What is Sampling? Types, Methods, and Examples
In research, the way you select your participants shapes everything: the validity of your findings, how far they can be generalized, and how well your methodology holds up under peer review. This guide walks you through what sampling is, the major types of sampling methods, how to choose the right one, how to avoid common errors, and how to report your approach in a research paper.
Jump to Contents
- What Is Sampling?
- Why Is Sampling Important?
- Two Major Types of Sampling Methods
- Probability Sampling Methods
- Non-Probability Sampling Methods
- All Sampling Methods at a Glance
- How to Choose the Right Sampling Method
- Sampling Errors and Bias: What to Watch Out For
- How to Report Sampling Methods in a Research Paper
- Frequently Asked Questions
What Is Sampling?
Sampling is the process of selecting a subset of individuals or units from a larger group, i.e., the population, to study. Since examining every member of a population is rarely practical (too costly, too time-consuming, or simply impossible), researchers use samples to draw inferences about the whole.
Three key terms underpin every sampling decision:
- Population: The entire group you want to draw conclusions about (e.g., all diabetic patients in India)
- Sample: The subset of the population you actually study
- Sampling frame: The list or source from which the sample is drawn (e.g., a hospital patient registry)
Why Is Sampling Important?
- Reduces the time and cost of data collection
- Makes large-scale and nationwide research feasible
- Allows statistical inference from a manageable group to a broader population
- Enables replication and comparison across studies
Two Major Types of Sampling Methods
All sampling methods fall into one of two broad categories: probability sampling and non-probability sampling.
| Feature | Probability Sampling | Non-Probability Sampling |
| Selection basis | Random; every unit has a known, non-zero chance of selection | Non-random; based on convenience, judgment, or researcher-set criteria |
| Representativeness | High; results are generalizable to the population | Variable; may not accurately represent the full population |
| Bias risk | Low | Higher |
| Best for | Quantitative, hypothesis-testing research | Qualitative, exploratory, or resource-limited research |
| Complexity | Higher; requires a complete sampling frame | Lower; easier and cheaper to implement |
Probability Sampling Methods
In probability sampling, every member of the population has a known, non-zero chance of being selected. This makes results generalizable and statistically robust.
1. Simple Random Sampling
Every individual in the population has an equal and independent chance of being selected — the equivalent of drawing names from a hat. In practice, researchers assign each member a number and use a random number generator to make selections.
Example
Nirmani et al. (2024) used simple random sampling to assess OTC drug usage among pregnant people.
Advantages:
- Eliminates researcher bias entirely
- Simple to conduct and analyze statistically
- No prior knowledge of population characteristics required
Disadvantages:
- Requires a complete, up-to-date list of the entire population
- Can, by chance, under-represent important subgroups
2. Systematic Sampling
The first participant is chosen randomly, then every nth individual on an ordered list is selected. For example, if your sample is 10% of a population of 500, you pick a random start between 1 and 10, then select every 10th person thereafter.
Example
A researcher surveys every 5th patient admitted to a clinic over a three-month period, starting from a randomly chosen first patient.
Advantages:
- Easier to implement than pure random sampling
- Ensures an even spread across the population list
Disadvantages:
- Risk of bias if there is a hidden pattern in the list that aligns with the sampling interval (e.g., roster lists that cluster participants by team or department)
3. Stratified Sampling
The population is divided into distinct subgroups called strata based on a relevant characteristic like age, gender, or disease severity, and random samples are then drawn from each stratum in proportion to its size in the population.
Example
Yarroo and Rathebe (2024) used stratified random sampling to examine the respiratory effects of solvents among workers in the Mauritius paint industry, sampling across different job categories.
Advantages:
- Guarantees proportional representation of all key subgroups
- Increases precision and reduces sampling error compared to simple random sampling
- Enables meaningful comparisons between subgroups
Disadvantages:
- Requires prior knowledge of population characteristics to define the strata
- More complex and time-consuming to design and execute
4. Cluster Sampling
Instead of sampling individuals directly, the population is divided into clusters, usually geographic or institutional groups, and entire clusters are randomly selected. In two-stage cluster sampling, individuals within the selected clusters are then sampled again.
