How Should Researchers Balance Inclusion With the Need for a Clearly Defined Study Population?

How Broad Should a Study Population Be?

Researchers are increasingly encouraged to include populations historically excluded or underrepresented in research. At the same time, good study design requires boundaries. A research question about a particular condition, developmental stage, intervention, occupation, experience, or setting cannot necessarily be answered by recruiting anyone who is willing to participate.

This creates a genuine design tension.

If eligibility becomes too restrictive, the study may exclude people whose experiences are relevant and produce evidence applicable to only a narrow subset of the intended population. If eligibility becomes too broad, researchers may combine participants for whom the phenomenon, intervention, measurement, or comparison does not mean the same thing.

The objective is therefore neither maximum inclusion nor maximum restriction. It is to define the broadest population that can answer the research question meaningfully and safely without introducing boundaries that the question itself does not require.

Include Relevant Variation Without Losing the Population the Question Is Actually About

Inclusion and Population Definition Are Complementary, Not Opposing, Goals

The apparent conflict often comes from treating “inclusive” as meaning “everyone can participate.” That is not a workable definition for most research.

Every study draws a boundary. The methodological question is whether that boundary is justified by the phenomenon and intended inference or merely inherited from how research is usually conducted.

Begin With the Population Implied by the Research Question

A useful population definition answers: who must be studied for this research question to make sense?

If the question concerns first-year university students' transition experiences, first-year status is part of the population definition. If researchers are evaluating an intervention intended for adults with a specific clinical condition, the relevant diagnosis may be essential. If the phenomenon is workplace experiences among nurses, occupational status may define the population.

These boundaries are not failures of inclusion. They are what make the question coherent.

The problem begins when additional restrictions appear that are not necessary to define the phenomenon. Researchers may then need to ask who is being left out and why.

Separate Essential Characteristics From Convenient Characteristics

Researchers can examine each eligibility requirement by asking what would happen if it were removed.

Would removing the criterion make the research question meaningless? Would it expose participants to unacceptable risk? Would the intervention no longer be appropriate? Would measurement become invalid? Would the resulting participants belong to a fundamentally different population?

If the answer is no, the criterion may deserve closer scrutiny.

A requirement that participants attend weekday sessions, own a smartphone, speak the research team's preferred language, live close to the university, or have no common comorbidities may sometimes be necessary. In other studies, these restrictions arise primarily because they make research easier to administer.

Operational restrictions may still be unavoidable. Recognizing them correctly allows researchers to evaluate their consequences instead of treating them as inherent characteristics of the population.

Safety Can Justify Narrower Eligibility

Some participants face risks that make particular procedures or interventions inappropriate. Exclusion can then be ethically necessary.

But “safety” should identify an actual concern rather than function as a general label for populations perceived as complicated to study.

CIOMS guidance on equitable participant selection states that excluding groups that need special protection should be justified and warns that categorical exclusion can perpetuate evidence gaps and disparities. Its guidance does not require researchers to ignore genuine risk. Instead, it encourages consideration of whether appropriate protections can make participation ethically acceptable.

The distinction is between protecting participants from a relevant risk and excluding a category of people because inclusion requires additional planning.

Internal Validity Does Not Always Require a Homogeneous Population

Researchers sometimes narrow eligibility to reduce variability. A homogeneous sample can simplify analysis and may be appropriate for certain mechanistic, early-stage, or explanatory questions.

Yet participant heterogeneity does not automatically invalidate a study. Modern designs and analytical approaches can often accommodate variation, and some heterogeneity is necessary when researchers want to understand how an intervention or phenomenon operates across the population likely to encounter it.

FDA's current clinical-trial guidance, for example, recommends broader enrollment across demographic and non-demographic characteristics when scientifically and clinically appropriate so that trial populations better reflect patients likely to use an approved drug. That recommendation applies to the regulated clinical-trial context addressed by the guidance, but it illustrates the broader trade-off between tightly controlled eligibility and real-world relevance.

The Appropriate Population Depends on the Stage and Purpose of the Research

The same eligibility boundary need not be optimal for every study of the same topic.

An early mechanistic investigation may deliberately use a narrowly defined population to isolate a particular process. A later effectiveness study may need broader eligibility to understand how an intervention performs under more heterogeneous real-world conditions. A qualitative investigation may purposively seek variation in experiences rather than attempting statistical population representation.

