Tests for paired lifetime data with frailty models

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

We consider the problem of hypothesis testing of the equality of marginal survival distributions observed from paired lifetime data. Usual procedures include the paired t-test, which may perform poor for certain types of data. We propose asymptotic tests based on gamma frailty models with Weibull conditional distributions, and investigate their theoretical properties using large sample theory. For finite samples, we conduct simulations to evaluate the powers of the associated tests. For moderate and less skewed data, the proposed tests are the most powerful among the commonly applied testing procedures. A data example is illustrated to demonstrate the methods.

Original languageEnglish (US)
Pages (from-to)3122-3139
Number of pages18
JournalCommunications in Statistics - Theory and Methods
Volume39
Issue number17
DOIs
StatePublished - Jan 2010
Externally publishedYes

Keywords

  • Bivariate distribution
  • Gamma frailty
  • Kendall's tau
  • Logistic distribution
  • Weibull distribution

ASJC Scopus subject areas

  • Statistics and Probability

Fingerprint

Dive into the research topics of 'Tests for paired lifetime data with frailty models'. Together they form a unique fingerprint.

Cite this