Measuring and detecting situations of need and deprivation using Graded Response Models

In the present work we exploit information on living conditions of individuals to build an estimation of their degree of need and deprivation. The information we use have often subjective components that help in identifying situations of need and we assume that deprivation is a latent variable that we estimate with IRT models on Italian data. We then relate this estimated trait to more objective and observable variables that thus could be used within policy actions and welfare programs to pinpoint situations where the estimated deprivation is high.