Purposive Sampling - Definition, Types, and Examples

Table of Contents

Purposive sampling is a non-probability sampling method in which researchers deliberately select people, cases, events, organisations, or documents because they can provide information directly relevant to the research question. It is widely used when specific experience, knowledge, variation, or depth matters more than statistical representation of an entire population.

Introduction

Researchers do not always need a random cross-section of a population. Sometimes only people with a particular experience, cases with particular characteristics, or settings that reveal important contrasts can answer the research question.

A study of how newly qualified nurses respond to night-shift stress, for example, would not benefit from randomly selecting members of the general public. It requires participants who are newly qualified nurses, have worked night shifts, and can describe the experience in sufficient detail.

Purposive sampling provides a systematic way to make such selections. However, it involves more than choosing people who appear relevant. A rigorous purposive sample is based on explicit criteria, a justified sampling strategy, an appropriate recruitment process, and transparent reporting.

This guide explains the meaning, types, uses, sample-size considerations, advantages, limitations, ethical issues, and reporting requirements of purposive sampling. It also distinguishes purposive sampling from convenience, quota, snowball, probability, and theoretical sampling.

Key Takeaways

Purposive sampling deliberately selects cases that are relevant to the research question.

It is a non-probability method because selection probabilities are not known.

The sampling strategy, eligibility criteria, and recruitment mechanism should be reported separately.

Sample size is guided by the study purpose, sample specificity, analytical approach, information power, and sometimes saturation—not by one universal formula.

Purposive samples support detailed, contextual, analytical, or theoretical conclusions but usually not population estimates with conventional margins of error.

Clear criteria, reflexivity, diverse recruitment routes, and an audit trail improve the credibility of the process.

What Is Purposive Sampling?

Purposive sampling, also called purposeful sampling, judgemental sampling, or judgment sampling, is the deliberate selection of information-rich cases that can help answer a defined research question.

The unit selected does not have to be a person. Researchers may purposively select:

Individuals
Households
Communities
Schools
Hospitals
Organisations
Policies
Events
Documents
Social-media posts
Geographic locations
Research sites

The logic is relevance rather than chance. Participants or cases are included because they possess a characteristic, experience, position, outcome, or perspective that matters to the study.

For example, a researcher examining barriers to implementing an electronic health-record system might select:

Clinicians who use the system daily
Administrators responsible for implementation
Staff members who resisted its adoption
Sites with unusually successful implementation
Sites with repeated implementation problems

Each selection serves a methodological purpose.

Purposive sampling is especially common in qualitative inquiry and mixed-methods research. It is also used in evaluations, case studies, expert consultations, implementation research, exploratory surveys, instrument development, and research involving rare or specialised populations (Palinkas et al., 2015).

Main Characteristics of Purposive Sampling

Purposive sampling has six defining characteristics.

1. Selection is intentional

Cases are selected according to their expected relevance to the study rather than through random selection.

2. The research question guides selection

A participant is not information-rich in the abstract. The participant is information-rich in relation to a particular question.

An experienced headteacher may be highly informative for a study of school leadership but not necessarily for a study of first-year pupils’ experiences.

3. Selection criteria should be explicit

Researchers should specify the characteristics required for inclusion and the conditions that lead to exclusion.

4. Different purposive strategies serve different goals

A researcher may seek similarity, variation, typicality, unusual outcomes, expertise, or emerging theoretical relevance. These purposes require different strategies.

5. Selection can be iterative

The initial sample may change as data collection and analysis reveal missing perspectives, contrasting cases, or emerging concepts.

6. Statistical selection probabilities are generally unknown

Because cases are not randomly drawn with known probabilities, conventional population-level margins of error cannot normally be calculated.

Sampling Strategy, Eligibility Criteria and Recruitment Are Not the Same

One of the most important distinctions in purposive sampling is the difference among:

Sampling strategy
Eligibility criteria
Recruitment mechanism

Sampling strategy

The strategy explains why particular kinds of cases are needed.

Examples include maximum variation, homogeneous, critical-case, and criterion sampling.

