Step‐Stress Testing with Multiple Samples: The Exponential Case

Abstract Step‐stress models are descriptions of special experiments in the field of accelerated life testing, where the test units are exposed to stress levels that change at intermediate time points during the experiment. The goal is to develop inference for the mean lifetime at each stress level. The time points of stress level change can either be fixed or random. Furthermore, the experiment can be terminated when a certain number of failures is reached or at a pre‐specified time point. These alternative assumptions of the type of the experiment lead to alternative models. Usually the step‐stress models are based on a single experiment. We develop inference for step stress models designed for multiple samples. The stress levels are the same applied to all samples but the duration of exposure under each stress level can vary among the experiments. The likelihood inference is then discussed in detail for the exponential case and different simple step‐stress experiments.