Abstract
Many phenomena of interest in official statistics are latent, meaning they cannot be observed directly. Instead, multiple indicators are typically required to measure, map, and monitor them over time. Examples of such phenomena include well-being, disability, social integration, poverty and social exclusion, and the digital divide. In this talk, I will focus on phenomena measured using binary or categorical variables and discuss the application of Item Response Theory models in this context. Additionally, I will address the challenge of obtaining reliable estimates for population subgroups where direct estimates lack sufficient precision, necessitating the use of small area estimation techniques. I will also present an application of these methods to an Italian household survey—The Activities of Everyday Life—to estimate social inclusion and educational poverty at the local level. Joint work with Gaia Bertarelli (Ca' Foscari University of Venice) and Simone Del Sarto (University of Perugia)
Collegamento Microsoft Teams
Organizzazione
Maria Ferrante