How Should Research Design Change When the Target Population Is Difficult to Identify or Contact?

What Should You Do When You Cannot Identify or Reach the People Your Study Needs?

Suppose you want to investigate access to online higher education among adults who have discontinued formal study and are not registered with any educational institution.

You can recruit students currently enrolled in universities, but the people most relevant to your research question are those outside institutional enrollment systems.

There is no complete list of eligible individuals. Some may have limited internet access, others may distrust recruitment messages, and many may not identify themselves using the terminology in your eligibility criteria.

You consider distributing an online survey through social media and asking respondents to share it with others. This may produce a substantial number of responses, but will those respondents adequately represent the population you intended to study?

When people are difficult to identify or contact, the problem extends beyond recruiting enough participants. The available recruitment channels may systematically determine who enters the study, which experiences are observed, and which conclusions can be defended.

Adapt the Sampling and Recruitment Design Before Increasing Recruitment Effort

How Population Accessibility Changes Sampling, Recruitment, and Inference

What is a hard-to-reach population?

A hard-to-reach population is one whose members are difficult to identify, contact, recruit, or retain through conventional research procedures.

The term describes a research access problem rather than an inherent characteristic of the people concerned.

For example, individuals may be difficult to contact because they are geographically dispersed, absent from institutional records, socially isolated, or reluctant to disclose sensitive experiences.

Others may be accessible in principle but excluded by language, disability-related barriers, limited connectivity, or recruitment procedures that do not accommodate their circumstances.

Researchers should therefore ask what makes the population difficult to reach and whether those barriers arise partly from the study's own design.

Describing participants as hard to reach should not obscure institutional or methodological responsibilities for making research accessible.

How is a hidden population different from a rare population?

A rare population contains relatively few eligible individuals within a defined scope.

A hidden population may be larger but difficult to enumerate because membership is not recorded in a complete accessible sampling frame or because disclosure is sensitive.

For example, a small professional specialty may have a limited but identifiable membership. By contrast, adults engaging in an unregistered activity may be numerous but difficult to identify.

Some populations are both rare and hidden.

The distinction matters because recruiting additional sites may help with a rare population, while network-based or community-informed recruitment may be more relevant when the principal limitation is identification.

Researchers should avoid treating all recruitment difficulties as evidence that the underlying population is small.

Why does the sampling frame matter?

A sampling frame is the operational list or mechanism from which research units can be selected.

For example, an institutional enrollment database may provide a sampling frame for currently registered students.

However, it cannot adequately represent adults who are not enrolled anywhere if those individuals are absent from the database.

This is a coverage problem.

Even a perfectly randomized sample from an incomplete frame may systematically exclude important parts of the intended population.

Researchers should therefore distinguish the target population from the frame population and the achieved sample.

The quality of a sampling design depends partly on whether the frame corresponds to the population about which conclusions are intended.

What are the main reasons people may be difficult to contact?

Recruitment barriers may arise from several different processes, and each has different methodological implications.

Researchers should identify the dominant barriers before selecting recruitment channels. A method that improves access to one subgroup may leave other subgroups systematically excluded.

Why can recruitment difficulty create selection bias?

Selection bias may arise when inclusion in the study is related to the variables or relationships being investigated.

For example, a study of barriers to online education that recruits exclusively through social media may disproportionately include people with reliable connectivity and digital confidence.

Those most affected by digital exclusion may be less likely to encounter the invitation.

Consequently, the observed prevalence of particular barriers may differ from that in the target population.

Selection can also distort associations when participation depends on combinations of exposures, outcomes, or related characteristics.

Increasing the number of respondents does not automatically eliminate this bias.

A large selectively recruited sample may remain less representative than a smaller sample obtained through a better-defined sampling process.

Should researchers broaden their eligibility criteria?

Eligibility criteria should be reviewed when they unnecessarily exclude individuals relevant to the research question.

For example, requiring a government-issued identification document may exclude otherwise eligible participants who cannot readily provide one.

Similarly, requiring participants to complete a lengthy online screening process may discourage individuals with limited connectivity or accessibility needs.

However, broadening eligibility can also change the target population and the estimand.

Researchers should distinguish restrictions necessary for scientific validity or participant safety from administrative requirements adopted for convenience.

Changes should be justified and documented, particularly when they occur after recruitment has begun.

See balancing eligibility rigor with inclusion for the related methodological decision.

Can snowball sampling help identify participants?

Snowball sampling recruits participants through referrals from individuals already identified or enrolled.

For example, a researcher may ask an eligible participant to share study information with others who might meet the criteria.

This can improve access when population membership is not publicly recorded.

However, referral networks may overrepresent socially connected individuals and underrepresent people who are isolated or outside the initial participants' networks.

The resulting sample ordinarily does not support conventional probability-based population estimates without additional assumptions or design features.

Researchers should therefore avoid claiming representativeness solely because participants were recruited through several referral chains.

The distinction is developed in snowball sampling versus respondent-driven sampling.

When might respondent-driven sampling be appropriate?

Respondent-driven sampling, or RDS, is a structured network-based recruitment approach developed for populations lacking conventional sampling frames.

