ABSTRACT
Many populations, including most social and demographic populations, are made up of individuals arranged in a naturally hierarchical structure in the sense that the individuals are nested within larger scale units, variously called groups, aggregates or clusters. The clusters may be households or family units, classes, year groups or schools, blocks, census districts, postcodes, regions, states or countries. The clusters may themselves be grouped within several higher layers in the hierarchy but, for simplicity, we discuss the simplest case where there is a single level of clustering and then indicate extensions to more complicated hierarchies. Typically, individuals within the same cluster tend to have more similar values for their survey variables than individuals within different clusters, and this feature needs to be taken into account, both in sampling clustered populations and in analyzing the resulting data.
