Glen Meeden

Title: Using prior information in finite population sampling

Abstract: In the standard design approach to survey sampling prior information, or perhaps the lack of such, is used when selecting the sampling design.Given the sample, each observed unit is assigned  a weight which depends on the design. Often these weights are adjusted to help account for missing observations and other information about the population. The resulting weights are used when making inferences about the population. In the Bayesian approach prior information is incorporated in a prior distribution and after the sample has been observed  the design plays no role when making inferences. In this talk we will discuss the different roles of prior information in the two approaches.