Inference for bivariate survival data by copula models adjusted for the boundary effect

Aidong Adam Ding*, Wei-Jing Wang

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

Copula models describe the dependence structure of two random variables separately from their marginal distributions and hence are particularly useful in studying the association for bivariate survival data. Semiparametric inference for bivariate survival data based on copula models has been studied for various types of data, including complete data, right-censored data, and current status data. This article discusses the boundary effect on these inference procedures, a problem that has been neglected in the previous literature. Specifically, asymptotic distribution of the association estimator on the boundary of parameter space is derived for one-dimensional copula models. The boundary properties are applied to test independence and to study the estimation efficiency. Simulation study is conducted for the bivariate right-censored data and current status data.

Original languageEnglish
Pages (from-to)2927-2936
Number of pages10
JournalCommunications in Statistics - Theory and Methods
Volume36
Issue number16
DOIs
StatePublished - 1 Dec 2007

Keywords

  • Copula model
  • Current status data
  • Independence test
  • Right-censored data
  • Semiparametric estimation

Fingerprint Dive into the research topics of 'Inference for bivariate survival data by copula models adjusted for the boundary effect'. Together they form a unique fingerprint.

Cite this