Principles of Proper Inferences From Complex Survey Data
Source: University of Maryland
This paper discusses commonly-encountered features of complex survey datasets that should be accounted for by employing the SURVEY family of SAS procedures. Concepts of clustering, stratification, unequal weighting, finite population corrections, and replication variances approximation methods new to SAS 9.2 will be presented. Potential pitfalls will be highlighted as will a few methods to trick SAS into performing techniques not yet available, such as DOMAIN analysis under replication variance approximations.