Nonparametric test for checking lack of fit of the quantile regression model under random censoring

The author proposes a nonparametric test for checking the lack of fit of the quantile function of survival time given the covariates; she assumes that survival time is subjected to random right censoring. Her test statistic is a kernel-based smoothing estimator of a moment condition. The test statistic is asymptotically Gaussian under the null hypothesis. The author investigates its behavior under local alternative sequences. She assesses its finite-sample power through simulations and illustrates its use with the Stanford heart transplant data.