Eligibility criteria

The criteria establish who or what qualifies.

For a study of remote supervision among doctoral students, the criteria might include:

Currently enrolled in a doctoral programme
Received primarily remote supervision for at least six months
Completed at least one formal progress review
Able to participate in an interview in the study language

Recruitment mechanism

The recruitment mechanism explains how eligible cases are found and approached.

Possible mechanisms include:

University mailing lists
Professional associations
Clinic databases
Community organisations
Online advertisements
Research panels
Gatekeeper referrals
Participant referrals
Direct invitations

A study can therefore use purposive eligibility criteria but convenience-based recruitment. For example, a researcher might advertise in one easily accessible university and accept every eligible volunteer. The participants meet purposive criteria, but the recruitment route may still overrepresent people connected to that institution.

Transparent researchers describe both parts rather than labelling the entire process “purposive” and leaving the recruitment pathway unexplained.

When Should Purposive Sampling Be Used?

Purposive sampling is appropriate when the study needs cases with particular information, experiences, characteristics, outcomes, or positions.

It is commonly used when:

Only a defined group can answer the research question.
The phenomenon is uncommon or specialised.
The study seeks detailed understanding rather than population prevalence.
Researchers want to compare contrasting experiences.
Particular cases can illuminate success, failure, normality, or variation.
Expert knowledge is required.
The research design involves interviews, focus groups, observation, case studies, or document analysis.
A qualitative phase is used to explain quantitative findings.
Initial findings guide later participant selection.
Resources need to be concentrated on the most relevant cases.

Example: education

A researcher wants to understand how teachers implement a newly introduced reading intervention.

The sample might include teachers who:

Completed formal intervention training
Used the programme for at least one term
Work in schools serving different socioeconomic communities
Report high, medium, or low levels of implementation

Randomly selecting all teachers would include many who had never used the intervention.

Example: healthcare

A study investigates communication difficulties after discharge from intensive care.

Eligible participants might include adults who:

Spent at least 72 hours in intensive care
Were discharged within the previous six months
Participated in a follow-up appointment
Can give informed consent

Example: business research

A study explores why enterprise customers discontinued a software subscription.

Researchers might select customers who:

Cancelled within the previous 12 months
Used the product for at least six months
Represent different industries and company sizes
Reported different cancellation reasons

When Is Purposive Sampling Not Appropriate?

Purposive sampling is usually unsuitable when the main objective is to estimate a population value with measurable sampling error.

It should not be the primary design when the study must reliably estimate:

Election support across a national population
Disease prevalence
Average household expenditure
National student achievement
Product market share
The percentage of employees holding a particular opinion

These objectives normally require a defensible probability-sampling design or another design that supports the intended population inference.

Purposive sampling is also weak when:

The researcher cannot justify why selected cases are especially relevant.
Criteria are created after seeing which participants are easiest to recruit.
Only favourable or confirming cases are selected.
The target population is poorly defined.
Selection depends entirely on one gatekeeper.
Researchers make statistical population claims unsupported by the design.

Types of Purposive Sampling

Purposive sampling is a family of strategies rather than one fixed procedure. Researchers should choose the strategy that matches the study’s purpose.

Type	Main purpose	Example
Criterion sampling	Include every selected case that meets predetermined criteria	Interviewing teachers who have delivered a programme for at least one year
Homogeneous sampling	Examine a relatively similar group in depth	Studying first-year international nursing students
Maximum-variation sampling	Capture a wide range of relevant perspectives	Selecting urban, suburban, and rural schools with different performance levels
Typical-case sampling	Understand an ordinary or average case	Studying a school considered typical in size, funding, and results
Extreme or deviant-case sampling	Examine unusual success, failure, or outliers	Comparing exceptionally high- and low-performing clinics
Critical-case sampling	Study a strategically important case with strong logical relevance	Testing a procedure in the most experienced site
Intensity sampling	Study strong but not extreme examples	Selecting organisations with substantial but not exceptional improvement
Expert sampling	Obtain specialised knowledge	Interviewing epidemiologists about outbreak surveillance
Total-population sampling	Include all members of a small, defined population	Inviting all certified specialists in a small region
Stratified purposeful sampling	Compare purposively defined subgroups	Selecting junior, mid-career, and senior teachers
Opportunistic or emergent sampling	Add relevant cases as opportunities arise	Adding a newly identified stakeholder group during fieldwork
Theoretical sampling	Select new data sources to develop an emerging grounded theory	Recruiting participants who can clarify an emerging conceptual category

Criterion sampling

Criterion sampling selects cases that meet one or more predetermined conditions directly connected to the research question.

