What If the Population You Can Recruit Differs From Your Planned Study Population?

What If the People Available for Your Study Are Not the Population You Planned to Study?

A study may begin with a clearly defined population and then encounter an uncomfortable practical reality: those are not the people you can actually recruit.

Perhaps the planned population is less accessible than expected. One subgroup rarely responds. A participating institution withdraws. Recruitment succeeds mainly among younger, healthier, more experienced, more motivated, or otherwise different participants. You may discover that the accessible population has characteristics that were not anticipated when the study was designed.

The obvious solution is often to recruit whoever is available and adjust the description later. That can create a mismatch between the question the study claims to answer and the people from whom its evidence actually comes.

The central issue is therefore whether the recruitable population is sufficiently aligned with the population relevant to your research question, and what claims remain justified if it is not.

A Different Recruitable Population May Require a Different Claim

Separate the Population You Want to Understand From the People You Can Actually Study

Target Population, Accessible Population, and Study Sample Are Not the Same Thing

Several populations can sit between a research question and the final dataset. Terminology varies somewhat across disciplines, but the distinction is methodologically useful.

These groups rarely correspond perfectly. That alone is not necessarily a flaw. Most empirical studies investigate a sample rather than every member of the population of interest. The important issue is whether the pathway from the target population to the observed sample permits the intended inference.

First Determine How the Recruitable Population Differs

“Different population” can mean many things. The accessible participants may differ in age, socioeconomic circumstances, geographic location, disease severity, educational background, institutional setting, prior exposure, language, motivation, technology access, or another characteristic relevant to the study.

Some differences may be largely incidental to the research question. Others may be directly related to the exposure, intervention, outcome, measurement process, or mechanism being studied.

Do not therefore ask only whether the populations are statistically different on some characteristics. Ask whether the differences are scientifically consequential for the inference you intend to make.

External Validity Is About Applying Findings to a Specified Population

Generalizability, often discussed under external validity, concerns whether findings can reasonably be applied beyond the particular participants studied. It is not an all-or-nothing property of a study.

STROBE guidance emphasizes that judgments about generalizability depend on the study setting, participant characteristics, exposures, outcomes, recruitment period, and other contextual features. This is why transparent descriptions of setting, eligibility criteria, recruitment, and participants matter.

The question should therefore be specific. Rather than asking, “Is this study generalizable?” ask, “Are these findings reasonably applicable to the population I originally intended to understand, given the differences between that population and the people who were actually studied?”

A Different Sample Is Not Automatically a Biased Sample

Researchers sometimes use “not representative” and “biased” as though they were interchangeable. They are not.

A sample can differ demographically from a target population without every study estimate necessarily being biased. Conversely, a sample that appears similar on several descriptive characteristics can still be affected by selection processes related to the exposure and outcome.

The methodological question is not solved by comparing a few demographic percentages. You need to understand how participants entered the study and whether the factors distinguishing participants from nonparticipants matter for the target inference.

Changing Eligibility Criteria Changes Who the Study Is About

If recruitment difficulties arise because few available participants satisfy the eligibility criteria, researchers may consider broadening those criteria. Sometimes this is defensible. Sometimes it substantially changes the study population.

Before removing an eligibility criterion, return to its rationale. Was it included for participant safety? To define the condition or phenomenon of interest? To reduce a known source of heterogeneity? To ensure that participants could meaningfully receive the intervention or complete the measurement? Or was it included largely by convention without a strong scientific reason?

The consequences of removing it depend on the answer.

If eligibility or recruitment changes are being considered primarily because enrolment is falling behind, first diagnose the underlying participant recruitment problem rather than assuming that the population definition must change.

Changing Recruitment Settings Can Also Change the Population

Suppose a study planned to recruit from public secondary schools but recruitment succeeds only after private schools are added. Or a health study designed around community clinics begins recruiting primarily from a specialist hospital. The eligibility criteria might remain identical, yet the source population can change meaningfully.

Settings influence who becomes available for recruitment. Institutions may serve populations with different resources, severity profiles, referral pathways, practices, exposures, or opportunities. Moving or expanding recruitment therefore deserves the same conceptual question as changing formal eligibility criteria: are the newly accessible participants still appropriate for the research question?

Underrepresentation Does Not Always Require Redesigning the Study

Suppose the target population contains several important subgroups, but one participates at a lower rate. Whether that requires changing the study depends on the research objective and sampling design.

If subgroup representation is essential to estimating a population quantity, the imbalance may be serious. If the research question concerns a relationship expected to operate similarly across groups, the implications may be different, although that similarity should not simply be assumed. In some designs, weighting, stratified recruitment, oversampling, model-based adjustment, or other approaches may be relevant, provided they are methodologically appropriate and supported by the required information.

No statistical adjustment can automatically manufacture evidence about a population that was effectively absent from the study. Analytical methods depend on assumptions and available data, so they should not be treated as substitutes for recruitment.

Changing the Population May Change More Than Generalizability

A population shift is often described only as an external-validity issue. Sometimes it reaches deeper.

If the intervention works differently according to participant characteristics, a changed population can alter the effect being estimated. If measurement behaves differently across groups, the validity of the instrument may change. If the exposure is distributed differently, the study may contain less informative variation. If contextual conditions shape the phenomenon, moving to another setting may change what is being studied.

This is why a major population change can become a design change rather than merely a limitation. At some point, you may need to ask whether changing the research sample turns the project into a different study.

Be Prepared to Narrow the Population Named in Your Conclusions

Sometimes the study itself remains methodologically useful, but the original generalization does not.

Suppose a project intended to study university students nationally but ultimately recruited participants almost entirely from urban institutions with particular characteristics. The appropriate conclusion may be narrower than the original ambition. The findings might still be informative for the observed settings or sufficiently similar populations while providing weaker evidence for students in settings that were poorly represented.

Narrowing the claim is not an admission that the data are useless. It is often the methodologically appropriate response to evidence whose reach is narrower than planned.

Population Changes Should Remain Visible in the Study Record

If the target, source population, eligibility criteria, recruitment settings, or sampling procedures change after the study begins, preserve the original plan and document the modification. Record why it occurred, when it took effect, and how it may affect the resulting sample and interpretation.

STROBE recommends reporting eligibility criteria, sources and methods of participant selection, study setting, relevant participant characteristics, and limitations concerning external validity. These details allow readers to judge how the study sample relates to populations of interest.

If the population change is unplanned, maintain a clear record of how and why the research change occurred rather than rewriting the final methods as though the revised population had always been intended.

A Study Planned for Several Types of Universities Recruits Mainly From One

The researchers should not describe the final sample simply as “university students” if the recruitment process produced a substantially narrower population and the institutional differences are relevant to interpretation. Nor should they assume that the findings are unusable. The defensible scope of the conclusion depends on what was actually studied.

Common Mistakes When the Available Population Differs From the Planned One

Ask Whether the Population Shift Changes the Answer You Can Defend

When the accessible population differs from the planned population, describe the mismatch concretely. Then connect each important difference to the research question. A characteristic matters methodologically because of what it may do to the inference, not simply because two populations have different descriptive profiles.

There is no methodological prize for preserving an ambitious population label after the data no longer support it. A carefully bounded conclusion is usually more informative than a broad claim built on an unexamined leap from the achieved sample.

If the population change is large enough to alter the study's central purpose or inferential logic, assess whether the adaptation remains methodologically defensible before proceeding.

When Your Recruitable Population Differs From the Plan, Check These Points

Questions About Differences Between Planned and Recruited Populations

Study the Population You Actually Have, and Claim No More Than It Can Support

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.