The sampling design for the Korean National Nutrition Survey

Korea has been conducting the National Nutrition Survey annually since 1969. Due to the magnitude and budget of the survey, its sample size was limited to 1,000. Since 1989, however, the government has increased the size to 2,000. We developed a new sampling design suitable when budget is limited and cases to be covered are relatively very small, which has been in use since 1987. The outline of the design is introduced. Basic sampling method adopted was Probability Proportional to Size (PPS) sampling. Using the Enumeration Districts (EDs) as primary sampling units (PSU), we selected 300 EDs and constructed six identical sample sets, each having 50 ED's. From each ED, we selected 20 households. Therefore, one set has 1,000 households. Currently, two sets are being used for the survey.