It is one of the most common and defensible forms of purposive sampling because eligibility rules can be stated clearly.

Suppose a study examines recovery after remote cardiac rehabilitation. Inclusion criteria might require participants to have completed at least eight remote sessions during the previous year.

Criterion sampling does not necessarily mean selecting every eligible person. Researchers may identify a larger pool of eligible people and then choose among them using another purposive strategy.

Homogeneous sampling

Homogeneous sampling selects participants who share important characteristics so that one subgroup can be examined in depth.

A study might focus exclusively on:

First-generation doctoral students
Newly appointed school principals
Parents of children receiving one specific treatment
Small-business owners operating for fewer than three years

Homogeneity can support focused comparison and detailed analysis. However, an excessively narrow sample may conceal variation that is important to the phenomenon.

Maximum-variation sampling

Maximum-variation sampling deliberately includes cases that differ across characteristics expected to influence the phenomenon.

The aim is not statistical proportionality. It is to examine how an experience varies and whether common patterns appear across contrasting contexts.

A researcher studying online learning might vary the sample by:

Age
Subject area
Study level
Disability status
Internet access
Domestic or international enrolment
Full-time or part-time status

The researcher should explain why each dimension is theoretically or practically relevant. Adding demographic diversity without a connection to the research question does not automatically improve the design.

Typical-case sampling

Typical-case sampling selects cases considered ordinary or illustrative of the usual situation.

Researchers must define how typicality was established. Possible evidence includes:

Administrative data
Prior surveys
Expert consultation
Median performance indicators
Established classifications
Multiple stakeholder assessments

Researchers should not simply call a convenient case “typical.”

Typical-case findings describe the selected case and may illuminate similar settings, but the method does not prove that the case statistically represents every setting.

Extreme or deviant-case sampling

Extreme-case sampling examines unusual, exceptional, or outlying cases to understand success, failure, risk, or unexpected outcomes.

Examples include:

Hospitals with exceptionally low readmission rates
Schools with unusually high attendance despite severe disadvantage
Projects completed far above or below budget
Patients who recovered much faster or slower than expected

The researcher needs a defensible benchmark for identifying an extreme case. Extreme should be defined through data, recognised criteria, or transparent expert judgement rather than personal impression.

Critical-case sampling

Critical-case sampling selects a strategically important case that can provide particularly strong logical insight.

The logic is often expressed as:

If a process fails under highly favourable conditions, it may fail elsewhere.
If a programme works under especially difficult conditions, it may be feasible elsewhere.
If highly experienced specialists cannot apply a rule, less experienced users may face even greater difficulty.

Critical-case findings require careful reasoning. A single critical case does not automatically establish statistical generalisability.

Intensity sampling

Intensity sampling selects cases that strongly demonstrate a phenomenon without being unusually extreme.

A researcher examining successful organisational change might select organisations with sustained, substantial improvement rather than only the single highest-performing organisation.

Intensity cases can provide rich examples while avoiding the possibility that extreme cases depend on rare circumstances.

Expert sampling

Expert sampling selects individuals with demonstrable specialised knowledge or experience.

Expert criteria may include:

Professional qualifications
Years of relevant experience
Leadership responsibility
Published work
Direct involvement in an event
Technical or community-recognised expertise

Expert status should be operationally defined. Holding a senior job title does not necessarily make a person knowledgeable about every aspect of the research topic.

Total-population sampling

Total-population sampling attempts to include every member of a small population that shares a relevant characteristic.

