Sampling, Access and Recruitment

Introduction

Qualitative health research depends not only on the quality of interview questions or analytic procedures, but also on whose experiences become available for interpretation. Sampling determines which perspectives enter a study; access determines whether the researcher can reach relevant settings and groups; recruitment determines who ultimately agrees to participate. These are therefore methodological, ethical and managerial activities rather than merely logistical tasks.

In quantitative research, sampling is commonly evaluated through representativeness and statistical generalisability. Qualitative research usually pursues a different objective: developing a rich, credible and theoretically useful understanding of meanings, practices, relationships and processes. A sample is consequently judged by its relevance, depth, diversity, adequacy and contribution to the research question. This does not mean that qualitative sampling is casual or unplanned. On the contrary, it requires explicit reasoning, reflexivity and continuous adjustment.

In health-care management, sampling decisions may involve patients, carers, clinicians, managers, administrators, policy actors and community representatives. Each group may understand the same service differently. A study of delayed discharge, for example, may produce misleading conclusions if it includes only hospital managers and excludes patients, families, ward staff and community providers.

Learning objectives

By the end of this chapter, the learner should be able to:

1. Explain the purpose and logic of qualitative sampling.
2. Distinguish purposive, criterion, maximum-variation, theoretical, snowball and convenience sampling.
3. Construct a defensible sampling strategy aligned with a research question and methodology.
4. Negotiate institutional and community access ethically.
5. Design inclusive recruitment procedures and manage gatekeeper influence.
6. Judge sample adequacy using saturation, information power and theoretical sufficiency.
7. Document sampling decisions, refusals, adaptations and limitations.

1. The logic of qualitative sampling

Qualitative sampling is generally non-probability sampling. Participants are selected because they can illuminate the phenomenon under investigation, not because each member of a population has a known and equal probability of selection. The central question is not “How many people represent the population?” but “Which people, experiences or cases can provide the knowledge required to answer this question convincingly?”

A sampling strategy should be aligned with four elements:

• The research question: Exploratory questions may require broad variation, while process questions may require participants with experience at particular stages.
• The methodological tradition: Ethnography may prioritise settings and social practices; phenomenology may seek people who have directly experienced a phenomenon; grounded theory uses theoretical sampling as analysis develops.
• The unit of analysis: The unit may be an individual, family, team, organisation, encounter, document or service pathway.
• The intended claim: A study examining one organisation cannot automatically claim to describe all organisations, but it may offer transferable conceptual insights.

Sampling occurs at several levels. Researchers may sample sites, departments, events, records, professional groups and individuals. “The sample” is therefore often a sample of cases within cases. A study of emergency-department flow might purposively select two hospitals, then select shifts, staff categories, patients and relevant observations within each hospital.

Population, sampling frame and eligibility

A target population is the broader group to which the research question relates. An accessible population is the portion realistically reachable by the research team. A sampling frame is the practical list or structure from which cases can be identified, such as a clinic register, professional association, community organisation or service-user database. Qualitative studies may not possess a complete sampling frame, especially when researching undocumented migrants, informal carers or stigmatised experiences.

Eligibility criteria should be sufficiently specific to identify relevant participants without unnecessarily excluding important perspectives. They commonly include:

• inclusion criteria, such as age, diagnosis, role, service use or experience;
• exclusion criteria, such as inability to provide informed consent or lack of direct experience;
• contextual criteria, such as having used a service during a defined period; and
• safety criteria, including whether participation could cause unacceptable distress or risk.

Criteria should be justified rather than treated as administrative details. For example, excluding people with limited proficiency in the dominant language may improve procedural simplicity but systematically remove important experiences. Translation, interpreters or multilingual recruitment may be preferable.

2. Major qualitative sampling strategies

Sampling strategies are not mutually exclusive. A study may begin with criterion purposive sampling, seek maximum variation, and later use theoretical sampling. The following categories describe dominant logics rather than rigid recipes.

Purposive sampling

Purposive sampling involves deliberately selecting information-rich cases relevant to the research purpose. Researchers use prior knowledge of the setting, literature, professional advice or preliminary findings to identify participants likely to provide substantive insight.

For example, a study of clinical incident reporting might purposively include nurses who have reported incidents, managers responsible for review, and staff who chose not to report them. The purpose is to obtain contrasting understandings of reporting culture.

