A New Approach To Inference From Accelerated Life Tests

In a recent paper, the authors have proposed a new approach for making inferences from accelerated life tests. Their approach is appreciably different from those that have been considered in the past, and is motivated by what is actually done in practice. Prior information which is generally available to the engineer is incorporated by adopting a Bayesian point of view. The usual assumptions about the failure distributions and the acceleration functions, which are appealing from a statistical point of view, have been sacrificed to achieve greater reality. We apply this new approach to some real life data arising from an accelerated life test. In the process, we explain our new approach in a manner which makes it easy to understand and apply by the reliability practitioner. Because of the nonparametric nature of the approach, the statistical precision of the results is less than that which can be achieved by making parametric assumptions. On the other hand, the chance of selecting an incorrect model is diminished.