The big picture
Probability sampling gives every unit a known chance of selection.
In probability sampling, selection is controlled by a random mechanism. That makes it possible to use a sample to learn about a larger population with a defensible statistical basis.
Designed to represent
Simple random, systematic, stratified, and cluster sampling.
Preferred for statistical inferenceDesigned for access or speed
Convenience and voluntary response samples are easier to collect but can be systematically biased.
Useful, but interpret carefullyProbability sampling
Simple Random Sample
Every possible sample of size n has the same chance of being selected. In practice, you need a list of the population and a random way to choose observations.
Randomly selected observations are highlighted.
A retailer has 12,000 loyalty-program members and randomly selects 300 customer IDs to estimate average satisfaction.
Best when: You have a complete population list and want a straightforward, unbiased design.
Watch for: It can be inefficient when important subgroups are small.
Probability sampling
Systematic Sample
Choose a random starting point, then select every kth observation. If you need 100 observations from a list of 1,000, you might select every 10th customer after a random start.
A quality analyst chooses a random starting item and inspects every 50th package leaving a production line.
Best when: Observations arrive in a natural sequence and you need a simple operational rule.
Watch for: Hidden patterns in the list can create bias if they line up with the sampling interval.
Probability sampling
Stratified Sample
Divide the population into meaningful subgroups called strata, then take a random sample from each stratum. The goal is to make sure important groups are represented.
A bank wants customer feedback and samples separately from checking, mortgage, and business-account customers so each segment is represented.
Best when: Subgroups differ in meaningful ways and each group should appear in the sample.
Watch for: You need good information about which stratum each population member belongs to.
Probability sampling
Cluster Sample
Divide the population into natural groups called clusters, randomly select some clusters, and then observe all—or a sample of—the units within those selected clusters.
A national retailer randomly selects 12 stores and surveys customers at those locations instead of building a list of every customer nationwide.
Best when: The population is geographically dispersed and collecting observations individually would be expensive.
Watch for: People inside the same cluster may be similar, which can reduce statistical precision.
Use with caution
Convenience and voluntary response samples
These methods are common because they are easy. The tradeoff is that the selection process can favor certain types of people, so the sample may not represent the population.
Convenience sample
Select the observations that are easiest to reach.
Main risk: Easy-to-reach people may systematically differ from everyone else.
Voluntary response sample
People decide for themselves whether to participate.
Main risk: People with strong opinions are often more likely to respond.
Compare
Which sampling method fits the situation?
| Method | Core idea | Good choice when… | Main concern |
|---|---|---|---|
| Simple random | Randomly select individuals | You have a complete population list | Small groups may be missed |
| Systematic | Select every kth unit | Units arrive in a sequence or ordered list | Periodic patterns |
| Stratified | Sample within every subgroup | Important subgroups must be represented | Requires subgroup information |
| Cluster | Randomly select groups | Population is dispersed or costly to reach | Clusters may be internally similar |
| Convenience | Use easiest observations | Speed matters more than representativeness | Selection bias |
| Voluntary response | Participants opt in | You want open feedback, not a representative estimate | Strong-opinion bias |
A useful shortcut
Need every subgroup represented?
Consider stratified sampling.
Population spread across locations?
Cluster sampling may reduce cost.
Have a complete list?
Simple random or systematic sampling may work well.
The sampling method determines who gets a chance to be observed—and that determines how confidently the sample can represent the population.Back to learning topics