ABSTRACT

The REG and GLM procedures in SAS® software will conduct model-based analyses for simple random samples. But with data from a complex survey, regression models need to be fit using the sampling weights, and properties of regression coefficients and predictions depend on the survey design. This chapter shows how to fit linear and logistic regression models to survey data using the SURVEYREG and SURVEYLOGISTIC procedures, and how to use the SURVEYREG procedure to compare means across subpopulations. Code and output for the examples in Chapter 11 of Sampling: Design and Analysis, Third Edition illustrate features of the procedures.