Sampling Method Selector
Answer a few questions about your population, research approach, sampling frame, and study priorities to identify sampling strategies that may fit your project.
Find a suitable sampling method
Recommendations update as you describe the study.
Describe your study
Your suggested sampling method and alternatives will appear here.
Choose sampling from the population and inference
Sampling determines who has a chance to enter the study and which conclusions the evidence can support.
Probability sampling is generally preferred when the goal is population estimation and a usable frame exists. Nonprobability and purposive approaches may be more appropriate for exploratory, qualitative, feasibility, or hard-to-reach population studies.
The selector provides a starting point, but the complete sampling plan should also specify eligibility, recruitment, allocation, nonresponse procedures, and analysis implications.
Sampling frame
The available list or structure determines which probability procedures are feasible.
Target inference
Population estimates require a different selection logic from in-depth qualitative understanding.
Population structure
Individuals, strata, clusters, institutions, and networks require different approaches.
Practical access
Cost, geography, gatekeepers, stigma, and recruitment burden affect feasibility.
Common sampling methods at a glance
Use these categories as starting points for a full recruitment plan.
| Sampling goal | Common method | Typical use |
|---|---|---|
| Equal random chance from a complete list | Simple random | Population estimates when a complete sampling frame is available |
| Select from an ordered frame efficiently | Systematic sampling | Every kth unit after a random start |
| Guarantee subgroup representation | Stratified random | Important demographic, geographic, institutional, or clinical strata |
| Reduce travel or fieldwork costs | Cluster samplingMultistage sampling | Schools, villages, clinics, workplaces, and dispersed populations |
| Recruit cases with relevant knowledge | PurposiveCriterion | Qualitative interviews, expert studies, and specialized experiences |
| Capture broad diversity | Maximum variation | Contrasting cases and perspectives across selected characteristics |
| Develop an emerging theory | Theoretical sampling | Grounded theory and iterative data collection |
| Fill category targets without a frame | Quota sampling | Nonprobability subgroup balancing |
| Reach hidden populations | SnowballRespondent-driven | Networked or stigmatized populations with limited frames |
| Test feasibility quickly | Convenience sampling | Pilots, classroom exercises, early prototypes, and recruitment testing |
Before finalizing the sampling plan
- Define the target population. State who should be represented and who is outside the study scope.
- Specify the sampling unit. Individuals, households, classes, schools, records, and organizations are not interchangeable.
- Evaluate frame coverage. Identify missing, duplicated, outdated, or ineligible entries.
- Plan for nonresponse. Track refusals, unreachable cases, exclusions, and differential participation.
- Align analysis with design. Stratification, clustering, unequal probabilities, and weights must be reflected in analysis.
Sampling labels do not guarantee quality
A probability label does not fix an incomplete frame, high nonresponse, poor implementation, or incorrect weighting. Likewise, a purposive sample is not automatically rigorous unless selection criteria and information needs are clearly justified.
Mixed-methods projects may legitimately use different sampling methods for each strand. For example, a stratified survey sample may be followed by purposive interviews selected from survey response patterns.
Frequently asked questions
Start with the target population, research approach, availability of a sampling frame, need for representativeness, subgroup requirements, access constraints, and the type of inference you intend to make.
Probability sampling uses a known selection mechanism so eligible units have a known or calculable chance of selection. Nonprobability sampling selects participants through access, judgment, quotas, referrals, or other methods without known inclusion probabilities.
Use simple random sampling when a reasonably complete list of the target population is available and each eligible unit can be selected independently using a random mechanism.
Use stratified sampling when important subgroups must be represented or compared. The population is divided into strata, and a probability sample is selected within each stratum.
Use cluster sampling when the population is naturally organized into groups such as schools, classes, villages, clinics, or workplaces and sampling individuals directly is costly or impractical.
Multistage sampling selects units in stages, such as provinces, then schools, then classes, then students. It is useful for large or geographically dispersed populations.
Purposive sampling is appropriate when participants are selected because they have specific experiences, roles, characteristics, or knowledge relevant to the research question.
Maximum variation sampling deliberately includes diverse cases across relevant characteristics so the study can examine shared patterns and meaningful differences.
Snowball sampling may be useful for hidden or difficult-to-reach populations when initial participants can refer other eligible participants. Referral chains can introduce network and selection bias.
Respondent-driven sampling is a structured peer-referral approach for networked hidden populations. It uses controlled recruitment and statistical weighting, but it requires strong assumptions and specialist implementation.
Convenience samples can support exploratory, pilot, educational, or feasibility studies, but broad population generalization is usually limited because inclusion probabilities are unknown and selection bias may be substantial.
Yes. The sampling method affects design effects, weighting, precision, subgroup allocation, recruitment feasibility, and the interpretation of the final sample size.