CLUSTER Keyword (CSPLAN command)

CLUSTER is used to sample groups of sampling units, such as states, counties, or school districts. Cluster sampling is often performed to reduce travel and/or interview costs in social surveys. For example, if census tracts are sampled within a particular city and each interviewer works within a particular tract, he or she would be able to conduct interviews within a small area, thus minimizing time and travel expenses.

• If CLUSTER is used, CSSELECT samples from values of the cluster variable as opposed to sampling elements (cases).
• If two or more cluster variables are specified, samples are drawn from among all combinations of values of the variables.
• CLUSTER is required for nonfinal stages of a sample or analysis plan.
• CLUSTER is required if any of the following sampling methods is specified: PPS_WOR, PPS_BREWER, PPS_MURTHY, or PPS_SAMPFORD.
• CLUSTER is required if the UNEQUAL_WOR estimator is specified.

varlist. Cluster variables.