For example, a researcher might invite all 18 paediatric surgeons qualified to perform a rare procedure in one country.

Although the entire defined group is approached, nonresponse may still produce a smaller achieved sample. Researchers should distinguish:

The defined total population
The number invited
The number eligible
The number who participated
The number who declined or could not be contacted

Stratified purposeful sampling

Stratified purposeful sampling divides the target population into relevant subgroups and purposively selects cases within each subgroup.

For example, a study of workplace mentoring might select employees at:

Entry level
Middle management
Senior management

Unlike stratified random sampling, selection within the strata is not necessarily random. The goal is an information-rich comparison rather than statistically representative estimation.

Opportunistic or emergent sampling

Opportunistic sampling adds cases that become relevant as the study develops.

During interviews about a public programme, researchers may discover that procurement officers played an unexpected role. The sampling plan could then be amended to include them.

This flexibility should not become arbitrary case selection. Researchers should document:

What new information prompted the change
Which cases were added
Why they were relevant
Whether ethical approval or consent materials required amendment

Theoretical sampling

Theoretical sampling is an iterative grounded-theory procedure in which emerging concepts determine what data should be collected next.

Researchers begin analysis during data collection. They then seek participants, events, or documents that can:

Develop a category
Compare emerging conditions
Examine variation
Test a relationship
Explore a negative case
Refine the developing theory

Theoretical sampling is related to purposive selection, but the terms should not be treated as interchangeable. Initial purposive sampling may identify participants with relevant experience; theoretical sampling subsequently follows the needs of the developing analysis.

How to Conduct Purposive Sampling

A rigorous purposive-sampling process can be organised into ten steps.

Step 1: Define the research question

State exactly what the study seeks to understand.

A vague question such as “What do students think about university?” does not provide a strong basis for purposive selection.

A clearer question is:

How do first-generation international master’s students experience academic feedback during their first semester at UK universities?

Step 2: Define the unit of analysis

Decide whether the study is selecting:

People
Groups
Organisations
Sites
Events
Documents
Digital records
Cases containing several types of evidence

Step 3: Define the target population and context

State the broader population to which the research question refers.

Include relevant boundaries such as:

Country or region
Institution
Time period
Programme
Professional role
Event exposure
Clinical condition
Digital platform

Step 4: Choose the purposive strategy

Select the strategy according to the information required.

Ask:

Do I need depth within one similar group?
Do I need contrasting perspectives?
Do I need ordinary cases?
Do I need unusual outcomes?
Do I need experts?
Do I need all members of a small population?
Will emerging analysis guide later selection?

Step 5: Write inclusion and exclusion criteria

Criteria should be:

Relevant to the research question
Observable or verifiable
Ethically defensible
Specific enough for consistent screening
Broad enough to avoid unnecessary exclusion

Avoid criteria based solely on expected agreement with the researcher.

Step 6: Create a sampling matrix

A sampling matrix records the dimensions that matter to the study.

Example sampling matrix

Dimension	Planned categories	Why it matters
Study status	Full-time and part-time	May affect access to supervisors
Stage	First and second year	Experiences may change over time
Location	On campus and remote	Changes communication opportunities
Discipline	Sciences, social sciences, humanities	Supervisory practices may differ
Funding	Funded and self-funded	May affect workload and time pressure

The matrix is a planning tool, not a statistical quota requirement. Categories can be adjusted when justified.

Step 7: Identify recruitment channels

Use channels capable of reaching the relevant range of cases.

Depending on one department, clinic, online group, or gatekeeper may create avoidable distortion. Multiple channels can improve coverage.

Step 8: Screen participants consistently

Use a short screener based on the approved criteria.

A screener should collect only information needed to determine eligibility and sampling relevance. Sensitive information should not be requested without a clear purpose and appropriate protection.