Example
A nationwide study on patient satisfaction randomly selects 20 hospitals, then surveys patients within those hospitals rather than attempting to recruit from every hospital in the country.
Advantages:
- Practical and cost-effective for large, geographically dispersed populations
- Does not require a complete list of all individuals, only a list of clusters
Disadvantages:
- Greater risk of sampling error if clusters are internally similar but differ from each other
- Less precise than stratified or simple random sampling
Non-Probability Sampling Methods
In non-probability sampling, participants are selected through non-random means. This category trades statistical rigor for speed, cost-efficiency, or access to hard-to-reach groups. It is the dominant approach in qualitative and exploratory research.
1. Convenience Sampling
Participants are selected based on their easy accessibility or proximity to the researcher, for example, students in one’s own department or patients at the hospital where the researcher works.
Example
Legros and Boyraz (2023) used convenience sampling to examine college students’ perceived mental health and help-seeking behaviors during the COVID-19 pandemic.
Advantages:
- Quick and cost-effective
- Ideal for pilot studies or preliminary research
- Accessible when resources are limited
Disadvantages:
- Highly susceptible to selection bias
- Findings are rarely generalizable to the broader population
2. Purposive (Judgmental) Sampling
The researcher deliberately selects participants based on their knowledge, expertise, or specific characteristics relevant to the research question. It is widely used in qualitative research where depth of insight matters more than representativeness.
Example
A researcher studying public policy selects participants with backgrounds in economics, law, and public administration specifically because of their relevant expertise.
Advantages:
- Targets exactly the type of participants needed for the study
- Efficient for niche topics or small, specialized populations
Disadvantages:
- Highly subjective and dependent on the researcher’s judgment
- Results cannot be generalized beyond the sample
3. Quota Sampling
The population is divided into subgroups, and the researcher recruits participants non-randomly until a pre-set quota for each subgroup is filled. It resembles stratified sampling in structure but lacks the random selection within strata.
Example
A survey on campus life sets quotas to ensure that 30% of respondents are engineering students, 30% humanities students, and 40% science students, mirroring the actual distribution of majors on campus.
Advantages:
- Ensures demographic subgroups are adequately represented without full randomization
- Practical when a complete sampling frame is unavailable
Disadvantages:
- Selection within each quota is non-random, introducing potential bias
- Researcher discretion in who gets recruited can skew results
4. Snowball Sampling
Initial participants recruit further participants from their own networks, creating a chain-referral effect. It is most useful when studying populations that are difficult to identify or access, such as marginalized or stigmatized groups.
Example
Nolan-Isles et al. (2021) used snowball sampling to examine barriers to healthcare among Aboriginal people in Australia, starting with a small initial group who referred others in their community.
Advantages:
- Effective for reaching hidden, rare, or hard-to-access populations
- Builds trust through peer referrals, which can improve participation rates
Disadvantages:
- Potential bias from non-random selection and dependence on initial participants
- Difficult to estimate population parameters or generalize findings
All Sampling Methods at a Glance
| Method | Type | Best Used When | Key Advantage | Key Limitation |
| Simple random sampling | Probability | Population is accessible and homogeneous | Eliminates researcher bias | Requires complete population list |
| Systematic sampling | Probability | Population has a natural order or roster | Easy to implement | Risk of pattern-based bias |
| Stratified sampling | Probability | Population has distinct, relevant subgroups | Ensures subgroup representation | Needs prior population knowledge |
| Cluster sampling | Probability | Population is large and geographically spread | Cost-effective at scale | Higher sampling error risk |
| Convenience sampling | Non-probability | Pilot or exploratory research | Fast and inexpensive | High selection bias |
| Purposive sampling | Non-probability | Qualitative or specialist research | Targets specific knowledge or traits | Subjective; low generalizability |
| Quota sampling | Non-probability | Subgroup representation without randomization | Ensures demographic spread | Non-random within quotas |
| Snowball sampling | Non-probability | Hidden or stigmatized populations | Accesses otherwise unreachable groups | Bias from initial recruitment seeds |
How to Choose the Right Sampling Method
Choosing a sampling method is not one-size-fits-all. Work through these four questions:
- What are your research objectives? If you need generalizable, population-level findings, probability sampling is essential. If you are exploring lived experience, attitudes, or phenomena in a specific group, non-probability methods are appropriate.
