Description
This paper is a discussion of Census 2000, focusing on planned use of sampling techniques for non-response followup and adjustment. Past experience with adjustment methods suggests that the design for Census 2000 is quite risky.