Abstract

During this talk, we will present ten important ideas in the theory of survey sampling. Survey sampling is a specific field in statistics because many classic recipes of statistics do not work in finite populations. Two points of view has been developed to base the inference: the design-based approach and the model-based approach. Although these approaches are based on completely different ideas, they can lead to the same estimates. In both case, the concept of ‘’representativeness’’ is misleading. The analysis of these two approaches will be the starting points to discuss the interest of the balanced sampling, calibration and the construction of a weighting system and their impact on the accuracy of the estimates. Then we state in which case the model-based and the design-based approaches coincide and how the accuracy can be estimated?