ABSTRACT
This chapter introduces the concept of adaptive survey design (ASD), which, contrary to nonadaptive or uniform survey design, allows for adaptation of survey design features before or during data collection to strata that are identified based on information that is auxiliary to the survey. It begins with an introduction to survey costs, survey errors, and statistical concepts such as bias and variance, as key factors motivating and informing ASDs. Surveys can have different objectives linked to one or more survey errors, such as cost minimization, response rate maximization, nonresponse bias reduction, or variance reduction, minimization of the risk of measurement bias, and minimization of mode-specific bias. As ASD is to help meet study objectives, it is imperative to understand that ASD needs to be customized to a particular survey. Furthermore, even the relative importance of each objective can vary across studies, leading to different designs. The National Survey of Family Growth focuses on fertility, marriage and cohabitation, and sexual activity.
