Autori: Gil, Yessenia A., Garay, Aldo M., Lachos, Victor H.
Titolo: Likelihood-based inference for interval censored regression models under heavy-tailed distributions
Periodico: Statistical methods & applications : Journal of the Italian Statistical Society
Anno: 2025 - Volume: 34 - Fascicolo: 3 - Pagina iniziale: 519 - Pagina finale: 544

Scale mixtures of skew-normal distributions form a class of asymmetric thick-tailed distributions that include skew-normal, skew-t, skew-contaminated normal, and the entire family of scale mixtures of normal distributions as special cases. This paper proposes an interval-censored linear regression model based on the class of scale mixtures of skew-normal distributions, providing an appealing, robust alternative to the usual Gaussian assumption in censored regression models. A novel Expectation/Conditional Maximization Either algorithm is proposed for maximum likelihood estimation, with analytical expressions at the E-step, as opposed to Monte Carlo simulations. These expressions rely on formulas for the mean and variance of truncated scale mixtures of skew-normal distributions that can be computed using the MomTrunc R package. The proposed methodology is illustrated through intensive simulations and the analysis of a real data set from the Household Survey OHS99 conducted by Statistics South Africa.




SICI: 1618-2510(2025)34:3<519:LIFICR>2.0.ZU;2-E

Esportazione dati in Refworks (solo per utenti abilitati)

Biblioteche ACNP che possiedono il periodico