How to Define a Population, Sample, and Sampling Method
Clearly defining the population, sample, and sampling method is essential for understanding who the findings represent, the limits within which results can be interpreted, and the extent to which conclusions may be generalized.
In quantitative research, sampling strategy can directly influence external validity. In qualitative research, the primary aim is often not statistical representation but the selection of participants who can provide rich and meaningful information about the phenomenon.
What Is a Population?
The population is the complete set of individuals, institutions, events, objects, or other units to which the research question and findings are intended to relate.
For example:
Research topic: Professional burnout among social workers in Türkiye
Population: Depending on the study definition, all individuals actively working as social workers in Türkiye.
Target Population and Accessible Population
A distinction may be made between:
- Target population: the broader population to which the study ultimately seeks to relate its conclusions.
- Accessible population: the population that can realistically be reached within the study's institutional, geographic, temporal, and logistical constraints.
For example, all nurses in a country may constitute the target population, while nurses working in public hospitals in one province may form the accessible population.
How Should the Population Be Defined?
A population definition should be as specific as necessary and may include:
- Who or what belongs to the population
- Geographic boundaries
- Institutional or sectoral setting
- Time period
- Professional or demographic characteristics
- Inclusion and exclusion conditions
A broad label such as “healthcare workers” is often insufficient without clarifying which professions, institutions, and time period are included.
What Is a Sample?
A sample is the subset of individuals or analytical units that actually participate in the study.
Its role differs across methodologies. Quantitative studies often use samples to make inferences about a wider population, whereas qualitative studies may deliberately select participants who can provide in-depth information about a specific experience or process.
What Is a Sampling Method?
A sampling method describes how the individuals or units included in the study are selected from the relevant population.
Sampling methods are commonly divided into:
- Probability sampling
- Non-probability sampling
What Is Probability Sampling?
In probability sampling, the probability that units in the population will be selected is known or can be calculated.
When appropriately implemented, probability sampling provides a stronger basis for statistical generalization to the population.
Simple Random Sampling
In simple random sampling, each population unit has an equal or known probability of selection. Selection may be performed using random numbers, computerized procedures, or equivalent methods.
A complete or sufficiently accurate sampling frame is generally required.
Example: Randomly selecting 300 employees from a list of 2,000 employees.
Systematic Sampling
Systematic sampling selects units at regular intervals after an initial starting point.
For example, if 200 participants are required from a list of 2,000, approximately every 10th person may be selected.
Care is required if the ordering of the list contains a periodic pattern that could introduce selection bias.
Stratified Sampling
In stratified sampling, the population is divided into meaningful subgroups and participants are selected from each stratum.
Strata may be defined by:
- Sex or gender
- Professional group
- Institution type
- Geographic region
- Educational level
This approach can improve representation of important subgroups within the sample.
Proportionate and Disproportionate Stratified Sampling
In proportionate stratified sampling, each stratum's share of the sample approximately reflects its share of the population.
In disproportionate stratified sampling, smaller groups may be sampled more heavily to obtain sufficient cases for analysis.
Disproportionate sampling may require weighting during analysis.
Cluster Sampling
Cluster sampling selects naturally occurring groups rather than selecting individuals one by one.
For example, schools may first be randomly selected and students within those schools may then be included in the study.
Cluster sampling can improve logistical feasibility, but similarities among individuals within the same cluster may need to be accounted for in sample size calculations and analysis.
Multistage Sampling
For large and geographically dispersed populations, several sampling stages may be combined.
For example:
Region selection → Province selection → Institution selection → Participant selection
What Is Non-Probability Sampling?
In non-probability sampling, the selection probability of each population unit is unknown. These approaches are common when access is limited and in exploratory or qualitative research.
When non-probability sampling is used, researchers should avoid overstating the extent to which the sample statistically represents the full population.
Convenience Sampling
Convenience sampling includes participants who are readily accessible and eligible to participate.
Examples may include students available to the researcher, employees of an accessible institution, or volunteers responding to an online survey.