Population breadth should therefore follow the study's objective rather than a universal rule that narrow or broad samples are inherently superior.

Broad Eligibility Does Not Guarantee Inclusive Participation

Researchers can write remarkably inclusive eligibility criteria and still recruit a narrow sample.

Recruitment channels may reach only particular communities. Study sites may be difficult to access. Consent materials may be available in one language. Procedures may require technology, transportation, flexible work schedules, or repeated visits.

This is why study procedures can unintentionally exclude participants even after formal eligibility has been broadened.

Inclusion therefore has at least two layers: who is allowed to participate and who can realistically do so.

Broadening Eligibility Can Introduce New Design Questions

Suppose researchers remove an upper age limit, include participants with common comorbidities, or broaden geographic recruitment. These changes may improve the relevance of the study population, but they can also introduce heterogeneity in baseline risk, intervention exposure, measurement, or follow-up.

That does not mean the broader population is wrong. It means the study may need corresponding design and analytical planning.

Researchers might consider stratification, prespecified subgroup analyses, covariate measurement, larger sample sizes, different recruitment allocations, or other methods appropriate to the question. The solution to heterogeneity is not automatically exclusion.

Do Not Confuse a Clearly Defined Population With a Demographically Homogeneous Population

A study population can be conceptually precise and demographically diverse at the same time.

“Undergraduate students enrolled in introductory programming courses at three universities during the study period” is a defined population. Within that population, participants may vary substantially in age, gender, socioeconomic circumstances, prior programming experience, language, disability, and other characteristics.

Researchers should ask which of those differences matter to the phenomenon rather than assuming that precision requires demographic sameness.

This is where sample diversity can contribute scientific information without changing the basic population the research question defines.

Sometimes the Target Population Itself Needs to Be Narrow

Concerns about inclusion should not pressure researchers into claiming broader populations than their question can support.

A study may genuinely concern one language, one occupation, one disease subtype, one educational stage, or one particular setting. Broadening participation beyond that boundary merely to appear more inclusive can dilute the construct and make interpretation less coherent.

The appropriate response is transparency: explain why the population is narrow and avoid extending the findings automatically beyond it.

Researchers Should Examine the Cumulative Effect of Eligibility Criteria

Individual criteria can look reasonable while collectively producing an unexpectedly narrow population.

An age restriction, language requirement, geographic boundary, technology requirement, exclusion for common comorbidities, and demanding schedule may each have a plausible explanation. Together, they can produce a sample quite unlike the population researchers ultimately hope to inform.

This is why reviewing when inclusion and exclusion criteria create an unrepresentative sample requires looking at the entire eligibility structure rather than defending each criterion separately.

Representation Goals May Require Deliberate Sampling Decisions

Broad eligibility does not ensure that smaller populations will enter the sample in sufficient numbers. If subgroup-specific inference is important, researchers may need targeted recruitment, stratification, or deliberate oversampling.

That does not necessarily mean the raw sample should mirror population proportions. Sometimes oversampling an underrepresented group provides the information needed for an important comparison.

The key is to separate the population definition from the sample allocation. Who belongs in the population and how many people to sample from different parts of that population are related but distinct design decisions.

Population Boundaries Should Constrain the Claims

Researchers sometimes conduct a carefully bounded study and then write conclusions that quietly expand beyond those boundaries.

A study of urban university students becomes a claim about “young adults.” A trial excluding older adults and people with common comorbidities becomes evidence about “patients” generally. A survey conducted only in one language is discussed as though it represents an entire multilingual community.

This is where population definition becomes part of interpretation. A narrow sample is not inherently problematic when the conclusions remain appropriately narrow.

How to Decide Whether a Restriction Defines the Population or Merely Simplifies the Study

The population did not become vague when unnecessary restrictions were removed. It became more closely aligned with the actual research question.

Common Mistakes When Balancing Inclusion and Population Definition

Build Eligibility Outward From the Question, Not Inward From Convenience

A practical approach is to begin with the broad population to which the research question genuinely applies and then add restrictions one at a time. Every additional boundary should have a reason.

The result should be a population that is neither artificially narrow nor conceptually vague. That balance is context-dependent, which is precisely why a universal checklist of “inclusive” eligibility criteria would be methodologically rather unhelpful.

Before Finalizing the Study Population, Test Every Boundary

Questions About Inclusive but Clearly Defined Study Populations

Define the Population the Question Requires, Then Remove Boundaries It Does Not

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.