zbMATH — the first resource for mathematics

Combined multiple testing of multivariate survival times by censored empirical likelihood. (English) Zbl 1454.62287
A class of linear rank statistics for the \(k\)-sample problem with right censored survival data was introduced by D. P. Harrington and T. R. Fleming [Biometrika 69, 553–565 (1982; Zbl 0532.62026)] and the asymptotic normality of test statistics was established. In this article the authors improved these classic individual test procedures. The developed multivariate method is asymptotically consistent and capable of testing various aspects of a dataset containing time-to-event observations. The method by A. Bathke et al. [J. Stat. Plann. Inference 139, No. 3, 814–827 (2009; Zbl 1156.62037)] based on censored, multiply-constrained empirical likelihood is generalised to the multivariate case. The efficiency of the proposed test is demonstrated using the retinopathy dataset from the R-package survival.
62N01 Censored data models
62N05 Reliability and life testing
62H15 Hypothesis testing in multivariate analysis
62J15 Paired and multiple comparisons; multiple testing
R; survival
Full Text: DOI