Step 9: Review sample adequacy during data collection

Track:

Which categories have been recruited
Which perspectives remain missing
Whether important variation has appeared
Whether new data continue to change the analysis
Whether recruitment channels are overrepresenting one group
Whether additional or contrasting cases are needed

Step 10: Document and report the final sample

Maintain an audit trail containing:

Planned strategy
Eligibility criteria
Recruitment sources
Screening decisions
Invitations
Participation and refusals
Changes to the sampling plan
Final participant characteristics
Sample-size rationale
Limitations

Worked Example

Research question

How do secondary-school teachers experience the implementation of generative-AI policies in classroom assessment?

Target population

Teachers in secondary schools that introduced a formal generative-AI assessment policy during the previous academic year.

Initial criterion sampling

Participants must:

Be employed as classroom teachers
Have taught at least one assessed course under the policy
Have used the policy when designing or marking an assessment
Be able to provide informed consent

Maximum-variation dimensions

The study seeks variation by:

Public and private school
Urban and rural location
Subject area
Years of teaching experience
Reported level of policy support
High and low access to educational technology

Recruitment

Participants are approached through:

School-system mailing lists
Teacher associations
Direct invitations to selected schools
Professional-development networks

Sampling review

After the first interviews, researchers notice that special-education teachers face distinctive issues. This group is added to the sampling matrix, with the change recorded and ethically approved where required.

Final interpretation

The findings are presented as an in-depth account of the sampled teachers and contexts. The study does not estimate the percentage of all teachers who support the policy.

How Many Participants Are Needed?

There is no universal sample-size formula for purposive sampling. The appropriate size depends on the research aim, sample specificity, population diversity, data quality, analytical approach, subgroup comparisons, resources, and the criterion used to judge adequacy.

Information power

The information-power principle proposes that fewer participants may be needed when:

The research aim is narrow.
Participants are highly specific to the aim.
The study is supported by an established theoretical framework.
Interviews or observations are rich and detailed.
The analysis is intensive and focused.

A broader aim, a more varied population, weak dialogue, cross-group comparison, or exploratory analysis generally requires more cases (Malterud et al., 2016).

Saturation

Saturation is often described as the point at which additional data no longer contribute relevant new information, codes, themes, meanings, or theoretical development.

Researchers should state which kind of saturation they mean. Possible forms include:

Code saturation: Few or no new issues or codes appear.
Meaning saturation: Existing issues are sufficiently understood in depth and variation.
Theoretical saturation: Categories in grounded theory are adequately developed and related.
Data adequacy: The material is sufficient to answer the research question even when saturation is not the study’s epistemological goal.

A systematic review found that saturation was often reached within 9–17 interviews or 4–8 focus-group discussions in studies with relatively homogeneous populations and narrowly defined objectives. The same review warned that multi-country studies, heterogeneous populations, metathemes, and deeper meaning-oriented analysis may require larger samples (Hennink & Kaiser, 2022).

These ranges are evidence from particular study designs—not universal minimums or guarantees.

Why “12 interviews are enough” is misleading

Guest, Bunce, and Johnson (2006) found saturation within 12 interviews in one relatively homogeneous dataset, with basic metathemes appearing earlier. The study is methodologically valuable, but it does not establish that every qualitative study should stop at 12 participants.

A defensible sample-size statement explains why the planned or achieved sample is adequate for the particular:

Research question
Population
sampling strategy
Data-collection method
Analytical approach
Comparison structure

Can Purposive Sampling Be Used in Quantitative Research?

Yes. Purposive sampling can be used in quantitative research, particularly for specialised, rare, expert, pilot, exploratory, or hard-to-reach populations. Its use limits design-based statistical generalisation because selection probabilities are unknown.

A researcher might purposively survey:

Certified specialists in a rare profession
Companies using a particular technology
Patients who received a specific intervention
Employees who completed a defined training programme
Experts evaluating a proposed instrument

Researchers may calculate percentages, means, correlations, or regression models for the achieved sample. However:

A conventional margin of sampling error is generally inappropriate.
A p-value does not remove selection bias.
A large sample does not automatically become representative.
Confidence intervals based on random-sampling assumptions may not describe uncertainty from non-random selection.
Population claims require explicit assumptions and careful qualification.

