What Decisions Should Be Made Before You Start the Research?

What actually needs to be decided before research begins?

You have a research question. Perhaps you also have a tentative method, a population you hope to study, and a deadline somewhere on the horizon. At this point, planning can become surprisingly difficult. How much must you decide before you can legitimately say that the research has started?

The problem is that research decisions do not all have the same consequences. Choosing a working file-naming convention is not equivalent to deciding who will be eligible for the study. Adjusting an internal meeting schedule is not equivalent to changing an outcome after seeing the data. Some choices are inexpensive to revise. Others affect ethics approval, recruitment, measurement, analysis, resources, or the interpretation of the eventual findings.

Good planning therefore does not mean deciding everything as early as possible. It means deciding the right things at the right time, while recognizing which choices can reasonably remain open.

Decide what could materially change the study before committing to it

Think in terms of commitments, not a giant pre-study checklist

A common planning mistake is to treat the beginning of a research project as a single boundary: before it, everything is planning; after it, everything is execution. Actual projects are messier. Literature searching may continue while instruments are being refined. Access negotiations may occur while the protocol develops. A pilot may reveal that an apparently sensible procedure is impractical.

A more useful question is: Which decisions must be sufficiently settled before the next consequential commitment?

That distinction matters because different decisions become consequential at different moments. Recruiting participants commits you to eligibility criteria and recruitment procedures. Collecting measurements commits you to particular operational definitions and instruments. Applying for ethics review may require a sufficiently developed protocol, consent materials, and study instruments. Purchasing equipment or contracting a service may commit part of the budget before data collection has even begun.

First, make sure the question can support a feasible study

A research question can be interesting without being realistically researchable under your circumstances. Before committing substantial time or resources, ask what evidence would actually be needed to answer the research question and whether you can reasonably obtain it.

That requires more than deciding whether the topic is worthwhile. Consider the population, cases, documents, datasets, settings, materials, or other sources of evidence the question requires. Then ask whether they are accessible, whether you have the necessary expertise and resources, and whether the work fits the available time.

This is where an ambitious idea often becomes a workable project. The goal is not to weaken the science until the project becomes convenient. It is to establish whether the proposed evidence can realistically support the question you intend to answer.

Decide the broad study logic before optimizing the details

You should be able to explain, at least provisionally, how the study will move from question to evidence to conclusion. That usually means identifying the broad research approach and design, the unit or object of study, the source of evidence, and the principal concepts or outcomes that need to be examined.

The required specificity varies considerably across methodologies. A randomized trial, an ethnography, a secondary analysis of an existing dataset, and a qualitative interview study do not require identical decisions at identical times. Qualitative designs may intentionally preserve opportunities for iterative sampling or refinement. Experimental studies may require much more to be specified before implementation. Secondary-data studies depend heavily on what variables, documentation, permissions, and data quality are actually available.

You often need the first before you need all of the second. Trying to finalize every procedure while the basic design is still unstable can produce impressive-looking plans for a study that should have been redesigned three meetings ago.

Know who or what will provide the evidence

Before the project advances too far, establish where the data are expected to come from. For human-participant research, this may involve defining the target population, setting, broad eligibility criteria, likely recruitment pathway, and realistic access to participants. For secondary research, it may mean identifying the dataset or records, confirming that the necessary variables exist, and determining whether access can actually be obtained.

For laboratory, computational, archival, documentary, or field research, equivalent questions arise about specimens, equipment, archives, software, databases, sites, or other research materials.

The distinction between available in principle and available to you is crucial. A hospital may have the records you need, but that does not mean you have permission to use them. A national dataset may exist, but the variables required for your analysis may be restricted. A school may seem willing to participate, yet recruitment may depend on several layers of institutional approval.

Identify ethical and regulatory requirements early

If the study involves human participants, identifiable information, sensitive data, animals, biological materials, or other regulated activities, determine what ethical and institutional requirements apply before beginning activities that require approval.

For example, under U.S. Department of Health and Human Services regulations, investigators conducting nonexempt human-subjects research must obtain Institutional Review Board approval before involving human subjects. Requirements vary by jurisdiction, institution, funder, and type of research, so researchers should verify the applicable rules rather than assuming that a study is exempt or that approval can be obtained retrospectively.

Check access before designing around something you may never obtain

Many otherwise reasonable projects depend on a gatekeeper: a school administrator, hospital, company, archive, laboratory, data custodian, community organization, government agency, commercial database, or collaborating institution.

Identify these dependencies before building the entire design around them. Determine what permission is required, who can grant it, what documentation is needed, whether there are fees or contractual restrictions, and how long access might take.

A useful early planning question is not simply, “What do I want to do?” but “What must happen before I am able to do it?” Mapping ethics approval, recruitment, data access, and other dependencies can reveal that the nominal first step of the study is actually preceded by several other processes.

Decide whether the project fits your actual resources

Feasibility includes time, but it is broader than time. Consider funding, personnel, expertise, equipment, software, travel, participant incentives where appropriate, transcription, laboratory costs, secure storage, statistical or methodological support, and access to specialized facilities or services.

Some resources are merely helpful. Others are design constraints. If the proposed analysis requires expertise that nobody on the team has, or the study requires equipment that cannot be obtained within the project period, those are not problems to discover after data collection.

