Multistage Sampling

Abstract

In probability sampling each unit in the finite population of interest has a known, non-zero, chance of selection, π i . In single stage sampling the units in the sample, s, are selected directly from the population and information is obtained from them. For example, the finite population of interest may consist of businesses and a sample of businesses is selected. In these cases the population units and sampling units are the same. To obtain a single stage sample a sampling frame consisting of a list of the population units and means of contacting them are usually required. Simple random sampling (SRS) can be used, in which each possible sample of a given size has the same chance of selection. SRS leads to each unit in the population having the same chance of selection and is an equal probability selection method (EPSEM). Other EPSEMs are available. A probability sampling method does not need to be an EPSEM. As long as the selection probabilities are known it is possible to produce an estimator that is design unbiased, that is unbiased over repeated sampling. For example the Horvitz Thompson estimator of the population total can be used,