The findings may still be useful for describing the participants, exploring relationships, testing an instrument, generating hypotheses, or studying a defined specialised group.

Purposive Sampling in Mixed-Methods Research

Purposive sampling is especially useful when qualitative and quantitative phases are connected.

Explanatory sequential design

A survey is completed first. Researchers then purposively select interviewees who can explain:

Unexpected results
Contrasting outcomes
High and low scores
Particular response patterns
Important subgroups

Exploratory sequential design

Qualitative research is conducted first. Information-rich participants help researchers identify concepts and language that inform the development of a later questionnaire or intervention.

Concurrent design

Quantitative and qualitative data are collected during the same period. Different sampling methods may be used for different components.

Researchers should report:

The sampling method for each component
How the samples are connected
Why cases moved from one phase to another
Whether the components answer the same or complementary questions

Purposive Sampling Compared With Other Methods

Method	How cases are selected	Main purpose	Main limitation
Purposive sampling	Deliberately selected for relevance	Depth, variation, expertise, or strategic insight	Unknown selection probabilities
Convenience sampling	Selected because they are accessible	Speed and feasibility	Strong accessibility and self-selection bias
Snowball sampling	Existing participants refer others	Reach networked or hidden populations	Network dependence and homophily
Quota sampling	Non-random recruitment fills specified category totals	Control sample composition	Selection within categories remains non-random
Simple random sampling	Cases selected by chance from a frame	Population estimation and statistical inference	Requires an adequate sampling frame
Stratified random sampling	Random selection within population strata	Improve subgroup representation and precision	Requires population information and random selection
Theoretical sampling	Emerging analysis determines what data are needed next	Develop grounded theory	Requires continuous analysis and methodological expertise

Purposive vs convenience sampling

The central difference is the basis of selection.

Purposive sampling asks: Who or what can best illuminate the research question?
Convenience sampling asks: Who or what can be reached most easily?

A sample can involve both. Researchers may purposively define eligible participants but recruit only those who are conveniently accessible.

Purposive vs snowball sampling

Snowball or chain-referral sampling explains how new participants are located through existing participants.

Purposive sampling explains why particular participants or cases are relevant.

A study may use both:

Establish purposive eligibility criteria.
Begin with several deliberately selected participants.
Ask them to refer others.
Screen referred individuals against the criteria.
Monitor whether one social network dominates recruitment.

Purposive vs quota sampling

Quota sampling establishes a required number or proportion within categories and fills each category non-randomly.

Stratified purposeful sampling selects information-rich cases from conceptually important subgroups. It does not necessarily seek population proportions or fixed numerical quotas.

Purposive vs probability sampling

Probability sampling gives members of the sampling frame a known, non-zero selection probability. Purposive sampling does not.

Probability sampling is generally stronger for estimating “how many” or “what percentage.” Purposive sampling is often stronger for examining “how,” “why,” “under what conditions,” and “what is the experience?”

Advantages of Purposive Sampling

Direct relevance

Researchers concentrate on cases capable of addressing the research question.

Depth of information

Participants can provide detailed accounts based on direct experience, responsibility, expertise, or exposure.

Flexibility

Researchers can seek similarity, variation, unusual outcomes, typical experiences, expertise, or emerging theoretical relevance.

Suitability for rare and specialised populations

Purposive criteria can identify cases that would be unlikely to appear in sufficient numbers in a general random sample.

Efficient use of resources

Time and funding are focused on cases with high potential informational value.

Support for comparative research

Researchers can intentionally include contrasting groups, sites, outcomes, or conditions.

Compatibility with iterative designs

The sampling plan can develop alongside data collection and analysis when the methodology permits this flexibility.

Limitations of Purposive Sampling

Researcher judgement can introduce bias

Researchers may consciously or unconsciously select cases that confirm expectations.

Population representativeness cannot usually be measured

Selection probabilities are unknown, making design-based population inference difficult.

Important perspectives may be excluded

Poorly designed criteria can omit people whose experiences challenge the initial assumptions.

Gatekeepers can shape the sample

Managers, clinicians, teachers, or community leaders may refer only cooperative, successful, visible, or favourable participants.

