Riccardo De Santis
DEPS, USiena
Lucio Barabesi
DEPS, USiena
Gianni Betti
DEPS, USiena
Abstract
The theme of variance estimation is central in sampling surveys, due to the necessity of furnishing a measure of accuracy for the estimates. In the ambit of social surveys, where we have to face with complex designs and complex statistics, it may be a major issue. To solve this matter, two main approaches can be found in the literature, and both have advantages and disadvantages. However, linearization methods can be safely used in a design-based approach. On the contrary, resampling methods are introduced only in a model-based approach, which means that the properties have to be assessed. Furthermore, some approximations are required. Therefore, we decide to conduce a simulation study by the use of a complete population available. We will focus on some poverty measures considered by the statistical office of the European Union
Keywords
variance estimation; poverty measures; simulation study
Jel Codes
C15, I32, C83