This method is easy to implement but may carry substantial risks of selection bias and limited representativeness.
Purposive Sampling
Purposive sampling deliberately selects individuals or cases that are particularly informative for the research question.
It is one of the most common sampling approaches in qualitative research.
For example, care staff who have worked with people with intellectual disabilities for at least five years may be purposively selected.
Criterion Sampling
Criterion sampling includes participants who meet one or more predefined conditions.
Example: Nurses who have worked in an intensive care unit for at least two years.
The criterion should have a clear scientific relationship to the research question.
Maximum Variation Sampling
Maximum variation sampling deliberately includes participants with diverse characteristics in order to explore different forms of the phenomenon.
Variation may be sought in age, professional experience, institution type, educational background, work unit, or other relevant characteristics.
Typical Case Sampling
Typical case sampling selects a case that represents relatively ordinary or common conditions.
The aim is not to investigate an exceptional case, but to understand a phenomenon in a more typical setting.
Extreme or Deviant Case Sampling
This approach selects unusual, extreme, or particularly informative cases.
For example, a program that produces exceptionally strong or unexpectedly poor outcomes may be selected for detailed investigation.
Snowball Sampling
In snowball sampling, initial participants help identify other individuals who meet the research criteria.
This method can be useful for hard-to-reach populations or groups without a formal sampling frame.
However, recruitment through social networks may produce greater homogeneity and selection bias.
Quota Sampling
Quota sampling establishes target numbers for predefined participant groups.
For example, a sample may be designed to include 50% women and 50% men.
Unlike stratified random sampling, individuals within each quota do not have to be selected randomly.
Volunteer Sampling
Participants who respond voluntarily to an open invitation or online survey form a self-selected or volunteer sample.
Researchers should consider whether those who choose to participate differ systematically from those who do not.
Which Sampling Method Should Be Used in Quantitative Research?
The choice depends on the research objective, population structure, accessibility, available resources, and the type of inference the study intends to make.
Where generalization to a population is important and a suitable sampling frame exists, probability sampling is often methodologically stronger.
Where non-probability sampling is necessary, the method should be reported transparently and its implications for generalizability considered in the Discussion.
How Is Sampling Determined in Qualitative Research?
The aim in qualitative research is generally not statistical representation but the selection of information-rich participants capable of illuminating the phenomenon in depth.
Participant selection should therefore be justified in relation to the research question, qualitative design, and the nature of the experience or process being investigated.
Sample Size in Qualitative Research
There is no universal participant number that applies to every qualitative study.
Sample adequacy may depend on:
- The breadth of the research question
- The qualitative design
- Homogeneity or diversity of the participant group
- Depth and richness of data
- Quality of interviews or observations
- Analytical approach
- Information power or saturation considerations
What Is Data Saturation?
Saturation is commonly used to describe a point at which additional data no longer make a substantial contribution to new themes, categories, or explanations.
However, saturation is not defined or assessed identically across all qualitative methodologies and should not be reported merely as an unsupported statement that “no new information emerged.”
What Is Information Power?
Some qualitative research uses the concept of information power to evaluate sample adequacy.
A narrower research aim, a highly specific sample, and rich, relevant data may allow a smaller sample to provide sufficient information.
Sampling in Phenomenological Research
Phenomenological studies generally require participants who have directly experienced the phenomenon being investigated.
The ability of participants to provide rich and detailed accounts of that lived experience may be an important selection criterion.
Sampling in Grounded Theory
Grounded theory studies may begin with purposive sampling and move toward theoretical sampling as analysis develops.
Emerging categories guide the selection of additional participants or cases needed to refine, compare, and develop the evolving theory.
Sampling in Case Study Research
In case study research, the primary sampling unit may be the case itself rather than an individual.
A case may be an institution, program, event, community, team, or process. The rationale for selecting the case should be stated clearly.
Sampling in Mixed Methods Research
Quantitative and qualitative components of a mixed methods study may use different sampling strategies.
For example, a large survey sample may be used in the quantitative phase, followed by purposive selection of particular participants for qualitative interviews.