Volunteer bias may remain

Eligible people who volunteer can differ from eligible people who decline.

Replication can be difficult

Another researcher may interpret subjective criteria differently unless the criteria and decision process are documented carefully.

Sample-size justification can be weak

Claims of saturation are sometimes made without explaining how saturation was assessed.

Access may still be difficult

A precisely defined expert, rare, or vulnerable population can be expensive and time-consuming to recruit.

How to Reduce Bias and Improve Rigour

Purposive sampling cannot be converted into random sampling through careful wording. However, researchers can make the selection process more credible, transparent, and aligned with the research purpose.

Predefine criteria

Write inclusion and exclusion criteria before recruitment wherever possible.

Link every criterion to the research question

Remove criteria that do not have a methodological or ethical purpose.

Use a sampling matrix

Track relevant variation and identify gaps in the achieved sample.

Recruit through more than one channel

Multiple channels can reduce dependence on one organisation, online community, or gatekeeper.

Seek disconfirming or negative cases

Include cases capable of challenging an emerging interpretation.

Keep reflexive notes

Record how the researchers’ positions, assumptions, access, and relationships may influence selection.

Review selection decisions as a team

Where appropriate, have more than one researcher review borderline eligibility and sampling decisions.

Document non-participation

Report the number invited, screened, eligible, enrolled, withdrawn, and analysed when these figures are available and ethically reportable.

Separate planned and emergent decisions

State which parts of the strategy were specified in advance and which changed during fieldwork.

Limit conclusions appropriately

Describe transferability, analytical relevance, and contextual boundaries rather than claiming unsupported population representativeness.

Generalisability and Transferability

Purposive sampling does not automatically make research ungeneralisable. It changes the type of inference that can be defended.

Statistical generalisation

Statistical generalisation estimates population values using known selection probabilities and appropriate statistical procedures. Purposive samples usually do not support this form of inference.

Analytical or theoretical generalisation

Researchers may examine how findings support, refine, challenge, or extend a concept, explanation, or theory.

Transferability

Readers assess whether findings may be relevant to another setting with sufficiently similar characteristics.

Researchers support transferability by providing:

Clear contextual description
Participant characteristics
Eligibility criteria
Recruitment processes
Organisational or cultural setting
Boundaries of the study
Variation within the sample
Evidence for interpretations

Ethical Considerations

Purposive selection creates ethical responsibilities because researchers decide who is included, excluded, visible, and heard.

Fair inclusion

Criteria should not exclude groups merely because they are difficult or expensive to recruit when their perspectives are important.

Vulnerable participants

Research involving illness, trauma, minority status, insecure employment, migration, or illegal or stigmatised behaviour may require additional privacy and consent safeguards.

Gatekeeper influence

Gatekeepers should not pressure people to participate or learn who declined.

Confidentiality in small populations

Experts and members of rare groups may be identifiable even after names are removed. Researchers should avoid publishing combinations of characteristics that reveal identity.

Incentives

Compensation should recognise participants’ time without becoming coercive, especially among financially vulnerable populations.

Consent for referrals

Participants using chain referral should not disclose another person’s sensitive status without permission. Safer procedures allow participants to pass study information to potential recruits rather than giving their contact details directly to the researcher.

Digital Research Tools and Artificial Intelligence

Digital tools can improve the organisation of purposive sampling, but they do not replace methodological judgement.

Useful digital functions

Researchers can use secure tools to manage:

Eligibility screeners
Recruitment databases
Sampling matrices
Invitation status
Appointment scheduling
Consent records
Participant characteristics
Recruitment channels
Subgroup coverage
Reasons for exclusion
Sampling changes

Spreadsheet software, survey platforms, research-data systems, and qualitative analysis programs can all support these tasks when approved security and privacy requirements are met.

AI-assisted screening

Artificial intelligence can help researchers draft screening questions, categorise open-ended responses, identify potential gaps in a sampling matrix, or organise recruitment records.

However, automated selection can reproduce bias, misclassify nuanced responses, obscure decision rules, or expose sensitive participant information.

