EMPIRICAL LIKELIHOOD ANALYSIS FOR THE HETEROSCEDASTIC ACCELERATED FAILURE TIME MODEL

We provide a general framework for the accelerated failure time (AFT) model that ties different data generating mechanisms, namely correlation and regression models, with different estimators that have been proposed for the AFT model. We develop empirical likelihood methods for inference that yield asymptotic chi-squared results by adequately reflecting the underlying data generating mechanisms in the construction of the likelihood. The results are also applicable to censored quantile regression, as illustrated in an example.