ABSTRACT
Perusal of the electronic codebook for any major survey data set quickly points to the fact that many key survey questions require only a simple “yes” or “no” answer. Consider these questions from the three data sets used for the examples in this text:
The responses to such questions are coded as binary variables (or dichotomous variables). This chapter is devoted to generalized linear models (GLMs) for a binary survey variable-focusing primarily on the application of logistic regression analysis to complex sample survey data. There are many excellent texts on standard simple random sample (SRS) methods for logistic regression analysis including comprehensive introductions to the theory, methods, and applications provided in texts by Hosmer and Lemeshow (2000), Agresti (2002), and Allison (1999). Readers wanting a full mathematical treatment of GLMs are referred to McCullagh and Nelder (1989) and McCulloch and Searle (2001).
