Power family of transformation for Cox's regression with random effects

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journalpeer-review

3 Scopus Citations
View graph of relations

Author(s)

Related Research Unit(s)

Detail(s)

Original languageEnglish
Pages (from-to)57-66
Journal / PublicationComputational Statistics and Data Analysis
Volume30
Issue number1
Publication statusPublished - 28 Mar 1999

Abstract

A random effect survival model with Box-Cox type relative risk function is proposed. It extends the model considered in McGilchrist (1993) by allowing the relative risk function to belong to a power family of transformations. This power family chooses the exponential relative risk function when the transformation parameter (k) tends to zero and the linear hazard when k equals one. Optimal value of k is obtained by maximizing an appropriate log-likelihood function. The proposed scheme is used to analyse two sets of multivariate failure time data. One data set confirms the choice of exponential relative risk function while the other concludes that the usual exponential specification is not optimal. © 1998 Elsevier Science B.V.

Research Area(s)

  • Cox's regression model, General relative risk function, GLMM, Multivariate failure time data, Random effect