Getting to Know the Complex Sample Design
The first step in any applied analysis of survey data involves defining the research questions that will be addressed using a given set of survey data. The next step is to study and understand the sampling design that generated the sample of elements (e.g., persons, businesses) from the target population of interest, given that the actual survey data with which the reader will be working were collected from the elements in this sample. This chapter discusses the complex sample designs in practice and identifies the features of the designs that have important implications for correct analyses of the survey data. Technical documentation for the sample design, weighting, and analysis procedures should be part of the "metadata" that are distributed with a survey data set. Survey research on natural populations in environmental (e.g., forestry or fisheries), geological, and some human and animal epidemiological studies is increasingly turning to adaptive sample designs to optimize observation, estimation, and inference.