ABSTRACT
Most survey practitioners accept design-based methods as the method of choice for simple descriptive estimation of population characteristics. This chapter provides a brief overview of important recent developments in several of these areas and describes areas of active and future research. It summarizes current work by survey statisticians who are looking more generally at the role Bayesian models and methods can play in the analysis of survey data sets. Furthermore, it considers multilevel models for cross-sectional complex sample survey data with multiple levels of substantive interest. The chapter provides a summary of recent work on the application of latent variable models, such as structural equation models (SEMs), to complex sample survey data, and presents examples of fitting SEMs to complex sample survey data using state-of-the-art software procedures. It also reviews recent developments in the applications of survey data to small area estimation problems. Finally, the chapter reviews the recent work on nonparametric methods.
