Selected Topics in Sample Design and Estimation Methodology

In this chapter we will discuss some sampling and estimation techniques that have been developed for special purposes. More specifically, we will first discuss some sampling methods that have been developed for use in health surveys conducted in developing countries, and then examine techniques that have been used in developed countries in order to meet special needs. The World Health Organization (WHO), through its Expanded Program on Immunization (EPI), aims to ensure the availability of immunization to all children of the world by the year 1990. As part of this program, it has developed methodology for estimating, by use of relatively quick and inexpensive sample surveys, immunization levels of children in areas that would be targeted for special immunization programs if the immunization levels were low. The methodology makes use of a modification of probability proportional to size (PPS) sampling developed originally in the United States [I] and later modified for use in the Smallpox Eradication Program in West Africa [2]. Although it can be adopted to meet other objectives, the major objectives of an EPI survey is to estimate the immunization coverage (i.e., the proportion of children having all their required immunizations) in a given target area (this could be a village, town, city, etc.). Over the years, the practice evolved in these surveys of selecting 30 sample clusters and 7 children within each cluster, which would yield a sample size of 210 children. In fact, these surveys became referred to as "30 x 7" surveys. Although hundreds of these "30 x 7" surveys have been conducted, the rationale for the choice of the sample size is rather unusual. Assuming that 50%