|
Abstract:
|
In many survey organizations, a large number of primary sampling units are drawn and their dwelling units listed close to decennial census years. Over the intercensus years, different surveys select dwelling units, households, subjects from this pool, and any new dwelling units constructed since the last listing through a well-defined probability selection scheme. Furthermore, the same questions may be asked in multiple surveys. Such sampling schemes provide a unique opportunity to combine information from multiple surveys to enhance the estimates of the population quantities and the temporal changes in the population quantities. This paper explores a Bayesian framework for the design and analysis that combines information from multiple multistage complex surveys. The methodology also will be used to inform on designs to increase precision and assess nonsampling errors.
|