Strengths include relevance, flexibility and efficient use of resources. Limitations include researcher judgement, possible selection bias and the risk of recruiting only articulate, visible or institutionally approved participants. Researchers should explain how cases were identified and why their perspectives were considered information-rich.

Criterion sampling

Criterion sampling selects all or a defined subset of cases meeting predetermined criteria. It is useful when direct experience is essential. A study of medication self-management might recruit adults who have used a prescribed medicine for at least six months and have experienced at least one medication-related difficulty.

Criterion sampling enhances conceptual coherence, but strict criteria can narrow the sample prematurely. Researchers should ask whether the criterion captures the relevant phenomenon or merely reflects administrative convenience.

Maximum-variation sampling

Maximum-variation sampling deliberately includes participants with substantially different characteristics or experiences. Variation may involve age, gender, ethnicity, socioeconomic position, geography, service setting, professional role, illness severity or outcome.

Its purpose is not to create statistical representativeness. Rather, it allows the researcher to identify common patterns across difference and examine how context shapes experience. A study of telehealth access might vary participants by rurality, age, digital confidence, disability and language.

Maximum variation requires careful interpretation. Differences observed between groups may arise from several intersecting conditions, and a small qualitative sample cannot establish the prevalence of those differences. Researchers should therefore describe patterns and mechanisms cautiously.

Theoretical sampling

Theoretical sampling is associated particularly with grounded theory. Participants, events or documents are selected sequentially because emerging analysis indicates that they can develop, challenge or refine a category. Sampling decisions are driven by the evolving explanation rather than fixed demographic quotas.

Suppose early interviews suggest that staff bypass a digital referral system when they distrust its triage rules. The researcher may then seek participants from departments with different levels of digital autonomy, or managers who redesigned the system, to examine the emerging category of “managed workarounds.”

Theoretical sampling requires concurrent data collection and analysis. It is not simply adding participants because the sample feels small. The researcher should state what analytic question prompted each new recruitment decision and how the resulting data altered the developing explanation.

Snowball or chain-referral sampling

Snowball sampling begins with initial participants who suggest or contact other potentially eligible participants. It can be useful for hidden, dispersed or trust-dependent populations, such as informal carers, people with stigmatised conditions or workers in insecure employment.

Its principal risk is network homogeneity. Participants tend to refer people who resemble themselves, belong to the same organisation or share a social network. To reduce this risk, researchers can begin with multiple entry points, ask participants to suggest people with different experiences, and combine snowballing with maximum-variation criteria. Participants should never be required to reveal another person’s identity without that person’s permission.

Convenience sampling

Convenience sampling recruits those who are easiest to access, such as people attending a clinic during the researcher’s visit or staff responding to an internal email. It may be appropriate in early exploratory work, feasibility studies or time-limited service evaluation, but it is methodologically weaker when used without acknowledgement.

Convenience samples may overrepresent highly engaged, available or satisfied participants and underrepresent people who work irregular hours, have caring responsibilities, distrust institutions or have negative experiences. Convenience should therefore be treated as a constraint to manage, not a justification that needs no explanation.

Sampling strategies compared

3. Designing a sampling plan

A defensible sampling plan normally specifies the phenomenon, setting, participant categories, eligibility criteria, recruitment route, anticipated variation and stopping logic. It should be sufficiently structured for ethical review while remaining adaptable to learning during fieldwork.

A useful planning sequence is:

1. Define what must be understood and whose experience is relevant.
2. Identify the unit or units of analysis.
3. Map stakeholders and potentially missing perspectives.
4. Select a primary sampling logic and any complementary strategies.
5. Establish inclusion, exclusion and safeguarding criteria.
6. Anticipate variation relevant to the question.
7. Identify access routes and gatekeepers.
8. Pilot recruitment materials and procedures.
9. Monitor who is and is not participating.
10. Revise recruitment transparently when the analysis justifies it.

A sampling matrix can make assumptions visible.

Matrices should not become quota systems unless that is genuinely appropriate. They are planning tools that help identify absences and prevent recruitment from being governed entirely by the easiest route.

Sample size is a design judgement

There is no universal correct number of qualitative participants. Sample size depends on the specificity of the sample, the complexity of the phenomenon, the quality and depth of data, the analytic approach, the number of comparisons and the researcher’s epistemological position.

