To increase customer satisfaction manufacturers should increase the expected life of their products. This can be done with a reliability experiment to know the various critical factors that are important for improving the reliability of the product. The main constraints in improving the reliability are cost and time factors. Because of this the data available will be randomized and censored. Many earlier studies have analyzed reliability experiments for randomized data and with subsampling as well. The article proposes an easy-to-implement modeling approach that handles reliability data containing random blocks and subsampling. The available methods do not properly incorporate this type of experimental protocol into the analysis for the Weibull distribution. This article provides a technique that is an improvement, the independence of all units, over the current approaches. (14 refs.)
展开▼