Sequential analysis of an accelerated life model

Acceptance sampling can be both time-consuming and destructive so that it is desirable to arrive at a sound lot disposition decision in a timely manner. Sequential sampling plans are attractive since they offer a lower average sample number than do matched single, double, or multiple sampling plans. Analogously, cumulative sum control charts offer the ability to detect moderate process shifts more rapidly than do Shewhart control charts applied to the same process. The inverse Gaussian distribution is flexible and is often the model of choice in accelerated life testing applications where early failure times predominate. Based on sequential probability ratio tests (SPRT), sequential sampling/ cumulative sum (CUSUM) plans provide timely, statistically based decisions. Presents SPRT and CUSUM results for the inverse Gaussian process mean. Also presents a simple goodness-of-fit test for the inverse Gaussian distribution which allows for model adequacy checking.