A narrowly focused study of a relatively homogeneous group may reach adequacy with fewer participants than a multi-site study involving several professional and social positions. However, “small sample” does not automatically mean “high quality,” and a large sample cannot compensate for weak data or poorly aligned recruitment.

Researchers should distinguish planned sample size from achieved sample size, report recruitment and refusal patterns, and explain why data collection stopped. Statements such as “interviews continued until saturation” are insufficient unless the researcher explains what saturation meant and how it was assessed.

4. Saturation, information power and theoretical sufficiency

Saturation

Saturation refers broadly to a point at which additional data no longer make a meaningful contribution to the analytic task. Several forms should be distinguished:

• Code saturation: no substantially new codes are appearing.
• Meaning saturation: existing codes are sufficiently elaborated, nuanced and understood.
• Theoretical saturation: relationships among categories are sufficiently developed for the intended theoretical account.
• Sampling saturation: further participants are unlikely to add relevant variation for the research question.

Code saturation can occur before meaning saturation. For instance, repeated statements that patients experience “poor communication” may establish a code, while further interviews are still needed to understand whether the problem concerns timing, language, authority, emotional support or conflicting professional messages.

Saturation is not a magical numerical threshold. It is an analytic judgement that must be documented through coding records, comparison of later interviews, case summaries and consideration of disconfirming evidence. Researchers should also acknowledge that absolute saturation is difficult to prove; a more defensible claim is that sufficient saturation was achieved for a specified analytic purpose.

Information power

The concept of information power proposes that the more relevant information a sample holds for the study, the fewer participants may be needed. Information power is increased by:

• a narrow and focused study aim;
• a specific and well-defined sample;
• strong use of established theory;
• high-quality, detailed dialogue;
• an analysis strategy involving case depth rather than broad comparison.

Information power is reduced by broad aims, heterogeneous samples, limited interview quality, weak theoretical grounding and complex multi-group comparisons. It is a useful planning heuristic, but it should not be used as a numerical formula.

Theoretical sufficiency

Theoretical sufficiency means that the data are adequate to support the level of conceptual explanation claimed. A study may be theoretically sufficient for describing how patients interpret waiting-room communication but insufficient for explaining the entire organisational system that produces waiting times.

A practical stopping decision can ask:

1. Are the central categories sufficiently developed?
2. Are important variations and negative cases represented?
3. Can the researcher explain relationships among categories with evidence?
4. Would new data likely alter the explanation, or merely repeat it?
5. Are claims appropriately limited to the sample, setting and analytic purpose?

5. Negotiating field access

Access is a social process involving authority, trust, institutional priorities and perceived risk. Formal permission does not guarantee meaningful access, and a gatekeeper’s approval does not substitute for individual consent.

Types of access

• Physical access: entering a clinic, ward, community site or workplace.
• Organisational access: permission to approach staff, use records or observe operations.
• Relational access: trust from people who can explain local norms and introduce the researcher.
• Discursive access: permission to ask about sensitive or contested matters.
• Analytic access: sufficient openness for participants to provide detailed accounts rather than official scripts.

The access negotiation process

An access proposal should explain the study purpose, methods, participant burden, confidentiality arrangements, risks, benefits, data management and anticipated outputs. It should also clarify what the organisation will not control. In particular, organisations should not be promised the power to remove findings merely because they are critical, although factual errors and confidentiality risks should be addressed.

Researchers should identify both formal and informal gatekeepers. A hospital research office, ward manager, community leader or professional association may facilitate contact, but each can also influence who feels able to participate. Access negotiations should therefore include strategies for independent recruitment, private contact and confidential consent where feasible.

Access is often iterative. A researcher may begin with one department, learn that night-shift workers are absent, and negotiate additional visits or remote interviews. Such adaptations should be recorded in a fieldwork log and reported as part of methodological reflexivity.

6. Recruitment and informed participation

Recruitment translates the sampling plan into actual participation. Effective recruitment is clear, respectful, inclusive and non-coercive.

A recruitment pack commonly includes an invitation, participant information sheet, consent form, eligibility questions, interview or observation details, confidentiality explanation and researcher contact information. Materials should use accessible language and, where necessary, translated or alternative formats.

Recruitment messages should state:

• why the person is being invited;
• what participation involves and how long it will take;
• whether participation is voluntary;
• the right to decline or withdraw within the stated limits;
• how privacy will be protected;
• whether compensation is offered and on what basis; and
• whom to contact with questions or complaints.

