Commentary: inference in simple step-stress models

This note considers simple step-stress models with type-II censored exponential data. For two stress-levels and a log-linear link between stress and mean life, there is a reparameterization of the parameters of the negative exponential distributions, so that one can equally well work with these parameters or those in the log-linear link. From a statistical perspective, it is preferable to work with the parameters of the negative exponential distributions, since the "lack-of-memory" property of this distribution allows the analysis to be sub-divided, with ensuing simplification.