CMStatistics 2021: Start Registration
View Submission - CMStatistics
Title: Simultaneously selection of balanced and spatially balanced samples by means of simulated annealing Authors:  Francesco Pantalone - University of Southampton (United Kingdom) [presenting]
Roberto Benedetti - University of Chieti - Pescara (Italy)
Maria Michela Dickson - University of Trento (Italy)
Giuseppe Espa - University of Trento (Italy)
Federica Piersimoni - ISTAT (Italy)
Abstract: A new sampling method for the selection of samples that are both balanced over a set of auxiliary variables and spatially balanced is proposed. A balanced sample is suited when correlation is present between the variable of interest and a set of auxiliary variables, while a spatially balanced sample is recommended when there is spatial correlation in the population. Indeed, a gain in efficiency in terms of variance of the Horvitz-Thompson estimator can be achieved in these situations by means of the aforementioned samples. The new method, which is based on a modified version of the simulated annealing, allows to face situations where a correlated set of auxiliary variables and spatial correlation are both present, since it can select samples that are balanced and spatially balanced simultaneously.