Avoiding coercion and undue influence

Power relationships are particularly important in health-care settings. Employees may fear managerial consequences, patients may believe participation affects treatment, and students may perceive academic pressure. Recruitment should be separated from decisions about employment, grades or clinical care wherever possible. A treating clinician should not normally obtain consent from a patient when an independent recruiter can do so.

Compensation should reimburse time and expenses without becoming so large that it compromises voluntariness. Researchers should explain that compensation is provided for participation, not for giving a particular answer.

Recruiting diverse participants

Diversity requires more than inviting everyone through one channel. Researchers should review recruitment by role, age, gender, ethnicity, disability, language, socioeconomic position, geography and other dimensions relevant to the study. Practical measures may include evening appointments, travel reimbursement, remote options, interpreters, accessible venues, childcare support and community partnerships.

Digital recruitment can expand reach through patient forums, professional networks and secure online platforms, but it may exclude people with limited connectivity or digital confidence. Researchers should not assume that online visibility equals population relevance.

7. Ethics, governance and confidentiality

Sampling and recruitment require approval through applicable research ethics and health-care governance systems. Important ethical issues include privacy, vulnerability, gatekeeper influence, re-identification, distress and incidental disclosures of risk.

Recruitment lists should be stored securely and accessed only by authorised personnel. Consent records should be separated from interview data where possible. In small professional communities, removing names may not prevent identification; role, location, event and quotation combinations can reveal a participant. Researchers should discuss these limits honestly.

Participants may disclose abuse, unsafe practice, suicidal thoughts or serious clinical risk. The information sheet should explain confidentiality boundaries and any safeguarding procedures before consent. Researchers need a proportionate response pathway and should not promise absolute secrecy.

In community-based research, collective interests may matter alongside individual consent. Community representatives can advise on culturally appropriate recruitment, but individual autonomy remains essential. Researchers should avoid treating communities as homogeneous or allowing a leader to speak for every member.

8. Reflexivity and documentation

Researchers are instruments in qualitative inquiry. Their professional identity, institutional affiliation, social position and assumptions can affect access, recruitment and what participants disclose. A senior clinician interviewing junior staff may receive deferential accounts; an external researcher may be viewed as safer but may lack contextual understanding.

Reflexive documentation should include:

• how the researcher was introduced;
• who controlled access;
• recruitment channels and refusals;
• changes to eligibility or sampling;
• participants who were difficult to reach;
• relationships and power dynamics;
• reasons for stopping recruitment; and
• how missing or contradictory perspectives affected interpretation.

An audit trail strengthens credibility. It demonstrates that sampling decisions were reasoned and allows readers to assess transferability.

9. Applied case: recruiting for a study of hospital discharge delays

The following is an illustrative case.

Background

A regional hospital reports increasing delays in discharging older adults. Managers initially attribute the problem to inadequate community capacity. A qualitative study is commissioned to understand how discharge decisions are made and experienced.

Problem

An initial proposal plans to interview ten ward managers. This would provide an organisational perspective but would omit patients, families, nurses, physicians, social workers and community providers. It would also risk reproducing the hospital’s preferred explanation.

Analysis and sampling approach

The research team defines the unit of analysis as the discharge pathway, including decisions and interactions from the identification of discharge readiness to post-discharge follow-up. It adopts criterion purposive sampling: participants must have been directly involved in at least one discharge within the previous three months. Maximum variation is then used across ward type, professional role, patient complexity and discharge outcome.

The team proposes:

• older patients with both timely and delayed discharges;
• family carers, including those who felt prepared and unprepared;
• nurses, physicians, therapists and discharge coordinators;
• ward and hospital managers; and
• community nurses and social-care practitioners.

Because some carers are known through existing participants, snowball recruitment is added, but the team begins with several independent routes: patient liaison services, community organisations and ward-based invitations. The hospital permits access but insists that managers distribute all invitations. The team negotiates an alternative: managers may circulate a general notice, while interested staff contact the research team directly. This reduces perceived pressure.

Recruitment and emerging analysis

Early interviews suggest that “delay” has different meanings. Managers define it as a medically unnecessary bed day; families describe uncertainty and lack of preparation; community staff describe incomplete information and unrealistic expectations. The team uses theoretical sampling to recruit a ward with rapid discharge performance and a community team serving a rural area. These cases test whether communication and transport, rather than bed capacity alone, explain variation.