The relationship between the two samples should be explained clearly.
What Are Inclusion Criteria?
Inclusion criteria define the characteristics required for an individual or unit to participate in the study.
Examples include:
- Being 18 years of age or older
- Currently working in a specified profession
- Having worked at the relevant institution for at least six months
- Being able to understand the language used in the study
What Are Exclusion Criteria?
Exclusion criteria identify scientific or methodological reasons for excluding some individuals who otherwise meet the basic eligibility requirements.
They should not be created after data collection merely to produce more favorable results.
What Is a Sampling Frame?
A sampling frame is an accessible list of population units from which a sample can be selected.
Employee rosters, student registers, patient lists, institutional databases, or other appropriate records may serve as sampling frames.
The frame should be sufficiently current and complete.
What Is Sampling Bias?
Sampling bias occurs when some members of the population have systematically different chances of being included, causing the sample to misrepresent the population.
For example, a health survey distributed only through social media may overrepresent individuals who are more active on digital platforms.
Nonresponse Bias
Results may also be affected when individuals who decline participation differ systematically from those who participate.
Where appropriate, researchers should report the number invited, the number participating, and the response rate.
How Is Representativeness Evaluated?
Representativeness does not depend on sample size alone.
Relevant considerations include:
- Sampling method
- Quality of the sampling frame
- Response rate
- Distribution of important subgroups
- Similarity between sample and population characteristics
- Missing data and participant loss
A very large convenience sample is not automatically more representative than a smaller, well-designed probability sample.
Sample Size and Sampling Method Are Different Concepts
Sample size addresses how many units should be included, whereas sampling method addresses how those units are selected.
A large sample does not automatically correct a biased or inappropriate selection method.
How Should Sampling Be Reported in the Methods Section?
The Methods section should clearly report, where relevant:
- Target or accessible population
- Approximate population size, if known
- Sampling method
- How sample size was determined
- Inclusion criteria
- Exclusion criteria
- How participants were approached or recruited
- Final analytical sample
Example Sampling Statements
Quantitative example: The accessible population consisted of 820 employees working at Institution X. Sample size was determined through power analysis, and participants were selected using stratified random sampling. Data from 312 eligible participants who consented to participate were included in the final analysis.
Qualitative example: Participants were selected using purposive criterion sampling from individuals with direct experience of the phenomenon. Eligibility required at least three years of work experience in the relevant service setting. Sampling continued with attention to data adequacy and analytical development.
Common Sampling Mistakes
- Confusing the population with the sample
- Calling an accessible group a random sample when no random selection occurred
- Describing a voluntary online survey as simple random sampling
- Failing to state the sampling method
- Failing to explain how sample size was determined
- Creating inclusion or exclusion criteria after data collection
- Generalizing a non-probability sample to the entire population without qualification
- Justifying qualitative sample size using a single numerical rule
- Claiming saturation without explaining how it was assessed
- Assuming that a very large sample is automatically representative
- Failing to report participant exclusions or losses
Which Sampling Method Should I Choose?
There is no universally best sampling method. The appropriate approach depends on the research question, study design, population, and intended inference.
As a practical guide:
- If strong population generalization is required: consider an appropriate probability sampling method.
- If important subgroups need representation: stratified sampling may be useful.
- If the population is geographically dispersed: cluster or multistage sampling may be appropriate.
- If the study is qualitative and experience-focused: purposive sampling approaches may be more suitable.
- If the population is difficult to reach: snowball sampling may be considered.
- If only readily accessible participants can be recruited: convenience sampling should be reported transparently.
Check Before Submission
- Is the target population clearly defined?
- Are target and accessible populations distinguished where necessary?
- Is it clear who or what constitutes the sample?
- Is the sampling method named correctly?
- Is the rationale for the sampling method explained?
- Are inclusion and exclusion criteria stated?
- Is the method used to determine sample size explained?
- Is the final analytical sample reported?
- Are implications of sampling for generalizability considered?
- In qualitative research, is sample adequacy justified in a manner appropriate to the design?