It typically begins with selected initial participants, often called seeds, who recruit eligible peers through controlled referral procedures.

Some RDS estimators use information about participants' network connections and recruitment patterns to estimate population characteristics.

However, these estimates depend on assumptions about the underlying social network, recruitment process, and participants' reported network sizes.

Violations of those assumptions can produce biased estimates.

RDS should not be treated as automatically representative simply because it uses recruitment coupons or multiple referral waves.

Researchers should evaluate when respondent-driven sampling is appropriate before adopting the approach.

Can community organizations improve recruitment?

Community organizations may help researchers understand population characteristics, identify acceptable recruitment procedures, and establish trust.

For example, organizations supporting adult learners may help researchers communicate with people who are not connected to universities.

However, organizations may reach only particular segments of the population.

Members who use formal services may differ from those who do not.

Researchers should therefore assess how organizational membership or service use relates to the study variables.

Partnerships should also preserve voluntary participation and prevent organizational representatives from pressuring individuals to enroll.

These concerns are examined in community recruitment and selection bias.

What role do gatekeepers play?

Gatekeepers are individuals or organizations that influence access to potential participants.

Examples include institutional administrators, community leaders, service providers, and organizational coordinators.

Gatekeepers may facilitate recruitment by distributing invitations or helping researchers understand appropriate contact procedures.

However, they may also filter invitations, select preferred participants, or discourage individuals whose experiences could reflect negatively on the organization.

Researchers should document the gatekeeper's role and consider whether the process systematically limits access.

Participants should have an appropriate opportunity to decide independently whether to participate.

See managing gatekeepers during hard-to-reach recruitment.

Can digital recruitment solve the problem?

Digital recruitment can reach individuals across geographic boundaries and may be useful when participants are not connected to a single institution.

For example, online communities, relevant social platforms, and digital advertisements may help researchers identify adults interested in returning to education.

However, digital recruitment is shaped by platform access, algorithms, language, digital literacy, and participants' willingness to disclose information online.

These mechanisms may produce systematic coverage differences.

Researchers should also consider duplicate participation, fraudulent responses, privacy, and the suitability of eligibility verification.

Online recruitment should be selected because it reaches the intended population appropriately, not merely because it can produce responses quickly.

See digital recruitment for hidden populations for further considerations.

Should multiple recruitment channels be combined?

Combining recruitment channels may improve coverage when different population segments are accessible through different routes.

For example, researchers might recruit through community organizations, online platforms, and local educational support services.

However, multiple channels can create overlapping recruitment frames and different probabilities of inclusion.

Participants may encounter the invitation through several channels, creating risks of duplicate enrollment.

Researchers should document recruitment sources and establish procedures for identifying overlapping invitations without unnecessarily compromising privacy.

Using several channels may improve diversity of access, but it does not automatically convert a nonprobability sample into a representative probability sample.

How should eligibility be verified?

Eligibility verification should be proportionate to the scientific requirements and risks of the study.

For example, researchers may need to confirm that participants belong to a particular age group or have experienced a specified educational circumstance.

However, demanding extensive documentation may exclude eligible individuals who lack formal records or are reluctant to disclose identifying information.

Researchers should consider whether self-report, appropriate screening questions, or alternative verification procedures can satisfy the study's needs.

Verification procedures should also protect confidentiality and avoid collecting unnecessary sensitive information.

The relevant trade-off is examined in eligibility verification when participants lack documentation.

How can researchers protect anonymity while preventing duplicate participation?

Anonymous recruitment can be important when participation could expose sensitive information.

However, online or network-based studies may face duplicate submissions, automated responses, or repeated enrollment.

Researchers should use proportionate safeguards consistent with the study's privacy commitments.

For example, eligibility checks, controlled invitation procedures, or carefully designed response-quality assessments may reduce duplicate participation.

However, collecting extensive identifying information solely for verification may discourage participation or create unnecessary privacy risks.

Researchers should distinguish anonymity from confidentiality and avoid promising complete anonymity when the actual procedures permit identification.

See protecting anonymity while preventing duplicate participation.

How do incentives affect recruitment?

Incentives may compensate participants for their time, transportation, connectivity, or other participation costs.

They may be particularly relevant when participation requires substantial effort.

However, incentives can affect who participates and may create undue influence in particular circumstances.

For example, a payment that is modest for one population may be substantial for individuals experiencing financial hardship.

Researchers should consider proportionality, local ethical guidance, and whether incentives are linked to completing every study activity.

Compensation should not be structured to discourage participants from exercising their right to withdraw.

Recruitment incentives may improve participation but cannot guarantee an unbiased sample.

Can qualitative research be more appropriate?

Qualitative research may be appropriate when the objective concerns experiences, meanings, social processes, or barriers that are not adequately understood.

For example, interviews may help explain why adults who discontinued higher education do not engage with conventional university recruitment systems.

Purposive sampling may then prioritize variation in relevant experiences rather than statistical representativeness.

However, qualitative research is not a universal fallback when probability sampling becomes difficult.

Its sampling strategy, data adequacy, analysis, and claims must be justified according to the qualitative research question.

Researchers should not use qualitative findings to estimate population prevalence without an appropriate inferential basis.

