Modelli di sopravvivenza con troncamenti variabili per l’analisi della durata dei governi della Repubblica Italiana

Abstract

In this paper we consider some different statistical models to explain the lifetime variability of the Italian governments from 1948 to 1994. One important feature of cabinet potential lifetimes is that they have an upper limit, wich obviously depend on how much time remains until next scheduled elections. The proposed duration models take into account these varying potential lifetimes considering truncated pdf’s and survival functions. We take into account constant in time and time varying covariates, whose effects are handled using piecewise hazard functions and exploiting the temporal factorization of the survival and likelihood functions. MCMC methods are employed to produce inference about the parameters posterior distributions in the models.