The same applies to workload. A project can be technically possible and still be unrealistic for the people expected to conduct it. Converting the idea into a realistic research plan requires estimating what the proposed design will demand rather than what you hope it will demand.

Establish roles when the project involves more than one person

Collaborative research creates another class of early decisions. Who is responsible for recruitment? Who maintains the master dataset? Who has access to identifiable information? Who performs particular analyses? Who communicates with external partners or oversight bodies? Who decides what happens when the protocol needs to change?

Not every responsibility needs a bureaucratic workflow. The important point is to prevent consequential tasks from existing in a vague collective space where everyone assumes somebody else owns them.

Plan how data will be handled, not only how they will be collected

Researchers naturally focus on obtaining data. The data must also remain understandable, secure, usable, and appropriately documented after collection.

Before collection begins, consider what data will be generated, how files and variables will be organized, where data will be stored, who can access them, how backups will work, what documentation is needed, how sensitive or identifiable information will be protected, and what retention or sharing requirements apply.

Formal requirements depend on the project and its funder. NIH-funded research that falls under its Data Management and Sharing Policy, for example, requires prospective planning for the management and sharing of scientific data. NIH guidance also recognizes that legal, ethical, technical, consent-related, and contractual considerations may constrain sharing. These are useful reminders even when a particular project is not governed by NIH policy: data management decisions are part of research planning, not housekeeping to be invented after the dataset arrives.

Separate decisions that must be fixed from decisions that can remain flexible

Planning does not improve merely because more decisions are frozen. Prematurely fixing uncertain details can make a study unnecessarily rigid.

A practical way to classify early decisions is by asking what happens if the decision changes later.

This is why the question of what can remain flexible at the beginning deserves deliberate attention. Flexibility is not necessarily evidence of poor planning. In some research traditions, carefully bounded flexibility is part of the methodology.

Use data collection as a major decision boundary

Although projects differ, data collection is an especially important threshold because some choices become harder to correct once observations have been made. If you discover halfway through a survey that a central construct was measured inadequately, a better analysis plan cannot create information that was never collected. If eligibility criteria shift during recruitment without methodological justification and appropriate documentation, the resulting sample may become difficult to interpret.

Before crossing that threshold, determine what needs to be fixed before data collection starts. Depending on the study, this can include sampling or recruitment procedures, instruments, operational definitions, consent procedures, data security arrangements, analysis-relevant variables, quality-control procedures, and required approvals.

Build a timeline around dependencies rather than dates alone

A timeline should show more than when you would like tasks to happen. Research activities often form a dependency network. Recruitment may depend on ethics approval. Data collection may depend on site access and instrument preparation. Analysis may depend on data cleaning. Submission may depend on coauthor review.

Breaking the project into manageable stages helps expose those relationships. It also makes it easier to identify the decisions that must precede each stage.

Do not assume that every dependency is under your control. Ethics review, contractual negotiations, recruitment, external data access, procurement, and collaborator feedback can introduce delays. A plausible plan therefore includes not only task duration but also waiting time and uncertainty.

You do not necessarily need a complete formal protocol on day one

A protocol can consolidate the scientific and operational decisions governing a study. WHO's recommended protocol format, for example, includes the rationale, objectives, research design, study population, sampling, instruments, data collection, analysis, ethical considerations, timeline, and budget. NIH likewise provides protocol templates for several forms of human research.

That does not mean every research idea needs a fully developed formal protocol before any planning activity can occur. The appropriate documentation depends on the type, scale, risk, institutional context, and stage of the project. The more consequential point is that important decisions should eventually become explicit and documented rather than remaining only in the researcher's memory.

Whether and when to consolidate those decisions into a formal research protocol is a separate question from whether the decisions themselves need to be made.

From a plausible idea to a study you can responsibly begin

The important feature of this example is not the particular methodology. It is the sequence of commitments. The researcher does not begin by filling every empty cell in a project plan. Decisions are prioritized according to what the next stage requires and what would be costly, unethical, scientifically damaging, or impossible to change later.

Planning mistakes often come from deciding too little or too much

Prioritize decisions by consequence and reversibility

When deciding what deserves attention now, consider two questions: How consequential is this decision, and how difficult will it be to change later?

A decision deserves early attention when getting it wrong could undermine the research question, harm participants, make the evidence uninterpretable, create an inaccessible dependency, consume substantial resources, or require repeating work. A low-consequence choice that can be revised cheaply may reasonably remain provisional.

This approach also helps when several aspects of a project remain uncertain. You do not necessarily have to solve every unresolved issue simultaneously. Identify which unanswered decision blocks other decisions, then work from that dependency. When several choices are interconnected, explicitly mapping decisions that depend on decisions not yet made can prevent circular planning.

Eventually, however, planning must give way to execution. The objective is not perfect foresight. It is enough clarity to proceed without knowingly carrying unresolved problems into stages where they become expensive or impossible to correct. Determining when you have planned enough to start is therefore partly a matter of checking whether the remaining uncertainty is acceptable for the next stage.

Before moving from an idea toward study execution, check these decisions

Questions about how much to decide before starting

Start with the decisions that become costly to change later

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.