A0927
Title: Spatial filtering techniques for the definition of spatial non-compensatory indices
Authors: Alfredo Cartone - University of Chieti-Pescara (Italy) [presenting]
Andrea Di Isidoro - University of Chieti-Pescara (Italy)
Paolo Postiglione - University of Chieti-Pescara (Italy)
Abstract: Multivariate statistical techniques are increasingly applied to spatial data for the definition of composite indicators. Unfortunately, spatial issues are commonly neglected in the construction of those indicators, leading to a loss of information for analysts. Meanwhile, recent literature has focused on spatial principal component analysis for the building of spatial composite measures. A spatial methodology for multidimensional non-compensatory indices is introduced. Non-compensatory indices are often used to limit substitutability between low/high achievements of different variables in the calculation of unit scores. Hence, spatial filtering techniques are exploited here to build a spatial version of the MPI non-compensatory measure and obtain two components, a spatial and a specific one. Results of an empirical application to Italian well-being are presented to show the potential use of this methodology.