Researchers using AI should:

Confirm that its use is permitted by the ethics approval, institution, funder, and data-management plan.
Avoid uploading identifiable or confidential participant information to unapproved systems.
Tell participants when AI materially affects screening, communication, transcription, or analysis.
Document the model, purpose, prompts, decision rules, and version where possible.
Retain meaningful human review of inclusion and exclusion decisions.
Test for systematic exclusion or misclassification.
Keep an audit trail of changes and overrides.
Report AI’s role transparently.

AI can assist the process, but accountability remains with the research team.

How to Report Purposive Sampling

A transparent methods section should explain:

Why purposive sampling was appropriate
Which purposive strategy was used
What population or setting was targeted
The unit of selection
Inclusion and exclusion criteria
How potentially eligible cases were identified
Recruitment channels
Who approached participants
Whether gatekeepers were involved
The planned and achieved sample
Participant refusals and withdrawals, where available
How sample size was justified
Whether sampling changed during the study
How adequacy or saturation was assessed
Important limitations

COREQ is specifically designed for reporting qualitative interview and focus-group studies, while SRQR provides broader standards for qualitative research (Tong et al., 2007; O’Brien et al., 2014).

Purposive Sampling Methodology Template

The following template can be adapted to a proposal, dissertation, thesis, or journal article:

Purposive sampling was used because the study required participants with direct experience of [phenomenon]. The target population comprised [population and setting]. Participants were eligible when they [inclusion criteria] and were excluded when they [exclusion criteria].

A [criterion/homogeneous/maximum-variation/other] purposive strategy was selected to [methodological purpose]. Potential participants were identified through [recruitment channels or sampling frame] and screened using [procedure]. The sampling matrix considered variation in [relevant dimensions].

The initial sample target was [number or range], based on [information power, analytical requirements, prior methodological evidence, subgroup structure, or feasibility]. Recruitment and analysis proceeded [concurrently/in stages], and sample adequacy was reviewed by [procedure]. The final sample consisted of [number] participants or cases.

Because selection was non-random, the findings are interpreted as [contextual, analytical, theoretical, or transferable insights] rather than statistically representative estimates of [larger population].

The final wording must describe what the study actually did. Researchers should not claim maximum variation, saturation, or theoretical sampling unless the design and records support that claim.

Common Mistakes

Calling the sample random

Purposive selection is non-random, even when researchers choose among eligible cases informally.

Reporting a conventional margin of error

A margin of error derived from simple-random-sampling assumptions does not describe unknown selection bias in a purposive sample.

Claiming statistical representativeness

A sample can contain important perspectives without being statistically representative of their population proportions.

Using vague criteria

Criteria such as “knowledgeable people,” “active students,” or “experienced staff” require operational definitions.

Selecting only supportive participants

Choosing participants likely to agree with the programme, policy, or researcher weakens credibility.

Confusing eligibility with selection

Everyone in the target population may meet broad eligibility criteria. Researchers must still explain how the achieved sample was selected or recruited.

Treating saturation as a magic number

Saturation should be defined, assessed, and documented rather than invoked as a generic justification.

Ignoring recruitment bias

A strong set of criteria cannot compensate for a recruitment channel that excludes important parts of the population.

Misusing theoretical sampling

Selecting participants because they initially appear relevant is purposive selection. Theoretical sampling specifically follows emerging grounded-theory analysis.

Hiding changes to the sampling plan

Emergent changes can strengthen qualitative research when they are methodologically justified and transparently reported.

Conclusion

Purposive sampling is a deliberate, question-driven method for selecting cases with high informational relevance. Its strength lies in depth, specificity, comparison, and strategic insight—not in random selection or automatic population representativeness.

A credible purposive sample requires more than a list of inclusion criteria. Researchers should justify the specific strategy, distinguish selection from recruitment, monitor sample adequacy, address bias and ethics, limit conclusions appropriately, and document the complete process. When these elements align with the research question and methodology, purposive sampling can produce rigorous and highly useful evidence.