Uno studio della non autosufficienza a partire dai dati dell'indagine multiscopo: il caso dell'Umbria

This paper proposes a methodology for the estimation of the number of people that are self insufficient – or dependent – using data coming from the Italian National Survey on Health conditions and Appeal to Medicare. Self insufficiency is treated as a latent trait hidden behind a set of items that survey difficulties in movements and in accomplishing everiday tasks (Activities of daily living). Latent class models are used to classify the population according to different levels of self sufficiency. The analysis provides a good classification using four classes. Looking at posterior probabilities, people belonging to each class may be labeled as: self sufficients, with light disability, initial self insufficients, self insufficients. The survey provides reliable estimates at regional – NUTS 2 – level. Estimating the amount of population within each latent class at sub-regional level, e.g. sanitary districts, requires small area estimation techniques. To this end, a multinomial unit level model is used with individual level covariates.