What Is the Minimum You Need to Know Before You Can Responsibly Start Designing the Study?

How much must you know before study design can responsibly begin?

There is an awkward point in research planning when the idea seems clear enough to move forward, yet many details remain unresolved. You may know the broad question but not the exact sample size. You may know whom you want to study but not the final recruitment procedure. You may have several possible measures, analytical strategies, or data sources still under consideration.

The difficulty is deciding which uncertainties are normal at this stage and which ones mean you are trying to design the study too early. Waiting until everything is known is unrealistic. Starting design while foundational assumptions are still vague can be equally problematic because apparently technical choices about sampling, measurement, comparison, and analysis depend on what the study is actually supposed to establish.

The useful question, then, is not whether the project is completely planned. It is whether you know enough to begin making design decisions without having to guess about the foundations on which those decisions depend.

You need clarity about the foundations, not every methodological detail

The minimum foundation for responsible study design

Study design translates a research question into a plan for producing and interpreting evidence. That translation cannot happen sensibly if the question, target of inference, or relevant evidence is still fundamentally unclear. Objectives and research questions provide the foundation from which design and analysis decisions follow, although the exact requirements differ substantially across experimental, observational, qualitative, mixed-methods, review, and other forms of research.

This does not mean that you must have a completed protocol before beginning design. That would reverse the process. It means that several foundational matters should be sufficiently clear for methodological choices to have a defensible rationale.

1. You need an answerable question or sufficiently precise objective

You should be able to explain what the study is trying to find out without relying on a broad topic as a substitute for a research question. “Artificial intelligence in higher education,” for example, identifies an area of interest. It does not yet tell you whether you want to estimate prevalence, understand experiences, compare outcomes, test an intervention, explain a relationship, develop an instrument, or accomplish something else.

Those purposes lead to quite different designs. Before choosing methods, the question should therefore be precise enough that you can identify what an informative answer would look like.

If you cannot yet describe the question with that degree of clarity, the project may still need more conceptual planning before it is ready to move into study design.

2. You need to know what kind of claim the study is intended to support

A research question carries an implied evidentiary burden. Estimating how common something is requires different evidence from estimating an association. Describing participants' experiences requires different evidence from evaluating whether an intervention caused an outcome. Predicting an outcome is not the same task as explaining why that outcome occurs.

You do not necessarily need to know the formal design label yet. In fact, choosing a label too early can make researchers force the question into a familiar method. What you do need is enough clarity about the intended claim to evaluate whether a proposed design could actually support it.

3. You need to know who or what the question is about

Study design requires a target. Depending on the research, that target might be people, classrooms, schools, organizations, documents, publications, datasets, events, communities, laboratory specimens, or another unit of interest.

You should have a defensible conception of the population, cases, setting, or material to which the question refers. This does not mean that the final sampling strategy must already be complete. It does mean that you should not be designing recruitment or sampling procedures while still being unsure about whom or what the resulting evidence is intended to represent or illuminate.

4. You need to know what must be observed, measured, elicited, or compared

You may not have selected the final instrument, operational definition, interview questions, coding framework, or data extraction form. Those can be design decisions. However, the important constructs, phenomena, exposures, outcomes, experiences, or other forms of evidence should be identifiable.

Suppose your question concerns whether an educational intervention improves “engagement.” Before designing the study, you need to understand what engagement means in the context of your question. Is it behavioral participation, time on task, emotional engagement, cognitive investment, attendance, interaction with a platform, or a multidimensional construct? A design cannot rescue a construct that has never been conceptually specified.

5. You need enough prior knowledge to avoid designing in an evidence vacuum

Responsible design requires some understanding of what is already known and how similar questions have been investigated. The amount of preliminary literature work varies by project, but you should normally know enough to identify established concepts, plausible measures or procedures, major methodological problems, relevant prior findings, and obvious sources of bias or confounding where applicable.

