Abstract:

The design-based estimation of finite population parameters such as totals usually relies on the knowledge of inclusion probabilities characterizing the sampling design. They are used in sampling weights and estimators. However for some useful sampling designs these probabilities may remain unknown. In such a situation they may often be estimated in a simulation experiment. It is carried out by repeatedly generating samples using the same sampling scheme and counting occurrences of individual units. By replacing unknown inclusion probabilities with such estimates, the design-based population total estimates may be computed. In this paper the properties of point estimators of parameters of finite and fixed population were investigated. These estimators were calculated by using inclusion probabilities established in the simulation experiment.