ISSUES ASSOCIATED WITH THE DESIGN OF A NATIONAL PROBABILITY SAMPLE FOR HUMAN EXPOSURE ASSESSMENT

Data obtained from national probability sample surveys provide importa nt information on the prevalence of various health conditions and dist ributions of physical and biochemical characteristics of the U.S. popu lation. The sample design of a survey specifies how sampling from a de signated population over a stated period is to be accomplished. A surv ey's analytical objectives and interests-in particular subpopulations- affect the sample design strategy. Selected subdomains of ?he populati on often must be oversampled so ?hat estimates can be made with accept able precision. This article addresses sample design considerations fo r a national probability sample for human tissue monitoring and specim en banking. Among the sampling issues addressed are the oversampling o f special populations e.g., minority groups and at-risk groups such as low income or elderly persons; geographic coverage; and sample size c onsiderations. The sample design for a major health survey, the Third National Health and Nutrition Examination Survey (NHANES III), is used to illustrate a complex, multistage probability sample design and to highlight some of the sampling issues discussed in this article.