ABSTRACT

A survey may be stratified with several stages of clustering and then be calibrated to population quantities. Design effects, which describe the effects of the survey design on variances of estimates, are commonly used when planning surveys. This chapter shows how to compute estimates of population means, totals, quantiles, and design effects for complex surveys with the SURVEYMEANS and SURVEYFREQ procedures in SAS® software, illustrating the procedures with code and output for analyzing data from the U.S. National Health and Nutrition Examination Survey. Example code shows how to use the SURVEYMEANS, SURVEYREG, and SGPLOT procedures to produce a variety of histograms, boxplots, and scatterplots for survey data.