Autori
Liu, Zhan
Tu, Chaofeng
Pan, Yingli

Titolo
Model-assisted calibration with SCAD to estimated control for non-probability samples
Periodico
Statistical methods & applications : Journal of the Italian Statistical Society
Anno: 2022 - Volume: 31 - Fascicolo: 4 - Pagina iniziale: 849 - Pagina finale: 879

Non-probability samples have been used in various fields in recent years. However, they usually can result in biased estimates. Calibration to estimated control has been proposed to reduce bias from non-probability samples. The relationship models between the study variable and covariates will help to improve the efficiency of calibration. Specifically, the selection of important covariates is a key issue in establishing the relationship models. In this paper, model-assisted calibration to estimated control using the smoothly clipped absolute deviation (SCAD) is proposed to make inference from non-probability samples. Instead of the traditional chi-square distance, the modified forward Kullback–Leibler distance is explored in the proposed method and the corresponding asymptotic properties are derived. Moreover, the classical variable selection approach SCAD is also implemented to conduct both variable selection and parameter estimation in establishing the relationship models for calibration. The performances of the proposed method are investigated through simulation studies, and an application to analyze a non-probability sample from the National Health Interview Survey in 2017.



SICI: 1618-2510(2022)31:4<849:MCWSTE>2.0.ZU;2-5

Esportazione dati in Refworks (solo per utenti abilitati)

Biblioteche ACNP che possiedono il periodico