deserve a place in the research question.
Do not add specificity merely to make the question look rigorous
A common mistake is to equate detail with rigor. Researchers may keep adding demographic characteristics, dates, instruments, sites, and methodological terminology because the longer question appears more scientific.
Consider:
“Among 18- to 20-year-old first-year undergraduate students enrolled in Section A of Introduction to Psychology at University X during the first semester of academic year Y, what is the association between daily self-reported minutes of social media use measured using Instrument Z and final examination scores?”
Some of those details might be necessary. Others may simply describe how the particular study happens to be conducted.
A more concise question might ask: “Among first-year undergraduate students, is social media use associated with academic performance?” The methods can then define the institutional setting, operationalization of social media use, outcome measure, sampling procedure, and study period, provided those choices do not alter the substantive question.
Specificity should remove consequential ambiguity, not advertise every decision the researcher has made.
Too little specificity can hide several different questions
A question may appear elegant precisely because it leaves all difficult decisions unstated.
“Does social media influence academic achievement?” could involve different platforms, forms of use, populations, measures of achievement, time periods, and causal interpretations. The researcher cannot sensibly choose a sample, measurement strategy, or analysis until some of those ambiguities are resolved.
This is closely related to recognizing when a research question is too broad. Lack of specificity is problematic when the omitted boundaries expand the question beyond what one coherent study can address.
Too much specificity can prematurely constrain the study
The opposite problem also occurs. Researchers sometimes specify details before they know whether those details are theoretically justified, feasible, or methodologically appropriate.
A question may name a particular instrument before the researcher has established whether that instrument validly measures the intended construct. It may restrict participants by age or program without a substantive reason. It may name one technology even though the research problem concerns a broader practice.
Such restrictions can shrink the eligible evidence without improving the question. In extreme cases, the attempt to make the question more precise produces a research question that is too narrow.
A useful rule is that every important restriction should be defensible. If you cannot explain why a boundary matters to the phenomenon, inference, design, or contribution, reconsider whether it needs to define the question.
Quantitative studies often need more decisions fixed before data collection
In many quantitative studies, important elements of the question should be settled before examining outcome data because those elements drive hypotheses, measurement, sampling, design, and analysis.
Research-question guidance commonly recommends refining quantitative questions by identifying relevant populations, interventions or exposures, comparisons, and outcomes when appropriate. FINER criteria then help assess whether the resulting question is feasible, interesting, novel, ethical, and relevant.
This does not mean all quantitative questions must follow PICO. Descriptive questions, diagnostic questions, prognostic questions, methodological studies, and other quantitative inquiries may require different structures. The broader principle is that the components necessary to interpret the intended analysis and inference should not remain ambiguous.
Qualitative studies may legitimately begin with more openness
Some qualitative approaches require a different balance. A qualitative question may initially be broad but clear, then become more focused as the researcher reads further, engages with the field, and develops understanding during early analysis. This iterative process is consistent with the cyclical character of many qualitative designs.
That flexibility should not be mistaken for absence of direction. Before collecting data, the researcher still needs a defensible phenomenon of interest, appropriate participants or cases, an ethical plan, and a methodological approach capable of generating relevant evidence.
The distinction is explored further in how qualitative research questions differ from quantitative ones. The appropriate form of specificity follows the logic of the methodology rather than a universal sentence template.
Specificity should increase as consequential decisions approach
One way to think about question development is as progressive commitment.
| Stage |
Reasonable level of specificity |
Main purpose |
| Initial idea |
Broad but identifiable problem or phenomenon |
Establish what interests you and why it may warrant investigation |
| Preliminary literature review |
Increasingly focused concepts, population or cases, and intended contribution |
Determine what is known, what remains uncertain, and which question is worth pursuing |
| Study planning |
Enough specificity to choose an appropriate design, evidence, sampling strategy, and analysis |
Test feasibility and methodological alignment |
| Before data collection |
Key elements that determine what evidence will answer the question should be settled, subject to the logic of the methodology |
Prevent ambiguity about what study is actually being conducted |
This progression is not perfectly linear. Literature may force you to reconsider the population. A feasibility assessment may reveal that the intended data are inaccessible. Methodological consultation may expose an inference that your design cannot support. Research-question formulation is commonly described as iterative for precisely this reason.
Ask whether the question now determines a plausible study
A practical stopping test is to hand the question to another researcher familiar with your field. They should not necessarily reproduce your exact protocol, but they should understand the central inquiry and be able to identify the general kind of evidence and design needed to answer it.
If several fundamentally different studies could still satisfy the wording, more specification may be necessary. If only your exact instrument code, recruitment dates, software version, and room number remain unstated, you have probably crossed from question formulation into protocol writing.
The final standard is whether the question has a credible path to an answer. Specificity is useful because it helps create that path, not because detailed questions are inherently more scholarly.
