What is the difference between cluster sampling and stratified sampling?

In stratified sampling, you divide the population into groups (strata) and randomly select some individuals from every stratum. In cluster sampling, you randomly select entire groups (clusters) and include all members of the chosen clusters. Stratified sampling guarantees representation from every group, while cluster sampling may miss some groups entirely but is often cheaper and more practical to carry out.

Cluster Sampling — Definition, Formula & Examples

Cluster sampling is a sampling method where you divide the entire population into groups called clusters, randomly select some of those clusters, and then collect data from every individual in the chosen clusters.

In a one-stage cluster sample, the population is partitioned into $k$ non-overlapping groups (clusters). A simple random sample of $c$ clusters is drawn, and every unit within each selected cluster is included in the sample. In two-stage cluster sampling, a random subset of units is sampled within each selected cluster rather than surveying all members.

How It Works

Start by identifying natural groupings in the population — schools in a district, city blocks in a town, or branches of a company. Number or list every cluster, then use a random process to choose which clusters to include. Once clusters are selected, you survey or measure every member inside them. This approach saves time and money because you only need to travel to or access a limited number of locations, rather than reaching individuals scattered across the entire population.

Worked Example

Problem: A school district has 40 elementary schools and wants to estimate the average reading score of all 4th graders. Each school has about 75 students in 4th grade. The district decides to use cluster sampling by randomly selecting 5 schools and testing every 4th grader in those schools.

Define the clusters: Each of the 40 elementary schools is one cluster. The population of interest is all 4th graders across the district.

k = 40 \text{ clusters}

Randomly select clusters: Use a random number generator to choose 5 schools from the 40. Suppose schools 3, 12, 19, 27, and 35 are selected.

c = 5 \text{ clusters selected}

Collect data from every member in chosen clusters: Administer the reading test to every 4th grader in those 5 schools. With approximately 75 students per school, the sample size is about 375.

n \approx 5 \times 75 = 375 \text{ students}

Compute the estimate: Suppose the 375 scores yield a mean of 82. This is the cluster-sample estimate of the district-wide 4th-grade reading average.

\bar{x} = 82

Answer: The estimated average reading score for all 4th graders in the district is 82, based on a cluster sample of 5 randomly chosen schools (375 students total).

Why It Matters

Cluster sampling appears regularly on the AP Statistics exam, where you need to identify sampling methods and explain their trade-offs. In real-world research — public health surveys, education studies, market research — it drastically cuts travel and administrative costs when populations are spread across many locations. Understanding when cluster sampling introduces higher variability compared to a simple random sample of the same size helps you design more efficient studies.

Common Mistakes

Mistake: Confusing cluster sampling with stratified sampling.

Correction: In stratified sampling you sample from every group; in cluster sampling you randomly select whole groups and skip the rest. Remember: stratified = sample within all groups, cluster = sample entire groups.

Mistake: Assuming cluster sampling is always as precise as a simple random sample of the same size.

Correction: Because individuals within a cluster often resemble each other, cluster samples typically have higher variability (larger standard errors) than a simple random sample with the same number of individuals. This trade-off is accepted for the practical savings in cost and logistics.