Analysis reaches code saturation for common barriers, but meaning saturation requires further interviews with families and community staff. The team stops when additional interviews repeat established mechanisms and no longer alter the developing explanation of discharge as a negotiated process rather than a single hospital decision.

Outcome

The study finds that delays arise through interaction among clinical uncertainty, incomplete medication information, family readiness, transport constraints and unclear accountability across organisational boundaries. The findings lead to a revised discharge checklist, earlier family conversations and a named cross-organisational coordinator.

Learning points

1. Sampling from one stakeholder group would have produced a partial explanation.
2. Maximum variation revealed how “delay” changed meaning across roles and settings.
3. Theoretical sampling enabled the team to test an emerging explanation.
4. Independent contact reduced gatekeeper influence.
5. Stopping was justified through analytic development rather than a fixed number alone.

10. Common challenges and best practices

Challenge: overreliance on gatekeepers

Gatekeepers may recruit cooperative or institutionally favourable participants. Use multiple access routes, protect private contact and monitor who is absent.

Challenge: under-recruitment

Build realistic timelines, offer flexible modes, maintain respectful reminders and avoid overpromising benefits. Examine whether eligibility criteria or procedures create unnecessary burden.

Challenge: sample imbalance

Maintain a recruitment monitoring table and compare achieved participation with the sampling matrix. Target underrepresented perspectives ethically rather than filling numerical quotas without analytic justification.

Challenge: participant distress

Use trauma-informed procedures, allow breaks, provide referral information and train interviewers to respond appropriately. Do not pressure participants to disclose more than they wish.

Challenge: recruitment through digital platforms

Verify eligibility without collecting unnecessary personal data, use secure systems, explain data practices and provide non-digital alternatives.

Best-practice principles

• Align sampling logic with the research question and methodology.
• Separate access permission from individual consent.
• Treat diversity as analytically relevant, not merely demographic decoration.
• Record refusals, exclusions and adaptations.
• Use saturation claims precisely and transparently.
• Search for negative cases and disconfirming accounts.
• Protect participants from organisational or clinical power.
• Limit conclusions to what the sample and data can support.

Chapter Summary

Qualitative sampling is a purposeful process of selecting people, settings, events or documents that can illuminate a research question. Purposive and criterion sampling establish relevance; maximum-variation sampling examines commonality and difference; theoretical sampling develops emerging concepts; snowball sampling can reach hidden networks; and convenience sampling offers access but carries substantial risk of bias. These strategies may be combined when their purposes are explicit.

Access is negotiated through formal permission, relationships and trust. Gatekeepers can facilitate entry but may also shape who participates, so independent and inclusive recruitment routes are important. Recruitment must be voluntary, accessible and sensitive to power, vulnerability and confidentiality.

Sample adequacy is a reasoned judgement. Saturation may concern codes, meanings, theoretical relationships or variation. Information power depends on the focus of the aim, specificity of the sample, quality of dialogue and strength of theory. Theoretical sufficiency concerns whether the data support the level of explanation claimed. Strong studies document sampling decisions, missing perspectives, refusals, adaptations and stopping rationales, thereby making their findings more credible and transferable.

Key Terms

Accessible population: The portion of the target population that can realistically be reached.

Criterion sampling: Selecting cases that meet predefined eligibility conditions.

Convenience sampling: Recruiting participants who are readily available.

Gatekeeper: A person or organisation that controls or facilitates access to a setting or group.

Information power: The amount of relevant knowledge a sample contributes to the study aim.

Maximum-variation sampling: Deliberately selecting contrasting cases or participant characteristics.

Negative case: Evidence that challenges or does not fit an emerging interpretation.

Purposive sampling: Deliberate selection of information-rich cases.

Recruitment: The process of inviting, informing and enrolling participants.

Sampling frame: A practical list or structure from which cases may be identified.

Sampling saturation: The point at which additional cases are unlikely to add relevant variation.

Snowball sampling: Recruiting through referrals from existing participants or networks.

Target population: The broader group to which the research question relates.

Theoretical sampling: Selecting further cases to develop or test emerging concepts.

Theoretical sufficiency: Adequacy of data for the conceptual explanation being claimed.