Can quantitative research still be conducted?

Yes. Quantitative research may remain feasible even when a complete sampling frame is unavailable.

Researchers may describe the achieved sample, examine measured relationships, or use specialized sampling methods under explicit assumptions.

However, population prevalence estimates and causal claims require particular care.

For example, the proportion of respondents reporting limited connectivity does not necessarily estimate the proportion among all eligible adults when recruitment is selective.

Weighting or statistical adjustment may reduce certain observed differences under suitable assumptions, but cannot automatically correct unknown coverage and selection mechanisms.

The intended estimand should determine whether the available sampling process can support the proposed quantitative inference.

What if population size is unknown?

When no complete sampling frame exists, researchers may not know how many eligible individuals are present in the target population.

This complicates recruitment planning, response-rate calculation, and some forms of population inference.

Researchers should not invent a population size or assume that the number of people reached through recruitment equals the total eligible population.

Specialized population estimation methods may be available in selected contexts, but they require additional data and assumptions.

For example, capture-recapture approaches may be considered when suitable overlapping sources exist, although dependence among sources and heterogeneous inclusion probabilities can create substantial bias.

Researchers should distinguish an unknown population size from a population known to be small.

How should sample size be justified?

Sample size justification should follow the research objective, intended inference, and realistic recruitment constraints.

For probability-based estimation, the desired precision and sampling design are central.

For hypothesis testing, effect size assumptions and statistical power may be relevant.

For qualitative research, sampling adequacy depends on the research question, case variation, and depth of evidence rather than conventional statistical power calculations.

When the achieved sample is determined largely by access constraints, researchers should explain those constraints transparently.

However, feasibility alone does not establish that the sample supports the original inferential objective.

If recruitment cannot provide the required information, the research question or design may need revision.

What should researchers do when recruitment remains unsuccessful?

Persistent recruitment difficulties should prompt a structured assessment rather than indefinite extension of the recruitment period.

Researchers should determine whether the problem concerns an inaccurate population definition, an incomplete frame, ineffective communication, burdensome participation requirements, or genuine population scarcity.

They should then evaluate whether feasible modifications would preserve the original estimand.

For example, adding community recruitment channels may improve coverage without changing the population definition.

By contrast, broadening eligibility to include substantially different individuals may change the research question.

The distinction is explored in when recruitment difficulties should lead to research design changes.

How should researchers assess generalizability?

Generalizability depends on the relationship between the achieved sample and the population to which conclusions are intended to apply.

For example, respondents recruited through online learner communities may differ from adults who are digitally disconnected or unaware of those communities.

Researchers should identify which population segments were likely to be included or excluded.

They should also describe the recruitment channels, eligibility procedures, participation rates where meaningful denominators exist, and relevant sample characteristics.

Claims about population prevalence should be limited when inclusion probabilities are unknown or selection mechanisms cannot be adequately addressed.

Transparent limitations are preferable to describing a convenience sample as representative without evidence.

What ethical considerations are particularly important?

Research involving difficult-to-contact populations may raise concerns about privacy, confidentiality, coercion, stigmatization, and participant safety.

For example, identifying individuals through sensitive networks may inadvertently disclose information about their circumstances.

Recruitment through employers, institutions, or community leaders may create perceived pressure to participate.

Researchers should establish appropriate consent procedures, minimize unnecessary identifying data, and protect participants from foreseeable harms.

Recruitment materials should be accessible and culturally appropriate without relying on stereotypes about the population.

Ethical review should consider both the risks of participation and the consequences of systematically excluding people whose experiences are important to the research question.

How should the final research design be documented?

The protocol should distinguish the target population, accessible population, recruitment frame, and achieved sample.

Researchers should describe how eligibility was established, which recruitment channels were used, and how participants entered the study.

Potential selection mechanisms should be identified before analysis where feasible.

If sampling probabilities are unknown, researchers should avoid presenting estimates as though they came from a conventional probability sample.

Changes to recruitment or eligibility procedures should be documented, particularly when introduced after recruitment has begun.

The final report should explain how accessibility constraints affect the conclusions and what additional evidence would be needed for broader inference.

Redesigning Recruitment for Adults Outside Formal Higher Education

The hypothetical example illustrates how recruitment design can influence the validity of the research question itself.

Improving access may produce a more diverse and informative sample, but the researchers must still evaluate which population-level conclusions the achieved sampling process can support.

Common Misconceptions About Hard-to-Reach Population Research

Design Around the Actual Access Mechanism, Not an Imaginary Complete Sampling Frame

Begin by identifying why the intended population is difficult to reach and which groups the proposed recruitment process may exclude.

Then choose recruitment and sampling methods according to the research objective, feasible access routes, and the inferential claims the study must support.

The most appropriate design may combine several recruitment approaches, but each should contribute to a clearly defined sampling and inferential strategy.

Access difficulties should be treated as methodological information that shapes the study, not merely an administrative obstacle to overcome.

Before Recruiting a Hard-to-Reach Population

Frequently Asked Questions About Research with Hard-to-Reach Populations

Recruitment Accessibility Is Part of Research Design Validity

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.