- What does your population look like? A diverse population with meaningful subgroups calls for stratified sampling. A geographically dispersed population is better suited to cluster sampling. A hidden population requires snowball or purposive sampling.
- What are your resource constraints? Probability sampling is generally more resource-intensive. If your budget, time, or access to a full sampling frame is limited, non-probability approaches may be the practical choice — provided you acknowledge the trade-offs.
- How far do you need to generalize? If your findings need to apply to a broader population, prioritize probability methods. If the goal is depth over breadth, non-probability methods serve you better.
Sampling Errors and Bias: What to Watch Out For
Using a sample is always a trade-off: you gain practicality but introduce some degree of uncertainty. The two main risks are:
- Sampling error: The natural difference between sample results and the true population value. It can be reduced by increasing sample size and using probability sampling.
- Sampling bias: A systematic distortion that occurs when the sample does not adequately represent the population. Unlike sampling error, this is not reduced by a larger sample; it is a design flaw.
Common sources of bias to guard against:
- Selection bias: Certain groups are more likely to be included than others (e.g., only recruiting participants who respond to an online survey)
- Volunteer bias: People who self-select into studies often differ systematically from those who do not
- Non-response bias: If a large portion of your sample declines to participate, those who respond may not represent the whole
Strategies to minimize bias:
- using a complete and accurate sampling frame,
- pre-registering your sampling procedure, and
- reporting response rates transparently.
Sample size and sampling method
Choosing a sampling method is closely linked to determining sample size. The right sample size depends on your population size, your desired confidence level (typically 95%), and your acceptable margin of error. Too small a sample inflates uncertainty; too large a sample wastes resources. Many online calculators can help you determine an appropriate size once you know these parameters.
How to Report Sampling Methods in a Research Paper
The methodology section of your research paper must describe your sampling approach clearly enough for a reader to replicate your study. Peer reviewers routinely flag vague or unjustified sampling descriptions.
Include the following elements:
- The sampling method used and the category it falls under (probability or non-probability)
- The sampling frame: what list or source you used to identify potential participants
- Inclusion and exclusion criteria: who qualified for the study and who was ruled out
- Sample size and how you determined it (e.g., power analysis, saturation in qualitative work)
- Justification: why this method was appropriate for your research question and population
Example write-up (quantitative)
“Participants were recruited using stratified random sampling. The population was stratified by age group (18–35, 36–55, 56+) and gender. Random samples were drawn from each stratum using a computer-generated random number sequence, yielding a final sample of 240 participants. Sample size was determined using G*Power with a significance level of 0.05 and power of 0.80.”
Example write-up (qualitative)
“Purposive sampling was used to recruit participants with direct experience of the phenomenon under study. Inclusion criteria required a minimum of five years’ experience in the relevant clinical role. Recruitment continued until theoretical saturation was reached at 18 participants.”
Frequently Asked Questions
What is the difference between sampling and a census?
A census collects data from every member of a population; sampling collects data from a selected subset. Censuses are comprehensive but expensive and time-consuming. Sampling is more practical for most research contexts.
Which sampling method is most accurate?
Simple random sampling and stratified sampling are generally considered the most accurate because they minimize bias and allow for statistical inference. Accuracy also depends heavily on sample size and how well the sampling frame covers the population.
What is the difference between cluster and stratified sampling?
Both divide the population into groups, but the logic differs. In stratified sampling, groups (strata) are internally homogeneous and the goal is to capture that diversity across the whole sample. In cluster sampling, groups (clusters) ideally mirror the whole population; you sample entire clusters, not individuals from each group.
When should I use non-probability sampling?
Non-probability sampling is appropriate in qualitative research, pilot studies, research on hidden populations, and situations where a complete sampling frame does not exist. Always acknowledge its limitations in your methodology section.
How do I justify my sampling method to a reviewer?
Link your method directly to your research question, population characteristics, and available resources. Explain why the chosen method is the most appropriate given these constraints, and acknowledge any resulting limitations on generalizability.
This article was originally published on April 14, 2024, and updated on May 31, 2026.
