Cluster sampling is a form of sampling in which the population sample is chosen by classifying groups within the overall population, then taking a sample from some of those groups randomly. As a result, the potential margin of error is higher in cluster sampling, but the practicality of cluster sampling in terms of cost can outweigh the margin of error in large-scale surveys.

see "Multistage Cluster Sampling"

Multistage cluster sampling is a form of cluster sampling in which the larger population is classified into groups, from which a random sample is taken, and then the process of classifying and pulling new random samples is continued until the researchers produce a sample with the desired characteristics.