Code saturation: The point at which substantially new codes cease to emerge.

Meaning saturation: The point at which existing codes are sufficiently elaborated.

Reflexivity: Critical examination of how the researcher influences the study.

Review Questions

Short Answer Questions

1. What is the main distinction between probability sampling and purposive qualitative sampling?
2. Define criterion sampling and give one health-research example.
3. Why can gatekeepers create a recruitment bias?
4. Distinguish code saturation from meaning saturation.
5. Identify two ethical risks associated with recruiting employees through their managers.

Long Answer Questions

1. Compare purposive, maximum-variation and theoretical sampling, including the type of research problem for which each is most appropriate.
2. Explain how a researcher should negotiate access to a hospital while preserving voluntary individual consent.
3. Discuss the concept of information power and its implications for qualitative sample size.
4. Explain how sampling matrices can improve inclusion without becoming rigid quota systems.
5. Critically evaluate snowball and convenience sampling in research involving marginalised populations.

Analytical / Application Questions

1. You are studying barriers to telehealth among older adults. Design a sampling and recruitment strategy that addresses age, disability, rurality, digital access and language.
2. During a study of medication errors, all participants are senior clinicians recruited by a medical director. Analyse the threats to credibility and propose corrective actions.
3. Interview data repeatedly identify “poor communication,” but the meaning differs across patients, nurses and community providers. Explain how you would use theoretical sampling and saturation assessment to develop the analysis.

Multiple Choice Questions with Answers

Question 1

Which sampling strategy deliberately selects participants who meet predefined eligibility conditions?

A. Criterion sampling

B. Convenience sampling

C. Snowball sampling

D. Theoretical sampling

Correct Answer: A. Criterion sampling

Question 2

The primary purpose of maximum-variation sampling is to:

A. Estimate population prevalence

B. Include contrasting experiences and identify patterns across difference

C. Recruit only the most accessible participants

D. Guarantee equal numbers in every subgroup

Correct Answer: B. Include contrasting experiences and identify patterns across difference

Question 3

What distinguishes theoretical sampling from ordinary purposive sampling?

A. It uses random selection after each interview

B. It selects cases in response to emerging concepts and analytic needs

C. It excludes participants with direct experience

D. It is used only for quantitative surveys

Correct Answer: B. It selects cases in response to emerging concepts and analytic needs

Question 4

Why can recruitment through a powerful organisational gatekeeper be problematic?

A. It always produces too many participants

B. It may pressure participation and favour institutionally approved perspectives

C. It prevents the researcher from obtaining ethics approval

D. It guarantees participant anonymity

Correct Answer: B. It may pressure participation and favour institutionally approved perspectives

Question 5

Meaning saturation is reached when:

A. The sample contains equal numbers of men and women

B. No participant refuses an invitation

C. Existing codes are sufficiently elaborated and understood

D. The researcher has completed exactly ten interviews

Correct Answer: C. Existing codes are sufficiently elaborated and understood

Question 6

Which condition generally increases information power?

A. A broad aim and highly heterogeneous sample

B. Brief, superficial interviews

C. A focused aim and a specific, relevant sample

D. Absence of an analytic framework

Correct Answer: C. A focused aim and a specific, relevant sample

Question 7

A major limitation of snowball sampling is that it may:

A. Prevent any contact with participants

B. Produce network homogeneity

C. Require probability calculations

D. Exclude all hidden populations

Correct Answer: B. Produce network homogeneity

Question 8

Which practice best protects employees from undue influence during recruitment?

A. Requiring managers to identify willing employees publicly

B. Making participation a condition of promotion

C. Providing independent contact with the research team

D. Allowing supervisors to review individual interview transcripts

Correct Answer: C. Providing independent contact with the research team

Question 9

Theoretical sufficiency refers primarily to whether:

A. The sample is statistically representative

B. The data support the level of conceptual explanation claimed

C. Every eligible person has participated

D. All interviews have identical duration

Correct Answer: B. The data support the level of conceptual explanation claimed

Question 10

Which statement best describes qualitative sample size?

A. It is determined by one universal numerical rule

B. It depends on the aim, sample specificity, data quality, variation and analytic purpose

C. It must always exceed the sample size of a quantitative study

D. It is irrelevant once ethics approval is obtained

Correct Answer: B. It depends on the aim, sample specificity, data quality, variation and analytic purpose