STAT 344 Chapter Notes - Chapter uni5: Simple Random Sample, Cluster Sampling, Implementation Force
Document Summary
With strati ed sampling, we saw a way to use structure in a population to possible advantage. Speci cally, we could collect data via separate simple random samples in each sub-population, and we could analyze the data appropriately to draw inferences about features of the whole population. Cluster sampling also uses sub-population structure, but does so di erently. Cluster sampling involves a random selection of only some sub- populations for which data will be collected. While both strati ed sampling and cluster sampling are based on a partition of the population of interest into sub-populations, we use di erent terms and notations for cluster sampling. We presume the partition involves n sub-populations, and these are referred to as clusters or primary sampling units (psu). And we presume the i-th psu contains mi secondary sampling units (ssu). Then yij is the value of the variable of interest y for the j-th ssu within the i-th psu.