CMStatistics 2018: Start Registration
View Submission - CMStatistics
B1127
Title: Modelling educational poverty by area-level SAE latent Markov models Authors:  Gaia Bertarelli - University of Pisa (Italy) [presenting]
Caterina Giusti - Centro Dagum c/o Dip. Economia e Management, University of Pisa (Italy)
Monica Pratesi - University of Pisa (Italy)
Francesco Bartolucci - University of Perugia (Italy)
Maria Giovanna Ranalli - University of Perugia (Italy)
Luciana Quattrociocchi - Istat (Italy)
Abstract: Educational Poverty (EP) is defined as deprivation, for children and adolescents, of the ability to learn, experiment, develop and freely flourish skills, talents and aspirations. EP is a latent trait, namely, only indirectly measurable through a collection of observable variables and indicators purposively selected as micro-aspects, contributing to the latent macro-dimension. It is generally measured by ISTAT by two multidimensional indices, the educational poverty index and the adjusted Mazziotta-Pareto index. A problem with these indices is that they are based on direct estimates, which are reliable only at the broad-areas level, while to intervene on the phenomenon it is important to obtain information at a finer geographical level. We use an adapted version of area-level SAE method that uses a Latent Markov Model (LMM) as linking model. In LMMs the characteristic of interest and its evolution in time is represented by a latent process that follows a discrete Markov chain. Therefore, areas are allowed to change their latent state across time. This model can handle both univariate and multivariate characteristics of interest and can provide a classification of the areas by the intensity of their EP. That is, we can use the area-level data of the single dimension or the value of the composite indices.