An adaptive test for two-sample accelerated life models

Abstract

Abstract We propose an adaptive testing method that is robust for two-sample scale models with censored observations. Motivated by [H. Uno, L. Tian, B. Claggett and L. J. Wei, A versatile test for equality of two survival functions based on weighted differences of Kaplan–Meier curves, Stat. Med. 34 2015, 28, 3680–3695], we propose simulation-based procedures to check model validity that exhibit robust performance across a broad range of alternative hypotheses. To evaluate the behavior of the proposed test, we conduct comprehensive simulations in some widely used survival functions. Simulation results indicate that the test exhibits strong performance in detecting scale difference between two samples, demonstrating adequate power. The proposed procedures are illustrated using a real-world dataset.