This is not the same as requiring a perfectly finished literature review before any design work begins. Literature review and design often develop iteratively. The problem arises when design choices are made without checking whether the field has already learned something important about how the phenomenon should be studied.

Reporting-guideline resources can also be useful surprisingly early. The EQUATOR Network, for example, organizes guidance for many study types, including randomized trials, observational studies, systematic reviews, diagnostic and prognostic studies, qualitative research, and study protocols. Looking at the relevant guidance while designing a project can expose methodological details that will eventually need to be specified, although a reporting checklist should not be treated as a substitute for methodological reasoning.

6. You need a plausible route to the evidence

A theoretically elegant study is not yet a viable study if the required participants, records, equipment, sites, permissions, expertise, or data cannot realistically be obtained. Before investing heavily in design, establish at least a plausible path to the resources on which the design depends.

This does not always require formal permission at this stage. It may require preliminary confirmation that the population exists in sufficient numbers, that a dataset contains the variables you need, that a collaborating site could potentially support recruitment, or that the required measurement procedure is available.

Where access depends on other organizations or people, distinguish an assumption from an actual indication of feasibility. Later, you may need to verify important feasibility assumptions with the people who control them.

7. You need to recognize major ethical and practical constraints

You do not necessarily need ethics approval before designing the study. Usually, the study must first be designed sufficiently for an ethics committee or institutional review board to evaluate it. Nevertheless, foreseeable ethical issues can shape the design from the beginning.

If the research involves vulnerable populations, sensitive information, deception, identifiable records, intrusive procedures, meaningful risks, or other ethically consequential features, these cannot be treated as administrative details to resolve after the methods have been chosen. Similarly, obvious constraints involving time, budget, researcher expertise, recruitment capacity, or institutional requirements may rule out otherwise attractive designs.

This is where a realistic research plan begins to matter. Feasibility is not something added to a scientifically ideal design afterward. Constraints can legitimately influence which scientifically defensible design is possible.

8. You need to know what you still do not know

One of the strongest signs that you are ready to design is not the absence of uncertainty. It is the ability to identify the remaining uncertainty accurately.

You might know, for example, that the population and outcome are settled but the most appropriate instrument requires evaluation. Or you may know that the conceptual framework is clear while recruitment rates remain uncertain. These are bounded problems that study design, consultation, feasibility work, or piloting may resolve.

By contrast, “we still aren't sure what we are trying to find out” is not merely an unresolved design detail.

The threshold differs by study type

There is no universal pre-design checklist that fits every research tradition. A randomized trial may require early attention to intervention, comparator, outcomes, allocation, sample size, and controls. An observational study may require careful thinking about exposure, outcome, confounding, selection, and available data. Qualitative research may begin with broader or deliberately evolving questions, particularly in traditions where data collection and analysis are iterative.

That variation matters. “Know enough before designing” should not be interpreted as “freeze every concept before research begins.” The appropriate degree of prior specification depends on the logic of inquiry. What remains constant is that your methodological choices should have reasons connected to the question and the kind of evidence you seek.

From an interesting idea to enough clarity for design

The important change is not that every decision has been made. It is that the unresolved decisions can now be evaluated against a reasonably stable question, target, evidentiary goal, and set of constraints. The researcher has crossed from conceptual planning into defensible study design.

Common mistakes when deciding whether you know enough to design

Use a dependency test rather than waiting for complete certainty

When deciding whether to begin study design, ask what your next methodological decisions depend on. If you want to choose a sampling strategy, do you know the population it must represent or illuminate? If you want to choose a measure, do you know the construct it must capture? If you want to choose a comparison, do you know what claim the comparison is supposed to support?

If the answer is yes, design work can often proceed even though the specific solution remains uncertain. If the answer is no, resolve the upstream issue first.

This approach avoids two opposite errors: premature commitment and perpetual planning. You do not need to solve the whole study before designing it, but you should avoid making downstream decisions on top of unstable foundations.

Are you ready to start designing the study?

Questions about moving from planning into study design

Know enough to justify the next decision

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.