Cluster Sampling
A sampling method where the entire population is divided into groups, or clusters, and a random sample of these clusters are selected.
What is Cluster Sampling?
Unlike stratified sampling, in cluster sampling the clusters themselves are randomly selected, and then either all elements within those clusters are surveyed (one-stage) or a random sample of elements within the selected clusters is taken (two-stage). Clusters are often naturally occurring, like schools or city blocks.
Why Cluster Sampling Matters
It is highly cost-effective and logistically feasible when the population is geographically dispersed, as it concentrates fieldwork.
Example
To study elementary school reading levels in a state, a researcher randomly selects 20 school districts (clusters) and then tests all students in those selected districts.
Common Mistakes
- Assuming clusters are homogenous; typically, clusters are internally diverse, which can increase standard error.
- Analyzing the data without accounting for the intra-class correlation (the tendency of subjects within a cluster to be similar).