Assessing influence on the estimated coefficients efficiency in a Cox regression model. Issue 7 (2nd May 2020)

• Record Type: Journal Article
• Authors: Alonso, R.; Pardo, M. C.
• Is Part Of: Journal of statistical computation and simulation. Volume 90:Issue 7(2020)
• Page Start: 1216
• Page End: 1229
• Publication Date: 2020-05-02
• Subjects: Deletion diagnostics -- Cox regression model -- influence -- Kullback–Leibler divergence; 14; Mathematical statistics -- Data processing -- Periodicals; Digital computer simulation -- Periodicals; 519.5028505
• DOI: 10.1080/00949655.2020.1720986
• Languages: English
• ISSNs: 0094-9655

Abstract

ABSTRACT: The proportional hazards model proposed by Cox [Regression models and life–tables [with discussion]. J Royal Stat Soc Ser B. 1972;34:187–220] has been widely used in modelling censored survival data. In order to make regression an useful and meaningful statistical tool, emphasis should be put not only on inference or fitting but also on diagnosing potential data problems. Most of the usual diagnostics for linear regression models have been generalized for the Cox model. However, global influence measures to assess the estimated coefficients efficiency are not available for Cox proportional hazard model. We propose new measures to detect influential subjects. The proposed measures are illustrated by an analysis of a set of real data and a simulation study.